[PATCH 0/9] drivers: cacheinfo support

[PATCH 0/9] drivers: cacheinfo support

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This series adds a generic cacheinfo support similar to topology. The
implementation is based on x86 cacheinfo support. Currently x86, powerpc,
ia64 and s390 have their own implementations. While adding similar support
to ARM and ARM64, here is the attempt to make it generic quite similar to
topology info support. It also adds the missing ABI documentation for
the cacheinfo sysfs which is already being used.

It moves all the existing different implementations on x86, ia64, powerpc
and s390 to use the generic cacheinfo infrastructure introduced here.
These changes on non-ARM platforms are only compile tested and tested on x86.

This series also adds support for ARM and ARM64 architectures based on
the generic support.

Since there was no objection to the idea in RFC, I am posting non-RFC
version here.

The code can be fetched from:
 git://linux-arm.org/linux-skn cacheinfo

Previous RFCs:
[1] https://lkml.org/lkml/2014/1/8/523
[2] https://lkml.org/lkml/2014/2/7/654
[3] https://lkml.org/lkml/2014/2/19/391

Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-ia64@vger.kernel.org
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-arm-kernel@lists.infradead.org

---
Sudeep Holla (9):
  drivers: base: add new class "cpu" to group cpu devices
  drivers: base: support cpu cache information interface to userspace
    via sysfs
  ia64: move cacheinfo sysfs to generic cacheinfo infrastructure
  s390: move cacheinfo sysfs to generic cacheinfo infrastructure
  x86: move cacheinfo sysfs to generic cacheinfo infrastructure
  powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure
  ARM64: kernel: add support for cpu cache information
  ARM: kernel: add support for cpu cache information
  ARM: kernel: add outer cache support for cacheinfo implementation

 Documentation/ABI/testing/sysfs-devices-system-cpu |  41 ++
 arch/arm/include/asm/outercache.h                  |  13 +
 arch/arm/kernel/Makefile                           |   1 +
 arch/arm/kernel/cacheinfo.c                        | 249 +++++++
 arch/arm/mm/Kconfig                                |  13 +
 arch/arm/mm/cache-l2x0.c                           |  10 +
 arch/arm/mm/cache-tauros2.c                        |  34 +
 arch/arm/mm/cache-xsc3l2.c                         |  15 +
 arch/arm64/kernel/Makefile                         |   3 +-
 arch/arm64/kernel/cacheinfo.c                      | 135 ++++
 arch/ia64/kernel/topology.c                        | 401 ++--------
 arch/powerpc/kernel/cacheinfo.c                    | 813 +++------------------
 arch/powerpc/kernel/cacheinfo.h                    |   8 -
 arch/powerpc/kernel/sysfs.c                        |  12 +-
 arch/s390/kernel/cache.c                           | 388 +++-------
 arch/x86/kernel/cpu/intel_cacheinfo.c              | 655 ++++-------------
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 564 ++++++++++++++
 drivers/base/core.c                                |  39 +-
 drivers/base/cpu.c                                 |   7 +
 include/linux/cacheinfo.h                          |  56 ++
 include/linux/cpu.h                                |   2 +
 22 files changed, 1590 insertions(+), 1871 deletions(-)
 create mode 100644 arch/arm/kernel/cacheinfo.c
 create mode 100644 arch/arm64/kernel/cacheinfo.c
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

-- 
1.8.3.2

[PATCH 0/9] drivers: cacheinfo support

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This series adds a generic cacheinfo support similar to topology. The
implementation is based on x86 cacheinfo support. Currently x86, powerpc,
ia64 and s390 have their own implementations. While adding similar support
to ARM and ARM64, here is the attempt to make it generic quite similar to
topology info support. It also adds the missing ABI documentation for
the cacheinfo sysfs which is already being used.

It moves all the existing different implementations on x86, ia64, powerpc
and s390 to use the generic cacheinfo infrastructure introduced here.
These changes on non-ARM platforms are only compile tested and tested on x86.

This series also adds support for ARM and ARM64 architectures based on
the generic support.

Since there was no objection to the idea in RFC, I am posting non-RFC
version here.

The code can be fetched from:
 git://linux-arm.org/linux-skn cacheinfo

Previous RFCs:
[1] https://lkml.org/lkml/2014/1/8/523
[2] https://lkml.org/lkml/2014/2/7/654
[3] https://lkml.org/lkml/2014/2/19/391

Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-ia64@vger.kernel.org
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-arm-kernel@lists.infradead.org

---
Sudeep Holla (9):
  drivers: base: add new class "cpu" to group cpu devices
  drivers: base: support cpu cache information interface to userspace
    via sysfs
  ia64: move cacheinfo sysfs to generic cacheinfo infrastructure
  s390: move cacheinfo sysfs to generic cacheinfo infrastructure
  x86: move cacheinfo sysfs to generic cacheinfo infrastructure
  powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure
  ARM64: kernel: add support for cpu cache information
  ARM: kernel: add support for cpu cache information
  ARM: kernel: add outer cache support for cacheinfo implementation

 Documentation/ABI/testing/sysfs-devices-system-cpu |  41 ++
 arch/arm/include/asm/outercache.h                  |  13 +
 arch/arm/kernel/Makefile                           |   1 +
 arch/arm/kernel/cacheinfo.c                        | 249 +++++++
 arch/arm/mm/Kconfig                                |  13 +
 arch/arm/mm/cache-l2x0.c                           |  10 +
 arch/arm/mm/cache-tauros2.c                        |  34 +
 arch/arm/mm/cache-xsc3l2.c                         |  15 +
 arch/arm64/kernel/Makefile                         |   3 +-
 arch/arm64/kernel/cacheinfo.c                      | 135 ++++
 arch/ia64/kernel/topology.c                        | 401 ++--------
 arch/powerpc/kernel/cacheinfo.c                    | 813 +++------------------
 arch/powerpc/kernel/cacheinfo.h                    |   8 -
 arch/powerpc/kernel/sysfs.c                        |  12 +-
 arch/s390/kernel/cache.c                           | 388 +++-------
 arch/x86/kernel/cpu/intel_cacheinfo.c              | 655 ++++-------------
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 564 ++++++++++++++
 drivers/base/core.c                                |  39 +-
 drivers/base/cpu.c                                 |   7 +
 include/linux/cacheinfo.h                          |  56 ++
 include/linux/cpu.h                                |   2 +
 22 files changed, 1590 insertions(+), 1871 deletions(-)
 create mode 100644 arch/arm/kernel/cacheinfo.c
 create mode 100644 arch/arm64/kernel/cacheinfo.c
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

-- 
1.8.3.2

[PATCH 0/9] drivers: cacheinfo support

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This series adds a generic cacheinfo support similar to topology. The
implementation is based on x86 cacheinfo support. Currently x86, powerpc,
ia64 and s390 have their own implementations. While adding similar support
to ARM and ARM64, here is the attempt to make it generic quite similar to
topology info support. It also adds the missing ABI documentation for
the cacheinfo sysfs which is already being used.

It moves all the existing different implementations on x86, ia64, powerpc
and s390 to use the generic cacheinfo infrastructure introduced here.
These changes on non-ARM platforms are only compile tested and tested on x86.

This series also adds support for ARM and ARM64 architectures based on
the generic support.

Since there was no objection to the idea in RFC, I am posting non-RFC
version here.

The code can be fetched from:
 git://linux-arm.org/linux-skn cacheinfo

Previous RFCs:
[1] https://lkml.org/lkml/2014/1/8/523
[2] https://lkml.org/lkml/2014/2/7/654
[3] https://lkml.org/lkml/2014/2/19/391

Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-ia64 at vger.kernel.org
Cc: linux390 at de.ibm.com
Cc: linux-s390 at vger.kernel.org
Cc: x86 at kernel.org
Cc: linuxppc-dev at lists.ozlabs.org
Cc: linux-arm-kernel at lists.infradead.org

---
Sudeep Holla (9):
  drivers: base: add new class "cpu" to group cpu devices
  drivers: base: support cpu cache information interface to userspace
    via sysfs
  ia64: move cacheinfo sysfs to generic cacheinfo infrastructure
  s390: move cacheinfo sysfs to generic cacheinfo infrastructure
  x86: move cacheinfo sysfs to generic cacheinfo infrastructure
  powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure
  ARM64: kernel: add support for cpu cache information
  ARM: kernel: add support for cpu cache information
  ARM: kernel: add outer cache support for cacheinfo implementation

 Documentation/ABI/testing/sysfs-devices-system-cpu |  41 ++
 arch/arm/include/asm/outercache.h                  |  13 +
 arch/arm/kernel/Makefile                           |   1 +
 arch/arm/kernel/cacheinfo.c                        | 249 +++++++
 arch/arm/mm/Kconfig                                |  13 +
 arch/arm/mm/cache-l2x0.c                           |  10 +
 arch/arm/mm/cache-tauros2.c                        |  34 +
 arch/arm/mm/cache-xsc3l2.c                         |  15 +
 arch/arm64/kernel/Makefile                         |   3 +-
 arch/arm64/kernel/cacheinfo.c                      | 135 ++++
 arch/ia64/kernel/topology.c                        | 401 ++--------
 arch/powerpc/kernel/cacheinfo.c                    | 813 +++------------------
 arch/powerpc/kernel/cacheinfo.h                    |   8 -
 arch/powerpc/kernel/sysfs.c                        |  12 +-
 arch/s390/kernel/cache.c                           | 388 +++-------
 arch/x86/kernel/cpu/intel_cacheinfo.c              | 655 ++++-------------
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 564 ++++++++++++++
 drivers/base/core.c                                |  39 +-
 drivers/base/cpu.c                                 |   7 +
 include/linux/cacheinfo.h                          |  56 ++
 include/linux/cpu.h                                |   2 +
 22 files changed, 1590 insertions(+), 1871 deletions(-)
 create mode 100644 arch/arm/kernel/cacheinfo.c
 create mode 100644 arch/arm64/kernel/cacheinfo.c
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

-- 
1.8.3.2

[PATCH 0/9] drivers: cacheinfo support

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This series adds a generic cacheinfo support similar to topology. The
implementation is based on x86 cacheinfo support. Currently x86, powerpc,
ia64 and s390 have their own implementations. While adding similar support
to ARM and ARM64, here is the attempt to make it generic quite similar to
topology info support. It also adds the missing ABI documentation for
the cacheinfo sysfs which is already being used.

It moves all the existing different implementations on x86, ia64, powerpc
and s390 to use the generic cacheinfo infrastructure introduced here.
These changes on non-ARM platforms are only compile tested and tested on x86.

This series also adds support for ARM and ARM64 architectures based on
the generic support.

Since there was no objection to the idea in RFC, I am posting non-RFC
version here.

The code can be fetched from:
 git://linux-arm.org/linux-skn cacheinfo

Previous RFCs:
[1] https://lkml.org/lkml/2014/1/8/523
[2] https://lkml.org/lkml/2014/2/7/654
[3] https://lkml.org/lkml/2014/2/19/391

Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-ia64@vger.kernel.org
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-arm-kernel@lists.infradead.org

---
Sudeep Holla (9):
  drivers: base: add new class "cpu" to group cpu devices
  drivers: base: support cpu cache information interface to userspace
    via sysfs
  ia64: move cacheinfo sysfs to generic cacheinfo infrastructure
  s390: move cacheinfo sysfs to generic cacheinfo infrastructure
  x86: move cacheinfo sysfs to generic cacheinfo infrastructure
  powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure
  ARM64: kernel: add support for cpu cache information
  ARM: kernel: add support for cpu cache information
  ARM: kernel: add outer cache support for cacheinfo implementation

 Documentation/ABI/testing/sysfs-devices-system-cpu |  41 ++
 arch/arm/include/asm/outercache.h                  |  13 +
 arch/arm/kernel/Makefile                           |   1 +
 arch/arm/kernel/cacheinfo.c                        | 249 +++++++
 arch/arm/mm/Kconfig                                |  13 +
 arch/arm/mm/cache-l2x0.c                           |  10 +
 arch/arm/mm/cache-tauros2.c                        |  34 +
 arch/arm/mm/cache-xsc3l2.c                         |  15 +
 arch/arm64/kernel/Makefile                         |   3 +-
 arch/arm64/kernel/cacheinfo.c                      | 135 ++++
 arch/ia64/kernel/topology.c                        | 401 ++--------
 arch/powerpc/kernel/cacheinfo.c                    | 813 +++------------------
 arch/powerpc/kernel/cacheinfo.h                    |   8 -
 arch/powerpc/kernel/sysfs.c                        |  12 +-
 arch/s390/kernel/cache.c                           | 388 +++-------
 arch/x86/kernel/cpu/intel_cacheinfo.c              | 655 ++++-------------
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 564 ++++++++++++++
 drivers/base/core.c                                |  39 +-
 drivers/base/cpu.c                                 |   7 +
 include/linux/cacheinfo.h                          |  56 ++
 include/linux/cpu.h                                |   2 +
 22 files changed, 1590 insertions(+), 1871 deletions(-)
 create mode 100644 arch/arm/kernel/cacheinfo.c
 create mode 100644 arch/arm64/kernel/cacheinfo.c
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

-- 
1.8.3.2

[PATCH 1/9] drivers: base: add new class "cpu" to group cpu devices

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch creates a new class called "cpu" and assigns it to all the
cpu devices. This helps in grouping all the cpu devices and associated
child devices under the same class.

This patch also:
1. modifies the get_parent_device to return the legacy path
   (/sys/devices/system/cpu/..) for the cpu class devices to support
   existing sysfs ABI
2. avoids creating link in the class directory pointing to the device as
   there would be per-cpu instance of these devices with the same name
3. makes sure subsystem symlink continues pointing to cpu bus instead of
   cpu class for cpu devices

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 drivers/base/core.c | 39 +++++++++++++++++++++++++++++++++------
 drivers/base/cpu.c  |  7 +++++++
 include/linux/cpu.h |  2 ++
 3 files changed, 42 insertions(+), 6 deletions(-)

diff --git a/drivers/base/core.c b/drivers/base/core.c
index 20da3ad..fe622e2 100644
--- a/drivers/base/core.c
+++ b/drivers/base/core.c
@@ -10,6 +10,7 @@
  *
  */
 
+#include <linux/cpu.h>
 #include <linux/device.h>
 #include <linux/err.h>
 #include <linux/init.h>
@@ -742,6 +743,12 @@ static struct kobject *get_device_parent(struct device *dev,
 			return &block_class.p->subsys.kobj;
 		}
 #endif
+		/*
+		 * if the device is in cpu class, then use the default/legacy
+		 * /sys/devices/system/cpu/.. path
+		 */
+		if (dev->class == cpu_class)
+			return &parent->kobj;
 
 		/*
 		 * If we have no parent, we live in "virtual".
@@ -808,11 +815,17 @@ static int device_add_class_symlinks(struct device *dev)
 	if (!dev->class)
 		return 0;
 
-	error = sysfs_create_link(&dev->kobj,
-				  &dev->class->p->subsys.kobj,
-				  "subsystem");
-	if (error)
-		goto out;
+	/*
+	 * the subsystem symlink in each cpu device needs to continue
+	 * pointing to cpu bus
+	 */
+	if (dev->bus != &cpu_subsys) {
+		error = sysfs_create_link(&dev->kobj,
+					  &dev->class->p->subsys.kobj,
+					  "subsystem");
+		if (error)
+			goto out;
+	}
 
 	if (dev->parent && device_is_not_partition(dev)) {
 		error = sysfs_create_link(&dev->kobj, &dev->parent->kobj,
@@ -826,6 +839,13 @@ static int device_add_class_symlinks(struct device *dev)
 	if (sysfs_deprecated && dev->class == &block_class)
 		return 0;
 #endif
+	/*
+	 * don't create a link in the cpu class directory pointing to the
+	 * device as there would be per-cpu instance of these devices with
+	 * the same name
+	 */
+	if (dev->class == cpu_class)
+		return 0;
 
 	/* link in the class directory pointing to the device */
 	error = sysfs_create_link(&dev->class->p->subsys.kobj,
@@ -851,11 +871,18 @@ static void device_remove_class_symlinks(struct device *dev)
 
 	if (dev->parent && device_is_not_partition(dev))
 		sysfs_remove_link(&dev->kobj, "device");
-	sysfs_remove_link(&dev->kobj, "subsystem");
+
+	/* if subsystem points to cpu bus, bus_remove_device will remove it */
+	if (dev->bus != &cpu_subsys)
+		sysfs_remove_link(&dev->kobj, "subsystem");
 #ifdef CONFIG_BLOCK
 	if (sysfs_deprecated && dev->class == &block_class)
 		return;
 #endif
+	/* symlinks are not created for cpu class devices, nothing to remove */
+	if (dev->class == cpu_class)
+		return;
+
 	sysfs_delete_link(&dev->class->p->subsys.kobj, &dev->kobj, dev_name(dev));
 }
 
diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c
index 006b1bc..056b885 100644
--- a/drivers/base/cpu.c
+++ b/drivers/base/cpu.c
@@ -322,6 +322,7 @@ static int cpu_uevent(struct device *dev, struct kobj_uevent_env *env)
 }
 #endif
 
+struct class *cpu_class;
 /*
  * register_cpu - Setup a sysfs device for a CPU.
  * @cpu - cpu->hotpluggable field set to 1 will generate a control file in
@@ -338,6 +339,8 @@ int register_cpu(struct cpu *cpu, int num)
 	memset(&cpu->dev, 0x00, sizeof(struct device));
 	cpu->dev.id = num;
 	cpu->dev.bus = &cpu_subsys;
+	cpu->dev.parent = cpu_subsys.dev_root;
+	cpu->dev.class = cpu_class;
 	cpu->dev.release = cpu_device_release;
 	cpu->dev.offline_disabled = !cpu->hotpluggable;
 	cpu->dev.offline = !cpu_online(num);
@@ -423,5 +426,9 @@ void __init cpu_dev_init(void)
 	if (subsys_system_register(&cpu_subsys, cpu_root_attr_groups))
 		panic("Failed to register CPU subsystem");
 
+	cpu_class = class_create(THIS_MODULE, "cpu");
+	if (IS_ERR(cpu_class))
+		panic("Failed to register CPU class");
+
 	cpu_dev_register_generic();
 }
diff --git a/include/linux/cpu.h b/include/linux/cpu.h
index 95978ad..8c0fc9b 100644
--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -39,6 +39,8 @@ extern void cpu_remove_dev_attr(struct device_attribute *attr);
 extern int cpu_add_dev_attr_group(struct attribute_group *attrs);
 extern void cpu_remove_dev_attr_group(struct attribute_group *attrs);
 
+extern struct class *cpu_class;
+
 #ifdef CONFIG_HOTPLUG_CPU
 extern void unregister_cpu(struct cpu *cpu);
 extern ssize_t arch_cpu_probe(const char *, size_t);
-- 
1.8.3.2

[PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds initial support for providing processor cache information
to userspace through sysfs interface. This is based on already existing
implementations(x86, ia64, s390 and powerpc) and hence the interface is
intended to be fully compatible.

The main purpose of this generic support is to avoid further code
duplication to support new architectures and also to unify all the existing
different implementations.

This implementation maintains the hierarchy of cache objects which reflects
the system's cache topology. Cache devices are instantiated as needed as
CPUs come online. The cache information is replicated per-cpu even if they are
shared. A per-cpu array of cache information maintained is used mainly for
sysfs-related book keeping.

It also implements the shared_cpu_map attribute, which is essential for
enabling both kernel and user-space to discover the system's overall cache
topology.

This patch also add the missing ABI documentation for the cacheinfo sysfs
interface already, which is well defined and widely used.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Rob Herring <robh@kernel.org>
Cc: linux-doc@vger.kernel.org
Cc: linux-ia64@vger.kernel.org
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-arm-kernel@lists.infradead.org
---
 Documentation/ABI/testing/sysfs-devices-system-cpu |  41 ++
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 564 +++++++++++++++++++++
 include/linux/cacheinfo.h                          |  56 ++
 4 files changed, 662 insertions(+), 1 deletion(-)
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

diff --git a/Documentation/ABI/testing/sysfs-devices-system-cpu b/Documentation/ABI/testing/sysfs-devices-system-cpu
index acb9bfc..5827f4e 100644
--- a/Documentation/ABI/testing/sysfs-devices-system-cpu
+++ b/Documentation/ABI/testing/sysfs-devices-system-cpu
@@ -224,3 +224,44 @@ Description:	Parameters for the Intel P-state driver
 		frequency range.
 
 		More details can be found in Documentation/cpu-freq/intel-pstate.txt
+
+What:		/sys/devices/system/cpu/cpu*/cache/index*/<set_of_attributes_mentioned_below>
+Date:		June 2014(documented, existed before August 2008)
+Contact:	Sudeep Holla <sudeep.holla@arm.com>
+		Linux kernel mailing list <linux-kernel@vger.kernel.org>
+Description:	Parameters for the CPU cache attributes
+
+		attributes:
+			- writethrough: data is written to both the cache line
+					and to the block in the lower-level memory
+			- writeback: data is written only to the cache line and
+				     the modified cache line is written to main
+				     memory only when it is replaced
+			- writeallocate: allocate a memory location to a cache line
+					 on a cache miss because of a write
+			- readallocate: allocate a memory location to a cache line
+					on a cache miss because of a read
+
+		coherency_line_size: the minimum amount of data that gets transferred
+
+		level: the cache hierarcy in the multi-level cache configuration
+
+		number_of_sets: total number of sets in the cache, a set is a
+				collection of cache lines with the same cache index
+
+		physical_line_partition: number of physical cache line per cache tag
+
+		shared_cpu_list: the list of cpus sharing the cache
+
+		shared_cpu_map: logical cpu mask containing the list of cpus sharing
+				the cache
+
+		size: the total cache size in kB
+
+		type:
+			- instruction: cache that only holds instructions
+			- data: cache that only caches data
+			- unified: cache that holds both data and instructions
+
+		ways_of_associativity: degree of freedom in placing a particular block
+					of memory in the cache
diff --git a/drivers/base/Makefile b/drivers/base/Makefile
index 04b314e..bad2ff8 100644
--- a/drivers/base/Makefile
+++ b/drivers/base/Makefile
@@ -4,7 +4,7 @@ obj-y			:= component.o core.o bus.o dd.o syscore.o \
 			   driver.o class.o platform.o \
 			   cpu.o firmware.o init.o map.o devres.o \
 			   attribute_container.o transport_class.o \
-			   topology.o container.o
+			   topology.o container.o cacheinfo.o
 obj-$(CONFIG_DEVTMPFS)	+= devtmpfs.o
 obj-$(CONFIG_DMA_CMA) += dma-contiguous.o
 obj-y			+= power/
diff --git a/drivers/base/cacheinfo.c b/drivers/base/cacheinfo.c
new file mode 100644
index 0000000..c12e03c
--- /dev/null
+++ b/drivers/base/cacheinfo.c
@@ -0,0 +1,564 @@
+/*
+ * cacheinfo support - processor cache information via sysfs
+ *
+ * Based on arch/x86/kernel/cpu/intel_cacheinfo.c
+ * Author: Sudeep Holla <sudeep.holla@arm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/compiler.h>
+#include <linux/cpu.h>
+#include <linux/device.h>
+#include <linux/init.h>
+#include <linux/of.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/smp.h>
+#include <linux/sysfs.h>
+
+/* pointer to per cpu cacheinfo */
+static DEFINE_PER_CPU(struct cpu_cacheinfo, ci_cpu_cacheinfo);
+#define ci_cacheinfo(cpu)	(&per_cpu(ci_cpu_cacheinfo, cpu))
+#define cache_leaves(cpu)	(ci_cacheinfo(cpu)->num_leaves)
+#define per_cpu_cacheinfo(cpu)	(ci_cacheinfo(cpu)->info_list)
+
+struct cpu_cacheinfo *get_cpu_cacheinfo(unsigned int cpu)
+{
+	return ci_cacheinfo(cpu);
+}
+
+#ifdef CONFIG_OF
+static int cache_setup_of_node(unsigned int cpu)
+{
+	struct device_node *np;
+	struct cacheinfo *this_leaf;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int index = 0;
+
+	/* skip if of_node is already populated */
+	if (this_cpu_ci->info_list->of_node)
+		return 0;
+
+	if (!cpu_dev) {
+		pr_err("No cpu device for CPU %d\n", cpu);
+		return -ENODEV;
+	}
+	np = cpu_dev->of_node;
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
+	}
+
+	while (np && index < cache_leaves(cpu)) {
+		this_leaf = this_cpu_ci->info_list + index;
+		if (this_leaf->level != 1)
+			np = of_find_next_cache_node(np);
+		else
+			np = of_node_get(np);/* cpu node itself */
+		this_leaf->of_node = np;
+		index++;
+	}
+	return 0;
+}
+
+static inline bool cache_leaves_are_shared(struct cacheinfo *this_leaf,
+					   struct cacheinfo *sib_leaf)
+{
+	return sib_leaf->of_node == this_leaf->of_node;
+}
+
+static int of_cache_shared_cpu_map_setup(unsigned int cpu)
+{
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sib_leaf;
+	unsigned int index;
+	int ret;
+
+	ret = cache_setup_of_node(cpu);
+	if (ret)
+		return ret;
+
+	for (index = 0; index < cache_leaves(cpu); index++) {
+		unsigned int i;
+
+		this_leaf = this_cpu_ci->info_list + index;
+		cpumask_set_cpu(cpu, &this_leaf->shared_cpu_map);
+
+		for_each_online_cpu(i) {
+			struct cpu_cacheinfo *sib_cpu_ci = get_cpu_cacheinfo(i);
+
+			if (i == cpu || !sib_cpu_ci->info_list)
+				continue;/* skip if itself or no cacheinfo */
+			sib_leaf = sib_cpu_ci->info_list + index;
+			if (cache_leaves_are_shared(this_leaf, sib_leaf)) {
+				cpumask_set_cpu(cpu, &sib_leaf->shared_cpu_map);
+				cpumask_set_cpu(i, &this_leaf->shared_cpu_map);
+			}
+		}
+	}
+
+	return 0;
+}
+#else
+static inline int of_cache_shared_cpu_map_setup(unsigned int cpu)
+{
+	return 0;
+}
+#endif
+
+static void cache_shared_cpu_map_remove(unsigned int cpu)
+{
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sib_leaf;
+	unsigned int sibling, index;
+
+	for (index = 0; index < cache_leaves(cpu); index++) {
+		this_leaf = this_cpu_ci->info_list + index;
+		for_each_cpu(sibling, &this_leaf->shared_cpu_map) {
+			struct cpu_cacheinfo *sib_cpu_ci;
+
+			if (sibling == cpu) /* skip itself */
+				continue;
+			sib_cpu_ci = get_cpu_cacheinfo(sibling);
+			sib_leaf = sib_cpu_ci->info_list + index;
+			cpumask_clear_cpu(cpu, &sib_leaf->shared_cpu_map);
+			cpumask_clear_cpu(sibling, &this_leaf->shared_cpu_map);
+		}
+		of_node_put(this_leaf->of_node);
+	}
+}
+
+int __weak init_cache_level(unsigned int cpu)
+{
+	return -ENOENT;
+}
+
+int __weak populate_cache_leaves(unsigned int cpu)
+{
+	return -ENOENT;
+}
+
+static void free_cache_attributes(unsigned int cpu)
+{
+	cache_shared_cpu_map_remove(cpu);
+
+	kfree(per_cpu_cacheinfo(cpu));
+	per_cpu_cacheinfo(cpu) = NULL;
+}
+
+/*
+ * Helpers to make sure "func" is executed on the cpu whose cache
+ * attributes are being detected
+ */
+#define DEFINE_SMP_CALL_FUNCTION(func)		\
+static void _##func(void *ret)			\
+{						\
+	int cpu = smp_processor_id();		\
+	*(int *)ret = func(cpu);		\
+}						\
+static int __##func(unsigned int cpu)		\
+{						\
+	int ret;				\
+	smp_call_function_single(cpu, _##func, &ret, true);	\
+	return ret;				\
+}
+DEFINE_SMP_CALL_FUNCTION(init_cache_level)
+DEFINE_SMP_CALL_FUNCTION(populate_cache_leaves)
+
+static int detect_cache_attributes(unsigned int cpu)
+{
+	int ret;
+
+	if (__init_cache_level(cpu))
+		return -ENOENT;
+
+	per_cpu_cacheinfo(cpu) = kzalloc(sizeof(struct cacheinfo) *
+					 cache_leaves(cpu), GFP_KERNEL);
+	if (per_cpu_cacheinfo(cpu) == NULL)
+		return -ENOMEM;
+
+	ret = __populate_cache_leaves(cpu);
+	if (ret)
+		goto free_ci;
+	/*
+	 * For systems using DT for cache hierarcy, of_node and shared_cpu_map
+	 * will be set up here. Otherwise populate_cache_leaves needs to set
+	 * shared_cpu_map and next-level-cache should not be specified in DT
+	 */
+	ret = of_cache_shared_cpu_map_setup(cpu);
+	if (ret)
+		goto free_ci;
+	return 0;
+
+free_ci:
+	free_cache_attributes(cpu);
+	return ret;
+}
+
+#ifdef CONFIG_SYSFS
+
+/* pointer to cpuX/cache device */
+static DEFINE_PER_CPU(struct device *, ci_cache_dev);
+#define per_cpu_cache_dev(cpu)	(per_cpu(ci_cache_dev, cpu))
+
+static cpumask_t cache_dev_map;
+
+/* pointer to array of devices for cpuX/cache/indexY */
+static DEFINE_PER_CPU(struct device **, ci_index_dev);
+#define per_cpu_index_dev(cpu)	(per_cpu(ci_index_dev, cpu))
+#define per_cache_index_dev(cpu, idx)	((per_cpu_index_dev(cpu))[idx])
+
+#define show_one(file_name, object)				\
+static ssize_t file_name##_show(struct device *dev,		\
+		struct device_attribute *attr, char *buf)	\
+{								\
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);	\
+	if (!this_leaf->object)					\
+		return -EINVAL;					\
+	return sprintf(buf, "%u\n", this_leaf->object);		\
+}
+
+show_one(level, level);
+show_one(coherency_line_size, coherency_line_size);
+show_one(number_of_sets, number_of_sets);
+show_one(physical_line_partition, physical_line_partition);
+
+static ssize_t ways_of_associativity_show(struct device *dev,
+					  struct device_attribute *attr,
+					  char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	/* will be zero for fully associative cache, but check for size */
+	if (!this_leaf->size)
+		return -EINVAL;
+	return sprintf(buf, "%u\n", this_leaf->ways_of_associativity);
+}
+
+static ssize_t size_show(struct device *dev,
+			 struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	if (!this_leaf->size)
+		return -EINVAL;
+	return sprintf(buf, "%uK\n", this_leaf->size >> 10);
+}
+
+static ssize_t shared_cpumap_show_func(struct device *dev, int type, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf;
+	int n = 0;
+
+	if (len > 1) {
+		const struct cpumask *mask = &this_leaf->shared_cpu_map;
+
+		n = type ? cpulist_scnprintf(buf, len - 2, mask) :
+			   cpumask_scnprintf(buf, len - 2, mask);
+		buf[n++] = '\n';
+		buf[n] = '\0';
+	}
+	return n;
+}
+
+static ssize_t shared_cpu_map_show(struct device *dev,
+				   struct device_attribute *attr, char *buf)
+{
+	return shared_cpumap_show_func(dev, 0, buf);
+}
+
+static ssize_t shared_cpu_list_show(struct device *dev,
+				    struct device_attribute *attr, char *buf)
+{
+	return shared_cpumap_show_func(dev, 1, buf);
+}
+
+static ssize_t type_show(struct device *dev,
+			 struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	switch (this_leaf->type) {
+	case CACHE_TYPE_DATA:
+		return sprintf(buf, "Data\n");
+	case CACHE_TYPE_INST:
+		return sprintf(buf, "Instruction\n");
+	case CACHE_TYPE_UNIFIED:
+		return sprintf(buf, "Unified\n");
+	default:
+		return -EINVAL;
+	}
+}
+
+static ssize_t attributes_show(struct device *dev,
+			       struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	unsigned int ci_attr = this_leaf->attributes;
+	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf - 2;
+	int n = 0;
+
+	if (!ci_attr)
+		return -EINVAL;
+
+	if (ci_attr & CACHE_WRITE_THROUGH)
+		n += snprintf(buf + n, len - n, "WriteThrough\n");
+	if (ci_attr & CACHE_WRITE_BACK)
+		n += snprintf(buf + n, len - n, "WriteBack\n");
+	if (ci_attr & CACHE_READ_ALLOCATE)
+		n += snprintf(buf + n, len - n, "ReadAllocate\n");
+	if (ci_attr & CACHE_WRITE_ALLOCATE)
+		n += snprintf(buf + n, len - n, "WriteAllocate\n");
+	buf[n] = '\0';
+	return n;
+}
+
+static DEVICE_ATTR_RO(level);
+static DEVICE_ATTR_RO(type);
+static DEVICE_ATTR_RO(coherency_line_size);
+static DEVICE_ATTR_RO(ways_of_associativity);
+static DEVICE_ATTR_RO(number_of_sets);
+static DEVICE_ATTR_RO(size);
+static DEVICE_ATTR_RO(attributes);
+static DEVICE_ATTR_RO(shared_cpu_map);
+static DEVICE_ATTR_RO(shared_cpu_list);
+static DEVICE_ATTR_RO(physical_line_partition);
+
+static struct attribute *cache_default_attrs[] = {
+	&dev_attr_type.attr,
+	&dev_attr_level.attr,
+	&dev_attr_shared_cpu_map.attr,
+	&dev_attr_shared_cpu_list.attr,
+	NULL
+};
+
+ATTRIBUTE_GROUPS(cache_default);
+
+static const struct device_attribute *cache_optional_attrs[] = {
+	&dev_attr_coherency_line_size,
+	&dev_attr_ways_of_associativity,
+	&dev_attr_number_of_sets,
+	&dev_attr_size,
+	&dev_attr_attributes,
+	&dev_attr_physical_line_partition,
+	NULL
+};
+
+static int device_add_attrs(struct device *dev,
+			    const struct device_attribute **dev_attrs)
+{
+	int i, error = 0;
+	struct device_attribute *dev_attr;
+	char *buf;
+
+	if (!dev_attrs)
+		return 0;
+
+	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+
+	for (i = 0; dev_attrs[i]; i++) {
+		dev_attr = (struct device_attribute *)dev_attrs[i];
+
+		/* create attributes that provides meaningful value */
+		if (dev_attr->show(dev, dev_attr, buf) < 0)
+			continue;
+
+		error = device_create_file(dev, dev_attrs[i]);
+		if (error) {
+			while (--i >= 0)
+				device_remove_file(dev, dev_attrs[i]);
+			break;
+		}
+	}
+
+	kfree(buf);
+	return error;
+}
+
+static void device_remove_attrs(struct device *dev,
+				const struct device_attribute **dev_attrs)
+{
+	int i;
+
+	if (!dev_attrs)
+		return;
+
+	for (i = 0; dev_attrs[i]; dev_attrs++, i++)
+		device_remove_file(dev, dev_attrs[i]);
+}
+
+const struct device_attribute **
+__weak cache_get_priv_attr(struct device *cache_idx_dev)
+{
+	return NULL;
+}
+
+/* Add/Remove cache interface for CPU device */
+static void cpu_cache_sysfs_exit(unsigned int cpu)
+{
+	int i;
+	struct device *tmp_dev;
+	const struct device_attribute **ci_priv_attr;
+
+	if (per_cpu_index_dev(cpu)) {
+		for (i = 0; i < cache_leaves(cpu); i++) {
+			tmp_dev = per_cache_index_dev(cpu, i);
+			if (!tmp_dev)
+				continue;
+			ci_priv_attr = cache_get_priv_attr(tmp_dev);
+			device_remove_attrs(tmp_dev, ci_priv_attr);
+			device_remove_attrs(tmp_dev, cache_optional_attrs);
+			device_unregister(tmp_dev);
+		}
+		kfree(per_cpu_index_dev(cpu));
+		per_cpu_index_dev(cpu) = NULL;
+	}
+	device_unregister(per_cpu_cache_dev(cpu));
+	per_cpu_cache_dev(cpu) = NULL;
+}
+
+static int cpu_cache_sysfs_init(unsigned int cpu)
+{
+	struct device *dev = get_cpu_device(cpu);
+
+	if (per_cpu_cacheinfo(cpu) == NULL)
+		return -ENOENT;
+
+	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
+					       NULL, "cache");
+	if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
+		return PTR_ERR(per_cpu_cache_dev(cpu));
+
+	/* Allocate all required memory */
+	per_cpu_index_dev(cpu) = kzalloc(sizeof(struct device *) *
+					 cache_leaves(cpu), GFP_KERNEL);
+	if (unlikely(per_cpu_index_dev(cpu) == NULL))
+		goto err_out;
+
+	return 0;
+
+err_out:
+	cpu_cache_sysfs_exit(cpu);
+	return -ENOMEM;
+}
+
+static int cache_add_dev(unsigned int cpu)
+{
+	unsigned short i;
+	int rc;
+	struct device *tmp_dev, *parent;
+	struct cacheinfo *this_leaf;
+	const struct device_attribute **ci_priv_attr;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	rc = cpu_cache_sysfs_init(cpu);
+	if (unlikely(rc < 0))
+		return rc;
+
+	parent = per_cpu_cache_dev(cpu);
+	for (i = 0; i < cache_leaves(cpu); i++) {
+		this_leaf = this_cpu_ci->info_list + i;
+		if (this_leaf->disable_sysfs)
+			continue;
+		tmp_dev = device_create_with_groups(parent->class, parent, i,
+						    this_leaf,
+						    cache_default_groups,
+						    "index%1u", i);
+		if (IS_ERR_OR_NULL(tmp_dev)) {
+			rc = PTR_ERR(tmp_dev);
+			goto err;
+		}
+
+		rc = device_add_attrs(tmp_dev, cache_optional_attrs);
+		if (unlikely(rc))
+			goto err;
+
+		ci_priv_attr = cache_get_priv_attr(tmp_dev);
+		rc = device_add_attrs(tmp_dev, ci_priv_attr);
+		if (unlikely(rc))
+			goto err;
+
+		per_cache_index_dev(cpu, i) = tmp_dev;
+	}
+	cpumask_set_cpu(cpu, &cache_dev_map);
+
+	return 0;
+err:
+	cpu_cache_sysfs_exit(cpu);
+	return rc;
+}
+
+static void cache_remove_dev(unsigned int cpu)
+{
+	if (!cpumask_test_cpu(cpu, &cache_dev_map))
+		return;
+	cpumask_clear_cpu(cpu, &cache_dev_map);
+
+	cpu_cache_sysfs_exit(cpu);
+}
+
+static int cacheinfo_cpu_callback(struct notifier_block *nfb,
+				  unsigned long action, void *hcpu)
+{
+	unsigned int cpu = (unsigned long)hcpu;
+	int rc = 0;
+
+	switch (action) {
+	case CPU_ONLINE:
+	case CPU_ONLINE_FROZEN:
+		rc = detect_cache_attributes(cpu);
+		if (!rc)
+			rc = cache_add_dev(cpu);
+		break;
+	case CPU_DEAD:
+	case CPU_DEAD_FROZEN:
+		cache_remove_dev(cpu);
+		if (per_cpu_cacheinfo(cpu))
+			free_cache_attributes(cpu);
+		break;
+	}
+	return notifier_from_errno(rc);
+}
+
+static int __init cacheinfo_sysfs_init(void)
+{
+	int cpu, rc = 0;
+
+	cpu_notifier_register_begin();
+
+	for_each_online_cpu(cpu) {
+		rc = detect_cache_attributes(cpu);
+		if (rc) {
+			pr_err("error detecting cacheinfo..cpu%d\n", cpu);
+			goto out;
+		}
+		rc = cache_add_dev(cpu);
+		if (rc) {
+			free_cache_attributes(cpu);
+			pr_err("error populating cacheinfo..cpu%d\n", cpu);
+			goto out;
+		}
+	}
+	__hotcpu_notifier(cacheinfo_cpu_callback, 0);
+
+out:
+	cpu_notifier_register_done();
+	return rc;
+}
+
+device_initcall(cacheinfo_sysfs_init);
+
+#endif	/* CONFIG_SYSFS */
diff --git a/include/linux/cacheinfo.h b/include/linux/cacheinfo.h
new file mode 100644
index 0000000..a9bd9f9
--- /dev/null
+++ b/include/linux/cacheinfo.h
@@ -0,0 +1,56 @@
+#ifndef _LINUX_CACHEINFO_H
+#define _LINUX_CACHEINFO_H
+
+#include <linux/bitops.h>
+#include <linux/compiler.h>
+#include <linux/cpumask.h>
+#include <linux/device.h>
+#include <linux/of.h>
+#include <linux/sysfs.h>
+
+enum cache_type {
+	CACHE_TYPE_NOCACHE = 0,
+	CACHE_TYPE_INST = BIT(0),
+	CACHE_TYPE_DATA = BIT(1),
+	CACHE_TYPE_SEPARATE = CACHE_TYPE_INST | CACHE_TYPE_DATA,
+	CACHE_TYPE_UNIFIED = BIT(2),
+};
+
+struct cacheinfo {
+	/* core properties */
+	enum cache_type type; /* data, inst or unified */
+	unsigned int level;
+	unsigned int coherency_line_size; /* cache line size  */
+	unsigned int number_of_sets; /* no. of sets per way */
+	unsigned int ways_of_associativity; /* no. of ways */
+	unsigned int physical_line_partition; /* no. of lines per tag */
+	unsigned int size; /* total cache size */
+	cpumask_t shared_cpu_map;
+	unsigned int attributes;
+#define CACHE_WRITE_THROUGH	BIT(0)
+#define CACHE_WRITE_BACK	BIT(1)
+#define CACHE_READ_ALLOCATE	BIT(2)
+#define CACHE_WRITE_ALLOCATE	BIT(3)
+
+	/* book keeping */
+	struct device_node *of_node;	/* cpu if no explicit cache node */
+	bool disable_sysfs; /* don't expose this leaf through sysfs */
+	void *priv;
+};
+
+struct cpu_cacheinfo {
+	struct cacheinfo *info_list;
+	unsigned int num_levels;
+	unsigned int num_leaves;
+};
+
+struct cpu_cacheinfo *get_cpu_cacheinfo(unsigned int cpu);
+int init_cache_level(unsigned int cpu);
+int populate_cache_leaves(unsigned int cpu);
+
+#ifdef CONFIG_SYSFS
+const struct device_attribute **
+cache_get_priv_attr(struct device *cache_idx_dev);
+#endif
+
+#endif /* _LINUX_CACHEINFO_H */
-- 
1.8.3.2

[PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds initial support for providing processor cache information
to userspace through sysfs interface. This is based on already existing
implementations(x86, ia64, s390 and powerpc) and hence the interface is
intended to be fully compatible.

The main purpose of this generic support is to avoid further code
duplication to support new architectures and also to unify all the existing
different implementations.

This implementation maintains the hierarchy of cache objects which reflects
the system's cache topology. Cache devices are instantiated as needed as
CPUs come online. The cache information is replicated per-cpu even if they are
shared. A per-cpu array of cache information maintained is used mainly for
sysfs-related book keeping.

It also implements the shared_cpu_map attribute, which is essential for
enabling both kernel and user-space to discover the system's overall cache
topology.

This patch also add the missing ABI documentation for the cacheinfo sysfs
interface already, which is well defined and widely used.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Rob Herring <robh@kernel.org>
Cc: linux-doc@vger.kernel.org
Cc: linux-ia64@vger.kernel.org
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-arm-kernel@lists.infradead.org
---
 Documentation/ABI/testing/sysfs-devices-system-cpu |  41 ++
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 564 +++++++++++++++++++++
 include/linux/cacheinfo.h                          |  56 ++
 4 files changed, 662 insertions(+), 1 deletion(-)
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

diff --git a/Documentation/ABI/testing/sysfs-devices-system-cpu b/Documentation/ABI/testing/sysfs-devices-system-cpu
index acb9bfc..5827f4e 100644
--- a/Documentation/ABI/testing/sysfs-devices-system-cpu
+++ b/Documentation/ABI/testing/sysfs-devices-system-cpu
@@ -224,3 +224,44 @@ Description:	Parameters for the Intel P-state driver
 		frequency range.
 
 		More details can be found in Documentation/cpu-freq/intel-pstate.txt
+
+What:		/sys/devices/system/cpu/cpu*/cache/index*/<set_of_attributes_mentioned_below>
+Date:		June 2014(documented, existed before August 2008)
+Contact:	Sudeep Holla <sudeep.holla@arm.com>
+		Linux kernel mailing list <linux-kernel@vger.kernel.org>
+Description:	Parameters for the CPU cache attributes
+
+		attributes:
+			- writethrough: data is written to both the cache line
+					and to the block in the lower-level memory
+			- writeback: data is written only to the cache line and
+				     the modified cache line is written to main
+				     memory only when it is replaced
+			- writeallocate: allocate a memory location to a cache line
+					 on a cache miss because of a write
+			- readallocate: allocate a memory location to a cache line
+					on a cache miss because of a read
+
+		coherency_line_size: the minimum amount of data that gets transferred
+
+		level: the cache hierarcy in the multi-level cache configuration
+
+		number_of_sets: total number of sets in the cache, a set is a
+				collection of cache lines with the same cache index
+
+		physical_line_partition: number of physical cache line per cache tag
+
+		shared_cpu_list: the list of cpus sharing the cache
+
+		shared_cpu_map: logical cpu mask containing the list of cpus sharing
+				the cache
+
+		size: the total cache size in kB
+
+		type:
+			- instruction: cache that only holds instructions
+			- data: cache that only caches data
+			- unified: cache that holds both data and instructions
+
+		ways_of_associativity: degree of freedom in placing a particular block
+					of memory in the cache
diff --git a/drivers/base/Makefile b/drivers/base/Makefile
index 04b314e..bad2ff8 100644
--- a/drivers/base/Makefile
+++ b/drivers/base/Makefile
@@ -4,7 +4,7 @@ obj-y			:= component.o core.o bus.o dd.o syscore.o \
 			   driver.o class.o platform.o \
 			   cpu.o firmware.o init.o map.o devres.o \
 			   attribute_container.o transport_class.o \
-			   topology.o container.o
+			   topology.o container.o cacheinfo.o
 obj-$(CONFIG_DEVTMPFS)	+= devtmpfs.o
 obj-$(CONFIG_DMA_CMA) += dma-contiguous.o
 obj-y			+= power/
diff --git a/drivers/base/cacheinfo.c b/drivers/base/cacheinfo.c
new file mode 100644
index 0000000..c12e03c
--- /dev/null
+++ b/drivers/base/cacheinfo.c
@@ -0,0 +1,564 @@
+/*
+ * cacheinfo support - processor cache information via sysfs
+ *
+ * Based on arch/x86/kernel/cpu/intel_cacheinfo.c
+ * Author: Sudeep Holla <sudeep.holla@arm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/compiler.h>
+#include <linux/cpu.h>
+#include <linux/device.h>
+#include <linux/init.h>
+#include <linux/of.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/smp.h>
+#include <linux/sysfs.h>
+
+/* pointer to per cpu cacheinfo */
+static DEFINE_PER_CPU(struct cpu_cacheinfo, ci_cpu_cacheinfo);
+#define ci_cacheinfo(cpu)	(&per_cpu(ci_cpu_cacheinfo, cpu))
+#define cache_leaves(cpu)	(ci_cacheinfo(cpu)->num_leaves)
+#define per_cpu_cacheinfo(cpu)	(ci_cacheinfo(cpu)->info_list)
+
+struct cpu_cacheinfo *get_cpu_cacheinfo(unsigned int cpu)
+{
+	return ci_cacheinfo(cpu);
+}
+
+#ifdef CONFIG_OF
+static int cache_setup_of_node(unsigned int cpu)
+{
+	struct device_node *np;
+	struct cacheinfo *this_leaf;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int index = 0;
+
+	/* skip if of_node is already populated */
+	if (this_cpu_ci->info_list->of_node)
+		return 0;
+
+	if (!cpu_dev) {
+		pr_err("No cpu device for CPU %d\n", cpu);
+		return -ENODEV;
+	}
+	np = cpu_dev->of_node;
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
+	}
+
+	while (np && index < cache_leaves(cpu)) {
+		this_leaf = this_cpu_ci->info_list + index;
+		if (this_leaf->level != 1)
+			np = of_find_next_cache_node(np);
+		else
+			np = of_node_get(np);/* cpu node itself */
+		this_leaf->of_node = np;
+		index++;
+	}
+	return 0;
+}
+
+static inline bool cache_leaves_are_shared(struct cacheinfo *this_leaf,
+					   struct cacheinfo *sib_leaf)
+{
+	return sib_leaf->of_node = this_leaf->of_node;
+}
+
+static int of_cache_shared_cpu_map_setup(unsigned int cpu)
+{
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sib_leaf;
+	unsigned int index;
+	int ret;
+
+	ret = cache_setup_of_node(cpu);
+	if (ret)
+		return ret;
+
+	for (index = 0; index < cache_leaves(cpu); index++) {
+		unsigned int i;
+
+		this_leaf = this_cpu_ci->info_list + index;
+		cpumask_set_cpu(cpu, &this_leaf->shared_cpu_map);
+
+		for_each_online_cpu(i) {
+			struct cpu_cacheinfo *sib_cpu_ci = get_cpu_cacheinfo(i);
+
+			if (i = cpu || !sib_cpu_ci->info_list)
+				continue;/* skip if itself or no cacheinfo */
+			sib_leaf = sib_cpu_ci->info_list + index;
+			if (cache_leaves_are_shared(this_leaf, sib_leaf)) {
+				cpumask_set_cpu(cpu, &sib_leaf->shared_cpu_map);
+				cpumask_set_cpu(i, &this_leaf->shared_cpu_map);
+			}
+		}
+	}
+
+	return 0;
+}
+#else
+static inline int of_cache_shared_cpu_map_setup(unsigned int cpu)
+{
+	return 0;
+}
+#endif
+
+static void cache_shared_cpu_map_remove(unsigned int cpu)
+{
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sib_leaf;
+	unsigned int sibling, index;
+
+	for (index = 0; index < cache_leaves(cpu); index++) {
+		this_leaf = this_cpu_ci->info_list + index;
+		for_each_cpu(sibling, &this_leaf->shared_cpu_map) {
+			struct cpu_cacheinfo *sib_cpu_ci;
+
+			if (sibling = cpu) /* skip itself */
+				continue;
+			sib_cpu_ci = get_cpu_cacheinfo(sibling);
+			sib_leaf = sib_cpu_ci->info_list + index;
+			cpumask_clear_cpu(cpu, &sib_leaf->shared_cpu_map);
+			cpumask_clear_cpu(sibling, &this_leaf->shared_cpu_map);
+		}
+		of_node_put(this_leaf->of_node);
+	}
+}
+
+int __weak init_cache_level(unsigned int cpu)
+{
+	return -ENOENT;
+}
+
+int __weak populate_cache_leaves(unsigned int cpu)
+{
+	return -ENOENT;
+}
+
+static void free_cache_attributes(unsigned int cpu)
+{
+	cache_shared_cpu_map_remove(cpu);
+
+	kfree(per_cpu_cacheinfo(cpu));
+	per_cpu_cacheinfo(cpu) = NULL;
+}
+
+/*
+ * Helpers to make sure "func" is executed on the cpu whose cache
+ * attributes are being detected
+ */
+#define DEFINE_SMP_CALL_FUNCTION(func)		\
+static void _##func(void *ret)			\
+{						\
+	int cpu = smp_processor_id();		\
+	*(int *)ret = func(cpu);		\
+}						\
+static int __##func(unsigned int cpu)		\
+{						\
+	int ret;				\
+	smp_call_function_single(cpu, _##func, &ret, true);	\
+	return ret;				\
+}
+DEFINE_SMP_CALL_FUNCTION(init_cache_level)
+DEFINE_SMP_CALL_FUNCTION(populate_cache_leaves)
+
+static int detect_cache_attributes(unsigned int cpu)
+{
+	int ret;
+
+	if (__init_cache_level(cpu))
+		return -ENOENT;
+
+	per_cpu_cacheinfo(cpu) = kzalloc(sizeof(struct cacheinfo) *
+					 cache_leaves(cpu), GFP_KERNEL);
+	if (per_cpu_cacheinfo(cpu) = NULL)
+		return -ENOMEM;
+
+	ret = __populate_cache_leaves(cpu);
+	if (ret)
+		goto free_ci;
+	/*
+	 * For systems using DT for cache hierarcy, of_node and shared_cpu_map
+	 * will be set up here. Otherwise populate_cache_leaves needs to set
+	 * shared_cpu_map and next-level-cache should not be specified in DT
+	 */
+	ret = of_cache_shared_cpu_map_setup(cpu);
+	if (ret)
+		goto free_ci;
+	return 0;
+
+free_ci:
+	free_cache_attributes(cpu);
+	return ret;
+}
+
+#ifdef CONFIG_SYSFS
+
+/* pointer to cpuX/cache device */
+static DEFINE_PER_CPU(struct device *, ci_cache_dev);
+#define per_cpu_cache_dev(cpu)	(per_cpu(ci_cache_dev, cpu))
+
+static cpumask_t cache_dev_map;
+
+/* pointer to array of devices for cpuX/cache/indexY */
+static DEFINE_PER_CPU(struct device **, ci_index_dev);
+#define per_cpu_index_dev(cpu)	(per_cpu(ci_index_dev, cpu))
+#define per_cache_index_dev(cpu, idx)	((per_cpu_index_dev(cpu))[idx])
+
+#define show_one(file_name, object)				\
+static ssize_t file_name##_show(struct device *dev,		\
+		struct device_attribute *attr, char *buf)	\
+{								\
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);	\
+	if (!this_leaf->object)					\
+		return -EINVAL;					\
+	return sprintf(buf, "%u\n", this_leaf->object);		\
+}
+
+show_one(level, level);
+show_one(coherency_line_size, coherency_line_size);
+show_one(number_of_sets, number_of_sets);
+show_one(physical_line_partition, physical_line_partition);
+
+static ssize_t ways_of_associativity_show(struct device *dev,
+					  struct device_attribute *attr,
+					  char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	/* will be zero for fully associative cache, but check for size */
+	if (!this_leaf->size)
+		return -EINVAL;
+	return sprintf(buf, "%u\n", this_leaf->ways_of_associativity);
+}
+
+static ssize_t size_show(struct device *dev,
+			 struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	if (!this_leaf->size)
+		return -EINVAL;
+	return sprintf(buf, "%uK\n", this_leaf->size >> 10);
+}
+
+static ssize_t shared_cpumap_show_func(struct device *dev, int type, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf;
+	int n = 0;
+
+	if (len > 1) {
+		const struct cpumask *mask = &this_leaf->shared_cpu_map;
+
+		n = type ? cpulist_scnprintf(buf, len - 2, mask) :
+			   cpumask_scnprintf(buf, len - 2, mask);
+		buf[n++] = '\n';
+		buf[n] = '\0';
+	}
+	return n;
+}
+
+static ssize_t shared_cpu_map_show(struct device *dev,
+				   struct device_attribute *attr, char *buf)
+{
+	return shared_cpumap_show_func(dev, 0, buf);
+}
+
+static ssize_t shared_cpu_list_show(struct device *dev,
+				    struct device_attribute *attr, char *buf)
+{
+	return shared_cpumap_show_func(dev, 1, buf);
+}
+
+static ssize_t type_show(struct device *dev,
+			 struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	switch (this_leaf->type) {
+	case CACHE_TYPE_DATA:
+		return sprintf(buf, "Data\n");
+	case CACHE_TYPE_INST:
+		return sprintf(buf, "Instruction\n");
+	case CACHE_TYPE_UNIFIED:
+		return sprintf(buf, "Unified\n");
+	default:
+		return -EINVAL;
+	}
+}
+
+static ssize_t attributes_show(struct device *dev,
+			       struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	unsigned int ci_attr = this_leaf->attributes;
+	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf - 2;
+	int n = 0;
+
+	if (!ci_attr)
+		return -EINVAL;
+
+	if (ci_attr & CACHE_WRITE_THROUGH)
+		n += snprintf(buf + n, len - n, "WriteThrough\n");
+	if (ci_attr & CACHE_WRITE_BACK)
+		n += snprintf(buf + n, len - n, "WriteBack\n");
+	if (ci_attr & CACHE_READ_ALLOCATE)
+		n += snprintf(buf + n, len - n, "ReadAllocate\n");
+	if (ci_attr & CACHE_WRITE_ALLOCATE)
+		n += snprintf(buf + n, len - n, "WriteAllocate\n");
+	buf[n] = '\0';
+	return n;
+}
+
+static DEVICE_ATTR_RO(level);
+static DEVICE_ATTR_RO(type);
+static DEVICE_ATTR_RO(coherency_line_size);
+static DEVICE_ATTR_RO(ways_of_associativity);
+static DEVICE_ATTR_RO(number_of_sets);
+static DEVICE_ATTR_RO(size);
+static DEVICE_ATTR_RO(attributes);
+static DEVICE_ATTR_RO(shared_cpu_map);
+static DEVICE_ATTR_RO(shared_cpu_list);
+static DEVICE_ATTR_RO(physical_line_partition);
+
+static struct attribute *cache_default_attrs[] = {
+	&dev_attr_type.attr,
+	&dev_attr_level.attr,
+	&dev_attr_shared_cpu_map.attr,
+	&dev_attr_shared_cpu_list.attr,
+	NULL
+};
+
+ATTRIBUTE_GROUPS(cache_default);
+
+static const struct device_attribute *cache_optional_attrs[] = {
+	&dev_attr_coherency_line_size,
+	&dev_attr_ways_of_associativity,
+	&dev_attr_number_of_sets,
+	&dev_attr_size,
+	&dev_attr_attributes,
+	&dev_attr_physical_line_partition,
+	NULL
+};
+
+static int device_add_attrs(struct device *dev,
+			    const struct device_attribute **dev_attrs)
+{
+	int i, error = 0;
+	struct device_attribute *dev_attr;
+	char *buf;
+
+	if (!dev_attrs)
+		return 0;
+
+	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+
+	for (i = 0; dev_attrs[i]; i++) {
+		dev_attr = (struct device_attribute *)dev_attrs[i];
+
+		/* create attributes that provides meaningful value */
+		if (dev_attr->show(dev, dev_attr, buf) < 0)
+			continue;
+
+		error = device_create_file(dev, dev_attrs[i]);
+		if (error) {
+			while (--i >= 0)
+				device_remove_file(dev, dev_attrs[i]);
+			break;
+		}
+	}
+
+	kfree(buf);
+	return error;
+}
+
+static void device_remove_attrs(struct device *dev,
+				const struct device_attribute **dev_attrs)
+{
+	int i;
+
+	if (!dev_attrs)
+		return;
+
+	for (i = 0; dev_attrs[i]; dev_attrs++, i++)
+		device_remove_file(dev, dev_attrs[i]);
+}
+
+const struct device_attribute **
+__weak cache_get_priv_attr(struct device *cache_idx_dev)
+{
+	return NULL;
+}
+
+/* Add/Remove cache interface for CPU device */
+static void cpu_cache_sysfs_exit(unsigned int cpu)
+{
+	int i;
+	struct device *tmp_dev;
+	const struct device_attribute **ci_priv_attr;
+
+	if (per_cpu_index_dev(cpu)) {
+		for (i = 0; i < cache_leaves(cpu); i++) {
+			tmp_dev = per_cache_index_dev(cpu, i);
+			if (!tmp_dev)
+				continue;
+			ci_priv_attr = cache_get_priv_attr(tmp_dev);
+			device_remove_attrs(tmp_dev, ci_priv_attr);
+			device_remove_attrs(tmp_dev, cache_optional_attrs);
+			device_unregister(tmp_dev);
+		}
+		kfree(per_cpu_index_dev(cpu));
+		per_cpu_index_dev(cpu) = NULL;
+	}
+	device_unregister(per_cpu_cache_dev(cpu));
+	per_cpu_cache_dev(cpu) = NULL;
+}
+
+static int cpu_cache_sysfs_init(unsigned int cpu)
+{
+	struct device *dev = get_cpu_device(cpu);
+
+	if (per_cpu_cacheinfo(cpu) = NULL)
+		return -ENOENT;
+
+	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
+					       NULL, "cache");
+	if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
+		return PTR_ERR(per_cpu_cache_dev(cpu));
+
+	/* Allocate all required memory */
+	per_cpu_index_dev(cpu) = kzalloc(sizeof(struct device *) *
+					 cache_leaves(cpu), GFP_KERNEL);
+	if (unlikely(per_cpu_index_dev(cpu) = NULL))
+		goto err_out;
+
+	return 0;
+
+err_out:
+	cpu_cache_sysfs_exit(cpu);
+	return -ENOMEM;
+}
+
+static int cache_add_dev(unsigned int cpu)
+{
+	unsigned short i;
+	int rc;
+	struct device *tmp_dev, *parent;
+	struct cacheinfo *this_leaf;
+	const struct device_attribute **ci_priv_attr;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	rc = cpu_cache_sysfs_init(cpu);
+	if (unlikely(rc < 0))
+		return rc;
+
+	parent = per_cpu_cache_dev(cpu);
+	for (i = 0; i < cache_leaves(cpu); i++) {
+		this_leaf = this_cpu_ci->info_list + i;
+		if (this_leaf->disable_sysfs)
+			continue;
+		tmp_dev = device_create_with_groups(parent->class, parent, i,
+						    this_leaf,
+						    cache_default_groups,
+						    "index%1u", i);
+		if (IS_ERR_OR_NULL(tmp_dev)) {
+			rc = PTR_ERR(tmp_dev);
+			goto err;
+		}
+
+		rc = device_add_attrs(tmp_dev, cache_optional_attrs);
+		if (unlikely(rc))
+			goto err;
+
+		ci_priv_attr = cache_get_priv_attr(tmp_dev);
+		rc = device_add_attrs(tmp_dev, ci_priv_attr);
+		if (unlikely(rc))
+			goto err;
+
+		per_cache_index_dev(cpu, i) = tmp_dev;
+	}
+	cpumask_set_cpu(cpu, &cache_dev_map);
+
+	return 0;
+err:
+	cpu_cache_sysfs_exit(cpu);
+	return rc;
+}
+
+static void cache_remove_dev(unsigned int cpu)
+{
+	if (!cpumask_test_cpu(cpu, &cache_dev_map))
+		return;
+	cpumask_clear_cpu(cpu, &cache_dev_map);
+
+	cpu_cache_sysfs_exit(cpu);
+}
+
+static int cacheinfo_cpu_callback(struct notifier_block *nfb,
+				  unsigned long action, void *hcpu)
+{
+	unsigned int cpu = (unsigned long)hcpu;
+	int rc = 0;
+
+	switch (action) {
+	case CPU_ONLINE:
+	case CPU_ONLINE_FROZEN:
+		rc = detect_cache_attributes(cpu);
+		if (!rc)
+			rc = cache_add_dev(cpu);
+		break;
+	case CPU_DEAD:
+	case CPU_DEAD_FROZEN:
+		cache_remove_dev(cpu);
+		if (per_cpu_cacheinfo(cpu))
+			free_cache_attributes(cpu);
+		break;
+	}
+	return notifier_from_errno(rc);
+}
+
+static int __init cacheinfo_sysfs_init(void)
+{
+	int cpu, rc = 0;
+
+	cpu_notifier_register_begin();
+
+	for_each_online_cpu(cpu) {
+		rc = detect_cache_attributes(cpu);
+		if (rc) {
+			pr_err("error detecting cacheinfo..cpu%d\n", cpu);
+			goto out;
+		}
+		rc = cache_add_dev(cpu);
+		if (rc) {
+			free_cache_attributes(cpu);
+			pr_err("error populating cacheinfo..cpu%d\n", cpu);
+			goto out;
+		}
+	}
+	__hotcpu_notifier(cacheinfo_cpu_callback, 0);
+
+out:
+	cpu_notifier_register_done();
+	return rc;
+}
+
+device_initcall(cacheinfo_sysfs_init);
+
+#endif	/* CONFIG_SYSFS */
diff --git a/include/linux/cacheinfo.h b/include/linux/cacheinfo.h
new file mode 100644
index 0000000..a9bd9f9
--- /dev/null
+++ b/include/linux/cacheinfo.h
@@ -0,0 +1,56 @@
+#ifndef _LINUX_CACHEINFO_H
+#define _LINUX_CACHEINFO_H
+
+#include <linux/bitops.h>
+#include <linux/compiler.h>
+#include <linux/cpumask.h>
+#include <linux/device.h>
+#include <linux/of.h>
+#include <linux/sysfs.h>
+
+enum cache_type {
+	CACHE_TYPE_NOCACHE = 0,
+	CACHE_TYPE_INST = BIT(0),
+	CACHE_TYPE_DATA = BIT(1),
+	CACHE_TYPE_SEPARATE = CACHE_TYPE_INST | CACHE_TYPE_DATA,
+	CACHE_TYPE_UNIFIED = BIT(2),
+};
+
+struct cacheinfo {
+	/* core properties */
+	enum cache_type type; /* data, inst or unified */
+	unsigned int level;
+	unsigned int coherency_line_size; /* cache line size  */
+	unsigned int number_of_sets; /* no. of sets per way */
+	unsigned int ways_of_associativity; /* no. of ways */
+	unsigned int physical_line_partition; /* no. of lines per tag */
+	unsigned int size; /* total cache size */
+	cpumask_t shared_cpu_map;
+	unsigned int attributes;
+#define CACHE_WRITE_THROUGH	BIT(0)
+#define CACHE_WRITE_BACK	BIT(1)
+#define CACHE_READ_ALLOCATE	BIT(2)
+#define CACHE_WRITE_ALLOCATE	BIT(3)
+
+	/* book keeping */
+	struct device_node *of_node;	/* cpu if no explicit cache node */
+	bool disable_sysfs; /* don't expose this leaf through sysfs */
+	void *priv;
+};
+
+struct cpu_cacheinfo {
+	struct cacheinfo *info_list;
+	unsigned int num_levels;
+	unsigned int num_leaves;
+};
+
+struct cpu_cacheinfo *get_cpu_cacheinfo(unsigned int cpu);
+int init_cache_level(unsigned int cpu);
+int populate_cache_leaves(unsigned int cpu);
+
+#ifdef CONFIG_SYSFS
+const struct device_attribute **
+cache_get_priv_attr(struct device *cache_idx_dev);
+#endif
+
+#endif /* _LINUX_CACHEINFO_H */
-- 
1.8.3.2

[PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds initial support for providing processor cache information
to userspace through sysfs interface. This is based on already existing
implementations(x86, ia64, s390 and powerpc) and hence the interface is
intended to be fully compatible.

The main purpose of this generic support is to avoid further code
duplication to support new architectures and also to unify all the existing
different implementations.

This implementation maintains the hierarchy of cache objects which reflects
the system's cache topology. Cache devices are instantiated as needed as
CPUs come online. The cache information is replicated per-cpu even if they are
shared. A per-cpu array of cache information maintained is used mainly for
sysfs-related book keeping.

It also implements the shared_cpu_map attribute, which is essential for
enabling both kernel and user-space to discover the system's overall cache
topology.

This patch also add the missing ABI documentation for the cacheinfo sysfs
interface already, which is well defined and widely used.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Rob Herring <robh@kernel.org>
Cc: linux-doc at vger.kernel.org
Cc: linux-ia64 at vger.kernel.org
Cc: linux390 at de.ibm.com
Cc: linux-s390 at vger.kernel.org
Cc: x86 at kernel.org
Cc: linuxppc-dev at lists.ozlabs.org
Cc: linux-arm-kernel at lists.infradead.org
---
 Documentation/ABI/testing/sysfs-devices-system-cpu |  41 ++
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 564 +++++++++++++++++++++
 include/linux/cacheinfo.h                          |  56 ++
 4 files changed, 662 insertions(+), 1 deletion(-)
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

diff --git a/Documentation/ABI/testing/sysfs-devices-system-cpu b/Documentation/ABI/testing/sysfs-devices-system-cpu
index acb9bfc..5827f4e 100644
--- a/Documentation/ABI/testing/sysfs-devices-system-cpu
+++ b/Documentation/ABI/testing/sysfs-devices-system-cpu
@@ -224,3 +224,44 @@ Description:	Parameters for the Intel P-state driver
 		frequency range.
 
 		More details can be found in Documentation/cpu-freq/intel-pstate.txt
+
+What:		/sys/devices/system/cpu/cpu*/cache/index*/<set_of_attributes_mentioned_below>
+Date:		June 2014(documented, existed before August 2008)
+Contact:	Sudeep Holla <sudeep.holla@arm.com>
+		Linux kernel mailing list <linux-kernel@vger.kernel.org>
+Description:	Parameters for the CPU cache attributes
+
+		attributes:
+			- writethrough: data is written to both the cache line
+					and to the block in the lower-level memory
+			- writeback: data is written only to the cache line and
+				     the modified cache line is written to main
+				     memory only when it is replaced
+			- writeallocate: allocate a memory location to a cache line
+					 on a cache miss because of a write
+			- readallocate: allocate a memory location to a cache line
+					on a cache miss because of a read
+
+		coherency_line_size: the minimum amount of data that gets transferred
+
+		level: the cache hierarcy in the multi-level cache configuration
+
+		number_of_sets: total number of sets in the cache, a set is a
+				collection of cache lines with the same cache index
+
+		physical_line_partition: number of physical cache line per cache tag
+
+		shared_cpu_list: the list of cpus sharing the cache
+
+		shared_cpu_map: logical cpu mask containing the list of cpus sharing
+				the cache
+
+		size: the total cache size in kB
+
+		type:
+			- instruction: cache that only holds instructions
+			- data: cache that only caches data
+			- unified: cache that holds both data and instructions
+
+		ways_of_associativity: degree of freedom in placing a particular block
+					of memory in the cache
diff --git a/drivers/base/Makefile b/drivers/base/Makefile
index 04b314e..bad2ff8 100644
--- a/drivers/base/Makefile
+++ b/drivers/base/Makefile
@@ -4,7 +4,7 @@ obj-y			:= component.o core.o bus.o dd.o syscore.o \
 			   driver.o class.o platform.o \
 			   cpu.o firmware.o init.o map.o devres.o \
 			   attribute_container.o transport_class.o \
-			   topology.o container.o
+			   topology.o container.o cacheinfo.o
 obj-$(CONFIG_DEVTMPFS)	+= devtmpfs.o
 obj-$(CONFIG_DMA_CMA) += dma-contiguous.o
 obj-y			+= power/
diff --git a/drivers/base/cacheinfo.c b/drivers/base/cacheinfo.c
new file mode 100644
index 0000000..c12e03c
--- /dev/null
+++ b/drivers/base/cacheinfo.c
@@ -0,0 +1,564 @@
+/*
+ * cacheinfo support - processor cache information via sysfs
+ *
+ * Based on arch/x86/kernel/cpu/intel_cacheinfo.c
+ * Author: Sudeep Holla <sudeep.holla@arm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/compiler.h>
+#include <linux/cpu.h>
+#include <linux/device.h>
+#include <linux/init.h>
+#include <linux/of.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/smp.h>
+#include <linux/sysfs.h>
+
+/* pointer to per cpu cacheinfo */
+static DEFINE_PER_CPU(struct cpu_cacheinfo, ci_cpu_cacheinfo);
+#define ci_cacheinfo(cpu)	(&per_cpu(ci_cpu_cacheinfo, cpu))
+#define cache_leaves(cpu)	(ci_cacheinfo(cpu)->num_leaves)
+#define per_cpu_cacheinfo(cpu)	(ci_cacheinfo(cpu)->info_list)
+
+struct cpu_cacheinfo *get_cpu_cacheinfo(unsigned int cpu)
+{
+	return ci_cacheinfo(cpu);
+}
+
+#ifdef CONFIG_OF
+static int cache_setup_of_node(unsigned int cpu)
+{
+	struct device_node *np;
+	struct cacheinfo *this_leaf;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int index = 0;
+
+	/* skip if of_node is already populated */
+	if (this_cpu_ci->info_list->of_node)
+		return 0;
+
+	if (!cpu_dev) {
+		pr_err("No cpu device for CPU %d\n", cpu);
+		return -ENODEV;
+	}
+	np = cpu_dev->of_node;
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
+	}
+
+	while (np && index < cache_leaves(cpu)) {
+		this_leaf = this_cpu_ci->info_list + index;
+		if (this_leaf->level != 1)
+			np = of_find_next_cache_node(np);
+		else
+			np = of_node_get(np);/* cpu node itself */
+		this_leaf->of_node = np;
+		index++;
+	}
+	return 0;
+}
+
+static inline bool cache_leaves_are_shared(struct cacheinfo *this_leaf,
+					   struct cacheinfo *sib_leaf)
+{
+	return sib_leaf->of_node == this_leaf->of_node;
+}
+
+static int of_cache_shared_cpu_map_setup(unsigned int cpu)
+{
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sib_leaf;
+	unsigned int index;
+	int ret;
+
+	ret = cache_setup_of_node(cpu);
+	if (ret)
+		return ret;
+
+	for (index = 0; index < cache_leaves(cpu); index++) {
+		unsigned int i;
+
+		this_leaf = this_cpu_ci->info_list + index;
+		cpumask_set_cpu(cpu, &this_leaf->shared_cpu_map);
+
+		for_each_online_cpu(i) {
+			struct cpu_cacheinfo *sib_cpu_ci = get_cpu_cacheinfo(i);
+
+			if (i == cpu || !sib_cpu_ci->info_list)
+				continue;/* skip if itself or no cacheinfo */
+			sib_leaf = sib_cpu_ci->info_list + index;
+			if (cache_leaves_are_shared(this_leaf, sib_leaf)) {
+				cpumask_set_cpu(cpu, &sib_leaf->shared_cpu_map);
+				cpumask_set_cpu(i, &this_leaf->shared_cpu_map);
+			}
+		}
+	}
+
+	return 0;
+}
+#else
+static inline int of_cache_shared_cpu_map_setup(unsigned int cpu)
+{
+	return 0;
+}
+#endif
+
+static void cache_shared_cpu_map_remove(unsigned int cpu)
+{
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sib_leaf;
+	unsigned int sibling, index;
+
+	for (index = 0; index < cache_leaves(cpu); index++) {
+		this_leaf = this_cpu_ci->info_list + index;
+		for_each_cpu(sibling, &this_leaf->shared_cpu_map) {
+			struct cpu_cacheinfo *sib_cpu_ci;
+
+			if (sibling == cpu) /* skip itself */
+				continue;
+			sib_cpu_ci = get_cpu_cacheinfo(sibling);
+			sib_leaf = sib_cpu_ci->info_list + index;
+			cpumask_clear_cpu(cpu, &sib_leaf->shared_cpu_map);
+			cpumask_clear_cpu(sibling, &this_leaf->shared_cpu_map);
+		}
+		of_node_put(this_leaf->of_node);
+	}
+}
+
+int __weak init_cache_level(unsigned int cpu)
+{
+	return -ENOENT;
+}
+
+int __weak populate_cache_leaves(unsigned int cpu)
+{
+	return -ENOENT;
+}
+
+static void free_cache_attributes(unsigned int cpu)
+{
+	cache_shared_cpu_map_remove(cpu);
+
+	kfree(per_cpu_cacheinfo(cpu));
+	per_cpu_cacheinfo(cpu) = NULL;
+}
+
+/*
+ * Helpers to make sure "func" is executed on the cpu whose cache
+ * attributes are being detected
+ */
+#define DEFINE_SMP_CALL_FUNCTION(func)		\
+static void _##func(void *ret)			\
+{						\
+	int cpu = smp_processor_id();		\
+	*(int *)ret = func(cpu);		\
+}						\
+static int __##func(unsigned int cpu)		\
+{						\
+	int ret;				\
+	smp_call_function_single(cpu, _##func, &ret, true);	\
+	return ret;				\
+}
+DEFINE_SMP_CALL_FUNCTION(init_cache_level)
+DEFINE_SMP_CALL_FUNCTION(populate_cache_leaves)
+
+static int detect_cache_attributes(unsigned int cpu)
+{
+	int ret;
+
+	if (__init_cache_level(cpu))
+		return -ENOENT;
+
+	per_cpu_cacheinfo(cpu) = kzalloc(sizeof(struct cacheinfo) *
+					 cache_leaves(cpu), GFP_KERNEL);
+	if (per_cpu_cacheinfo(cpu) == NULL)
+		return -ENOMEM;
+
+	ret = __populate_cache_leaves(cpu);
+	if (ret)
+		goto free_ci;
+	/*
+	 * For systems using DT for cache hierarcy, of_node and shared_cpu_map
+	 * will be set up here. Otherwise populate_cache_leaves needs to set
+	 * shared_cpu_map and next-level-cache should not be specified in DT
+	 */
+	ret = of_cache_shared_cpu_map_setup(cpu);
+	if (ret)
+		goto free_ci;
+	return 0;
+
+free_ci:
+	free_cache_attributes(cpu);
+	return ret;
+}
+
+#ifdef CONFIG_SYSFS
+
+/* pointer to cpuX/cache device */
+static DEFINE_PER_CPU(struct device *, ci_cache_dev);
+#define per_cpu_cache_dev(cpu)	(per_cpu(ci_cache_dev, cpu))
+
+static cpumask_t cache_dev_map;
+
+/* pointer to array of devices for cpuX/cache/indexY */
+static DEFINE_PER_CPU(struct device **, ci_index_dev);
+#define per_cpu_index_dev(cpu)	(per_cpu(ci_index_dev, cpu))
+#define per_cache_index_dev(cpu, idx)	((per_cpu_index_dev(cpu))[idx])
+
+#define show_one(file_name, object)				\
+static ssize_t file_name##_show(struct device *dev,		\
+		struct device_attribute *attr, char *buf)	\
+{								\
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);	\
+	if (!this_leaf->object)					\
+		return -EINVAL;					\
+	return sprintf(buf, "%u\n", this_leaf->object);		\
+}
+
+show_one(level, level);
+show_one(coherency_line_size, coherency_line_size);
+show_one(number_of_sets, number_of_sets);
+show_one(physical_line_partition, physical_line_partition);
+
+static ssize_t ways_of_associativity_show(struct device *dev,
+					  struct device_attribute *attr,
+					  char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	/* will be zero for fully associative cache, but check for size */
+	if (!this_leaf->size)
+		return -EINVAL;
+	return sprintf(buf, "%u\n", this_leaf->ways_of_associativity);
+}
+
+static ssize_t size_show(struct device *dev,
+			 struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	if (!this_leaf->size)
+		return -EINVAL;
+	return sprintf(buf, "%uK\n", this_leaf->size >> 10);
+}
+
+static ssize_t shared_cpumap_show_func(struct device *dev, int type, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf;
+	int n = 0;
+
+	if (len > 1) {
+		const struct cpumask *mask = &this_leaf->shared_cpu_map;
+
+		n = type ? cpulist_scnprintf(buf, len - 2, mask) :
+			   cpumask_scnprintf(buf, len - 2, mask);
+		buf[n++] = '\n';
+		buf[n] = '\0';
+	}
+	return n;
+}
+
+static ssize_t shared_cpu_map_show(struct device *dev,
+				   struct device_attribute *attr, char *buf)
+{
+	return shared_cpumap_show_func(dev, 0, buf);
+}
+
+static ssize_t shared_cpu_list_show(struct device *dev,
+				    struct device_attribute *attr, char *buf)
+{
+	return shared_cpumap_show_func(dev, 1, buf);
+}
+
+static ssize_t type_show(struct device *dev,
+			 struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	switch (this_leaf->type) {
+	case CACHE_TYPE_DATA:
+		return sprintf(buf, "Data\n");
+	case CACHE_TYPE_INST:
+		return sprintf(buf, "Instruction\n");
+	case CACHE_TYPE_UNIFIED:
+		return sprintf(buf, "Unified\n");
+	default:
+		return -EINVAL;
+	}
+}
+
+static ssize_t attributes_show(struct device *dev,
+			       struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	unsigned int ci_attr = this_leaf->attributes;
+	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf - 2;
+	int n = 0;
+
+	if (!ci_attr)
+		return -EINVAL;
+
+	if (ci_attr & CACHE_WRITE_THROUGH)
+		n += snprintf(buf + n, len - n, "WriteThrough\n");
+	if (ci_attr & CACHE_WRITE_BACK)
+		n += snprintf(buf + n, len - n, "WriteBack\n");
+	if (ci_attr & CACHE_READ_ALLOCATE)
+		n += snprintf(buf + n, len - n, "ReadAllocate\n");
+	if (ci_attr & CACHE_WRITE_ALLOCATE)
+		n += snprintf(buf + n, len - n, "WriteAllocate\n");
+	buf[n] = '\0';
+	return n;
+}
+
+static DEVICE_ATTR_RO(level);
+static DEVICE_ATTR_RO(type);
+static DEVICE_ATTR_RO(coherency_line_size);
+static DEVICE_ATTR_RO(ways_of_associativity);
+static DEVICE_ATTR_RO(number_of_sets);
+static DEVICE_ATTR_RO(size);
+static DEVICE_ATTR_RO(attributes);
+static DEVICE_ATTR_RO(shared_cpu_map);
+static DEVICE_ATTR_RO(shared_cpu_list);
+static DEVICE_ATTR_RO(physical_line_partition);
+
+static struct attribute *cache_default_attrs[] = {
+	&dev_attr_type.attr,
+	&dev_attr_level.attr,
+	&dev_attr_shared_cpu_map.attr,
+	&dev_attr_shared_cpu_list.attr,
+	NULL
+};
+
+ATTRIBUTE_GROUPS(cache_default);
+
+static const struct device_attribute *cache_optional_attrs[] = {
+	&dev_attr_coherency_line_size,
+	&dev_attr_ways_of_associativity,
+	&dev_attr_number_of_sets,
+	&dev_attr_size,
+	&dev_attr_attributes,
+	&dev_attr_physical_line_partition,
+	NULL
+};
+
+static int device_add_attrs(struct device *dev,
+			    const struct device_attribute **dev_attrs)
+{
+	int i, error = 0;
+	struct device_attribute *dev_attr;
+	char *buf;
+
+	if (!dev_attrs)
+		return 0;
+
+	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+
+	for (i = 0; dev_attrs[i]; i++) {
+		dev_attr = (struct device_attribute *)dev_attrs[i];
+
+		/* create attributes that provides meaningful value */
+		if (dev_attr->show(dev, dev_attr, buf) < 0)
+			continue;
+
+		error = device_create_file(dev, dev_attrs[i]);
+		if (error) {
+			while (--i >= 0)
+				device_remove_file(dev, dev_attrs[i]);
+			break;
+		}
+	}
+
+	kfree(buf);
+	return error;
+}
+
+static void device_remove_attrs(struct device *dev,
+				const struct device_attribute **dev_attrs)
+{
+	int i;
+
+	if (!dev_attrs)
+		return;
+
+	for (i = 0; dev_attrs[i]; dev_attrs++, i++)
+		device_remove_file(dev, dev_attrs[i]);
+}
+
+const struct device_attribute **
+__weak cache_get_priv_attr(struct device *cache_idx_dev)
+{
+	return NULL;
+}
+
+/* Add/Remove cache interface for CPU device */
+static void cpu_cache_sysfs_exit(unsigned int cpu)
+{
+	int i;
+	struct device *tmp_dev;
+	const struct device_attribute **ci_priv_attr;
+
+	if (per_cpu_index_dev(cpu)) {
+		for (i = 0; i < cache_leaves(cpu); i++) {
+			tmp_dev = per_cache_index_dev(cpu, i);
+			if (!tmp_dev)
+				continue;
+			ci_priv_attr = cache_get_priv_attr(tmp_dev);
+			device_remove_attrs(tmp_dev, ci_priv_attr);
+			device_remove_attrs(tmp_dev, cache_optional_attrs);
+			device_unregister(tmp_dev);
+		}
+		kfree(per_cpu_index_dev(cpu));
+		per_cpu_index_dev(cpu) = NULL;
+	}
+	device_unregister(per_cpu_cache_dev(cpu));
+	per_cpu_cache_dev(cpu) = NULL;
+}
+
+static int cpu_cache_sysfs_init(unsigned int cpu)
+{
+	struct device *dev = get_cpu_device(cpu);
+
+	if (per_cpu_cacheinfo(cpu) == NULL)
+		return -ENOENT;
+
+	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
+					       NULL, "cache");
+	if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
+		return PTR_ERR(per_cpu_cache_dev(cpu));
+
+	/* Allocate all required memory */
+	per_cpu_index_dev(cpu) = kzalloc(sizeof(struct device *) *
+					 cache_leaves(cpu), GFP_KERNEL);
+	if (unlikely(per_cpu_index_dev(cpu) == NULL))
+		goto err_out;
+
+	return 0;
+
+err_out:
+	cpu_cache_sysfs_exit(cpu);
+	return -ENOMEM;
+}
+
+static int cache_add_dev(unsigned int cpu)
+{
+	unsigned short i;
+	int rc;
+	struct device *tmp_dev, *parent;
+	struct cacheinfo *this_leaf;
+	const struct device_attribute **ci_priv_attr;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	rc = cpu_cache_sysfs_init(cpu);
+	if (unlikely(rc < 0))
+		return rc;
+
+	parent = per_cpu_cache_dev(cpu);
+	for (i = 0; i < cache_leaves(cpu); i++) {
+		this_leaf = this_cpu_ci->info_list + i;
+		if (this_leaf->disable_sysfs)
+			continue;
+		tmp_dev = device_create_with_groups(parent->class, parent, i,
+						    this_leaf,
+						    cache_default_groups,
+						    "index%1u", i);
+		if (IS_ERR_OR_NULL(tmp_dev)) {
+			rc = PTR_ERR(tmp_dev);
+			goto err;
+		}
+
+		rc = device_add_attrs(tmp_dev, cache_optional_attrs);
+		if (unlikely(rc))
+			goto err;
+
+		ci_priv_attr = cache_get_priv_attr(tmp_dev);
+		rc = device_add_attrs(tmp_dev, ci_priv_attr);
+		if (unlikely(rc))
+			goto err;
+
+		per_cache_index_dev(cpu, i) = tmp_dev;
+	}
+	cpumask_set_cpu(cpu, &cache_dev_map);
+
+	return 0;
+err:
+	cpu_cache_sysfs_exit(cpu);
+	return rc;
+}
+
+static void cache_remove_dev(unsigned int cpu)
+{
+	if (!cpumask_test_cpu(cpu, &cache_dev_map))
+		return;
+	cpumask_clear_cpu(cpu, &cache_dev_map);
+
+	cpu_cache_sysfs_exit(cpu);
+}
+
+static int cacheinfo_cpu_callback(struct notifier_block *nfb,
+				  unsigned long action, void *hcpu)
+{
+	unsigned int cpu = (unsigned long)hcpu;
+	int rc = 0;
+
+	switch (action) {
+	case CPU_ONLINE:
+	case CPU_ONLINE_FROZEN:
+		rc = detect_cache_attributes(cpu);
+		if (!rc)
+			rc = cache_add_dev(cpu);
+		break;
+	case CPU_DEAD:
+	case CPU_DEAD_FROZEN:
+		cache_remove_dev(cpu);
+		if (per_cpu_cacheinfo(cpu))
+			free_cache_attributes(cpu);
+		break;
+	}
+	return notifier_from_errno(rc);
+}
+
+static int __init cacheinfo_sysfs_init(void)
+{
+	int cpu, rc = 0;
+
+	cpu_notifier_register_begin();
+
+	for_each_online_cpu(cpu) {
+		rc = detect_cache_attributes(cpu);
+		if (rc) {
+			pr_err("error detecting cacheinfo..cpu%d\n", cpu);
+			goto out;
+		}
+		rc = cache_add_dev(cpu);
+		if (rc) {
+			free_cache_attributes(cpu);
+			pr_err("error populating cacheinfo..cpu%d\n", cpu);
+			goto out;
+		}
+	}
+	__hotcpu_notifier(cacheinfo_cpu_callback, 0);
+
+out:
+	cpu_notifier_register_done();
+	return rc;
+}
+
+device_initcall(cacheinfo_sysfs_init);
+
+#endif	/* CONFIG_SYSFS */
diff --git a/include/linux/cacheinfo.h b/include/linux/cacheinfo.h
new file mode 100644
index 0000000..a9bd9f9
--- /dev/null
+++ b/include/linux/cacheinfo.h
@@ -0,0 +1,56 @@
+#ifndef _LINUX_CACHEINFO_H
+#define _LINUX_CACHEINFO_H
+
+#include <linux/bitops.h>
+#include <linux/compiler.h>
+#include <linux/cpumask.h>
+#include <linux/device.h>
+#include <linux/of.h>
+#include <linux/sysfs.h>
+
+enum cache_type {
+	CACHE_TYPE_NOCACHE = 0,
+	CACHE_TYPE_INST = BIT(0),
+	CACHE_TYPE_DATA = BIT(1),
+	CACHE_TYPE_SEPARATE = CACHE_TYPE_INST | CACHE_TYPE_DATA,
+	CACHE_TYPE_UNIFIED = BIT(2),
+};
+
+struct cacheinfo {
+	/* core properties */
+	enum cache_type type; /* data, inst or unified */
+	unsigned int level;
+	unsigned int coherency_line_size; /* cache line size  */
+	unsigned int number_of_sets; /* no. of sets per way */
+	unsigned int ways_of_associativity; /* no. of ways */
+	unsigned int physical_line_partition; /* no. of lines per tag */
+	unsigned int size; /* total cache size */
+	cpumask_t shared_cpu_map;
+	unsigned int attributes;
+#define CACHE_WRITE_THROUGH	BIT(0)
+#define CACHE_WRITE_BACK	BIT(1)
+#define CACHE_READ_ALLOCATE	BIT(2)
+#define CACHE_WRITE_ALLOCATE	BIT(3)
+
+	/* book keeping */
+	struct device_node *of_node;	/* cpu if no explicit cache node */
+	bool disable_sysfs; /* don't expose this leaf through sysfs */
+	void *priv;
+};
+
+struct cpu_cacheinfo {
+	struct cacheinfo *info_list;
+	unsigned int num_levels;
+	unsigned int num_leaves;
+};
+
+struct cpu_cacheinfo *get_cpu_cacheinfo(unsigned int cpu);
+int init_cache_level(unsigned int cpu);
+int populate_cache_leaves(unsigned int cpu);
+
+#ifdef CONFIG_SYSFS
+const struct device_attribute **
+cache_get_priv_attr(struct device *cache_idx_dev);
+#endif
+
+#endif /* _LINUX_CACHEINFO_H */
-- 
1.8.3.2

[PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds initial support for providing processor cache information
to userspace through sysfs interface. This is based on already existing
implementations(x86, ia64, s390 and powerpc) and hence the interface is
intended to be fully compatible.

The main purpose of this generic support is to avoid further code
duplication to support new architectures and also to unify all the existing
different implementations.

This implementation maintains the hierarchy of cache objects which reflects
the system's cache topology. Cache devices are instantiated as needed as
CPUs come online. The cache information is replicated per-cpu even if they are
shared. A per-cpu array of cache information maintained is used mainly for
sysfs-related book keeping.

It also implements the shared_cpu_map attribute, which is essential for
enabling both kernel and user-space to discover the system's overall cache
topology.

This patch also add the missing ABI documentation for the cacheinfo sysfs
interface already, which is well defined and widely used.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Rob Herring <robh@kernel.org>
Cc: linux-doc@vger.kernel.org
Cc: linux-ia64@vger.kernel.org
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-arm-kernel@lists.infradead.org
---
 Documentation/ABI/testing/sysfs-devices-system-cpu |  41 ++
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 564 +++++++++++++++++++++
 include/linux/cacheinfo.h                          |  56 ++
 4 files changed, 662 insertions(+), 1 deletion(-)
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

diff --git a/Documentation/ABI/testing/sysfs-devices-system-cpu b/Documentation/ABI/testing/sysfs-devices-system-cpu
index acb9bfc..5827f4e 100644
--- a/Documentation/ABI/testing/sysfs-devices-system-cpu
+++ b/Documentation/ABI/testing/sysfs-devices-system-cpu
@@ -224,3 +224,44 @@ Description:	Parameters for the Intel P-state driver
 		frequency range.
 
 		More details can be found in Documentation/cpu-freq/intel-pstate.txt
+
+What:		/sys/devices/system/cpu/cpu*/cache/index*/<set_of_attributes_mentioned_below>
+Date:		June 2014(documented, existed before August 2008)
+Contact:	Sudeep Holla <sudeep.holla@arm.com>
+		Linux kernel mailing list <linux-kernel@vger.kernel.org>
+Description:	Parameters for the CPU cache attributes
+
+		attributes:
+			- writethrough: data is written to both the cache line
+					and to the block in the lower-level memory
+			- writeback: data is written only to the cache line and
+				     the modified cache line is written to main
+				     memory only when it is replaced
+			- writeallocate: allocate a memory location to a cache line
+					 on a cache miss because of a write
+			- readallocate: allocate a memory location to a cache line
+					on a cache miss because of a read
+
+		coherency_line_size: the minimum amount of data that gets transferred
+
+		level: the cache hierarcy in the multi-level cache configuration
+
+		number_of_sets: total number of sets in the cache, a set is a
+				collection of cache lines with the same cache index
+
+		physical_line_partition: number of physical cache line per cache tag
+
+		shared_cpu_list: the list of cpus sharing the cache
+
+		shared_cpu_map: logical cpu mask containing the list of cpus sharing
+				the cache
+
+		size: the total cache size in kB
+
+		type:
+			- instruction: cache that only holds instructions
+			- data: cache that only caches data
+			- unified: cache that holds both data and instructions
+
+		ways_of_associativity: degree of freedom in placing a particular block
+					of memory in the cache
diff --git a/drivers/base/Makefile b/drivers/base/Makefile
index 04b314e..bad2ff8 100644
--- a/drivers/base/Makefile
+++ b/drivers/base/Makefile
@@ -4,7 +4,7 @@ obj-y			:= component.o core.o bus.o dd.o syscore.o \
 			   driver.o class.o platform.o \
 			   cpu.o firmware.o init.o map.o devres.o \
 			   attribute_container.o transport_class.o \
-			   topology.o container.o
+			   topology.o container.o cacheinfo.o
 obj-$(CONFIG_DEVTMPFS)	+= devtmpfs.o
 obj-$(CONFIG_DMA_CMA) += dma-contiguous.o
 obj-y			+= power/
diff --git a/drivers/base/cacheinfo.c b/drivers/base/cacheinfo.c
new file mode 100644
index 0000000..c12e03c
--- /dev/null
+++ b/drivers/base/cacheinfo.c
@@ -0,0 +1,564 @@
+/*
+ * cacheinfo support - processor cache information via sysfs
+ *
+ * Based on arch/x86/kernel/cpu/intel_cacheinfo.c
+ * Author: Sudeep Holla <sudeep.holla@arm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/compiler.h>
+#include <linux/cpu.h>
+#include <linux/device.h>
+#include <linux/init.h>
+#include <linux/of.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/smp.h>
+#include <linux/sysfs.h>
+
+/* pointer to per cpu cacheinfo */
+static DEFINE_PER_CPU(struct cpu_cacheinfo, ci_cpu_cacheinfo);
+#define ci_cacheinfo(cpu)	(&per_cpu(ci_cpu_cacheinfo, cpu))
+#define cache_leaves(cpu)	(ci_cacheinfo(cpu)->num_leaves)
+#define per_cpu_cacheinfo(cpu)	(ci_cacheinfo(cpu)->info_list)
+
+struct cpu_cacheinfo *get_cpu_cacheinfo(unsigned int cpu)
+{
+	return ci_cacheinfo(cpu);
+}
+
+#ifdef CONFIG_OF
+static int cache_setup_of_node(unsigned int cpu)
+{
+	struct device_node *np;
+	struct cacheinfo *this_leaf;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int index = 0;
+
+	/* skip if of_node is already populated */
+	if (this_cpu_ci->info_list->of_node)
+		return 0;
+
+	if (!cpu_dev) {
+		pr_err("No cpu device for CPU %d\n", cpu);
+		return -ENODEV;
+	}
+	np = cpu_dev->of_node;
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
+	}
+
+	while (np && index < cache_leaves(cpu)) {
+		this_leaf = this_cpu_ci->info_list + index;
+		if (this_leaf->level != 1)
+			np = of_find_next_cache_node(np);
+		else
+			np = of_node_get(np);/* cpu node itself */
+		this_leaf->of_node = np;
+		index++;
+	}
+	return 0;
+}
+
+static inline bool cache_leaves_are_shared(struct cacheinfo *this_leaf,
+					   struct cacheinfo *sib_leaf)
+{
+	return sib_leaf->of_node == this_leaf->of_node;
+}
+
+static int of_cache_shared_cpu_map_setup(unsigned int cpu)
+{
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sib_leaf;
+	unsigned int index;
+	int ret;
+
+	ret = cache_setup_of_node(cpu);
+	if (ret)
+		return ret;
+
+	for (index = 0; index < cache_leaves(cpu); index++) {
+		unsigned int i;
+
+		this_leaf = this_cpu_ci->info_list + index;
+		cpumask_set_cpu(cpu, &this_leaf->shared_cpu_map);
+
+		for_each_online_cpu(i) {
+			struct cpu_cacheinfo *sib_cpu_ci = get_cpu_cacheinfo(i);
+
+			if (i == cpu || !sib_cpu_ci->info_list)
+				continue;/* skip if itself or no cacheinfo */
+			sib_leaf = sib_cpu_ci->info_list + index;
+			if (cache_leaves_are_shared(this_leaf, sib_leaf)) {
+				cpumask_set_cpu(cpu, &sib_leaf->shared_cpu_map);
+				cpumask_set_cpu(i, &this_leaf->shared_cpu_map);
+			}
+		}
+	}
+
+	return 0;
+}
+#else
+static inline int of_cache_shared_cpu_map_setup(unsigned int cpu)
+{
+	return 0;
+}
+#endif
+
+static void cache_shared_cpu_map_remove(unsigned int cpu)
+{
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sib_leaf;
+	unsigned int sibling, index;
+
+	for (index = 0; index < cache_leaves(cpu); index++) {
+		this_leaf = this_cpu_ci->info_list + index;
+		for_each_cpu(sibling, &this_leaf->shared_cpu_map) {
+			struct cpu_cacheinfo *sib_cpu_ci;
+
+			if (sibling == cpu) /* skip itself */
+				continue;
+			sib_cpu_ci = get_cpu_cacheinfo(sibling);
+			sib_leaf = sib_cpu_ci->info_list + index;
+			cpumask_clear_cpu(cpu, &sib_leaf->shared_cpu_map);
+			cpumask_clear_cpu(sibling, &this_leaf->shared_cpu_map);
+		}
+		of_node_put(this_leaf->of_node);
+	}
+}
+
+int __weak init_cache_level(unsigned int cpu)
+{
+	return -ENOENT;
+}
+
+int __weak populate_cache_leaves(unsigned int cpu)
+{
+	return -ENOENT;
+}
+
+static void free_cache_attributes(unsigned int cpu)
+{
+	cache_shared_cpu_map_remove(cpu);
+
+	kfree(per_cpu_cacheinfo(cpu));
+	per_cpu_cacheinfo(cpu) = NULL;
+}
+
+/*
+ * Helpers to make sure "func" is executed on the cpu whose cache
+ * attributes are being detected
+ */
+#define DEFINE_SMP_CALL_FUNCTION(func)		\
+static void _##func(void *ret)			\
+{						\
+	int cpu = smp_processor_id();		\
+	*(int *)ret = func(cpu);		\
+}						\
+static int __##func(unsigned int cpu)		\
+{						\
+	int ret;				\
+	smp_call_function_single(cpu, _##func, &ret, true);	\
+	return ret;				\
+}
+DEFINE_SMP_CALL_FUNCTION(init_cache_level)
+DEFINE_SMP_CALL_FUNCTION(populate_cache_leaves)
+
+static int detect_cache_attributes(unsigned int cpu)
+{
+	int ret;
+
+	if (__init_cache_level(cpu))
+		return -ENOENT;
+
+	per_cpu_cacheinfo(cpu) = kzalloc(sizeof(struct cacheinfo) *
+					 cache_leaves(cpu), GFP_KERNEL);
+	if (per_cpu_cacheinfo(cpu) == NULL)
+		return -ENOMEM;
+
+	ret = __populate_cache_leaves(cpu);
+	if (ret)
+		goto free_ci;
+	/*
+	 * For systems using DT for cache hierarcy, of_node and shared_cpu_map
+	 * will be set up here. Otherwise populate_cache_leaves needs to set
+	 * shared_cpu_map and next-level-cache should not be specified in DT
+	 */
+	ret = of_cache_shared_cpu_map_setup(cpu);
+	if (ret)
+		goto free_ci;
+	return 0;
+
+free_ci:
+	free_cache_attributes(cpu);
+	return ret;
+}
+
+#ifdef CONFIG_SYSFS
+
+/* pointer to cpuX/cache device */
+static DEFINE_PER_CPU(struct device *, ci_cache_dev);
+#define per_cpu_cache_dev(cpu)	(per_cpu(ci_cache_dev, cpu))
+
+static cpumask_t cache_dev_map;
+
+/* pointer to array of devices for cpuX/cache/indexY */
+static DEFINE_PER_CPU(struct device **, ci_index_dev);
+#define per_cpu_index_dev(cpu)	(per_cpu(ci_index_dev, cpu))
+#define per_cache_index_dev(cpu, idx)	((per_cpu_index_dev(cpu))[idx])
+
+#define show_one(file_name, object)				\
+static ssize_t file_name##_show(struct device *dev,		\
+		struct device_attribute *attr, char *buf)	\
+{								\
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);	\
+	if (!this_leaf->object)					\
+		return -EINVAL;					\
+	return sprintf(buf, "%u\n", this_leaf->object);		\
+}
+
+show_one(level, level);
+show_one(coherency_line_size, coherency_line_size);
+show_one(number_of_sets, number_of_sets);
+show_one(physical_line_partition, physical_line_partition);
+
+static ssize_t ways_of_associativity_show(struct device *dev,
+					  struct device_attribute *attr,
+					  char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	/* will be zero for fully associative cache, but check for size */
+	if (!this_leaf->size)
+		return -EINVAL;
+	return sprintf(buf, "%u\n", this_leaf->ways_of_associativity);
+}
+
+static ssize_t size_show(struct device *dev,
+			 struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	if (!this_leaf->size)
+		return -EINVAL;
+	return sprintf(buf, "%uK\n", this_leaf->size >> 10);
+}
+
+static ssize_t shared_cpumap_show_func(struct device *dev, int type, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf;
+	int n = 0;
+
+	if (len > 1) {
+		const struct cpumask *mask = &this_leaf->shared_cpu_map;
+
+		n = type ? cpulist_scnprintf(buf, len - 2, mask) :
+			   cpumask_scnprintf(buf, len - 2, mask);
+		buf[n++] = '\n';
+		buf[n] = '\0';
+	}
+	return n;
+}
+
+static ssize_t shared_cpu_map_show(struct device *dev,
+				   struct device_attribute *attr, char *buf)
+{
+	return shared_cpumap_show_func(dev, 0, buf);
+}
+
+static ssize_t shared_cpu_list_show(struct device *dev,
+				    struct device_attribute *attr, char *buf)
+{
+	return shared_cpumap_show_func(dev, 1, buf);
+}
+
+static ssize_t type_show(struct device *dev,
+			 struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	switch (this_leaf->type) {
+	case CACHE_TYPE_DATA:
+		return sprintf(buf, "Data\n");
+	case CACHE_TYPE_INST:
+		return sprintf(buf, "Instruction\n");
+	case CACHE_TYPE_UNIFIED:
+		return sprintf(buf, "Unified\n");
+	default:
+		return -EINVAL;
+	}
+}
+
+static ssize_t attributes_show(struct device *dev,
+			       struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	unsigned int ci_attr = this_leaf->attributes;
+	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf - 2;
+	int n = 0;
+
+	if (!ci_attr)
+		return -EINVAL;
+
+	if (ci_attr & CACHE_WRITE_THROUGH)
+		n += snprintf(buf + n, len - n, "WriteThrough\n");
+	if (ci_attr & CACHE_WRITE_BACK)
+		n += snprintf(buf + n, len - n, "WriteBack\n");
+	if (ci_attr & CACHE_READ_ALLOCATE)
+		n += snprintf(buf + n, len - n, "ReadAllocate\n");
+	if (ci_attr & CACHE_WRITE_ALLOCATE)
+		n += snprintf(buf + n, len - n, "WriteAllocate\n");
+	buf[n] = '\0';
+	return n;
+}
+
+static DEVICE_ATTR_RO(level);
+static DEVICE_ATTR_RO(type);
+static DEVICE_ATTR_RO(coherency_line_size);
+static DEVICE_ATTR_RO(ways_of_associativity);
+static DEVICE_ATTR_RO(number_of_sets);
+static DEVICE_ATTR_RO(size);
+static DEVICE_ATTR_RO(attributes);
+static DEVICE_ATTR_RO(shared_cpu_map);
+static DEVICE_ATTR_RO(shared_cpu_list);
+static DEVICE_ATTR_RO(physical_line_partition);
+
+static struct attribute *cache_default_attrs[] = {
+	&dev_attr_type.attr,
+	&dev_attr_level.attr,
+	&dev_attr_shared_cpu_map.attr,
+	&dev_attr_shared_cpu_list.attr,
+	NULL
+};
+
+ATTRIBUTE_GROUPS(cache_default);
+
+static const struct device_attribute *cache_optional_attrs[] = {
+	&dev_attr_coherency_line_size,
+	&dev_attr_ways_of_associativity,
+	&dev_attr_number_of_sets,
+	&dev_attr_size,
+	&dev_attr_attributes,
+	&dev_attr_physical_line_partition,
+	NULL
+};
+
+static int device_add_attrs(struct device *dev,
+			    const struct device_attribute **dev_attrs)
+{
+	int i, error = 0;
+	struct device_attribute *dev_attr;
+	char *buf;
+
+	if (!dev_attrs)
+		return 0;
+
+	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+
+	for (i = 0; dev_attrs[i]; i++) {
+		dev_attr = (struct device_attribute *)dev_attrs[i];
+
+		/* create attributes that provides meaningful value */
+		if (dev_attr->show(dev, dev_attr, buf) < 0)
+			continue;
+
+		error = device_create_file(dev, dev_attrs[i]);
+		if (error) {
+			while (--i >= 0)
+				device_remove_file(dev, dev_attrs[i]);
+			break;
+		}
+	}
+
+	kfree(buf);
+	return error;
+}
+
+static void device_remove_attrs(struct device *dev,
+				const struct device_attribute **dev_attrs)
+{
+	int i;
+
+	if (!dev_attrs)
+		return;
+
+	for (i = 0; dev_attrs[i]; dev_attrs++, i++)
+		device_remove_file(dev, dev_attrs[i]);
+}
+
+const struct device_attribute **
+__weak cache_get_priv_attr(struct device *cache_idx_dev)
+{
+	return NULL;
+}
+
+/* Add/Remove cache interface for CPU device */
+static void cpu_cache_sysfs_exit(unsigned int cpu)
+{
+	int i;
+	struct device *tmp_dev;
+	const struct device_attribute **ci_priv_attr;
+
+	if (per_cpu_index_dev(cpu)) {
+		for (i = 0; i < cache_leaves(cpu); i++) {
+			tmp_dev = per_cache_index_dev(cpu, i);
+			if (!tmp_dev)
+				continue;
+			ci_priv_attr = cache_get_priv_attr(tmp_dev);
+			device_remove_attrs(tmp_dev, ci_priv_attr);
+			device_remove_attrs(tmp_dev, cache_optional_attrs);
+			device_unregister(tmp_dev);
+		}
+		kfree(per_cpu_index_dev(cpu));
+		per_cpu_index_dev(cpu) = NULL;
+	}
+	device_unregister(per_cpu_cache_dev(cpu));
+	per_cpu_cache_dev(cpu) = NULL;
+}
+
+static int cpu_cache_sysfs_init(unsigned int cpu)
+{
+	struct device *dev = get_cpu_device(cpu);
+
+	if (per_cpu_cacheinfo(cpu) == NULL)
+		return -ENOENT;
+
+	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
+					       NULL, "cache");
+	if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
+		return PTR_ERR(per_cpu_cache_dev(cpu));
+
+	/* Allocate all required memory */
+	per_cpu_index_dev(cpu) = kzalloc(sizeof(struct device *) *
+					 cache_leaves(cpu), GFP_KERNEL);
+	if (unlikely(per_cpu_index_dev(cpu) == NULL))
+		goto err_out;
+
+	return 0;
+
+err_out:
+	cpu_cache_sysfs_exit(cpu);
+	return -ENOMEM;
+}
+
+static int cache_add_dev(unsigned int cpu)
+{
+	unsigned short i;
+	int rc;
+	struct device *tmp_dev, *parent;
+	struct cacheinfo *this_leaf;
+	const struct device_attribute **ci_priv_attr;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	rc = cpu_cache_sysfs_init(cpu);
+	if (unlikely(rc < 0))
+		return rc;
+
+	parent = per_cpu_cache_dev(cpu);
+	for (i = 0; i < cache_leaves(cpu); i++) {
+		this_leaf = this_cpu_ci->info_list + i;
+		if (this_leaf->disable_sysfs)
+			continue;
+		tmp_dev = device_create_with_groups(parent->class, parent, i,
+						    this_leaf,
+						    cache_default_groups,
+						    "index%1u", i);
+		if (IS_ERR_OR_NULL(tmp_dev)) {
+			rc = PTR_ERR(tmp_dev);
+			goto err;
+		}
+
+		rc = device_add_attrs(tmp_dev, cache_optional_attrs);
+		if (unlikely(rc))
+			goto err;
+
+		ci_priv_attr = cache_get_priv_attr(tmp_dev);
+		rc = device_add_attrs(tmp_dev, ci_priv_attr);
+		if (unlikely(rc))
+			goto err;
+
+		per_cache_index_dev(cpu, i) = tmp_dev;
+	}
+	cpumask_set_cpu(cpu, &cache_dev_map);
+
+	return 0;
+err:
+	cpu_cache_sysfs_exit(cpu);
+	return rc;
+}
+
+static void cache_remove_dev(unsigned int cpu)
+{
+	if (!cpumask_test_cpu(cpu, &cache_dev_map))
+		return;
+	cpumask_clear_cpu(cpu, &cache_dev_map);
+
+	cpu_cache_sysfs_exit(cpu);
+}
+
+static int cacheinfo_cpu_callback(struct notifier_block *nfb,
+				  unsigned long action, void *hcpu)
+{
+	unsigned int cpu = (unsigned long)hcpu;
+	int rc = 0;
+
+	switch (action) {
+	case CPU_ONLINE:
+	case CPU_ONLINE_FROZEN:
+		rc = detect_cache_attributes(cpu);
+		if (!rc)
+			rc = cache_add_dev(cpu);
+		break;
+	case CPU_DEAD:
+	case CPU_DEAD_FROZEN:
+		cache_remove_dev(cpu);
+		if (per_cpu_cacheinfo(cpu))
+			free_cache_attributes(cpu);
+		break;
+	}
+	return notifier_from_errno(rc);
+}
+
+static int __init cacheinfo_sysfs_init(void)
+{
+	int cpu, rc = 0;
+
+	cpu_notifier_register_begin();
+
+	for_each_online_cpu(cpu) {
+		rc = detect_cache_attributes(cpu);
+		if (rc) {
+			pr_err("error detecting cacheinfo..cpu%d\n", cpu);
+			goto out;
+		}
+		rc = cache_add_dev(cpu);
+		if (rc) {
+			free_cache_attributes(cpu);
+			pr_err("error populating cacheinfo..cpu%d\n", cpu);
+			goto out;
+		}
+	}
+	__hotcpu_notifier(cacheinfo_cpu_callback, 0);
+
+out:
+	cpu_notifier_register_done();
+	return rc;
+}
+
+device_initcall(cacheinfo_sysfs_init);
+
+#endif	/* CONFIG_SYSFS */
diff --git a/include/linux/cacheinfo.h b/include/linux/cacheinfo.h
new file mode 100644
index 0000000..a9bd9f9
--- /dev/null
+++ b/include/linux/cacheinfo.h
@@ -0,0 +1,56 @@
+#ifndef _LINUX_CACHEINFO_H
+#define _LINUX_CACHEINFO_H
+
+#include <linux/bitops.h>
+#include <linux/compiler.h>
+#include <linux/cpumask.h>
+#include <linux/device.h>
+#include <linux/of.h>
+#include <linux/sysfs.h>
+
+enum cache_type {
+	CACHE_TYPE_NOCACHE = 0,
+	CACHE_TYPE_INST = BIT(0),
+	CACHE_TYPE_DATA = BIT(1),
+	CACHE_TYPE_SEPARATE = CACHE_TYPE_INST | CACHE_TYPE_DATA,
+	CACHE_TYPE_UNIFIED = BIT(2),
+};
+
+struct cacheinfo {
+	/* core properties */
+	enum cache_type type; /* data, inst or unified */
+	unsigned int level;
+	unsigned int coherency_line_size; /* cache line size  */
+	unsigned int number_of_sets; /* no. of sets per way */
+	unsigned int ways_of_associativity; /* no. of ways */
+	unsigned int physical_line_partition; /* no. of lines per tag */
+	unsigned int size; /* total cache size */
+	cpumask_t shared_cpu_map;
+	unsigned int attributes;
+#define CACHE_WRITE_THROUGH	BIT(0)
+#define CACHE_WRITE_BACK	BIT(1)
+#define CACHE_READ_ALLOCATE	BIT(2)
+#define CACHE_WRITE_ALLOCATE	BIT(3)
+
+	/* book keeping */
+	struct device_node *of_node;	/* cpu if no explicit cache node */
+	bool disable_sysfs; /* don't expose this leaf through sysfs */
+	void *priv;
+};
+
+struct cpu_cacheinfo {
+	struct cacheinfo *info_list;
+	unsigned int num_levels;
+	unsigned int num_leaves;
+};
+
+struct cpu_cacheinfo *get_cpu_cacheinfo(unsigned int cpu);
+int init_cache_level(unsigned int cpu);
+int populate_cache_leaves(unsigned int cpu);
+
+#ifdef CONFIG_SYSFS
+const struct device_attribute **
+cache_get_priv_attr(struct device *cache_idx_dev);
+#endif
+
+#endif /* _LINUX_CACHEINFO_H */
-- 
1.8.3.2

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Russell King - ARM Linux]

On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
> +		coherency_line_size: the minimum amount of data that gets transferred

So, what value to do envision this taking for a CPU where the cache
line size is 32 bytes, but each cache line has two dirty bits which
allow it to only evict either the upper or lower 16 bytes depending
on which are dirty?

-- 
FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly
improving, and getting towards what was expected from it.

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Russell King - ARM Linux]

On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
> +		coherency_line_size: the minimum amount of data that gets transferred

So, what value to do envision this taking for a CPU where the cache
line size is 32 bytes, but each cache line has two dirty bits which
allow it to only evict either the upper or lower 16 bytes depending
on which are dirty?

-- 
FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly
improving, and getting towards what was expected from it.

[PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Russell King - ARM Linux]

On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
> +		coherency_line_size: the minimum amount of data that gets transferred

So, what value to do envision this taking for a CPU where the cache
line size is 32 bytes, but each cache line has two dirty bits which
allow it to only evict either the upper or lower 16 bytes depending
on which are dirty?

-- 
FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly
improving, and getting towards what was expected from it.

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Russell King - ARM Linux]

On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
> +		coherency_line_size: the minimum amount of data that gets transferred

So, what value to do envision this taking for a CPU where the cache
line size is 32 bytes, but each cache line has two dirty bits which
allow it to only evict either the upper or lower 16 bytes depending
on which are dirty?

-- 
FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly
improving, and getting towards what was expected from it.

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

Hi,

On 25/06/14 23:23, Russell King - ARM Linux wrote:
> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>> +		coherency_line_size: the minimum amount of data that gets transferred
>
> So, what value to do envision this taking for a CPU where the cache
> line size is 32 bytes, but each cache line has two dirty bits which
> allow it to only evict either the upper or lower 16 bytes depending
> on which are dirty?
>

IIUC most of existing implementations of cacheinfo on various architectures
are representing the cache line size as coherency_line_size, in which case I
need fix the definition in this file.

BTW will there be any architectural way of finding such configuration ?

Regards,
Sudeep

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

Hi,

On 25/06/14 23:23, Russell King - ARM Linux wrote:
> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>> +		coherency_line_size: the minimum amount of data that gets transferred
>
> So, what value to do envision this taking for a CPU where the cache
> line size is 32 bytes, but each cache line has two dirty bits which
> allow it to only evict either the upper or lower 16 bytes depending
> on which are dirty?
>

IIUC most of existing implementations of cacheinfo on various architectures
are representing the cache line size as coherency_line_size, in which case I
need fix the definition in this file.

BTW will there be any architectural way of finding such configuration ?

Regards,
Sudeep

[PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

Hi,

On 25/06/14 23:23, Russell King - ARM Linux wrote:
> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>> +		coherency_line_size: the minimum amount of data that gets transferred
>
> So, what value to do envision this taking for a CPU where the cache
> line size is 32 bytes, but each cache line has two dirty bits which
> allow it to only evict either the upper or lower 16 bytes depending
> on which are dirty?
>

IIUC most of existing implementations of cacheinfo on various architectures
are representing the cache line size as coherency_line_size, in which case I
need fix the definition in this file.

BTW will there be any architectural way of finding such configuration ?

Regards,
Sudeep

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

Hi,

On 25/06/14 23:23, Russell King - ARM Linux wrote:
> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>> +=09=09coherency_line_size: the minimum amount of data that gets transfe=
rred
>
> So, what value to do envision this taking for a CPU where the cache
> line size is 32 bytes, but each cache line has two dirty bits which
> allow it to only evict either the upper or lower 16 bytes depending
> on which are dirty?
>

IIUC most of existing implementations of cacheinfo on various architectures
are representing the cache line size as coherency_line_size, in which case =
I
need fix the definition in this file.

BTW will there be any architectural way of finding such configuration ?

Regards,
Sudeep

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

Hi,

On 25/06/14 23:23, Russell King - ARM Linux wrote:
> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>> +		coherency_line_size: the minimum amount of data that gets transferred
>
> So, what value to do envision this taking for a CPU where the cache
> line size is 32 bytes, but each cache line has two dirty bits which
> allow it to only evict either the upper or lower 16 bytes depending
> on which are dirty?
>

IIUC most of existing implementations of cacheinfo on various architectures
are representing the cache line size as coherency_line_size, in which case I
need fix the definition in this file.

BTW will there be any architectural way of finding such configuration ?

Regards,
Sudeep

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Russell King - ARM Linux]

On Thu, Jun 26, 2014 at 07:41:32PM +0100, Sudeep Holla wrote:
> Hi,
>
> On 25/06/14 23:23, Russell King - ARM Linux wrote:
>> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>>> +		coherency_line_size: the minimum amount of data that gets transferred
>>
>> So, what value to do envision this taking for a CPU where the cache
>> line size is 32 bytes, but each cache line has two dirty bits which
>> allow it to only evict either the upper or lower 16 bytes depending
>> on which are dirty?
>>
>
> IIUC most of existing implementations of cacheinfo on various architectures
> are representing the cache line size as coherency_line_size, in which case I
> need fix the definition in this file.

As an example, here's an extract from the SA110 TRM:

StrongARM contains a 16KByte writeback data cache. The DC has 512 lines
of 32 bytes (8 words), arranged as a 32 way set associative cache, and
uses the virtual addresses generated by the processor. A line also
contains the physical address the block was fetched from and two dirty
bits. There is a dirty bit associated with both the first and second
half of the block. When a store hits in the cache the dirty bit
associated with it is set. When a block is evicted from the cache the
dirty bits are used to decide if all, half, or none of the block will
be written back to memory using the physical address stored with the
block. The DC is always reloaded a line at a time (8 words).

> BTW will there be any architectural way of finding such configuration ?

Not that I know of.

-- 
FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly
improving, and getting towards what was expected from it.

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Russell King - ARM Linux]

On Thu, Jun 26, 2014 at 07:41:32PM +0100, Sudeep Holla wrote:
> Hi,
>
> On 25/06/14 23:23, Russell King - ARM Linux wrote:
>> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>>> +		coherency_line_size: the minimum amount of data that gets transferred
>>
>> So, what value to do envision this taking for a CPU where the cache
>> line size is 32 bytes, but each cache line has two dirty bits which
>> allow it to only evict either the upper or lower 16 bytes depending
>> on which are dirty?
>>
>
> IIUC most of existing implementations of cacheinfo on various architectures
> are representing the cache line size as coherency_line_size, in which case I
> need fix the definition in this file.

As an example, here's an extract from the SA110 TRM:

StrongARM contains a 16KByte writeback data cache. The DC has 512 lines
of 32 bytes (8 words), arranged as a 32 way set associative cache, and
uses the virtual addresses generated by the processor. A line also
contains the physical address the block was fetched from and two dirty
bits. There is a dirty bit associated with both the first and second
half of the block. When a store hits in the cache the dirty bit
associated with it is set. When a block is evicted from the cache the
dirty bits are used to decide if all, half, or none of the block will
be written back to memory using the physical address stored with the
block. The DC is always reloaded a line at a time (8 words).

> BTW will there be any architectural way of finding such configuration ?

Not that I know of.

-- 
FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly
improving, and getting towards what was expected from it.

[PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Russell King - ARM Linux]

On Thu, Jun 26, 2014 at 07:41:32PM +0100, Sudeep Holla wrote:
> Hi,
>
> On 25/06/14 23:23, Russell King - ARM Linux wrote:
>> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>>> +		coherency_line_size: the minimum amount of data that gets transferred
>>
>> So, what value to do envision this taking for a CPU where the cache
>> line size is 32 bytes, but each cache line has two dirty bits which
>> allow it to only evict either the upper or lower 16 bytes depending
>> on which are dirty?
>>
>
> IIUC most of existing implementations of cacheinfo on various architectures
> are representing the cache line size as coherency_line_size, in which case I
> need fix the definition in this file.

As an example, here's an extract from the SA110 TRM:

StrongARM contains a 16KByte writeback data cache. The DC has 512 lines
of 32 bytes (8 words), arranged as a 32 way set associative cache, and
uses the virtual addresses generated by the processor. A line also
contains the physical address the block was fetched from and two dirty
bits. There is a dirty bit associated with both the first and second
half of the block. When a store hits in the cache the dirty bit
associated with it is set. When a block is evicted from the cache the
dirty bits are used to decide if all, half, or none of the block will
be written back to memory using the physical address stored with the
block. The DC is always reloaded a line at a time (8 words).

> BTW will there be any architectural way of finding such configuration ?

Not that I know of.

-- 
FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly
improving, and getting towards what was expected from it.

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Russell King - ARM Linux]

On Thu, Jun 26, 2014 at 07:41:32PM +0100, Sudeep Holla wrote:
> Hi,
>
> On 25/06/14 23:23, Russell King - ARM Linux wrote:
>> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>>> +		coherency_line_size: the minimum amount of data that gets transferred
>>
>> So, what value to do envision this taking for a CPU where the cache
>> line size is 32 bytes, but each cache line has two dirty bits which
>> allow it to only evict either the upper or lower 16 bytes depending
>> on which are dirty?
>>
>
> IIUC most of existing implementations of cacheinfo on various architectures
> are representing the cache line size as coherency_line_size, in which case I
> need fix the definition in this file.

As an example, here's an extract from the SA110 TRM:

StrongARM contains a 16KByte writeback data cache. The DC has 512 lines
of 32 bytes (8 words), arranged as a 32 way set associative cache, and
uses the virtual addresses generated by the processor. A line also
contains the physical address the block was fetched from and two dirty
bits. There is a dirty bit associated with both the first and second
half of the block. When a store hits in the cache the dirty bit
associated with it is set. When a block is evicted from the cache the
dirty bits are used to decide if all, half, or none of the block will
be written back to memory using the physical address stored with the
block. The DC is always reloaded a line at a time (8 words).

> BTW will there be any architectural way of finding such configuration ?

Not that I know of.

-- 
FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly
improving, and getting towards what was expected from it.

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Russell King - ARM Linux]

On Thu, Jun 26, 2014 at 07:41:32PM +0100, Sudeep Holla wrote:
> Hi,
>
> On 25/06/14 23:23, Russell King - ARM Linux wrote:
>> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>>> +		coherency_line_size: the minimum amount of data that gets transferred
>>
>> So, what value to do envision this taking for a CPU where the cache
>> line size is 32 bytes, but each cache line has two dirty bits which
>> allow it to only evict either the upper or lower 16 bytes depending
>> on which are dirty?
>>
>
> IIUC most of existing implementations of cacheinfo on various architectures
> are representing the cache line size as coherency_line_size, in which case I
> need fix the definition in this file.

As an example, here's an extract from the SA110 TRM:

StrongARM contains a 16KByte writeback data cache. The DC has 512 lines
of 32 bytes (8 words), arranged as a 32 way set associative cache, and
uses the virtual addresses generated by the processor. A line also
contains the physical address the block was fetched from and two dirty
bits. There is a dirty bit associated with both the first and second
half of the block. When a store hits in the cache the dirty bit
associated with it is set. When a block is evicted from the cache the
dirty bits are used to decide if all, half, or none of the block will
be written back to memory using the physical address stored with the
block. The DC is always reloaded a line at a time (8 words).

> BTW will there be any architectural way of finding such configuration ?

Not that I know of.

-- 
FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly
improving, and getting towards what was expected from it.

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

On 26/06/14 19:50, Russell King - ARM Linux wrote:
> On Thu, Jun 26, 2014 at 07:41:32PM +0100, Sudeep Holla wrote:
>> Hi,
>>
>> On 25/06/14 23:23, Russell King - ARM Linux wrote:
>>> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>>>> +		coherency_line_size: the minimum amount of data that gets transferred
>>>
>>> So, what value to do envision this taking for a CPU where the cache
>>> line size is 32 bytes, but each cache line has two dirty bits which
>>> allow it to only evict either the upper or lower 16 bytes depending
>>> on which are dirty?
>>>
>>
>> IIUC most of existing implementations of cacheinfo on various architectures
>> are representing the cache line size as coherency_line_size, in which case I
>> need fix the definition in this file.
>
> As an example, here's an extract from the SA110 TRM:
>
> StrongARM contains a 16KByte writeback data cache. The DC has 512 lines
> of 32 bytes (8 words), arranged as a 32 way set associative cache, and
> uses the virtual addresses generated by the processor. A line also
> contains the physical address the block was fetched from and two dirty
> bits. There is a dirty bit associated with both the first and second
> half of the block. When a store hits in the cache the dirty bit
> associated with it is set. When a block is evicted from the cache the
> dirty bits are used to decide if all, half, or none of the block will
> be written back to memory using the physical address stored with the
> block. The DC is always reloaded a line at a time (8 words).
>

Thanks for the information. It's interesting that line is referred as block
when referring to 2 dirty bits. I am not sure if this can be mapped to 
physical_line_partition = 2. Thoughts ?

>> BTW will there be any architectural way of finding such configuration ?
>
> Not that I know of.
>
That's bad :)

Regards,
Sudeep

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

On 26/06/14 19:50, Russell King - ARM Linux wrote:
> On Thu, Jun 26, 2014 at 07:41:32PM +0100, Sudeep Holla wrote:
>> Hi,
>>
>> On 25/06/14 23:23, Russell King - ARM Linux wrote:
>>> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>>>> +		coherency_line_size: the minimum amount of data that gets transferred
>>>
>>> So, what value to do envision this taking for a CPU where the cache
>>> line size is 32 bytes, but each cache line has two dirty bits which
>>> allow it to only evict either the upper or lower 16 bytes depending
>>> on which are dirty?
>>>
>>
>> IIUC most of existing implementations of cacheinfo on various architectures
>> are representing the cache line size as coherency_line_size, in which case I
>> need fix the definition in this file.
>
> As an example, here's an extract from the SA110 TRM:
>
> StrongARM contains a 16KByte writeback data cache. The DC has 512 lines
> of 32 bytes (8 words), arranged as a 32 way set associative cache, and
> uses the virtual addresses generated by the processor. A line also
> contains the physical address the block was fetched from and two dirty
> bits. There is a dirty bit associated with both the first and second
> half of the block. When a store hits in the cache the dirty bit
> associated with it is set. When a block is evicted from the cache the
> dirty bits are used to decide if all, half, or none of the block will
> be written back to memory using the physical address stored with the
> block. The DC is always reloaded a line at a time (8 words).
>

Thanks for the information. It's interesting that line is referred as block
when referring to 2 dirty bits. I am not sure if this can be mapped to 
physical_line_partition = 2. Thoughts ?

>> BTW will there be any architectural way of finding such configuration ?
>
> Not that I know of.
>
That's bad :)

Regards,
Sudeep

[PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

On 26/06/14 19:50, Russell King - ARM Linux wrote:
> On Thu, Jun 26, 2014 at 07:41:32PM +0100, Sudeep Holla wrote:
>> Hi,
>>
>> On 25/06/14 23:23, Russell King - ARM Linux wrote:
>>> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>>>> +		coherency_line_size: the minimum amount of data that gets transferred
>>>
>>> So, what value to do envision this taking for a CPU where the cache
>>> line size is 32 bytes, but each cache line has two dirty bits which
>>> allow it to only evict either the upper or lower 16 bytes depending
>>> on which are dirty?
>>>
>>
>> IIUC most of existing implementations of cacheinfo on various architectures
>> are representing the cache line size as coherency_line_size, in which case I
>> need fix the definition in this file.
>
> As an example, here's an extract from the SA110 TRM:
>
> StrongARM contains a 16KByte writeback data cache. The DC has 512 lines
> of 32 bytes (8 words), arranged as a 32 way set associative cache, and
> uses the virtual addresses generated by the processor. A line also
> contains the physical address the block was fetched from and two dirty
> bits. There is a dirty bit associated with both the first and second
> half of the block. When a store hits in the cache the dirty bit
> associated with it is set. When a block is evicted from the cache the
> dirty bits are used to decide if all, half, or none of the block will
> be written back to memory using the physical address stored with the
> block. The DC is always reloaded a line at a time (8 words).
>

Thanks for the information. It's interesting that line is referred as block
when referring to 2 dirty bits. I am not sure if this can be mapped to 
physical_line_partition = 2. Thoughts ?

>> BTW will there be any architectural way of finding such configuration ?
>
> Not that I know of.
>
That's bad :)

Regards,
Sudeep

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

On 26/06/14 19:50, Russell King - ARM Linux wrote:
> On Thu, Jun 26, 2014 at 07:41:32PM +0100, Sudeep Holla wrote:
>> Hi,
>>
>> On 25/06/14 23:23, Russell King - ARM Linux wrote:
>>> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>>>> +=09=09coherency_line_size: the minimum amount of data that gets trans=
ferred
>>>
>>> So, what value to do envision this taking for a CPU where the cache
>>> line size is 32 bytes, but each cache line has two dirty bits which
>>> allow it to only evict either the upper or lower 16 bytes depending
>>> on which are dirty?
>>>
>>
>> IIUC most of existing implementations of cacheinfo on various architectu=
res
>> are representing the cache line size as coherency_line_size, in which ca=
se I
>> need fix the definition in this file.
>
> As an example, here's an extract from the SA110 TRM:
>
> StrongARM contains a 16KByte writeback data cache. The DC has 512 lines
> of 32 bytes (8 words), arranged as a 32 way set associative cache, and
> uses the virtual addresses generated by the processor. A line also
> contains the physical address the block was fetched from and two dirty
> bits. There is a dirty bit associated with both the first and second
> half of the block. When a store hits in the cache the dirty bit
> associated with it is set. When a block is evicted from the cache the
> dirty bits are used to decide if all, half, or none of the block will
> be written back to memory using the physical address stored with the
> block. The DC is always reloaded a line at a time (8 words).
>

Thanks for the information. It's interesting that line is referred as block
when referring to 2 dirty bits. I am not sure if this can be mapped to=20
physical_line_partition =3D 2. Thoughts ?

>> BTW will there be any architectural way of finding such configuration ?
>
> Not that I know of.
>
That's bad :)

Regards,
Sudeep

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

On 26/06/14 19:50, Russell King - ARM Linux wrote:
> On Thu, Jun 26, 2014 at 07:41:32PM +0100, Sudeep Holla wrote:
>> Hi,
>>
>> On 25/06/14 23:23, Russell King - ARM Linux wrote:
>>> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>>>> +		coherency_line_size: the minimum amount of data that gets transferred
>>>
>>> So, what value to do envision this taking for a CPU where the cache
>>> line size is 32 bytes, but each cache line has two dirty bits which
>>> allow it to only evict either the upper or lower 16 bytes depending
>>> on which are dirty?
>>>
>>
>> IIUC most of existing implementations of cacheinfo on various architectures
>> are representing the cache line size as coherency_line_size, in which case I
>> need fix the definition in this file.
>
> As an example, here's an extract from the SA110 TRM:
>
> StrongARM contains a 16KByte writeback data cache. The DC has 512 lines
> of 32 bytes (8 words), arranged as a 32 way set associative cache, and
> uses the virtual addresses generated by the processor. A line also
> contains the physical address the block was fetched from and two dirty
> bits. There is a dirty bit associated with both the first and second
> half of the block. When a store hits in the cache the dirty bit
> associated with it is set. When a block is evicted from the cache the
> dirty bits are used to decide if all, half, or none of the block will
> be written back to memory using the physical address stored with the
> block. The DC is always reloaded a line at a time (8 words).
>

Thanks for the information. It's interesting that line is referred as block
when referring to 2 dirty bits. I am not sure if this can be mapped to 
physical_line_partition = 2. Thoughts ?

>> BTW will there be any architectural way of finding such configuration ?
>
> Not that I know of.
>
That's bad :)

Regards,
Sudeep

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Greg Kroah-Hartman]

On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
> +static const struct device_attribute *cache_optional_attrs[] = {
> +	&dev_attr_coherency_line_size,
> +	&dev_attr_ways_of_associativity,
> +	&dev_attr_number_of_sets,
> +	&dev_attr_size,
> +	&dev_attr_attributes,
> +	&dev_attr_physical_line_partition,
> +	NULL
> +};
> +
> +static int device_add_attrs(struct device *dev,
> +			    const struct device_attribute **dev_attrs)
> +{
> +	int i, error = 0;
> +	struct device_attribute *dev_attr;
> +	char *buf;
> +
> +	if (!dev_attrs)
> +		return 0;
> +
> +	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
> +	if (!buf)
> +		return -ENOMEM;
> +
> +	for (i = 0; dev_attrs[i]; i++) {
> +		dev_attr = (struct device_attribute *)dev_attrs[i];
> +
> +		/* create attributes that provides meaningful value */
> +		if (dev_attr->show(dev, dev_attr, buf) < 0)
> +			continue;
> +
> +		error = device_create_file(dev, dev_attrs[i]);
> +		if (error) {
> +			while (--i >= 0)
> +				device_remove_file(dev, dev_attrs[i]);
> +			break;
> +		}
> +	}
> +
> +	kfree(buf);
> +	return error;
> +}

Ick, why create your own function for this when the driver core has this
functionality built into it?  Look at the is_visible() callback, and how
it is use for an attribute group please.

> +static void device_remove_attrs(struct device *dev,
> +				const struct device_attribute **dev_attrs)
> +{
> +	int i;
> +
> +	if (!dev_attrs)
> +		return;
> +
> +	for (i = 0; dev_attrs[i]; dev_attrs++, i++)
> +		device_remove_file(dev, dev_attrs[i]);
> +}

You should just remove a whole group at once, not individually.

> +
> +const struct device_attribute **
> +__weak cache_get_priv_attr(struct device *cache_idx_dev)
> +{
> +	return NULL;
> +}
> +
> +/* Add/Remove cache interface for CPU device */
> +static void cpu_cache_sysfs_exit(unsigned int cpu)
> +{
> +	int i;
> +	struct device *tmp_dev;
> +	const struct device_attribute **ci_priv_attr;
> +
> +	if (per_cpu_index_dev(cpu)) {
> +		for (i = 0; i < cache_leaves(cpu); i++) {
> +			tmp_dev = per_cache_index_dev(cpu, i);
> +			if (!tmp_dev)
> +				continue;
> +			ci_priv_attr = cache_get_priv_attr(tmp_dev);
> +			device_remove_attrs(tmp_dev, ci_priv_attr);
> +			device_remove_attrs(tmp_dev, cache_optional_attrs);
> +			device_unregister(tmp_dev);
> +		}
> +		kfree(per_cpu_index_dev(cpu));
> +		per_cpu_index_dev(cpu) = NULL;
> +	}
> +	device_unregister(per_cpu_cache_dev(cpu));
> +	per_cpu_cache_dev(cpu) = NULL;
> +}
> +
> +static int cpu_cache_sysfs_init(unsigned int cpu)
> +{
> +	struct device *dev = get_cpu_device(cpu);
> +
> +	if (per_cpu_cacheinfo(cpu) == NULL)
> +		return -ENOENT;
> +
> +	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
> +					       NULL, "cache");
> +	if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
> +		return PTR_ERR(per_cpu_cache_dev(cpu));
> +
> +	/* Allocate all required memory */
> +	per_cpu_index_dev(cpu) = kzalloc(sizeof(struct device *) *
> +					 cache_leaves(cpu), GFP_KERNEL);
> +	if (unlikely(per_cpu_index_dev(cpu) == NULL))
> +		goto err_out;
> +
> +	return 0;
> +
> +err_out:
> +	cpu_cache_sysfs_exit(cpu);
> +	return -ENOMEM;
> +}
> +
> +static int cache_add_dev(unsigned int cpu)
> +{
> +	unsigned short i;
> +	int rc;
> +	struct device *tmp_dev, *parent;
> +	struct cacheinfo *this_leaf;
> +	const struct device_attribute **ci_priv_attr;
> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
> +
> +	rc = cpu_cache_sysfs_init(cpu);
> +	if (unlikely(rc < 0))
> +		return rc;
> +
> +	parent = per_cpu_cache_dev(cpu);
> +	for (i = 0; i < cache_leaves(cpu); i++) {
> +		this_leaf = this_cpu_ci->info_list + i;
> +		if (this_leaf->disable_sysfs)
> +			continue;
> +		tmp_dev = device_create_with_groups(parent->class, parent, i,
> +						    this_leaf,
> +						    cache_default_groups,
> +						    "index%1u", i);
> +		if (IS_ERR_OR_NULL(tmp_dev)) {
> +			rc = PTR_ERR(tmp_dev);
> +			goto err;
> +		}
> +
> +		rc = device_add_attrs(tmp_dev, cache_optional_attrs);
> +		if (unlikely(rc))
> +			goto err;
> +
> +		ci_priv_attr = cache_get_priv_attr(tmp_dev);
> +		rc = device_add_attrs(tmp_dev, ci_priv_attr);
> +		if (unlikely(rc))
> +			goto err;

You just raced with userspace here, creating these files _after_ the
device was announced to userspace, causing problems with anyone wanting
to read these attributes :(

I think if you fix up the is_visible() thing above, these calls will go
away, right?

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Greg Kroah-Hartman]

On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
> +static const struct device_attribute *cache_optional_attrs[] = {
> +	&dev_attr_coherency_line_size,
> +	&dev_attr_ways_of_associativity,
> +	&dev_attr_number_of_sets,
> +	&dev_attr_size,
> +	&dev_attr_attributes,
> +	&dev_attr_physical_line_partition,
> +	NULL
> +};
> +
> +static int device_add_attrs(struct device *dev,
> +			    const struct device_attribute **dev_attrs)
> +{
> +	int i, error = 0;
> +	struct device_attribute *dev_attr;
> +	char *buf;
> +
> +	if (!dev_attrs)
> +		return 0;
> +
> +	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
> +	if (!buf)
> +		return -ENOMEM;
> +
> +	for (i = 0; dev_attrs[i]; i++) {
> +		dev_attr = (struct device_attribute *)dev_attrs[i];
> +
> +		/* create attributes that provides meaningful value */
> +		if (dev_attr->show(dev, dev_attr, buf) < 0)
> +			continue;
> +
> +		error = device_create_file(dev, dev_attrs[i]);
> +		if (error) {
> +			while (--i >= 0)
> +				device_remove_file(dev, dev_attrs[i]);
> +			break;
> +		}
> +	}
> +
> +	kfree(buf);
> +	return error;
> +}

Ick, why create your own function for this when the driver core has this
functionality built into it?  Look at the is_visible() callback, and how
it is use for an attribute group please.

> +static void device_remove_attrs(struct device *dev,
> +				const struct device_attribute **dev_attrs)
> +{
> +	int i;
> +
> +	if (!dev_attrs)
> +		return;
> +
> +	for (i = 0; dev_attrs[i]; dev_attrs++, i++)
> +		device_remove_file(dev, dev_attrs[i]);
> +}

You should just remove a whole group at once, not individually.

> +
> +const struct device_attribute **
> +__weak cache_get_priv_attr(struct device *cache_idx_dev)
> +{
> +	return NULL;
> +}
> +
> +/* Add/Remove cache interface for CPU device */
> +static void cpu_cache_sysfs_exit(unsigned int cpu)
> +{
> +	int i;
> +	struct device *tmp_dev;
> +	const struct device_attribute **ci_priv_attr;
> +
> +	if (per_cpu_index_dev(cpu)) {
> +		for (i = 0; i < cache_leaves(cpu); i++) {
> +			tmp_dev = per_cache_index_dev(cpu, i);
> +			if (!tmp_dev)
> +				continue;
> +			ci_priv_attr = cache_get_priv_attr(tmp_dev);
> +			device_remove_attrs(tmp_dev, ci_priv_attr);
> +			device_remove_attrs(tmp_dev, cache_optional_attrs);
> +			device_unregister(tmp_dev);
> +		}
> +		kfree(per_cpu_index_dev(cpu));
> +		per_cpu_index_dev(cpu) = NULL;
> +	}
> +	device_unregister(per_cpu_cache_dev(cpu));
> +	per_cpu_cache_dev(cpu) = NULL;
> +}
> +
> +static int cpu_cache_sysfs_init(unsigned int cpu)
> +{
> +	struct device *dev = get_cpu_device(cpu);
> +
> +	if (per_cpu_cacheinfo(cpu) = NULL)
> +		return -ENOENT;
> +
> +	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
> +					       NULL, "cache");
> +	if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
> +		return PTR_ERR(per_cpu_cache_dev(cpu));
> +
> +	/* Allocate all required memory */
> +	per_cpu_index_dev(cpu) = kzalloc(sizeof(struct device *) *
> +					 cache_leaves(cpu), GFP_KERNEL);
> +	if (unlikely(per_cpu_index_dev(cpu) = NULL))
> +		goto err_out;
> +
> +	return 0;
> +
> +err_out:
> +	cpu_cache_sysfs_exit(cpu);
> +	return -ENOMEM;
> +}
> +
> +static int cache_add_dev(unsigned int cpu)
> +{
> +	unsigned short i;
> +	int rc;
> +	struct device *tmp_dev, *parent;
> +	struct cacheinfo *this_leaf;
> +	const struct device_attribute **ci_priv_attr;
> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
> +
> +	rc = cpu_cache_sysfs_init(cpu);
> +	if (unlikely(rc < 0))
> +		return rc;
> +
> +	parent = per_cpu_cache_dev(cpu);
> +	for (i = 0; i < cache_leaves(cpu); i++) {
> +		this_leaf = this_cpu_ci->info_list + i;
> +		if (this_leaf->disable_sysfs)
> +			continue;
> +		tmp_dev = device_create_with_groups(parent->class, parent, i,
> +						    this_leaf,
> +						    cache_default_groups,
> +						    "index%1u", i);
> +		if (IS_ERR_OR_NULL(tmp_dev)) {
> +			rc = PTR_ERR(tmp_dev);
> +			goto err;
> +		}
> +
> +		rc = device_add_attrs(tmp_dev, cache_optional_attrs);
> +		if (unlikely(rc))
> +			goto err;
> +
> +		ci_priv_attr = cache_get_priv_attr(tmp_dev);
> +		rc = device_add_attrs(tmp_dev, ci_priv_attr);
> +		if (unlikely(rc))
> +			goto err;

You just raced with userspace here, creating these files _after_ the
device was announced to userspace, causing problems with anyone wanting
to read these attributes :(

I think if you fix up the is_visible() thing above, these calls will go
away, right?

[PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Greg Kroah-Hartman]

On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
> +static const struct device_attribute *cache_optional_attrs[] = {
> +	&dev_attr_coherency_line_size,
> +	&dev_attr_ways_of_associativity,
> +	&dev_attr_number_of_sets,
> +	&dev_attr_size,
> +	&dev_attr_attributes,
> +	&dev_attr_physical_line_partition,
> +	NULL
> +};
> +
> +static int device_add_attrs(struct device *dev,
> +			    const struct device_attribute **dev_attrs)
> +{
> +	int i, error = 0;
> +	struct device_attribute *dev_attr;
> +	char *buf;
> +
> +	if (!dev_attrs)
> +		return 0;
> +
> +	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
> +	if (!buf)
> +		return -ENOMEM;
> +
> +	for (i = 0; dev_attrs[i]; i++) {
> +		dev_attr = (struct device_attribute *)dev_attrs[i];
> +
> +		/* create attributes that provides meaningful value */
> +		if (dev_attr->show(dev, dev_attr, buf) < 0)
> +			continue;
> +
> +		error = device_create_file(dev, dev_attrs[i]);
> +		if (error) {
> +			while (--i >= 0)
> +				device_remove_file(dev, dev_attrs[i]);
> +			break;
> +		}
> +	}
> +
> +	kfree(buf);
> +	return error;
> +}

Ick, why create your own function for this when the driver core has this
functionality built into it?  Look at the is_visible() callback, and how
it is use for an attribute group please.

> +static void device_remove_attrs(struct device *dev,
> +				const struct device_attribute **dev_attrs)
> +{
> +	int i;
> +
> +	if (!dev_attrs)
> +		return;
> +
> +	for (i = 0; dev_attrs[i]; dev_attrs++, i++)
> +		device_remove_file(dev, dev_attrs[i]);
> +}

You should just remove a whole group at once, not individually.

> +
> +const struct device_attribute **
> +__weak cache_get_priv_attr(struct device *cache_idx_dev)
> +{
> +	return NULL;
> +}
> +
> +/* Add/Remove cache interface for CPU device */
> +static void cpu_cache_sysfs_exit(unsigned int cpu)
> +{
> +	int i;
> +	struct device *tmp_dev;
> +	const struct device_attribute **ci_priv_attr;
> +
> +	if (per_cpu_index_dev(cpu)) {
> +		for (i = 0; i < cache_leaves(cpu); i++) {
> +			tmp_dev = per_cache_index_dev(cpu, i);
> +			if (!tmp_dev)
> +				continue;
> +			ci_priv_attr = cache_get_priv_attr(tmp_dev);
> +			device_remove_attrs(tmp_dev, ci_priv_attr);
> +			device_remove_attrs(tmp_dev, cache_optional_attrs);
> +			device_unregister(tmp_dev);
> +		}
> +		kfree(per_cpu_index_dev(cpu));
> +		per_cpu_index_dev(cpu) = NULL;
> +	}
> +	device_unregister(per_cpu_cache_dev(cpu));
> +	per_cpu_cache_dev(cpu) = NULL;
> +}
> +
> +static int cpu_cache_sysfs_init(unsigned int cpu)
> +{
> +	struct device *dev = get_cpu_device(cpu);
> +
> +	if (per_cpu_cacheinfo(cpu) == NULL)
> +		return -ENOENT;
> +
> +	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
> +					       NULL, "cache");
> +	if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
> +		return PTR_ERR(per_cpu_cache_dev(cpu));
> +
> +	/* Allocate all required memory */
> +	per_cpu_index_dev(cpu) = kzalloc(sizeof(struct device *) *
> +					 cache_leaves(cpu), GFP_KERNEL);
> +	if (unlikely(per_cpu_index_dev(cpu) == NULL))
> +		goto err_out;
> +
> +	return 0;
> +
> +err_out:
> +	cpu_cache_sysfs_exit(cpu);
> +	return -ENOMEM;
> +}
> +
> +static int cache_add_dev(unsigned int cpu)
> +{
> +	unsigned short i;
> +	int rc;
> +	struct device *tmp_dev, *parent;
> +	struct cacheinfo *this_leaf;
> +	const struct device_attribute **ci_priv_attr;
> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
> +
> +	rc = cpu_cache_sysfs_init(cpu);
> +	if (unlikely(rc < 0))
> +		return rc;
> +
> +	parent = per_cpu_cache_dev(cpu);
> +	for (i = 0; i < cache_leaves(cpu); i++) {
> +		this_leaf = this_cpu_ci->info_list + i;
> +		if (this_leaf->disable_sysfs)
> +			continue;
> +		tmp_dev = device_create_with_groups(parent->class, parent, i,
> +						    this_leaf,
> +						    cache_default_groups,
> +						    "index%1u", i);
> +		if (IS_ERR_OR_NULL(tmp_dev)) {
> +			rc = PTR_ERR(tmp_dev);
> +			goto err;
> +		}
> +
> +		rc = device_add_attrs(tmp_dev, cache_optional_attrs);
> +		if (unlikely(rc))
> +			goto err;
> +
> +		ci_priv_attr = cache_get_priv_attr(tmp_dev);
> +		rc = device_add_attrs(tmp_dev, ci_priv_attr);
> +		if (unlikely(rc))
> +			goto err;

You just raced with userspace here, creating these files _after_ the
device was announced to userspace, causing problems with anyone wanting
to read these attributes :(

I think if you fix up the is_visible() thing above, these calls will go
away, right?

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Greg Kroah-Hartman]

On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
> +static const struct device_attribute *cache_optional_attrs[] = {
> +	&dev_attr_coherency_line_size,
> +	&dev_attr_ways_of_associativity,
> +	&dev_attr_number_of_sets,
> +	&dev_attr_size,
> +	&dev_attr_attributes,
> +	&dev_attr_physical_line_partition,
> +	NULL
> +};
> +
> +static int device_add_attrs(struct device *dev,
> +			    const struct device_attribute **dev_attrs)
> +{
> +	int i, error = 0;
> +	struct device_attribute *dev_attr;
> +	char *buf;
> +
> +	if (!dev_attrs)
> +		return 0;
> +
> +	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
> +	if (!buf)
> +		return -ENOMEM;
> +
> +	for (i = 0; dev_attrs[i]; i++) {
> +		dev_attr = (struct device_attribute *)dev_attrs[i];
> +
> +		/* create attributes that provides meaningful value */
> +		if (dev_attr->show(dev, dev_attr, buf) < 0)
> +			continue;
> +
> +		error = device_create_file(dev, dev_attrs[i]);
> +		if (error) {
> +			while (--i >= 0)
> +				device_remove_file(dev, dev_attrs[i]);
> +			break;
> +		}
> +	}
> +
> +	kfree(buf);
> +	return error;
> +}

Ick, why create your own function for this when the driver core has this
functionality built into it?  Look at the is_visible() callback, and how
it is use for an attribute group please.

> +static void device_remove_attrs(struct device *dev,
> +				const struct device_attribute **dev_attrs)
> +{
> +	int i;
> +
> +	if (!dev_attrs)
> +		return;
> +
> +	for (i = 0; dev_attrs[i]; dev_attrs++, i++)
> +		device_remove_file(dev, dev_attrs[i]);
> +}

You should just remove a whole group at once, not individually.

> +
> +const struct device_attribute **
> +__weak cache_get_priv_attr(struct device *cache_idx_dev)
> +{
> +	return NULL;
> +}
> +
> +/* Add/Remove cache interface for CPU device */
> +static void cpu_cache_sysfs_exit(unsigned int cpu)
> +{
> +	int i;
> +	struct device *tmp_dev;
> +	const struct device_attribute **ci_priv_attr;
> +
> +	if (per_cpu_index_dev(cpu)) {
> +		for (i = 0; i < cache_leaves(cpu); i++) {
> +			tmp_dev = per_cache_index_dev(cpu, i);
> +			if (!tmp_dev)
> +				continue;
> +			ci_priv_attr = cache_get_priv_attr(tmp_dev);
> +			device_remove_attrs(tmp_dev, ci_priv_attr);
> +			device_remove_attrs(tmp_dev, cache_optional_attrs);
> +			device_unregister(tmp_dev);
> +		}
> +		kfree(per_cpu_index_dev(cpu));
> +		per_cpu_index_dev(cpu) = NULL;
> +	}
> +	device_unregister(per_cpu_cache_dev(cpu));
> +	per_cpu_cache_dev(cpu) = NULL;
> +}
> +
> +static int cpu_cache_sysfs_init(unsigned int cpu)
> +{
> +	struct device *dev = get_cpu_device(cpu);
> +
> +	if (per_cpu_cacheinfo(cpu) == NULL)
> +		return -ENOENT;
> +
> +	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
> +					       NULL, "cache");
> +	if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
> +		return PTR_ERR(per_cpu_cache_dev(cpu));
> +
> +	/* Allocate all required memory */
> +	per_cpu_index_dev(cpu) = kzalloc(sizeof(struct device *) *
> +					 cache_leaves(cpu), GFP_KERNEL);
> +	if (unlikely(per_cpu_index_dev(cpu) == NULL))
> +		goto err_out;
> +
> +	return 0;
> +
> +err_out:
> +	cpu_cache_sysfs_exit(cpu);
> +	return -ENOMEM;
> +}
> +
> +static int cache_add_dev(unsigned int cpu)
> +{
> +	unsigned short i;
> +	int rc;
> +	struct device *tmp_dev, *parent;
> +	struct cacheinfo *this_leaf;
> +	const struct device_attribute **ci_priv_attr;
> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
> +
> +	rc = cpu_cache_sysfs_init(cpu);
> +	if (unlikely(rc < 0))
> +		return rc;
> +
> +	parent = per_cpu_cache_dev(cpu);
> +	for (i = 0; i < cache_leaves(cpu); i++) {
> +		this_leaf = this_cpu_ci->info_list + i;
> +		if (this_leaf->disable_sysfs)
> +			continue;
> +		tmp_dev = device_create_with_groups(parent->class, parent, i,
> +						    this_leaf,
> +						    cache_default_groups,
> +						    "index%1u", i);
> +		if (IS_ERR_OR_NULL(tmp_dev)) {
> +			rc = PTR_ERR(tmp_dev);
> +			goto err;
> +		}
> +
> +		rc = device_add_attrs(tmp_dev, cache_optional_attrs);
> +		if (unlikely(rc))
> +			goto err;
> +
> +		ci_priv_attr = cache_get_priv_attr(tmp_dev);
> +		rc = device_add_attrs(tmp_dev, ci_priv_attr);
> +		if (unlikely(rc))
> +			goto err;

You just raced with userspace here, creating these files _after_ the
device was announced to userspace, causing problems with anyone wanting
to read these attributes :(

I think if you fix up the is_visible() thing above, these calls will go
away, right?

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

Hi Greg,

Thanks for reviewing this.

On 10/07/14 01:09, Greg Kroah-Hartman wrote:
> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>> +static const struct device_attribute *cache_optional_attrs[] = {
>> +	&dev_attr_coherency_line_size,
>> +	&dev_attr_ways_of_associativity,
>> +	&dev_attr_number_of_sets,
>> +	&dev_attr_size,
>> +	&dev_attr_attributes,
>> +	&dev_attr_physical_line_partition,
>> +	NULL
>> +};
>> +
>> +static int device_add_attrs(struct device *dev,
>> +			    const struct device_attribute **dev_attrs)
>> +{
>> +	int i, error = 0;
>> +	struct device_attribute *dev_attr;
>> +	char *buf;
>> +
>> +	if (!dev_attrs)
>> +		return 0;
>> +
>> +	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
>> +	if (!buf)
>> +		return -ENOMEM;
>> +
>> +	for (i = 0; dev_attrs[i]; i++) {
>> +		dev_attr = (struct device_attribute *)dev_attrs[i];
>> +
>> +		/* create attributes that provides meaningful value */
>> +		if (dev_attr->show(dev, dev_attr, buf) < 0)
>> +			continue;
>> +
>> +		error = device_create_file(dev, dev_attrs[i]);
>> +		if (error) {
>> +			while (--i >= 0)
>> +				device_remove_file(dev, dev_attrs[i]);
>> +			break;
>> +		}
>> +	}
>> +
>> +	kfree(buf);
>> +	return error;
>> +}
>
> Ick, why create your own function for this when the driver core has this
> functionality built into it?  Look at the is_visible() callback, and how
> it is use for an attribute group please.
>

I agree even I added this function hesitantly as didn't realize that I can use
is_visible for this purpose. Thanks for pointing that out I will have a look
at it.

>> +static void device_remove_attrs(struct device *dev,
>> +				const struct device_attribute **dev_attrs)
>> +{
>> +	int i;
>> +
>> +	if (!dev_attrs)
>> +		return;
>> +
>> +	for (i = 0; dev_attrs[i]; dev_attrs++, i++)
>> +		device_remove_file(dev, dev_attrs[i]);
>> +}
>
> You should just remove a whole group at once, not individually.
>

Right, I must be able to get rid of these 2 functions once I use
is_visible callback.

>> +
>> +const struct device_attribute **
>> +__weak cache_get_priv_attr(struct device *cache_idx_dev)
>> +{
>> +	return NULL;
>> +}
>> +
>> +/* Add/Remove cache interface for CPU device */
>> +static void cpu_cache_sysfs_exit(unsigned int cpu)
>> +{
>> +	int i;
>> +	struct device *tmp_dev;
>> +	const struct device_attribute **ci_priv_attr;
>> +
>> +	if (per_cpu_index_dev(cpu)) {
>> +		for (i = 0; i < cache_leaves(cpu); i++) {
>> +			tmp_dev = per_cache_index_dev(cpu, i);
>> +			if (!tmp_dev)
>> +				continue;
>> +			ci_priv_attr = cache_get_priv_attr(tmp_dev);
>> +			device_remove_attrs(tmp_dev, ci_priv_attr);
>> +			device_remove_attrs(tmp_dev, cache_optional_attrs);
>> +			device_unregister(tmp_dev);
>> +		}
>> +		kfree(per_cpu_index_dev(cpu));
>> +		per_cpu_index_dev(cpu) = NULL;
>> +	}
>> +	device_unregister(per_cpu_cache_dev(cpu));
>> +	per_cpu_cache_dev(cpu) = NULL;
>> +}
>> +
>> +static int cpu_cache_sysfs_init(unsigned int cpu)
>> +{
>> +	struct device *dev = get_cpu_device(cpu);
>> +
>> +	if (per_cpu_cacheinfo(cpu) == NULL)
>> +		return -ENOENT;
>> +
>> +	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
>> +					       NULL, "cache");
>> +	if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
>> +		return PTR_ERR(per_cpu_cache_dev(cpu));
>> +
>> +	/* Allocate all required memory */
>> +	per_cpu_index_dev(cpu) = kzalloc(sizeof(struct device *) *
>> +					 cache_leaves(cpu), GFP_KERNEL);
>> +	if (unlikely(per_cpu_index_dev(cpu) == NULL))
>> +		goto err_out;
>> +
>> +	return 0;
>> +
>> +err_out:
>> +	cpu_cache_sysfs_exit(cpu);
>> +	return -ENOMEM;
>> +}
>> +
>> +static int cache_add_dev(unsigned int cpu)
>> +{
>> +	unsigned short i;
>> +	int rc;
>> +	struct device *tmp_dev, *parent;
>> +	struct cacheinfo *this_leaf;
>> +	const struct device_attribute **ci_priv_attr;
>> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
>> +
>> +	rc = cpu_cache_sysfs_init(cpu);
>> +	if (unlikely(rc < 0))
>> +		return rc;
>> +
>> +	parent = per_cpu_cache_dev(cpu);
>> +	for (i = 0; i < cache_leaves(cpu); i++) {
>> +		this_leaf = this_cpu_ci->info_list + i;
>> +		if (this_leaf->disable_sysfs)
>> +			continue;
>> +		tmp_dev = device_create_with_groups(parent->class, parent, i,
>> +						    this_leaf,
>> +						    cache_default_groups,
>> +						    "index%1u", i);
>> +		if (IS_ERR_OR_NULL(tmp_dev)) {
>> +			rc = PTR_ERR(tmp_dev);
>> +			goto err;
>> +		}
>> +
>> +		rc = device_add_attrs(tmp_dev, cache_optional_attrs);
>> +		if (unlikely(rc))
>> +			goto err;
>> +
>> +		ci_priv_attr = cache_get_priv_attr(tmp_dev);
>> +		rc = device_add_attrs(tmp_dev, ci_priv_attr);
>> +		if (unlikely(rc))
>> +			goto err;
>
> You just raced with userspace here, creating these files _after_ the
> device was announced to userspace, causing problems with anyone wanting
> to read these attributes :(
>
> I think if you fix up the is_visible() thing above, these calls will go
> away, right?
>

Yes I agree.

Regards,
Sudeep

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

Hi Greg,

Thanks for reviewing this.

On 10/07/14 01:09, Greg Kroah-Hartman wrote:
> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>> +static const struct device_attribute *cache_optional_attrs[] = {
>> +	&dev_attr_coherency_line_size,
>> +	&dev_attr_ways_of_associativity,
>> +	&dev_attr_number_of_sets,
>> +	&dev_attr_size,
>> +	&dev_attr_attributes,
>> +	&dev_attr_physical_line_partition,
>> +	NULL
>> +};
>> +
>> +static int device_add_attrs(struct device *dev,
>> +			    const struct device_attribute **dev_attrs)
>> +{
>> +	int i, error = 0;
>> +	struct device_attribute *dev_attr;
>> +	char *buf;
>> +
>> +	if (!dev_attrs)
>> +		return 0;
>> +
>> +	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
>> +	if (!buf)
>> +		return -ENOMEM;
>> +
>> +	for (i = 0; dev_attrs[i]; i++) {
>> +		dev_attr = (struct device_attribute *)dev_attrs[i];
>> +
>> +		/* create attributes that provides meaningful value */
>> +		if (dev_attr->show(dev, dev_attr, buf) < 0)
>> +			continue;
>> +
>> +		error = device_create_file(dev, dev_attrs[i]);
>> +		if (error) {
>> +			while (--i >= 0)
>> +				device_remove_file(dev, dev_attrs[i]);
>> +			break;
>> +		}
>> +	}
>> +
>> +	kfree(buf);
>> +	return error;
>> +}
>
> Ick, why create your own function for this when the driver core has this
> functionality built into it?  Look at the is_visible() callback, and how
> it is use for an attribute group please.
>

I agree even I added this function hesitantly as didn't realize that I can use
is_visible for this purpose. Thanks for pointing that out I will have a look
at it.

>> +static void device_remove_attrs(struct device *dev,
>> +				const struct device_attribute **dev_attrs)
>> +{
>> +	int i;
>> +
>> +	if (!dev_attrs)
>> +		return;
>> +
>> +	for (i = 0; dev_attrs[i]; dev_attrs++, i++)
>> +		device_remove_file(dev, dev_attrs[i]);
>> +}
>
> You should just remove a whole group at once, not individually.
>

Right, I must be able to get rid of these 2 functions once I use
is_visible callback.

>> +
>> +const struct device_attribute **
>> +__weak cache_get_priv_attr(struct device *cache_idx_dev)
>> +{
>> +	return NULL;
>> +}
>> +
>> +/* Add/Remove cache interface for CPU device */
>> +static void cpu_cache_sysfs_exit(unsigned int cpu)
>> +{
>> +	int i;
>> +	struct device *tmp_dev;
>> +	const struct device_attribute **ci_priv_attr;
>> +
>> +	if (per_cpu_index_dev(cpu)) {
>> +		for (i = 0; i < cache_leaves(cpu); i++) {
>> +			tmp_dev = per_cache_index_dev(cpu, i);
>> +			if (!tmp_dev)
>> +				continue;
>> +			ci_priv_attr = cache_get_priv_attr(tmp_dev);
>> +			device_remove_attrs(tmp_dev, ci_priv_attr);
>> +			device_remove_attrs(tmp_dev, cache_optional_attrs);
>> +			device_unregister(tmp_dev);
>> +		}
>> +		kfree(per_cpu_index_dev(cpu));
>> +		per_cpu_index_dev(cpu) = NULL;
>> +	}
>> +	device_unregister(per_cpu_cache_dev(cpu));
>> +	per_cpu_cache_dev(cpu) = NULL;
>> +}
>> +
>> +static int cpu_cache_sysfs_init(unsigned int cpu)
>> +{
>> +	struct device *dev = get_cpu_device(cpu);
>> +
>> +	if (per_cpu_cacheinfo(cpu) = NULL)
>> +		return -ENOENT;
>> +
>> +	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
>> +					       NULL, "cache");
>> +	if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
>> +		return PTR_ERR(per_cpu_cache_dev(cpu));
>> +
>> +	/* Allocate all required memory */
>> +	per_cpu_index_dev(cpu) = kzalloc(sizeof(struct device *) *
>> +					 cache_leaves(cpu), GFP_KERNEL);
>> +	if (unlikely(per_cpu_index_dev(cpu) = NULL))
>> +		goto err_out;
>> +
>> +	return 0;
>> +
>> +err_out:
>> +	cpu_cache_sysfs_exit(cpu);
>> +	return -ENOMEM;
>> +}
>> +
>> +static int cache_add_dev(unsigned int cpu)
>> +{
>> +	unsigned short i;
>> +	int rc;
>> +	struct device *tmp_dev, *parent;
>> +	struct cacheinfo *this_leaf;
>> +	const struct device_attribute **ci_priv_attr;
>> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
>> +
>> +	rc = cpu_cache_sysfs_init(cpu);
>> +	if (unlikely(rc < 0))
>> +		return rc;
>> +
>> +	parent = per_cpu_cache_dev(cpu);
>> +	for (i = 0; i < cache_leaves(cpu); i++) {
>> +		this_leaf = this_cpu_ci->info_list + i;
>> +		if (this_leaf->disable_sysfs)
>> +			continue;
>> +		tmp_dev = device_create_with_groups(parent->class, parent, i,
>> +						    this_leaf,
>> +						    cache_default_groups,
>> +						    "index%1u", i);
>> +		if (IS_ERR_OR_NULL(tmp_dev)) {
>> +			rc = PTR_ERR(tmp_dev);
>> +			goto err;
>> +		}
>> +
>> +		rc = device_add_attrs(tmp_dev, cache_optional_attrs);
>> +		if (unlikely(rc))
>> +			goto err;
>> +
>> +		ci_priv_attr = cache_get_priv_attr(tmp_dev);
>> +		rc = device_add_attrs(tmp_dev, ci_priv_attr);
>> +		if (unlikely(rc))
>> +			goto err;
>
> You just raced with userspace here, creating these files _after_ the
> device was announced to userspace, causing problems with anyone wanting
> to read these attributes :(
>
> I think if you fix up the is_visible() thing above, these calls will go
> away, right?
>

Yes I agree.

Regards,
Sudeep

[PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

Hi Greg,

Thanks for reviewing this.

On 10/07/14 01:09, Greg Kroah-Hartman wrote:
> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>> +static const struct device_attribute *cache_optional_attrs[] = {
>> +	&dev_attr_coherency_line_size,
>> +	&dev_attr_ways_of_associativity,
>> +	&dev_attr_number_of_sets,
>> +	&dev_attr_size,
>> +	&dev_attr_attributes,
>> +	&dev_attr_physical_line_partition,
>> +	NULL
>> +};
>> +
>> +static int device_add_attrs(struct device *dev,
>> +			    const struct device_attribute **dev_attrs)
>> +{
>> +	int i, error = 0;
>> +	struct device_attribute *dev_attr;
>> +	char *buf;
>> +
>> +	if (!dev_attrs)
>> +		return 0;
>> +
>> +	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
>> +	if (!buf)
>> +		return -ENOMEM;
>> +
>> +	for (i = 0; dev_attrs[i]; i++) {
>> +		dev_attr = (struct device_attribute *)dev_attrs[i];
>> +
>> +		/* create attributes that provides meaningful value */
>> +		if (dev_attr->show(dev, dev_attr, buf) < 0)
>> +			continue;
>> +
>> +		error = device_create_file(dev, dev_attrs[i]);
>> +		if (error) {
>> +			while (--i >= 0)
>> +				device_remove_file(dev, dev_attrs[i]);
>> +			break;
>> +		}
>> +	}
>> +
>> +	kfree(buf);
>> +	return error;
>> +}
>
> Ick, why create your own function for this when the driver core has this
> functionality built into it?  Look at the is_visible() callback, and how
> it is use for an attribute group please.
>

I agree even I added this function hesitantly as didn't realize that I can use
is_visible for this purpose. Thanks for pointing that out I will have a look
at it.

>> +static void device_remove_attrs(struct device *dev,
>> +				const struct device_attribute **dev_attrs)
>> +{
>> +	int i;
>> +
>> +	if (!dev_attrs)
>> +		return;
>> +
>> +	for (i = 0; dev_attrs[i]; dev_attrs++, i++)
>> +		device_remove_file(dev, dev_attrs[i]);
>> +}
>
> You should just remove a whole group at once, not individually.
>

Right, I must be able to get rid of these 2 functions once I use
is_visible callback.

>> +
>> +const struct device_attribute **
>> +__weak cache_get_priv_attr(struct device *cache_idx_dev)
>> +{
>> +	return NULL;
>> +}
>> +
>> +/* Add/Remove cache interface for CPU device */
>> +static void cpu_cache_sysfs_exit(unsigned int cpu)
>> +{
>> +	int i;
>> +	struct device *tmp_dev;
>> +	const struct device_attribute **ci_priv_attr;
>> +
>> +	if (per_cpu_index_dev(cpu)) {
>> +		for (i = 0; i < cache_leaves(cpu); i++) {
>> +			tmp_dev = per_cache_index_dev(cpu, i);
>> +			if (!tmp_dev)
>> +				continue;
>> +			ci_priv_attr = cache_get_priv_attr(tmp_dev);
>> +			device_remove_attrs(tmp_dev, ci_priv_attr);
>> +			device_remove_attrs(tmp_dev, cache_optional_attrs);
>> +			device_unregister(tmp_dev);
>> +		}
>> +		kfree(per_cpu_index_dev(cpu));
>> +		per_cpu_index_dev(cpu) = NULL;
>> +	}
>> +	device_unregister(per_cpu_cache_dev(cpu));
>> +	per_cpu_cache_dev(cpu) = NULL;
>> +}
>> +
>> +static int cpu_cache_sysfs_init(unsigned int cpu)
>> +{
>> +	struct device *dev = get_cpu_device(cpu);
>> +
>> +	if (per_cpu_cacheinfo(cpu) == NULL)
>> +		return -ENOENT;
>> +
>> +	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
>> +					       NULL, "cache");
>> +	if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
>> +		return PTR_ERR(per_cpu_cache_dev(cpu));
>> +
>> +	/* Allocate all required memory */
>> +	per_cpu_index_dev(cpu) = kzalloc(sizeof(struct device *) *
>> +					 cache_leaves(cpu), GFP_KERNEL);
>> +	if (unlikely(per_cpu_index_dev(cpu) == NULL))
>> +		goto err_out;
>> +
>> +	return 0;
>> +
>> +err_out:
>> +	cpu_cache_sysfs_exit(cpu);
>> +	return -ENOMEM;
>> +}
>> +
>> +static int cache_add_dev(unsigned int cpu)
>> +{
>> +	unsigned short i;
>> +	int rc;
>> +	struct device *tmp_dev, *parent;
>> +	struct cacheinfo *this_leaf;
>> +	const struct device_attribute **ci_priv_attr;
>> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
>> +
>> +	rc = cpu_cache_sysfs_init(cpu);
>> +	if (unlikely(rc < 0))
>> +		return rc;
>> +
>> +	parent = per_cpu_cache_dev(cpu);
>> +	for (i = 0; i < cache_leaves(cpu); i++) {
>> +		this_leaf = this_cpu_ci->info_list + i;
>> +		if (this_leaf->disable_sysfs)
>> +			continue;
>> +		tmp_dev = device_create_with_groups(parent->class, parent, i,
>> +						    this_leaf,
>> +						    cache_default_groups,
>> +						    "index%1u", i);
>> +		if (IS_ERR_OR_NULL(tmp_dev)) {
>> +			rc = PTR_ERR(tmp_dev);
>> +			goto err;
>> +		}
>> +
>> +		rc = device_add_attrs(tmp_dev, cache_optional_attrs);
>> +		if (unlikely(rc))
>> +			goto err;
>> +
>> +		ci_priv_attr = cache_get_priv_attr(tmp_dev);
>> +		rc = device_add_attrs(tmp_dev, ci_priv_attr);
>> +		if (unlikely(rc))
>> +			goto err;
>
> You just raced with userspace here, creating these files _after_ the
> device was announced to userspace, causing problems with anyone wanting
> to read these attributes :(
>
> I think if you fix up the is_visible() thing above, these calls will go
> away, right?
>

Yes I agree.

Regards,
Sudeep

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

Hi Greg,

Thanks for reviewing this.

On 10/07/14 01:09, Greg Kroah-Hartman wrote:
> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>> +static const struct device_attribute *cache_optional_attrs[] =3D {
>> +=09&dev_attr_coherency_line_size,
>> +=09&dev_attr_ways_of_associativity,
>> +=09&dev_attr_number_of_sets,
>> +=09&dev_attr_size,
>> +=09&dev_attr_attributes,
>> +=09&dev_attr_physical_line_partition,
>> +=09NULL
>> +};
>> +
>> +static int device_add_attrs(struct device *dev,
>> +=09=09=09    const struct device_attribute **dev_attrs)
>> +{
>> +=09int i, error =3D 0;
>> +=09struct device_attribute *dev_attr;
>> +=09char *buf;
>> +
>> +=09if (!dev_attrs)
>> +=09=09return 0;
>> +
>> +=09buf =3D kmalloc(PAGE_SIZE, GFP_KERNEL);
>> +=09if (!buf)
>> +=09=09return -ENOMEM;
>> +
>> +=09for (i =3D 0; dev_attrs[i]; i++) {
>> +=09=09dev_attr =3D (struct device_attribute *)dev_attrs[i];
>> +
>> +=09=09/* create attributes that provides meaningful value */
>> +=09=09if (dev_attr->show(dev, dev_attr, buf) < 0)
>> +=09=09=09continue;
>> +
>> +=09=09error =3D device_create_file(dev, dev_attrs[i]);
>> +=09=09if (error) {
>> +=09=09=09while (--i >=3D 0)
>> +=09=09=09=09device_remove_file(dev, dev_attrs[i]);
>> +=09=09=09break;
>> +=09=09}
>> +=09}
>> +
>> +=09kfree(buf);
>> +=09return error;
>> +}
>
> Ick, why create your own function for this when the driver core has this
> functionality built into it?  Look at the is_visible() callback, and how
> it is use for an attribute group please.
>

I agree even I added this function hesitantly as didn't realize that I can =
use
is_visible for this purpose. Thanks for pointing that out I will have a loo=
k
at it.

>> +static void device_remove_attrs(struct device *dev,
>> +=09=09=09=09const struct device_attribute **dev_attrs)
>> +{
>> +=09int i;
>> +
>> +=09if (!dev_attrs)
>> +=09=09return;
>> +
>> +=09for (i =3D 0; dev_attrs[i]; dev_attrs++, i++)
>> +=09=09device_remove_file(dev, dev_attrs[i]);
>> +}
>
> You should just remove a whole group at once, not individually.
>

Right, I must be able to get rid of these 2 functions once I use
is_visible callback.

>> +
>> +const struct device_attribute **
>> +__weak cache_get_priv_attr(struct device *cache_idx_dev)
>> +{
>> +=09return NULL;
>> +}
>> +
>> +/* Add/Remove cache interface for CPU device */
>> +static void cpu_cache_sysfs_exit(unsigned int cpu)
>> +{
>> +=09int i;
>> +=09struct device *tmp_dev;
>> +=09const struct device_attribute **ci_priv_attr;
>> +
>> +=09if (per_cpu_index_dev(cpu)) {
>> +=09=09for (i =3D 0; i < cache_leaves(cpu); i++) {
>> +=09=09=09tmp_dev =3D per_cache_index_dev(cpu, i);
>> +=09=09=09if (!tmp_dev)
>> +=09=09=09=09continue;
>> +=09=09=09ci_priv_attr =3D cache_get_priv_attr(tmp_dev);
>> +=09=09=09device_remove_attrs(tmp_dev, ci_priv_attr);
>> +=09=09=09device_remove_attrs(tmp_dev, cache_optional_attrs);
>> +=09=09=09device_unregister(tmp_dev);
>> +=09=09}
>> +=09=09kfree(per_cpu_index_dev(cpu));
>> +=09=09per_cpu_index_dev(cpu) =3D NULL;
>> +=09}
>> +=09device_unregister(per_cpu_cache_dev(cpu));
>> +=09per_cpu_cache_dev(cpu) =3D NULL;
>> +}
>> +
>> +static int cpu_cache_sysfs_init(unsigned int cpu)
>> +{
>> +=09struct device *dev =3D get_cpu_device(cpu);
>> +
>> +=09if (per_cpu_cacheinfo(cpu) =3D=3D NULL)
>> +=09=09return -ENOENT;
>> +
>> +=09per_cpu_cache_dev(cpu) =3D device_create(dev->class, dev, cpu,
>> +=09=09=09=09=09       NULL, "cache");
>> +=09if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
>> +=09=09return PTR_ERR(per_cpu_cache_dev(cpu));
>> +
>> +=09/* Allocate all required memory */
>> +=09per_cpu_index_dev(cpu) =3D kzalloc(sizeof(struct device *) *
>> +=09=09=09=09=09 cache_leaves(cpu), GFP_KERNEL);
>> +=09if (unlikely(per_cpu_index_dev(cpu) =3D=3D NULL))
>> +=09=09goto err_out;
>> +
>> +=09return 0;
>> +
>> +err_out:
>> +=09cpu_cache_sysfs_exit(cpu);
>> +=09return -ENOMEM;
>> +}
>> +
>> +static int cache_add_dev(unsigned int cpu)
>> +{
>> +=09unsigned short i;
>> +=09int rc;
>> +=09struct device *tmp_dev, *parent;
>> +=09struct cacheinfo *this_leaf;
>> +=09const struct device_attribute **ci_priv_attr;
>> +=09struct cpu_cacheinfo *this_cpu_ci =3D get_cpu_cacheinfo(cpu);
>> +
>> +=09rc =3D cpu_cache_sysfs_init(cpu);
>> +=09if (unlikely(rc < 0))
>> +=09=09return rc;
>> +
>> +=09parent =3D per_cpu_cache_dev(cpu);
>> +=09for (i =3D 0; i < cache_leaves(cpu); i++) {
>> +=09=09this_leaf =3D this_cpu_ci->info_list + i;
>> +=09=09if (this_leaf->disable_sysfs)
>> +=09=09=09continue;
>> +=09=09tmp_dev =3D device_create_with_groups(parent->class, parent, i,
>> +=09=09=09=09=09=09    this_leaf,
>> +=09=09=09=09=09=09    cache_default_groups,
>> +=09=09=09=09=09=09    "index%1u", i);
>> +=09=09if (IS_ERR_OR_NULL(tmp_dev)) {
>> +=09=09=09rc =3D PTR_ERR(tmp_dev);
>> +=09=09=09goto err;
>> +=09=09}
>> +
>> +=09=09rc =3D device_add_attrs(tmp_dev, cache_optional_attrs);
>> +=09=09if (unlikely(rc))
>> +=09=09=09goto err;
>> +
>> +=09=09ci_priv_attr =3D cache_get_priv_attr(tmp_dev);
>> +=09=09rc =3D device_add_attrs(tmp_dev, ci_priv_attr);
>> +=09=09if (unlikely(rc))
>> +=09=09=09goto err;
>
> You just raced with userspace here, creating these files _after_ the
> device was announced to userspace, causing problems with anyone wanting
> to read these attributes :(
>
> I think if you fix up the is_visible() thing above, these calls will go
> away, right?
>

Yes I agree.

Regards,
Sudeep

Re: [PATCH 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

Hi Greg,

Thanks for reviewing this.

On 10/07/14 01:09, Greg Kroah-Hartman wrote:
> On Wed, Jun 25, 2014 at 06:30:37PM +0100, Sudeep Holla wrote:
>> +static const struct device_attribute *cache_optional_attrs[] = {
>> +	&dev_attr_coherency_line_size,
>> +	&dev_attr_ways_of_associativity,
>> +	&dev_attr_number_of_sets,
>> +	&dev_attr_size,
>> +	&dev_attr_attributes,
>> +	&dev_attr_physical_line_partition,
>> +	NULL
>> +};
>> +
>> +static int device_add_attrs(struct device *dev,
>> +			    const struct device_attribute **dev_attrs)
>> +{
>> +	int i, error = 0;
>> +	struct device_attribute *dev_attr;
>> +	char *buf;
>> +
>> +	if (!dev_attrs)
>> +		return 0;
>> +
>> +	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
>> +	if (!buf)
>> +		return -ENOMEM;
>> +
>> +	for (i = 0; dev_attrs[i]; i++) {
>> +		dev_attr = (struct device_attribute *)dev_attrs[i];
>> +
>> +		/* create attributes that provides meaningful value */
>> +		if (dev_attr->show(dev, dev_attr, buf) < 0)
>> +			continue;
>> +
>> +		error = device_create_file(dev, dev_attrs[i]);
>> +		if (error) {
>> +			while (--i >= 0)
>> +				device_remove_file(dev, dev_attrs[i]);
>> +			break;
>> +		}
>> +	}
>> +
>> +	kfree(buf);
>> +	return error;
>> +}
>
> Ick, why create your own function for this when the driver core has this
> functionality built into it?  Look at the is_visible() callback, and how
> it is use for an attribute group please.
>

I agree even I added this function hesitantly as didn't realize that I can use
is_visible for this purpose. Thanks for pointing that out I will have a look
at it.

>> +static void device_remove_attrs(struct device *dev,
>> +				const struct device_attribute **dev_attrs)
>> +{
>> +	int i;
>> +
>> +	if (!dev_attrs)
>> +		return;
>> +
>> +	for (i = 0; dev_attrs[i]; dev_attrs++, i++)
>> +		device_remove_file(dev, dev_attrs[i]);
>> +}
>
> You should just remove a whole group at once, not individually.
>

Right, I must be able to get rid of these 2 functions once I use
is_visible callback.

>> +
>> +const struct device_attribute **
>> +__weak cache_get_priv_attr(struct device *cache_idx_dev)
>> +{
>> +	return NULL;
>> +}
>> +
>> +/* Add/Remove cache interface for CPU device */
>> +static void cpu_cache_sysfs_exit(unsigned int cpu)
>> +{
>> +	int i;
>> +	struct device *tmp_dev;
>> +	const struct device_attribute **ci_priv_attr;
>> +
>> +	if (per_cpu_index_dev(cpu)) {
>> +		for (i = 0; i < cache_leaves(cpu); i++) {
>> +			tmp_dev = per_cache_index_dev(cpu, i);
>> +			if (!tmp_dev)
>> +				continue;
>> +			ci_priv_attr = cache_get_priv_attr(tmp_dev);
>> +			device_remove_attrs(tmp_dev, ci_priv_attr);
>> +			device_remove_attrs(tmp_dev, cache_optional_attrs);
>> +			device_unregister(tmp_dev);
>> +		}
>> +		kfree(per_cpu_index_dev(cpu));
>> +		per_cpu_index_dev(cpu) = NULL;
>> +	}
>> +	device_unregister(per_cpu_cache_dev(cpu));
>> +	per_cpu_cache_dev(cpu) = NULL;
>> +}
>> +
>> +static int cpu_cache_sysfs_init(unsigned int cpu)
>> +{
>> +	struct device *dev = get_cpu_device(cpu);
>> +
>> +	if (per_cpu_cacheinfo(cpu) == NULL)
>> +		return -ENOENT;
>> +
>> +	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
>> +					       NULL, "cache");
>> +	if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
>> +		return PTR_ERR(per_cpu_cache_dev(cpu));
>> +
>> +	/* Allocate all required memory */
>> +	per_cpu_index_dev(cpu) = kzalloc(sizeof(struct device *) *
>> +					 cache_leaves(cpu), GFP_KERNEL);
>> +	if (unlikely(per_cpu_index_dev(cpu) == NULL))
>> +		goto err_out;
>> +
>> +	return 0;
>> +
>> +err_out:
>> +	cpu_cache_sysfs_exit(cpu);
>> +	return -ENOMEM;
>> +}
>> +
>> +static int cache_add_dev(unsigned int cpu)
>> +{
>> +	unsigned short i;
>> +	int rc;
>> +	struct device *tmp_dev, *parent;
>> +	struct cacheinfo *this_leaf;
>> +	const struct device_attribute **ci_priv_attr;
>> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
>> +
>> +	rc = cpu_cache_sysfs_init(cpu);
>> +	if (unlikely(rc < 0))
>> +		return rc;
>> +
>> +	parent = per_cpu_cache_dev(cpu);
>> +	for (i = 0; i < cache_leaves(cpu); i++) {
>> +		this_leaf = this_cpu_ci->info_list + i;
>> +		if (this_leaf->disable_sysfs)
>> +			continue;
>> +		tmp_dev = device_create_with_groups(parent->class, parent, i,
>> +						    this_leaf,
>> +						    cache_default_groups,
>> +						    "index%1u", i);
>> +		if (IS_ERR_OR_NULL(tmp_dev)) {
>> +			rc = PTR_ERR(tmp_dev);
>> +			goto err;
>> +		}
>> +
>> +		rc = device_add_attrs(tmp_dev, cache_optional_attrs);
>> +		if (unlikely(rc))
>> +			goto err;
>> +
>> +		ci_priv_attr = cache_get_priv_attr(tmp_dev);
>> +		rc = device_add_attrs(tmp_dev, ci_priv_attr);
>> +		if (unlikely(rc))
>> +			goto err;
>
> You just raced with userspace here, creating these files _after_ the
> device was announced to userspace, causing problems with anyone wanting
> to read these attributes :(
>
> I think if you fix up the is_visible() thing above, these calls will go
> away, right?
>

Yes I agree.

Regards,
Sudeep

[PATCH 3/9] ia64: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: linux-ia64@vger.kernel.org
---
 arch/ia64/kernel/topology.c | 401 +++++++++-----------------------------------
 1 file changed, 79 insertions(+), 322 deletions(-)

diff --git a/arch/ia64/kernel/topology.c b/arch/ia64/kernel/topology.c
index f295f9a..48dc46d 100644
--- a/arch/ia64/kernel/topology.c
+++ b/arch/ia64/kernel/topology.c
@@ -13,6 +13,7 @@
  *	Populate cpu cache entries in sysfs for cpu cache info
  */
 
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
 #include <linux/kernel.h>
 #include <linux/mm.h>
@@ -21,7 +22,6 @@
 #include <linux/init.h>
 #include <linux/bootmem.h>
 #include <linux/nodemask.h>
-#include <linux/notifier.h>
 #include <linux/export.h>
 #include <asm/mmzone.h>
 #include <asm/numa.h>
@@ -103,60 +103,25 @@ subsys_initcall(topology_init);
 /*
  * Export cpu cache information through sysfs
  */
-
-/*
- *  A bunch of string array to get pretty printing
- */
-static const char *cache_types[] = {
-	"",			/* not used */
-	"Instruction",
-	"Data",
-	"Unified"	/* unified */
-};
-
-static const char *cache_mattrib[]={
-	"WriteThrough",
-	"WriteBack",
-	"",		/* reserved */
-	""		/* reserved */
-};
-
-struct cache_info {
-	pal_cache_config_info_t	cci;
-	cpumask_t shared_cpu_map;
-	int level;
-	int type;
-	struct kobject kobj;
-};
-
-struct cpu_cache_info {
-	struct cache_info *cache_leaves;
-	int	num_cache_leaves;
-	struct kobject kobj;
-};
-
-static struct cpu_cache_info	all_cpu_cache_info[NR_CPUS];
-#define LEAF_KOBJECT_PTR(x,y)    (&all_cpu_cache_info[x].cache_leaves[y])
-
 #ifdef CONFIG_SMP
-static void cache_shared_cpu_map_setup(unsigned int cpu,
-		struct cache_info * this_leaf)
+static int __cache_cpumap_setup(unsigned int cpu, struct cacheinfo *this_leaf)
 {
 	pal_cache_shared_info_t	csi;
-	int num_shared, i = 0;
-	unsigned int j;
+	int num_shared, i = 0, j;
+	enum cache_type type = this_leaf->type;
 
 	if (cpu_data(cpu)->threads_per_core <= 1 &&
 		cpu_data(cpu)->cores_per_socket <= 1) {
 		cpu_set(cpu, this_leaf->shared_cpu_map);
-		return;
+		return 0;
 	}
 
-	if (ia64_pal_cache_shared_info(this_leaf->level,
-					this_leaf->type,
-					0,
-					&csi) != PAL_STATUS_SUCCESS)
-		return;
+	if (type == CACHE_TYPE_UNIFIED)
+		type = CACHE_TYPE_DATA;
+
+	if (ia64_pal_cache_shared_info(this_leaf->level, type, 0,
+				       &csi) != PAL_STATUS_SUCCESS)
+		return -EIO;
 
 	num_shared = (int) csi.num_shared;
 	do {
@@ -168,305 +133,97 @@ static void cache_shared_cpu_map_setup(unsigned int cpu,
 
 		i++;
 	} while (i < num_shared &&
-		ia64_pal_cache_shared_info(this_leaf->level,
-				this_leaf->type,
-				i,
-				&csi) == PAL_STATUS_SUCCESS);
-}
-#else
-static void cache_shared_cpu_map_setup(unsigned int cpu,
-		struct cache_info * this_leaf)
-{
-	cpu_set(cpu, this_leaf->shared_cpu_map);
-	return;
-}
-#endif
-
-static ssize_t show_coherency_line_size(struct cache_info *this_leaf,
-					char *buf)
-{
-	return sprintf(buf, "%u\n", 1 << this_leaf->cci.pcci_line_size);
-}
-
-static ssize_t show_ways_of_associativity(struct cache_info *this_leaf,
-					char *buf)
-{
-	return sprintf(buf, "%u\n", this_leaf->cci.pcci_assoc);
-}
-
-static ssize_t show_attributes(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf,
-			"%s\n",
-			cache_mattrib[this_leaf->cci.pcci_cache_attr]);
-}
-
-static ssize_t show_size(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf, "%uK\n", this_leaf->cci.pcci_cache_size / 1024);
-}
-
-static ssize_t show_number_of_sets(struct cache_info *this_leaf, char *buf)
-{
-	unsigned number_of_sets = this_leaf->cci.pcci_cache_size;
-	number_of_sets /= this_leaf->cci.pcci_assoc;
-	number_of_sets /= 1 << this_leaf->cci.pcci_line_size;
-
-	return sprintf(buf, "%u\n", number_of_sets);
-}
-
-static ssize_t show_shared_cpu_map(struct cache_info *this_leaf, char *buf)
-{
-	ssize_t	len;
-	cpumask_t shared_cpu_map;
-
-	cpumask_and(&shared_cpu_map,
-				&this_leaf->shared_cpu_map, cpu_online_mask);
-	len = cpumask_scnprintf(buf, NR_CPUS+1, &shared_cpu_map);
-	len += sprintf(buf+len, "\n");
-	return len;
-}
-
-static ssize_t show_type(struct cache_info *this_leaf, char *buf)
-{
-	int type = this_leaf->type + this_leaf->cci.pcci_unified;
-	return sprintf(buf, "%s\n", cache_types[type]);
-}
-
-static ssize_t show_level(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf, "%u\n", this_leaf->level);
+		 ia64_pal_cache_shared_info(this_leaf->level, type, i,
+					    &csi) == PAL_STATUS_SUCCESS);
+	return 0;
 }
 
-struct cache_attr {
-	struct attribute attr;
-	ssize_t (*show)(struct cache_info *, char *);
-	ssize_t (*store)(struct cache_info *, const char *, size_t count);
-};
-
-#ifdef define_one_ro
-	#undef define_one_ro
-#endif
-#define define_one_ro(_name) \
-	static struct cache_attr _name = \
-__ATTR(_name, 0444, show_##_name, NULL)
-
-define_one_ro(level);
-define_one_ro(type);
-define_one_ro(coherency_line_size);
-define_one_ro(ways_of_associativity);
-define_one_ro(size);
-define_one_ro(number_of_sets);
-define_one_ro(shared_cpu_map);
-define_one_ro(attributes);
-
-static struct attribute * cache_default_attrs[] = {
-	&type.attr,
-	&level.attr,
-	&coherency_line_size.attr,
-	&ways_of_associativity.attr,
-	&attributes.attr,
-	&size.attr,
-	&number_of_sets.attr,
-	&shared_cpu_map.attr,
-	NULL
-};
-
-#define to_object(k) container_of(k, struct cache_info, kobj)
-#define to_attr(a) container_of(a, struct cache_attr, attr)
-
-static ssize_t ia64_cache_show(struct kobject * kobj, struct attribute * attr, char * buf)
+static int cache_shared_cpu_map_setup(unsigned int cpu)
 {
-	struct cache_attr *fattr = to_attr(attr);
-	struct cache_info *this_leaf = to_object(kobj);
-	ssize_t ret;
+	unsigned int idx;
+	int ret = 0;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
 
-	ret = fattr->show ? fattr->show(this_leaf, buf) : 0;
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++, this_leaf++) {
+		ret = __cache_cpumap_setup(cpu, this_leaf);
+		if (ret)
+			break;
+	}
 	return ret;
 }
-
-static const struct sysfs_ops cache_sysfs_ops = {
-	.show   = ia64_cache_show
-};
-
-static struct kobj_type cache_ktype = {
-	.sysfs_ops	= &cache_sysfs_ops,
-	.default_attrs	= cache_default_attrs,
-};
-
-static struct kobj_type cache_ktype_percpu_entry = {
-	.sysfs_ops	= &cache_sysfs_ops,
-};
-
-static void cpu_cache_sysfs_exit(unsigned int cpu)
-{
-	kfree(all_cpu_cache_info[cpu].cache_leaves);
-	all_cpu_cache_info[cpu].cache_leaves = NULL;
-	all_cpu_cache_info[cpu].num_cache_leaves = 0;
-	memset(&all_cpu_cache_info[cpu].kobj, 0, sizeof(struct kobject));
-	return;
-}
-
-static int cpu_cache_sysfs_init(unsigned int cpu)
+#else
+static int cache_shared_cpu_map_setup(unsigned int cpu)
 {
-	unsigned long i, levels, unique_caches;
-	pal_cache_config_info_t cci;
-	int j;
-	long status;
-	struct cache_info *this_cache;
-	int num_cache_leaves = 0;
-
-	if ((status = ia64_pal_cache_summary(&levels, &unique_caches)) != 0) {
-		printk(KERN_ERR "ia64_pal_cache_summary=%ld\n", status);
-		return -1;
-	}
-
-	this_cache=kzalloc(sizeof(struct cache_info)*unique_caches,
-			GFP_KERNEL);
-	if (this_cache == NULL)
-		return -ENOMEM;
-
-	for (i=0; i < levels; i++) {
-		for (j=2; j >0 ; j--) {
-			if ((status=ia64_pal_cache_config_info(i,j, &cci)) !=
-					PAL_STATUS_SUCCESS)
-				continue;
-
-			this_cache[num_cache_leaves].cci = cci;
-			this_cache[num_cache_leaves].level = i + 1;
-			this_cache[num_cache_leaves].type = j;
-
-			cache_shared_cpu_map_setup(cpu,
-					&this_cache[num_cache_leaves]);
-			num_cache_leaves ++;
-		}
-	}
-
-	all_cpu_cache_info[cpu].cache_leaves = this_cache;
-	all_cpu_cache_info[cpu].num_cache_leaves = num_cache_leaves;
-
-	memset(&all_cpu_cache_info[cpu].kobj, 0, sizeof(struct kobject));
+	int idx;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
 
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++, this_leaf++)
+		cpu_set(cpu, this_leaf->shared_cpu_map);
 	return 0;
 }
+#endif
 
-/* Add cache interface for CPU device */
-static int cache_add_dev(struct device *sys_dev)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 pal_cache_config_info_t *cci,
+			 enum cache_type type, unsigned int level)
 {
-	unsigned int cpu = sys_dev->id;
-	unsigned long i, j;
-	struct cache_info *this_object;
-	int retval = 0;
-	cpumask_t oldmask;
-
-	if (all_cpu_cache_info[cpu].kobj.parent)
-		return 0;
-
-	oldmask = current->cpus_allowed;
-	retval = set_cpus_allowed_ptr(current, cpumask_of(cpu));
-	if (unlikely(retval))
-		return retval;
+	unsigned number_of_sets = cci->pcci_cache_size;
 
-	retval = cpu_cache_sysfs_init(cpu);
-	set_cpus_allowed_ptr(current, &oldmask);
-	if (unlikely(retval < 0))
-		return retval;
-
-	retval = kobject_init_and_add(&all_cpu_cache_info[cpu].kobj,
-				      &cache_ktype_percpu_entry, &sys_dev->kobj,
-				      "%s", "cache");
-	if (unlikely(retval < 0)) {
-		cpu_cache_sysfs_exit(cpu);
-		return retval;
-	}
-
-	for (i = 0; i < all_cpu_cache_info[cpu].num_cache_leaves; i++) {
-		this_object = LEAF_KOBJECT_PTR(cpu,i);
-		retval = kobject_init_and_add(&(this_object->kobj),
-					      &cache_ktype,
-					      &all_cpu_cache_info[cpu].kobj,
-					      "index%1lu", i);
-		if (unlikely(retval)) {
-			for (j = 0; j < i; j++) {
-				kobject_put(&(LEAF_KOBJECT_PTR(cpu,j)->kobj));
-			}
-			kobject_put(&all_cpu_cache_info[cpu].kobj);
-			cpu_cache_sysfs_exit(cpu);
-			return retval;
-		}
-		kobject_uevent(&(this_object->kobj), KOBJ_ADD);
-	}
-	kobject_uevent(&all_cpu_cache_info[cpu].kobj, KOBJ_ADD);
-	return retval;
+	number_of_sets /= cci->pcci_assoc;
+	number_of_sets /= 1 << cci->pcci_line_size;
+	this_leaf->level = level;
+	this_leaf->type = cci->pcci_unified ? CACHE_TYPE_UNIFIED : type;
+	this_leaf->coherency_line_size = cci->pcci_line_size;
+	this_leaf->ways_of_associativity = cci->pcci_assoc;
+	this_leaf->size = cci->pcci_cache_size;
+	this_leaf->attributes = cci->pcci_cache_attr;
+	this_leaf->number_of_sets = number_of_sets;
 }
 
-/* Remove cache interface for CPU device */
-static int cache_remove_dev(struct device *sys_dev)
+int init_cache_level(unsigned int cpu)
 {
-	unsigned int cpu = sys_dev->id;
-	unsigned long i;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned long levels, unique_caches;
+	long status;
 
-	for (i = 0; i < all_cpu_cache_info[cpu].num_cache_leaves; i++)
-		kobject_put(&(LEAF_KOBJECT_PTR(cpu,i)->kobj));
+	if (!this_cpu_ci)
+		return -EINVAL;
 
-	if (all_cpu_cache_info[cpu].kobj.parent) {
-		kobject_put(&all_cpu_cache_info[cpu].kobj);
-		memset(&all_cpu_cache_info[cpu].kobj,
-			0,
-			sizeof(struct kobject));
+	status = ia64_pal_cache_summary(&levels, &unique_caches);
+	if (status != PAL_STATUS_SUCCESS) {
+		pr_err("ia64_pal_cache_summary = %ld\n", status);
+		return -EIO;
 	}
-
-	cpu_cache_sysfs_exit(cpu);
+	this_cpu_ci->num_levels = levels;
+	this_cpu_ci->num_leaves = unique_caches;
 
 	return 0;
 }
 
-/*
- * When a cpu is hot-plugged, do a check and initiate
- * cache kobject if necessary
- */
-static int cache_cpu_callback(struct notifier_block *nfb,
-		unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (unsigned long)hcpu;
-	struct device *sys_dev;
-
-	sys_dev = get_cpu_device(cpu);
-	switch (action) {
-	case CPU_ONLINE:
-	case CPU_ONLINE_FROZEN:
-		cache_add_dev(sys_dev);
-		break;
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		cache_remove_dev(sys_dev);
-		break;
-	}
-	return NOTIFY_OK;
-}
-
-static struct notifier_block cache_cpu_notifier =
-{
-	.notifier_call = cache_cpu_callback
-};
-
-static int __init cache_sysfs_init(void)
+int populate_cache_leaves(unsigned int cpu)
 {
-	int i;
-
-	cpu_notifier_register_begin();
-
-	for_each_online_cpu(i) {
-		struct device *sys_dev = get_cpu_device((unsigned int)i);
-		cache_add_dev(sys_dev);
+	unsigned int level, idx;
+	s64 status;
+	pal_cache_config_info_t cci;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		type = CACHE_TYPE_INST;
+		status = ia64_pal_cache_config_info(level - 1, type, &cci);
+		if (status == PAL_STATUS_SUCCESS)
+			ci_leaf_init(this_leaf++, &cci, type, level);
+		type = CACHE_TYPE_DATA;
+		status = ia64_pal_cache_config_info(level - 1, type, &cci);
+		if (status == PAL_STATUS_SUCCESS)
+			ci_leaf_init(this_leaf++, &cci, type, level);
 	}
-
-	__register_hotcpu_notifier(&cache_cpu_notifier);
-
-	cpu_notifier_register_done();
-
-	return 0;
+	return cache_shared_cpu_map_setup(cpu);
 }
-
-device_initcall(cache_sysfs_init);
-
-- 
1.8.3.2

[PATCH 3/9] ia64: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: linux-ia64@vger.kernel.org
---
 arch/ia64/kernel/topology.c | 401 +++++++++-----------------------------------
 1 file changed, 79 insertions(+), 322 deletions(-)

diff --git a/arch/ia64/kernel/topology.c b/arch/ia64/kernel/topology.c
index f295f9a..48dc46d 100644
--- a/arch/ia64/kernel/topology.c
+++ b/arch/ia64/kernel/topology.c
@@ -13,6 +13,7 @@
  *	Populate cpu cache entries in sysfs for cpu cache info
  */
 
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
 #include <linux/kernel.h>
 #include <linux/mm.h>
@@ -21,7 +22,6 @@
 #include <linux/init.h>
 #include <linux/bootmem.h>
 #include <linux/nodemask.h>
-#include <linux/notifier.h>
 #include <linux/export.h>
 #include <asm/mmzone.h>
 #include <asm/numa.h>
@@ -103,60 +103,25 @@ subsys_initcall(topology_init);
 /*
  * Export cpu cache information through sysfs
  */
-
-/*
- *  A bunch of string array to get pretty printing
- */
-static const char *cache_types[] = {
-	"",			/* not used */
-	"Instruction",
-	"Data",
-	"Unified"	/* unified */
-};
-
-static const char *cache_mattrib[]={
-	"WriteThrough",
-	"WriteBack",
-	"",		/* reserved */
-	""		/* reserved */
-};
-
-struct cache_info {
-	pal_cache_config_info_t	cci;
-	cpumask_t shared_cpu_map;
-	int level;
-	int type;
-	struct kobject kobj;
-};
-
-struct cpu_cache_info {
-	struct cache_info *cache_leaves;
-	int	num_cache_leaves;
-	struct kobject kobj;
-};
-
-static struct cpu_cache_info	all_cpu_cache_info[NR_CPUS];
-#define LEAF_KOBJECT_PTR(x,y)    (&all_cpu_cache_info[x].cache_leaves[y])
-
 #ifdef CONFIG_SMP
-static void cache_shared_cpu_map_setup(unsigned int cpu,
-		struct cache_info * this_leaf)
+static int __cache_cpumap_setup(unsigned int cpu, struct cacheinfo *this_leaf)
 {
 	pal_cache_shared_info_t	csi;
-	int num_shared, i = 0;
-	unsigned int j;
+	int num_shared, i = 0, j;
+	enum cache_type type = this_leaf->type;
 
 	if (cpu_data(cpu)->threads_per_core <= 1 &&
 		cpu_data(cpu)->cores_per_socket <= 1) {
 		cpu_set(cpu, this_leaf->shared_cpu_map);
-		return;
+		return 0;
 	}
 
-	if (ia64_pal_cache_shared_info(this_leaf->level,
-					this_leaf->type,
-					0,
-					&csi) != PAL_STATUS_SUCCESS)
-		return;
+	if (type = CACHE_TYPE_UNIFIED)
+		type = CACHE_TYPE_DATA;
+
+	if (ia64_pal_cache_shared_info(this_leaf->level, type, 0,
+				       &csi) != PAL_STATUS_SUCCESS)
+		return -EIO;
 
 	num_shared = (int) csi.num_shared;
 	do {
@@ -168,305 +133,97 @@ static void cache_shared_cpu_map_setup(unsigned int cpu,
 
 		i++;
 	} while (i < num_shared &&
-		ia64_pal_cache_shared_info(this_leaf->level,
-				this_leaf->type,
-				i,
-				&csi) = PAL_STATUS_SUCCESS);
-}
-#else
-static void cache_shared_cpu_map_setup(unsigned int cpu,
-		struct cache_info * this_leaf)
-{
-	cpu_set(cpu, this_leaf->shared_cpu_map);
-	return;
-}
-#endif
-
-static ssize_t show_coherency_line_size(struct cache_info *this_leaf,
-					char *buf)
-{
-	return sprintf(buf, "%u\n", 1 << this_leaf->cci.pcci_line_size);
-}
-
-static ssize_t show_ways_of_associativity(struct cache_info *this_leaf,
-					char *buf)
-{
-	return sprintf(buf, "%u\n", this_leaf->cci.pcci_assoc);
-}
-
-static ssize_t show_attributes(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf,
-			"%s\n",
-			cache_mattrib[this_leaf->cci.pcci_cache_attr]);
-}
-
-static ssize_t show_size(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf, "%uK\n", this_leaf->cci.pcci_cache_size / 1024);
-}
-
-static ssize_t show_number_of_sets(struct cache_info *this_leaf, char *buf)
-{
-	unsigned number_of_sets = this_leaf->cci.pcci_cache_size;
-	number_of_sets /= this_leaf->cci.pcci_assoc;
-	number_of_sets /= 1 << this_leaf->cci.pcci_line_size;
-
-	return sprintf(buf, "%u\n", number_of_sets);
-}
-
-static ssize_t show_shared_cpu_map(struct cache_info *this_leaf, char *buf)
-{
-	ssize_t	len;
-	cpumask_t shared_cpu_map;
-
-	cpumask_and(&shared_cpu_map,
-				&this_leaf->shared_cpu_map, cpu_online_mask);
-	len = cpumask_scnprintf(buf, NR_CPUS+1, &shared_cpu_map);
-	len += sprintf(buf+len, "\n");
-	return len;
-}
-
-static ssize_t show_type(struct cache_info *this_leaf, char *buf)
-{
-	int type = this_leaf->type + this_leaf->cci.pcci_unified;
-	return sprintf(buf, "%s\n", cache_types[type]);
-}
-
-static ssize_t show_level(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf, "%u\n", this_leaf->level);
+		 ia64_pal_cache_shared_info(this_leaf->level, type, i,
+					    &csi) = PAL_STATUS_SUCCESS);
+	return 0;
 }
 
-struct cache_attr {
-	struct attribute attr;
-	ssize_t (*show)(struct cache_info *, char *);
-	ssize_t (*store)(struct cache_info *, const char *, size_t count);
-};
-
-#ifdef define_one_ro
-	#undef define_one_ro
-#endif
-#define define_one_ro(_name) \
-	static struct cache_attr _name = \
-__ATTR(_name, 0444, show_##_name, NULL)
-
-define_one_ro(level);
-define_one_ro(type);
-define_one_ro(coherency_line_size);
-define_one_ro(ways_of_associativity);
-define_one_ro(size);
-define_one_ro(number_of_sets);
-define_one_ro(shared_cpu_map);
-define_one_ro(attributes);
-
-static struct attribute * cache_default_attrs[] = {
-	&type.attr,
-	&level.attr,
-	&coherency_line_size.attr,
-	&ways_of_associativity.attr,
-	&attributes.attr,
-	&size.attr,
-	&number_of_sets.attr,
-	&shared_cpu_map.attr,
-	NULL
-};
-
-#define to_object(k) container_of(k, struct cache_info, kobj)
-#define to_attr(a) container_of(a, struct cache_attr, attr)
-
-static ssize_t ia64_cache_show(struct kobject * kobj, struct attribute * attr, char * buf)
+static int cache_shared_cpu_map_setup(unsigned int cpu)
 {
-	struct cache_attr *fattr = to_attr(attr);
-	struct cache_info *this_leaf = to_object(kobj);
-	ssize_t ret;
+	unsigned int idx;
+	int ret = 0;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
 
-	ret = fattr->show ? fattr->show(this_leaf, buf) : 0;
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++, this_leaf++) {
+		ret = __cache_cpumap_setup(cpu, this_leaf);
+		if (ret)
+			break;
+	}
 	return ret;
 }
-
-static const struct sysfs_ops cache_sysfs_ops = {
-	.show   = ia64_cache_show
-};
-
-static struct kobj_type cache_ktype = {
-	.sysfs_ops	= &cache_sysfs_ops,
-	.default_attrs	= cache_default_attrs,
-};
-
-static struct kobj_type cache_ktype_percpu_entry = {
-	.sysfs_ops	= &cache_sysfs_ops,
-};
-
-static void cpu_cache_sysfs_exit(unsigned int cpu)
-{
-	kfree(all_cpu_cache_info[cpu].cache_leaves);
-	all_cpu_cache_info[cpu].cache_leaves = NULL;
-	all_cpu_cache_info[cpu].num_cache_leaves = 0;
-	memset(&all_cpu_cache_info[cpu].kobj, 0, sizeof(struct kobject));
-	return;
-}
-
-static int cpu_cache_sysfs_init(unsigned int cpu)
+#else
+static int cache_shared_cpu_map_setup(unsigned int cpu)
 {
-	unsigned long i, levels, unique_caches;
-	pal_cache_config_info_t cci;
-	int j;
-	long status;
-	struct cache_info *this_cache;
-	int num_cache_leaves = 0;
-
-	if ((status = ia64_pal_cache_summary(&levels, &unique_caches)) != 0) {
-		printk(KERN_ERR "ia64_pal_cache_summary=%ld\n", status);
-		return -1;
-	}
-
-	this_cache=kzalloc(sizeof(struct cache_info)*unique_caches,
-			GFP_KERNEL);
-	if (this_cache = NULL)
-		return -ENOMEM;
-
-	for (i=0; i < levels; i++) {
-		for (j=2; j >0 ; j--) {
-			if ((status=ia64_pal_cache_config_info(i,j, &cci)) !-					PAL_STATUS_SUCCESS)
-				continue;
-
-			this_cache[num_cache_leaves].cci = cci;
-			this_cache[num_cache_leaves].level = i + 1;
-			this_cache[num_cache_leaves].type = j;
-
-			cache_shared_cpu_map_setup(cpu,
-					&this_cache[num_cache_leaves]);
-			num_cache_leaves ++;
-		}
-	}
-
-	all_cpu_cache_info[cpu].cache_leaves = this_cache;
-	all_cpu_cache_info[cpu].num_cache_leaves = num_cache_leaves;
-
-	memset(&all_cpu_cache_info[cpu].kobj, 0, sizeof(struct kobject));
+	int idx;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
 
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++, this_leaf++)
+		cpu_set(cpu, this_leaf->shared_cpu_map);
 	return 0;
 }
+#endif
 
-/* Add cache interface for CPU device */
-static int cache_add_dev(struct device *sys_dev)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 pal_cache_config_info_t *cci,
+			 enum cache_type type, unsigned int level)
 {
-	unsigned int cpu = sys_dev->id;
-	unsigned long i, j;
-	struct cache_info *this_object;
-	int retval = 0;
-	cpumask_t oldmask;
-
-	if (all_cpu_cache_info[cpu].kobj.parent)
-		return 0;
-
-	oldmask = current->cpus_allowed;
-	retval = set_cpus_allowed_ptr(current, cpumask_of(cpu));
-	if (unlikely(retval))
-		return retval;
+	unsigned number_of_sets = cci->pcci_cache_size;
 
-	retval = cpu_cache_sysfs_init(cpu);
-	set_cpus_allowed_ptr(current, &oldmask);
-	if (unlikely(retval < 0))
-		return retval;
-
-	retval = kobject_init_and_add(&all_cpu_cache_info[cpu].kobj,
-				      &cache_ktype_percpu_entry, &sys_dev->kobj,
-				      "%s", "cache");
-	if (unlikely(retval < 0)) {
-		cpu_cache_sysfs_exit(cpu);
-		return retval;
-	}
-
-	for (i = 0; i < all_cpu_cache_info[cpu].num_cache_leaves; i++) {
-		this_object = LEAF_KOBJECT_PTR(cpu,i);
-		retval = kobject_init_and_add(&(this_object->kobj),
-					      &cache_ktype,
-					      &all_cpu_cache_info[cpu].kobj,
-					      "index%1lu", i);
-		if (unlikely(retval)) {
-			for (j = 0; j < i; j++) {
-				kobject_put(&(LEAF_KOBJECT_PTR(cpu,j)->kobj));
-			}
-			kobject_put(&all_cpu_cache_info[cpu].kobj);
-			cpu_cache_sysfs_exit(cpu);
-			return retval;
-		}
-		kobject_uevent(&(this_object->kobj), KOBJ_ADD);
-	}
-	kobject_uevent(&all_cpu_cache_info[cpu].kobj, KOBJ_ADD);
-	return retval;
+	number_of_sets /= cci->pcci_assoc;
+	number_of_sets /= 1 << cci->pcci_line_size;
+	this_leaf->level = level;
+	this_leaf->type = cci->pcci_unified ? CACHE_TYPE_UNIFIED : type;
+	this_leaf->coherency_line_size = cci->pcci_line_size;
+	this_leaf->ways_of_associativity = cci->pcci_assoc;
+	this_leaf->size = cci->pcci_cache_size;
+	this_leaf->attributes = cci->pcci_cache_attr;
+	this_leaf->number_of_sets = number_of_sets;
 }
 
-/* Remove cache interface for CPU device */
-static int cache_remove_dev(struct device *sys_dev)
+int init_cache_level(unsigned int cpu)
 {
-	unsigned int cpu = sys_dev->id;
-	unsigned long i;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned long levels, unique_caches;
+	long status;
 
-	for (i = 0; i < all_cpu_cache_info[cpu].num_cache_leaves; i++)
-		kobject_put(&(LEAF_KOBJECT_PTR(cpu,i)->kobj));
+	if (!this_cpu_ci)
+		return -EINVAL;
 
-	if (all_cpu_cache_info[cpu].kobj.parent) {
-		kobject_put(&all_cpu_cache_info[cpu].kobj);
-		memset(&all_cpu_cache_info[cpu].kobj,
-			0,
-			sizeof(struct kobject));
+	status = ia64_pal_cache_summary(&levels, &unique_caches);
+	if (status != PAL_STATUS_SUCCESS) {
+		pr_err("ia64_pal_cache_summary = %ld\n", status);
+		return -EIO;
 	}
-
-	cpu_cache_sysfs_exit(cpu);
+	this_cpu_ci->num_levels = levels;
+	this_cpu_ci->num_leaves = unique_caches;
 
 	return 0;
 }
 
-/*
- * When a cpu is hot-plugged, do a check and initiate
- * cache kobject if necessary
- */
-static int cache_cpu_callback(struct notifier_block *nfb,
-		unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (unsigned long)hcpu;
-	struct device *sys_dev;
-
-	sys_dev = get_cpu_device(cpu);
-	switch (action) {
-	case CPU_ONLINE:
-	case CPU_ONLINE_FROZEN:
-		cache_add_dev(sys_dev);
-		break;
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		cache_remove_dev(sys_dev);
-		break;
-	}
-	return NOTIFY_OK;
-}
-
-static struct notifier_block cache_cpu_notifier -{
-	.notifier_call = cache_cpu_callback
-};
-
-static int __init cache_sysfs_init(void)
+int populate_cache_leaves(unsigned int cpu)
 {
-	int i;
-
-	cpu_notifier_register_begin();
-
-	for_each_online_cpu(i) {
-		struct device *sys_dev = get_cpu_device((unsigned int)i);
-		cache_add_dev(sys_dev);
+	unsigned int level, idx;
+	s64 status;
+	pal_cache_config_info_t cci;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		type = CACHE_TYPE_INST;
+		status = ia64_pal_cache_config_info(level - 1, type, &cci);
+		if (status = PAL_STATUS_SUCCESS)
+			ci_leaf_init(this_leaf++, &cci, type, level);
+		type = CACHE_TYPE_DATA;
+		status = ia64_pal_cache_config_info(level - 1, type, &cci);
+		if (status = PAL_STATUS_SUCCESS)
+			ci_leaf_init(this_leaf++, &cci, type, level);
 	}
-
-	__register_hotcpu_notifier(&cache_cpu_notifier);
-
-	cpu_notifier_register_done();
-
-	return 0;
+	return cache_shared_cpu_map_setup(cpu);
 }
-
-device_initcall(cache_sysfs_init);
-
-- 
1.8.3.2

[PATCH 4/9] s390: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
---
 arch/s390/kernel/cache.c | 388 +++++++++++------------------------------------
 1 file changed, 92 insertions(+), 296 deletions(-)

diff --git a/arch/s390/kernel/cache.c b/arch/s390/kernel/cache.c
index c0b03c2..fe21f07 100644
--- a/arch/s390/kernel/cache.c
+++ b/arch/s390/kernel/cache.c
@@ -5,37 +5,11 @@
  *    Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>
  */
 
-#include <linux/notifier.h>
 #include <linux/seq_file.h>
-#include <linux/init.h>
-#include <linux/list.h>
-#include <linux/slab.h>
 #include <linux/cpu.h>
+#include <linux/cacheinfo.h>
 #include <asm/facility.h>
 
-struct cache {
-	unsigned long size;
-	unsigned int line_size;
-	unsigned int associativity;
-	unsigned int nr_sets;
-	unsigned int level   : 3;
-	unsigned int type    : 2;
-	unsigned int private : 1;
-	struct list_head list;
-};
-
-struct cache_dir {
-	struct kobject *kobj;
-	struct cache_index_dir *index;
-};
-
-struct cache_index_dir {
-	struct kobject kobj;
-	int cpu;
-	struct cache *cache;
-	struct cache_index_dir *next;
-};
-
 enum {
 	CACHE_SCOPE_NOTEXISTS,
 	CACHE_SCOPE_PRIVATE,
@@ -44,10 +18,10 @@ enum {
 };
 
 enum {
-	CACHE_TYPE_SEPARATE,
-	CACHE_TYPE_DATA,
-	CACHE_TYPE_INSTRUCTION,
-	CACHE_TYPE_UNIFIED,
+	CTYPE_SEPARATE,
+	CTYPE_DATA,
+	CTYPE_INSTRUCTION,
+	CTYPE_UNIFIED,
 };
 
 enum {
@@ -70,39 +44,59 @@ struct cache_info {
 };
 
 #define CACHE_MAX_LEVEL 8
-
 union cache_topology {
 	struct cache_info ci[CACHE_MAX_LEVEL];
 	unsigned long long raw;
 };
 
 static const char * const cache_type_string[] = {
-	"Data",
+	"",
 	"Instruction",
+	"Data",
+	"",
 	"Unified",
 };
 
-static struct cache_dir *cache_dir_cpu[NR_CPUS];
-static LIST_HEAD(cache_list);
+static const enum cache_type cache_type_map[] = {
+	[CTYPE_SEPARATE] = CACHE_TYPE_SEPARATE,
+	[CTYPE_DATA] = CACHE_TYPE_DATA,
+	[CTYPE_INSTRUCTION] = CACHE_TYPE_INST,
+	[CTYPE_UNIFIED] = CACHE_TYPE_UNIFIED,
+};
 
 void show_cacheinfo(struct seq_file *m)
 {
-	struct cache *cache;
-	int index = 0;
+	int cpu = smp_processor_id(), idx;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *cache;
 
-	list_for_each_entry(cache, &cache_list, list) {
-		seq_printf(m, "cache%-11d: ", index);
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++) {
+		cache = this_cpu_ci->info_list + idx;
+		seq_printf(m, "cache%-11d: ", idx);
 		seq_printf(m, "level=%d ", cache->level);
 		seq_printf(m, "type=%s ", cache_type_string[cache->type]);
-		seq_printf(m, "scope=%s ", cache->private ? "Private" : "Shared");
-		seq_printf(m, "size=%luK ", cache->size >> 10);
-		seq_printf(m, "line_size=%u ", cache->line_size);
-		seq_printf(m, "associativity=%d", cache->associativity);
+		seq_printf(m, "scope=%s ",
+			   cache->disable_sysfs ? "Shared" : "Private");
+		seq_printf(m, "size=%dK ", cache->size >> 10);
+		seq_printf(m, "line_size=%u ", cache->coherency_line_size);
+		seq_printf(m, "associativity=%d", cache->ways_of_associativity);
 		seq_puts(m, "\n");
-		index++;
 	}
 }
 
+static inline enum cache_type get_cache_type(struct cache_info *ci, int level)
+{
+	if (level >= CACHE_MAX_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+
+	ci += level;
+
+	if (ci->scope != CACHE_SCOPE_SHARED && ci->scope != CACHE_SCOPE_PRIVATE)
+		return CACHE_TYPE_NOCACHE;
+
+	return cache_type_map[ci->type];
+}
+
 static inline unsigned long ecag(int ai, int li, int ti)
 {
 	unsigned long cmd, val;
@@ -113,277 +107,79 @@ static inline unsigned long ecag(int ai, int li, int ti)
 	return val;
 }
 
-static int __init cache_add(int level, int private, int type)
+static void ci_leaf_init(struct cacheinfo *this_leaf, int private,
+			 enum cache_type type, unsigned int level)
 {
-	struct cache *cache;
-	int ti;
+	int ti, num_sets;
+	int cpu = smp_processor_id();
 
-	cache = kzalloc(sizeof(*cache), GFP_KERNEL);
-	if (!cache)
-		return -ENOMEM;
-	if (type == CACHE_TYPE_INSTRUCTION)
+	if (type == CACHE_TYPE_INST)
 		ti = CACHE_TI_INSTRUCTION;
 	else
 		ti = CACHE_TI_UNIFIED;
-	cache->size = ecag(EXTRACT_SIZE, level, ti);
-	cache->line_size = ecag(EXTRACT_LINE_SIZE, level, ti);
-	cache->associativity = ecag(EXTRACT_ASSOCIATIVITY, level, ti);
-	cache->nr_sets = cache->size / cache->associativity;
-	cache->nr_sets /= cache->line_size;
-	cache->private = private;
-	cache->level = level + 1;
-	cache->type = type - 1;
-	list_add_tail(&cache->list, &cache_list);
-	return 0;
-}
-
-static void __init cache_build_info(void)
-{
-	struct cache *cache, *next;
-	union cache_topology ct;
-	int level, private, rc;
-
-	ct.raw = ecag(EXTRACT_TOPOLOGY, 0, 0);
-	for (level = 0; level < CACHE_MAX_LEVEL; level++) {
-		switch (ct.ci[level].scope) {
-		case CACHE_SCOPE_SHARED:
-			private = 0;
-			break;
-		case CACHE_SCOPE_PRIVATE:
-			private = 1;
-			break;
-		default:
-			return;
-		}
-		if (ct.ci[level].type == CACHE_TYPE_SEPARATE) {
-			rc  = cache_add(level, private, CACHE_TYPE_DATA);
-			rc |= cache_add(level, private, CACHE_TYPE_INSTRUCTION);
-		} else {
-			rc = cache_add(level, private, ct.ci[level].type);
-		}
-		if (rc)
-			goto error;
-	}
-	return;
-error:
-	list_for_each_entry_safe(cache, next, &cache_list, list) {
-		list_del(&cache->list);
-		kfree(cache);
-	}
-}
-
-static struct cache_dir *cache_create_cache_dir(int cpu)
-{
-	struct cache_dir *cache_dir;
-	struct kobject *kobj = NULL;
-	struct device *dev;
-
-	dev = get_cpu_device(cpu);
-	if (!dev)
-		goto out;
-	kobj = kobject_create_and_add("cache", &dev->kobj);
-	if (!kobj)
-		goto out;
-	cache_dir = kzalloc(sizeof(*cache_dir), GFP_KERNEL);
-	if (!cache_dir)
-		goto out;
-	cache_dir->kobj = kobj;
-	cache_dir_cpu[cpu] = cache_dir;
-	return cache_dir;
-out:
-	kobject_put(kobj);
-	return NULL;
-}
-
-static struct cache_index_dir *kobj_to_cache_index_dir(struct kobject *kobj)
-{
-	return container_of(kobj, struct cache_index_dir, kobj);
-}
-
-static void cache_index_release(struct kobject *kobj)
-{
-	struct cache_index_dir *index;
-
-	index = kobj_to_cache_index_dir(kobj);
-	kfree(index);
-}
-
-static ssize_t cache_index_show(struct kobject *kobj,
-				struct attribute *attr, char *buf)
-{
-	struct kobj_attribute *kobj_attr;
-
-	kobj_attr = container_of(attr, struct kobj_attribute, attr);
-	return kobj_attr->show(kobj, kobj_attr, buf);
-}
-
-#define DEFINE_CACHE_ATTR(_name, _format, _value)			\
-static ssize_t cache_##_name##_show(struct kobject *kobj,		\
-				    struct kobj_attribute *attr,	\
-				    char *buf)				\
-{									\
-	struct cache_index_dir *index;					\
-									\
-	index = kobj_to_cache_index_dir(kobj);				\
-	return sprintf(buf, _format, _value);				\
-}									\
-static struct kobj_attribute cache_##_name##_attr =			\
-	__ATTR(_name, 0444, cache_##_name##_show, NULL);
 
-DEFINE_CACHE_ATTR(size, "%luK\n", index->cache->size >> 10);
-DEFINE_CACHE_ATTR(coherency_line_size, "%u\n", index->cache->line_size);
-DEFINE_CACHE_ATTR(number_of_sets, "%u\n", index->cache->nr_sets);
-DEFINE_CACHE_ATTR(ways_of_associativity, "%u\n", index->cache->associativity);
-DEFINE_CACHE_ATTR(type, "%s\n", cache_type_string[index->cache->type]);
-DEFINE_CACHE_ATTR(level, "%d\n", index->cache->level);
+	this_leaf->level = level + 1;
+	this_leaf->type = type;
+	this_leaf->coherency_line_size = ecag(EXTRACT_LINE_SIZE, level, ti);
+	this_leaf->ways_of_associativity = ecag(EXTRACT_ASSOCIATIVITY,
+						level, ti);
+	this_leaf->size = ecag(EXTRACT_SIZE, level, ti);
 
-static ssize_t shared_cpu_map_func(struct kobject *kobj, int type, char *buf)
-{
-	struct cache_index_dir *index;
-	int len;
-
-	index = kobj_to_cache_index_dir(kobj);
-	len = type ?
-		cpulist_scnprintf(buf, PAGE_SIZE - 2, cpumask_of(index->cpu)) :
-		cpumask_scnprintf(buf, PAGE_SIZE - 2, cpumask_of(index->cpu));
-	len += sprintf(&buf[len], "\n");
-	return len;
-}
-
-static ssize_t shared_cpu_map_show(struct kobject *kobj,
-				   struct kobj_attribute *attr, char *buf)
-{
-	return shared_cpu_map_func(kobj, 0, buf);
+	num_sets = this_leaf->size / this_leaf->coherency_line_size;
+	num_sets /= this_leaf->ways_of_associativity;
+	this_leaf->number_of_sets = num_sets;
+	cpumask_set_cpu(cpu, &this_leaf->shared_cpu_map);
+	if (!private)
+		this_leaf->disable_sysfs = true;
 }
-static struct kobj_attribute cache_shared_cpu_map_attr =
-	__ATTR(shared_cpu_map, 0444, shared_cpu_map_show, NULL);
 
-static ssize_t shared_cpu_list_show(struct kobject *kobj,
-				    struct kobj_attribute *attr, char *buf)
+int init_cache_level(unsigned int cpu)
 {
-	return shared_cpu_map_func(kobj, 1, buf);
-}
-static struct kobj_attribute cache_shared_cpu_list_attr =
-	__ATTR(shared_cpu_list, 0444, shared_cpu_list_show, NULL);
-
-static struct attribute *cache_index_default_attrs[] = {
-	&cache_type_attr.attr,
-	&cache_size_attr.attr,
-	&cache_number_of_sets_attr.attr,
-	&cache_ways_of_associativity_attr.attr,
-	&cache_level_attr.attr,
-	&cache_coherency_line_size_attr.attr,
-	&cache_shared_cpu_map_attr.attr,
-	&cache_shared_cpu_list_attr.attr,
-	NULL,
-};
-
-static const struct sysfs_ops cache_index_ops = {
-	.show = cache_index_show,
-};
-
-static struct kobj_type cache_index_type = {
-	.sysfs_ops = &cache_index_ops,
-	.release = cache_index_release,
-	.default_attrs = cache_index_default_attrs,
-};
-
-static int cache_create_index_dir(struct cache_dir *cache_dir,
-				  struct cache *cache, int index, int cpu)
-{
-	struct cache_index_dir *index_dir;
-	int rc;
-
-	index_dir = kzalloc(sizeof(*index_dir), GFP_KERNEL);
-	if (!index_dir)
-		return -ENOMEM;
-	index_dir->cache = cache;
-	index_dir->cpu = cpu;
-	rc = kobject_init_and_add(&index_dir->kobj, &cache_index_type,
-				  cache_dir->kobj, "index%d", index);
-	if (rc)
-		goto out;
-	index_dir->next = cache_dir->index;
-	cache_dir->index = index_dir;
-	return 0;
-out:
-	kfree(index_dir);
-	return rc;
-}
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int level = 0, leaves = 0;
+	union cache_topology ct;
+	enum cache_type ctype;
 
-static int cache_add_cpu(int cpu)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-	int rc, index = 0;
+	if (!this_cpu_ci)
+		return -EINVAL;
 
-	if (list_empty(&cache_list))
-		return 0;
-	cache_dir = cache_create_cache_dir(cpu);
-	if (!cache_dir)
-		return -ENOMEM;
-	list_for_each_entry(cache, &cache_list, list) {
-		if (!cache->private)
+	ct.raw = ecag(EXTRACT_TOPOLOGY, 0, 0);
+	do {
+		ctype = get_cache_type(&ct.ci[0], level);
+		if (ctype == CACHE_TYPE_NOCACHE)
 			break;
-		rc = cache_create_index_dir(cache_dir, cache, index, cpu);
-		if (rc)
-			return rc;
-		index++;
-	}
-	return 0;
-}
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	} while (++level < CACHE_MAX_LEVEL);
 
-static void cache_remove_cpu(int cpu)
-{
-	struct cache_index_dir *index, *next;
-	struct cache_dir *cache_dir;
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
 
-	cache_dir = cache_dir_cpu[cpu];
-	if (!cache_dir)
-		return;
-	index = cache_dir->index;
-	while (index) {
-		next = index->next;
-		kobject_put(&index->kobj);
-		index = next;
-	}
-	kobject_put(cache_dir->kobj);
-	kfree(cache_dir);
-	cache_dir_cpu[cpu] = NULL;
+	return 0;
 }
 
-static int cache_hotplug(struct notifier_block *nfb, unsigned long action,
-			 void *hcpu)
+int populate_cache_leaves(unsigned int cpu)
 {
-	int cpu = (long)hcpu;
-	int rc = 0;
+	unsigned int level, idx, pvt;
+	union cache_topology ct;
+	enum cache_type ctype;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_ONLINE:
-		rc = cache_add_cpu(cpu);
-		if (rc)
-			cache_remove_cpu(cpu);
-		break;
-	case CPU_DEAD:
-		cache_remove_cpu(cpu);
-		break;
+	ct.raw = ecag(EXTRACT_TOPOLOGY, 0, 0);
+	for (idx = 0, level = 0; level < this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		pvt = (ct.ci[level].scope == CACHE_SCOPE_PRIVATE) ? 1 : 0;
+		ctype = get_cache_type(&ct.ci[0], level);
+		if (ctype == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, pvt, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, pvt, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, pvt, ctype, level);
+		}
 	}
-	return rc ? NOTIFY_BAD : NOTIFY_OK;
-}
-
-static int __init cache_init(void)
-{
-	int cpu;
-
-	if (!test_facility(34))
-		return 0;
-	cache_build_info();
-
-	cpu_notifier_register_begin();
-	for_each_online_cpu(cpu)
-		cache_add_cpu(cpu);
-	__hotcpu_notifier(cache_hotplug, 0);
-	cpu_notifier_register_done();
 	return 0;
 }
-device_initcall(cache_init);
-- 
1.8.3.2

[PATCH 5/9] x86: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

The private pointer provided by the cacheinfo is used to implement
the AMD L3 cache specific attributes.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: x86@kernel.org
---
 arch/x86/kernel/cpu/intel_cacheinfo.c | 655 ++++++++--------------------------
 1 file changed, 152 insertions(+), 503 deletions(-)

diff --git a/arch/x86/kernel/cpu/intel_cacheinfo.c b/arch/x86/kernel/cpu/intel_cacheinfo.c
index a952e9c..621312c 100644
--- a/arch/x86/kernel/cpu/intel_cacheinfo.c
+++ b/arch/x86/kernel/cpu/intel_cacheinfo.c
@@ -7,16 +7,14 @@
  *	Andi Kleen / Andreas Herrmann	: CPUID4 emulation on AMD.
  */
 
-#include <linux/init.h>
 #include <linux/slab.h>
-#include <linux/device.h>
-#include <linux/compiler.h>
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
 #include <linux/sched.h>
+#include <linux/sysfs.h>
 #include <linux/pci.h>
 
 #include <asm/processor.h>
-#include <linux/smp.h>
 #include <asm/amd_nb.h>
 #include <asm/smp.h>
 
@@ -116,10 +114,10 @@ static const struct _cache_table cache_table[] =
 
 
 enum _cache_type {
-	CACHE_TYPE_NULL	= 0,
-	CACHE_TYPE_DATA = 1,
-	CACHE_TYPE_INST = 2,
-	CACHE_TYPE_UNIFIED = 3
+	CTYPE_NULL = 0,
+	CTYPE_DATA = 1,
+	CTYPE_INST = 2,
+	CTYPE_UNIFIED = 3
 };
 
 union _cpuid4_leaf_eax {
@@ -159,12 +157,7 @@ struct _cpuid4_info_regs {
 	struct amd_northbridge *nb;
 };
 
-struct _cpuid4_info {
-	struct _cpuid4_info_regs base;
-	DECLARE_BITMAP(shared_cpu_map, NR_CPUS);
-};
-
-unsigned short			num_cache_leaves;
+unsigned short                 num_cache_leaves;
 
 /* AMD doesn't have CPUID4. Emulate it here to report the same
    information to the user.  This makes some assumptions about the machine:
@@ -220,6 +213,13 @@ static const unsigned short assocs[] = {
 static const unsigned char levels[] = { 1, 1, 2, 3 };
 static const unsigned char types[] = { 1, 2, 3, 3 };
 
+static const enum cache_type cache_type_map[] = {
+	[CTYPE_NULL] = CACHE_TYPE_NOCACHE,
+	[CTYPE_DATA] = CACHE_TYPE_DATA,
+	[CTYPE_INST] = CACHE_TYPE_INST,
+	[CTYPE_UNIFIED] = CACHE_TYPE_UNIFIED,
+};
+
 static void
 amd_cpuid4(int leaf, union _cpuid4_leaf_eax *eax,
 		     union _cpuid4_leaf_ebx *ebx,
@@ -291,14 +291,9 @@ amd_cpuid4(int leaf, union _cpuid4_leaf_eax *eax,
 		(ebx->split.ways_of_associativity + 1) - 1;
 }
 
-struct _cache_attr {
-	struct attribute attr;
-	ssize_t (*show)(struct _cpuid4_info *, char *, unsigned int);
-	ssize_t (*store)(struct _cpuid4_info *, const char *, size_t count,
-			 unsigned int);
-};
-
 #if defined(CONFIG_AMD_NB) && defined(CONFIG_SYSFS)
+
+static struct device_attribute **amd_l3_attrs;
 /*
  * L3 cache descriptors
  */
@@ -325,6 +320,7 @@ static void amd_calc_l3_indices(struct amd_northbridge *nb)
 	l3->indices = (max(max3(sc0, sc1, sc2), sc3) << 10) - 1;
 }
 
+static void init_amd_l3_attrs(void);
 static void amd_init_l3_cache(struct _cpuid4_info_regs *this_leaf, int index)
 {
 	int node;
@@ -335,8 +331,10 @@ static void amd_init_l3_cache(struct _cpuid4_info_regs *this_leaf, int index)
 
 	node = amd_get_nb_id(smp_processor_id());
 	this_leaf->nb = node_to_amd_nb(node);
-	if (this_leaf->nb && !this_leaf->nb->l3_cache.indices)
+	if (this_leaf->nb && !this_leaf->nb->l3_cache.indices) {
 		amd_calc_l3_indices(this_leaf->nb);
+		init_amd_l3_attrs();
+	}
 }
 
 /*
@@ -359,15 +357,16 @@ int amd_get_l3_disable_slot(struct amd_northbridge *nb, unsigned slot)
 	return -1;
 }
 
-static ssize_t show_cache_disable(struct _cpuid4_info *this_leaf, char *buf,
+static ssize_t show_cache_disable(struct cacheinfo *this_leaf, char *buf,
 				  unsigned int slot)
 {
 	int index;
+	struct amd_northbridge *nb = this_leaf->priv;
 
-	if (!this_leaf->base.nb || !amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
+	if (!nb || !amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
 		return -EINVAL;
 
-	index = amd_get_l3_disable_slot(this_leaf->base.nb, slot);
+	index = amd_get_l3_disable_slot(nb, slot);
 	if (index >= 0)
 		return sprintf(buf, "%d\n", index);
 
@@ -376,9 +375,10 @@ static ssize_t show_cache_disable(struct _cpuid4_info *this_leaf, char *buf,
 
 #define SHOW_CACHE_DISABLE(slot)					\
 static ssize_t								\
-show_cache_disable_##slot(struct _cpuid4_info *this_leaf, char *buf,	\
-			  unsigned int cpu)				\
+cache_disable_##slot##_show(struct device *dev,				\
+			    struct device_attribute *attr, char *buf)	\
 {									\
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);		\
 	return show_cache_disable(this_leaf, buf, slot);		\
 }
 SHOW_CACHE_DISABLE(0)
@@ -446,25 +446,26 @@ int amd_set_l3_disable_slot(struct amd_northbridge *nb, int cpu, unsigned slot,
 	return 0;
 }
 
-static ssize_t store_cache_disable(struct _cpuid4_info *this_leaf,
-				  const char *buf, size_t count,
-				  unsigned int slot)
+static ssize_t store_cache_disable(struct cacheinfo *this_leaf,
+				   const char *buf, size_t count,
+				   unsigned int slot)
 {
 	unsigned long val = 0;
 	int cpu, err = 0;
+	struct amd_northbridge *nb = this_leaf->priv;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
-	if (!this_leaf->base.nb || !amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
+	if (!nb || !amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
 		return -EINVAL;
 
-	cpu = cpumask_first(to_cpumask(this_leaf->shared_cpu_map));
+	cpu = cpumask_first(&this_leaf->shared_cpu_map);
 
 	if (strict_strtoul(buf, 10, &val) < 0)
 		return -EINVAL;
 
-	err = amd_set_l3_disable_slot(this_leaf->base.nb, cpu, slot, val);
+	err = amd_set_l3_disable_slot(nb, cpu, slot, val);
 	if (err) {
 		if (err == -EEXIST)
 			pr_warning("L3 slot %d in use/index already disabled!\n",
@@ -476,39 +477,39 @@ static ssize_t store_cache_disable(struct _cpuid4_info *this_leaf,
 
 #define STORE_CACHE_DISABLE(slot)					\
 static ssize_t								\
-store_cache_disable_##slot(struct _cpuid4_info *this_leaf,		\
-			   const char *buf, size_t count,		\
-			   unsigned int cpu)				\
+cache_disable_##slot##_store(struct device *dev,			\
+			     struct device_attribute *attr,		\
+			     const char *buf, size_t count)		\
 {									\
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);		\
 	return store_cache_disable(this_leaf, buf, count, slot);	\
 }
 STORE_CACHE_DISABLE(0)
 STORE_CACHE_DISABLE(1)
 
-static struct _cache_attr cache_disable_0 = __ATTR(cache_disable_0, 0644,
-		show_cache_disable_0, store_cache_disable_0);
-static struct _cache_attr cache_disable_1 = __ATTR(cache_disable_1, 0644,
-		show_cache_disable_1, store_cache_disable_1);
-
-static ssize_t
-show_subcaches(struct _cpuid4_info *this_leaf, char *buf, unsigned int cpu)
+static ssize_t subcaches_show(struct device *dev,
+			      struct device_attribute *attr, char *buf)
 {
-	if (!this_leaf->base.nb || !amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
-		return -EINVAL;
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	int cpu = cpumask_first(&this_leaf->shared_cpu_map);
 
+	if (!this_leaf->priv || !amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
+		return -EINVAL;
 	return sprintf(buf, "%x\n", amd_get_subcaches(cpu));
 }
 
-static ssize_t
-store_subcaches(struct _cpuid4_info *this_leaf, const char *buf, size_t count,
-		unsigned int cpu)
+static ssize_t subcaches_store(struct device *dev,
+			       struct device_attribute *attr,
+			       const char *buf, size_t count)
 {
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	int cpu = cpumask_first(&this_leaf->shared_cpu_map);
 	unsigned long val;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
-	if (!this_leaf->base.nb || !amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
+	if (!this_leaf->priv || !amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
 		return -EINVAL;
 
 	if (strict_strtoul(buf, 16, &val) < 0)
@@ -520,9 +521,42 @@ store_subcaches(struct _cpuid4_info *this_leaf, const char *buf, size_t count,
 	return count;
 }
 
-static struct _cache_attr subcaches =
-	__ATTR(subcaches, 0644, show_subcaches, store_subcaches);
+static DEVICE_ATTR_RW(cache_disable_0);
+static DEVICE_ATTR_RW(cache_disable_1);
+static DEVICE_ATTR_RW(subcaches);
+
+static void init_amd_l3_attrs(void)
+{
+	int n = 1;
+
+	if (amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
+		n += 2;
+	if (amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
+		n += 1;
+
+	amd_l3_attrs = kcalloc(n, sizeof(struct device_attribute *),
+			       GFP_KERNEL);
+	if (!amd_l3_attrs)
+		return;
+
+	n = 0;
+	if (amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE)) {
+		amd_l3_attrs[n++] = &dev_attr_cache_disable_0;
+		amd_l3_attrs[n++] = &dev_attr_cache_disable_1;
+	}
+	if (amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
+		amd_l3_attrs[n++] = &dev_attr_subcaches;
+}
+
+const struct device_attribute **
+cache_get_priv_attr(struct device *cache_idx_dev)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(cache_idx_dev);
 
+	if (!this_leaf->priv || !amd_l3_attrs)
+		return NULL;
+	return (const struct device_attribute **)amd_l3_attrs;
+}
 #else
 #define amd_init_l3_cache(x, y)
 #endif  /* CONFIG_AMD_NB && CONFIG_SYSFS */
@@ -546,7 +580,7 @@ cpuid4_cache_lookup_regs(int index, struct _cpuid4_info_regs *this_leaf)
 		cpuid_count(4, index, &eax.full, &ebx.full, &ecx.full, &edx);
 	}
 
-	if (eax.split.type == CACHE_TYPE_NULL)
+	if (eax.split.type == CTYPE_NULL)
 		return -EIO; /* better error ? */
 
 	this_leaf->eax = eax;
@@ -575,7 +609,7 @@ static int find_num_cache_leaves(struct cpuinfo_x86 *c)
 		/* Do cpuid(op) loop to find out num_cache_leaves */
 		cpuid_count(op, i, &eax, &ebx, &ecx, &edx);
 		cache_eax.full = eax;
-	} while (cache_eax.split.type != CACHE_TYPE_NULL);
+	} while (cache_eax.split.type != CTYPE_NULL);
 	return i;
 }
 
@@ -626,9 +660,9 @@ unsigned int init_intel_cacheinfo(struct cpuinfo_x86 *c)
 
 			switch (this_leaf.eax.split.level) {
 			case 1:
-				if (this_leaf.eax.split.type == CACHE_TYPE_DATA)
+				if (this_leaf.eax.split.type == CTYPE_DATA)
 					new_l1d = this_leaf.size/1024;
-				else if (this_leaf.eax.split.type == CACHE_TYPE_INST)
+				else if (this_leaf.eax.split.type == CTYPE_INST)
 					new_l1i = this_leaf.size/1024;
 				break;
 			case 2:
@@ -735,55 +769,46 @@ unsigned int init_intel_cacheinfo(struct cpuinfo_x86 *c)
 	return l2;
 }
 
-#ifdef CONFIG_SYSFS
-
-/* pointer to _cpuid4_info array (for each cache leaf) */
-static DEFINE_PER_CPU(struct _cpuid4_info *, ici_cpuid4_info);
-#define CPUID4_INFO_IDX(x, y)	(&((per_cpu(ici_cpuid4_info, x))[y]))
-
-#ifdef CONFIG_SMP
-
-static int cache_shared_amd_cpu_map_setup(unsigned int cpu, int index)
+static int __cache_amd_cpumap_setup(unsigned int cpu, int index,
+				    struct _cpuid4_info_regs *base)
 {
-	struct _cpuid4_info *this_leaf;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf;
 	int i, sibling;
 
 	if (cpu_has_topoext) {
 		unsigned int apicid, nshared, first, last;
 
-		if (!per_cpu(ici_cpuid4_info, cpu))
-			return 0;
-
-		this_leaf = CPUID4_INFO_IDX(cpu, index);
-		nshared = this_leaf->base.eax.split.num_threads_sharing + 1;
+		this_leaf = this_cpu_ci->info_list + index;
+		nshared = base->eax.split.num_threads_sharing + 1;
 		apicid = cpu_data(cpu).apicid;
 		first = apicid - (apicid % nshared);
 		last = first + nshared - 1;
 
 		for_each_online_cpu(i) {
+			this_cpu_ci = get_cpu_cacheinfo(i);
 			apicid = cpu_data(i).apicid;
 			if ((apicid < first) || (apicid > last))
 				continue;
-			if (!per_cpu(ici_cpuid4_info, i))
-				continue;
-			this_leaf = CPUID4_INFO_IDX(i, index);
+			this_leaf = this_cpu_ci->info_list + index;
 
 			for_each_online_cpu(sibling) {
 				apicid = cpu_data(sibling).apicid;
 				if ((apicid < first) || (apicid > last))
 					continue;
-				set_bit(sibling, this_leaf->shared_cpu_map);
+				cpumask_set_cpu(sibling,
+						&this_leaf->shared_cpu_map);
 			}
 		}
 	} else if (index == 3) {
 		for_each_cpu(i, cpu_llc_shared_mask(cpu)) {
-			if (!per_cpu(ici_cpuid4_info, i))
-				continue;
-			this_leaf = CPUID4_INFO_IDX(i, index);
+			this_cpu_ci = get_cpu_cacheinfo(i);
+			this_leaf = this_cpu_ci->info_list + index;
 			for_each_cpu(sibling, cpu_llc_shared_mask(cpu)) {
 				if (!cpu_online(sibling))
 					continue;
-				set_bit(sibling, this_leaf->shared_cpu_map);
+				cpumask_set_cpu(sibling,
+						&this_leaf->shared_cpu_map);
 			}
 		}
 	} else
@@ -792,459 +817,83 @@ static int cache_shared_amd_cpu_map_setup(unsigned int cpu, int index)
 	return 1;
 }
 
-static void cache_shared_cpu_map_setup(unsigned int cpu, int index)
+static void __cache_cpumap_setup(unsigned int cpu, int index,
+				 struct _cpuid4_info_regs *base)
 {
-	struct _cpuid4_info *this_leaf, *sibling_leaf;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sibling_leaf;
 	unsigned long num_threads_sharing;
 	int index_msb, i;
 	struct cpuinfo_x86 *c = &cpu_data(cpu);
 
 	if (c->x86_vendor == X86_VENDOR_AMD) {
-		if (cache_shared_amd_cpu_map_setup(cpu, index))
+		if (__cache_amd_cpumap_setup(cpu, index, base))
 			return;
 	}
 
-	this_leaf = CPUID4_INFO_IDX(cpu, index);
-	num_threads_sharing = 1 + this_leaf->base.eax.split.num_threads_sharing;
+	this_leaf = this_cpu_ci->info_list + index;
+	num_threads_sharing = 1 + base->eax.split.num_threads_sharing;
 
+	cpumask_set_cpu(cpu, &this_leaf->shared_cpu_map);
 	if (num_threads_sharing == 1)
-		cpumask_set_cpu(cpu, to_cpumask(this_leaf->shared_cpu_map));
-	else {
-		index_msb = get_count_order(num_threads_sharing);
-
-		for_each_online_cpu(i) {
-			if (cpu_data(i).apicid >> index_msb ==
-			    c->apicid >> index_msb) {
-				cpumask_set_cpu(i,
-					to_cpumask(this_leaf->shared_cpu_map));
-				if (i != cpu && per_cpu(ici_cpuid4_info, i))  {
-					sibling_leaf =
-						CPUID4_INFO_IDX(i, index);
-					cpumask_set_cpu(cpu, to_cpumask(
-						sibling_leaf->shared_cpu_map));
-				}
-			}
-		}
-	}
-}
-static void cache_remove_shared_cpu_map(unsigned int cpu, int index)
-{
-	struct _cpuid4_info	*this_leaf, *sibling_leaf;
-	int sibling;
-
-	this_leaf = CPUID4_INFO_IDX(cpu, index);
-	for_each_cpu(sibling, to_cpumask(this_leaf->shared_cpu_map)) {
-		sibling_leaf = CPUID4_INFO_IDX(sibling, index);
-		cpumask_clear_cpu(cpu,
-				  to_cpumask(sibling_leaf->shared_cpu_map));
-	}
-}
-#else
-static void cache_shared_cpu_map_setup(unsigned int cpu, int index)
-{
-}
-
-static void cache_remove_shared_cpu_map(unsigned int cpu, int index)
-{
-}
-#endif
-
-static void free_cache_attributes(unsigned int cpu)
-{
-	int i;
-
-	for (i = 0; i < num_cache_leaves; i++)
-		cache_remove_shared_cpu_map(cpu, i);
-
-	kfree(per_cpu(ici_cpuid4_info, cpu));
-	per_cpu(ici_cpuid4_info, cpu) = NULL;
-}
-
-static void get_cpu_leaves(void *_retval)
-{
-	int j, *retval = _retval, cpu = smp_processor_id();
+		return;
 
-	/* Do cpuid and store the results */
-	for (j = 0; j < num_cache_leaves; j++) {
-		struct _cpuid4_info *this_leaf = CPUID4_INFO_IDX(cpu, j);
+	index_msb = get_count_order(num_threads_sharing);
 
-		*retval = cpuid4_cache_lookup_regs(j, &this_leaf->base);
-		if (unlikely(*retval < 0)) {
-			int i;
+	for_each_online_cpu(i)
+		if (cpu_data(i).apicid >> index_msb == c->apicid >> index_msb) {
+			struct cpu_cacheinfo *sib_cpu_ci = get_cpu_cacheinfo(i);
 
-			for (i = 0; i < j; i++)
-				cache_remove_shared_cpu_map(cpu, i);
-			break;
+			if (i == cpu || !sib_cpu_ci->info_list)
+				continue;/* skip if itself or no cacheinfo */
+			sibling_leaf = sib_cpu_ci->info_list + index;
+			cpumask_set_cpu(i, &this_leaf->shared_cpu_map);
+			cpumask_set_cpu(cpu, &sibling_leaf->shared_cpu_map);
 		}
-		cache_shared_cpu_map_setup(cpu, j);
-	}
-}
-
-static int detect_cache_attributes(unsigned int cpu)
-{
-	int			retval;
-
-	if (num_cache_leaves == 0)
-		return -ENOENT;
-
-	per_cpu(ici_cpuid4_info, cpu) = kzalloc(
-	    sizeof(struct _cpuid4_info) * num_cache_leaves, GFP_KERNEL);
-	if (per_cpu(ici_cpuid4_info, cpu) == NULL)
-		return -ENOMEM;
-
-	smp_call_function_single(cpu, get_cpu_leaves, &retval, true);
-	if (retval) {
-		kfree(per_cpu(ici_cpuid4_info, cpu));
-		per_cpu(ici_cpuid4_info, cpu) = NULL;
-	}
-
-	return retval;
-}
-
-#include <linux/kobject.h>
-#include <linux/sysfs.h>
-#include <linux/cpu.h>
-
-/* pointer to kobject for cpuX/cache */
-static DEFINE_PER_CPU(struct kobject *, ici_cache_kobject);
-
-struct _index_kobject {
-	struct kobject kobj;
-	unsigned int cpu;
-	unsigned short index;
-};
-
-/* pointer to array of kobjects for cpuX/cache/indexY */
-static DEFINE_PER_CPU(struct _index_kobject *, ici_index_kobject);
-#define INDEX_KOBJECT_PTR(x, y)		(&((per_cpu(ici_index_kobject, x))[y]))
-
-#define show_one_plus(file_name, object, val)				\
-static ssize_t show_##file_name(struct _cpuid4_info *this_leaf, char *buf, \
-				unsigned int cpu)			\
-{									\
-	return sprintf(buf, "%lu\n", (unsigned long)this_leaf->object + val); \
-}
-
-show_one_plus(level, base.eax.split.level, 0);
-show_one_plus(coherency_line_size, base.ebx.split.coherency_line_size, 1);
-show_one_plus(physical_line_partition, base.ebx.split.physical_line_partition, 1);
-show_one_plus(ways_of_associativity, base.ebx.split.ways_of_associativity, 1);
-show_one_plus(number_of_sets, base.ecx.split.number_of_sets, 1);
-
-static ssize_t show_size(struct _cpuid4_info *this_leaf, char *buf,
-			 unsigned int cpu)
-{
-	return sprintf(buf, "%luK\n", this_leaf->base.size / 1024);
-}
-
-static ssize_t show_shared_cpu_map_func(struct _cpuid4_info *this_leaf,
-					int type, char *buf)
-{
-	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf;
-	int n = 0;
-
-	if (len > 1) {
-		const struct cpumask *mask;
-
-		mask = to_cpumask(this_leaf->shared_cpu_map);
-		n = type ?
-			cpulist_scnprintf(buf, len-2, mask) :
-			cpumask_scnprintf(buf, len-2, mask);
-		buf[n++] = '\n';
-		buf[n] = '\0';
-	}
-	return n;
-}
-
-static inline ssize_t show_shared_cpu_map(struct _cpuid4_info *leaf, char *buf,
-					  unsigned int cpu)
-{
-	return show_shared_cpu_map_func(leaf, 0, buf);
-}
-
-static inline ssize_t show_shared_cpu_list(struct _cpuid4_info *leaf, char *buf,
-					   unsigned int cpu)
-{
-	return show_shared_cpu_map_func(leaf, 1, buf);
-}
-
-static ssize_t show_type(struct _cpuid4_info *this_leaf, char *buf,
-			 unsigned int cpu)
-{
-	switch (this_leaf->base.eax.split.type) {
-	case CACHE_TYPE_DATA:
-		return sprintf(buf, "Data\n");
-	case CACHE_TYPE_INST:
-		return sprintf(buf, "Instruction\n");
-	case CACHE_TYPE_UNIFIED:
-		return sprintf(buf, "Unified\n");
-	default:
-		return sprintf(buf, "Unknown\n");
-	}
-}
-
-#define to_object(k)	container_of(k, struct _index_kobject, kobj)
-#define to_attr(a)	container_of(a, struct _cache_attr, attr)
-
-#define define_one_ro(_name) \
-static struct _cache_attr _name = \
-	__ATTR(_name, 0444, show_##_name, NULL)
-
-define_one_ro(level);
-define_one_ro(type);
-define_one_ro(coherency_line_size);
-define_one_ro(physical_line_partition);
-define_one_ro(ways_of_associativity);
-define_one_ro(number_of_sets);
-define_one_ro(size);
-define_one_ro(shared_cpu_map);
-define_one_ro(shared_cpu_list);
-
-static struct attribute *default_attrs[] = {
-	&type.attr,
-	&level.attr,
-	&coherency_line_size.attr,
-	&physical_line_partition.attr,
-	&ways_of_associativity.attr,
-	&number_of_sets.attr,
-	&size.attr,
-	&shared_cpu_map.attr,
-	&shared_cpu_list.attr,
-	NULL
-};
-
-#ifdef CONFIG_AMD_NB
-static struct attribute **amd_l3_attrs(void)
-{
-	static struct attribute **attrs;
-	int n;
-
-	if (attrs)
-		return attrs;
-
-	n = ARRAY_SIZE(default_attrs);
-
-	if (amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
-		n += 2;
-
-	if (amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
-		n += 1;
-
-	attrs = kzalloc(n * sizeof (struct attribute *), GFP_KERNEL);
-	if (attrs == NULL)
-		return attrs = default_attrs;
-
-	for (n = 0; default_attrs[n]; n++)
-		attrs[n] = default_attrs[n];
-
-	if (amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE)) {
-		attrs[n++] = &cache_disable_0.attr;
-		attrs[n++] = &cache_disable_1.attr;
-	}
-
-	if (amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
-		attrs[n++] = &subcaches.attr;
-
-	return attrs;
-}
-#endif
-
-static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
-{
-	struct _cache_attr *fattr = to_attr(attr);
-	struct _index_kobject *this_leaf = to_object(kobj);
-	ssize_t ret;
-
-	ret = fattr->show ?
-		fattr->show(CPUID4_INFO_IDX(this_leaf->cpu, this_leaf->index),
-			buf, this_leaf->cpu) :
-		0;
-	return ret;
 }
 
-static ssize_t store(struct kobject *kobj, struct attribute *attr,
-		     const char *buf, size_t count)
-{
-	struct _cache_attr *fattr = to_attr(attr);
-	struct _index_kobject *this_leaf = to_object(kobj);
-	ssize_t ret;
-
-	ret = fattr->store ?
-		fattr->store(CPUID4_INFO_IDX(this_leaf->cpu, this_leaf->index),
-			buf, count, this_leaf->cpu) :
-		0;
-	return ret;
-}
-
-static const struct sysfs_ops sysfs_ops = {
-	.show   = show,
-	.store  = store,
-};
-
-static struct kobj_type ktype_cache = {
-	.sysfs_ops	= &sysfs_ops,
-	.default_attrs	= default_attrs,
-};
-
-static struct kobj_type ktype_percpu_entry = {
-	.sysfs_ops	= &sysfs_ops,
-};
-
-static void cpuid4_cache_sysfs_exit(unsigned int cpu)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 struct _cpuid4_info_regs *base)
 {
-	kfree(per_cpu(ici_cache_kobject, cpu));
-	kfree(per_cpu(ici_index_kobject, cpu));
-	per_cpu(ici_cache_kobject, cpu) = NULL;
-	per_cpu(ici_index_kobject, cpu) = NULL;
-	free_cache_attributes(cpu);
+	this_leaf->level = base->eax.split.level;
+	this_leaf->type = cache_type_map[base->eax.split.type];
+	this_leaf->coherency_line_size =
+				base->ebx.split.coherency_line_size + 1;
+	this_leaf->ways_of_associativity =
+				base->ebx.split.ways_of_associativity + 1;
+	this_leaf->size = base->size;
+	this_leaf->number_of_sets = base->ecx.split.number_of_sets + 1;
+	this_leaf->physical_line_partition =
+				base->ebx.split.physical_line_partition + 1;
+	this_leaf->priv = base->nb;
 }
 
-static int cpuid4_cache_sysfs_init(unsigned int cpu)
+int init_cache_level(unsigned int cpu)
 {
-	int err;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
 
-	if (num_cache_leaves == 0)
+	if (!num_cache_leaves)
 		return -ENOENT;
-
-	err = detect_cache_attributes(cpu);
-	if (err)
-		return err;
-
-	/* Allocate all required memory */
-	per_cpu(ici_cache_kobject, cpu) =
-		kzalloc(sizeof(struct kobject), GFP_KERNEL);
-	if (unlikely(per_cpu(ici_cache_kobject, cpu) == NULL))
-		goto err_out;
-
-	per_cpu(ici_index_kobject, cpu) = kzalloc(
-	    sizeof(struct _index_kobject) * num_cache_leaves, GFP_KERNEL);
-	if (unlikely(per_cpu(ici_index_kobject, cpu) == NULL))
-		goto err_out;
-
+	if (!this_cpu_ci)
+		return -EINVAL;
+	this_cpu_ci->num_levels = 3;
+	this_cpu_ci->num_leaves = num_cache_leaves;
 	return 0;
-
-err_out:
-	cpuid4_cache_sysfs_exit(cpu);
-	return -ENOMEM;
 }
 
-static DECLARE_BITMAP(cache_dev_map, NR_CPUS);
-
-/* Add/Remove cache interface for CPU device */
-static int cache_add_dev(struct device *dev)
+int populate_cache_leaves(unsigned int cpu)
 {
-	unsigned int cpu = dev->id;
-	unsigned long i, j;
-	struct _index_kobject *this_object;
-	struct _cpuid4_info   *this_leaf;
-	int retval;
-
-	retval = cpuid4_cache_sysfs_init(cpu);
-	if (unlikely(retval < 0))
-		return retval;
-
-	retval = kobject_init_and_add(per_cpu(ici_cache_kobject, cpu),
-				      &ktype_percpu_entry,
-				      &dev->kobj, "%s", "cache");
-	if (retval < 0) {
-		cpuid4_cache_sysfs_exit(cpu);
-		return retval;
-	}
-
-	for (i = 0; i < num_cache_leaves; i++) {
-		this_object = INDEX_KOBJECT_PTR(cpu, i);
-		this_object->cpu = cpu;
-		this_object->index = i;
-
-		this_leaf = CPUID4_INFO_IDX(cpu, i);
+	unsigned int idx, ret;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+	struct _cpuid4_info_regs id4_regs = {};
 
-		ktype_cache.default_attrs = default_attrs;
-#ifdef CONFIG_AMD_NB
-		if (this_leaf->base.nb)
-			ktype_cache.default_attrs = amd_l3_attrs();
-#endif
-		retval = kobject_init_and_add(&(this_object->kobj),
-					      &ktype_cache,
-					      per_cpu(ici_cache_kobject, cpu),
-					      "index%1lu", i);
-		if (unlikely(retval)) {
-			for (j = 0; j < i; j++)
-				kobject_put(&(INDEX_KOBJECT_PTR(cpu, j)->kobj));
-			kobject_put(per_cpu(ici_cache_kobject, cpu));
-			cpuid4_cache_sysfs_exit(cpu);
-			return retval;
-		}
-		kobject_uevent(&(this_object->kobj), KOBJ_ADD);
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++) {
+		ret = cpuid4_cache_lookup_regs(idx, &id4_regs);
+		if (ret)
+			return ret;
+		ci_leaf_init(this_leaf++, &id4_regs);
+		__cache_cpumap_setup(cpu, idx, &id4_regs);
 	}
-	cpumask_set_cpu(cpu, to_cpumask(cache_dev_map));
-
-	kobject_uevent(per_cpu(ici_cache_kobject, cpu), KOBJ_ADD);
 	return 0;
 }
-
-static void cache_remove_dev(struct device *dev)
-{
-	unsigned int cpu = dev->id;
-	unsigned long i;
-
-	if (per_cpu(ici_cpuid4_info, cpu) == NULL)
-		return;
-	if (!cpumask_test_cpu(cpu, to_cpumask(cache_dev_map)))
-		return;
-	cpumask_clear_cpu(cpu, to_cpumask(cache_dev_map));
-
-	for (i = 0; i < num_cache_leaves; i++)
-		kobject_put(&(INDEX_KOBJECT_PTR(cpu, i)->kobj));
-	kobject_put(per_cpu(ici_cache_kobject, cpu));
-	cpuid4_cache_sysfs_exit(cpu);
-}
-
-static int cacheinfo_cpu_callback(struct notifier_block *nfb,
-				  unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (unsigned long)hcpu;
-	struct device *dev;
-
-	dev = get_cpu_device(cpu);
-	switch (action) {
-	case CPU_ONLINE:
-	case CPU_ONLINE_FROZEN:
-		cache_add_dev(dev);
-		break;
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		cache_remove_dev(dev);
-		break;
-	}
-	return NOTIFY_OK;
-}
-
-static struct notifier_block cacheinfo_cpu_notifier = {
-	.notifier_call = cacheinfo_cpu_callback,
-};
-
-static int __init cache_sysfs_init(void)
-{
-	int i, err = 0;
-
-	if (num_cache_leaves == 0)
-		return 0;
-
-	cpu_notifier_register_begin();
-	for_each_online_cpu(i) {
-		struct device *dev = get_cpu_device(i);
-
-		err = cache_add_dev(dev);
-		if (err)
-			goto out;
-	}
-	__register_hotcpu_notifier(&cacheinfo_cpu_notifier);
-
-out:
-	cpu_notifier_register_done();
-	return err;
-}
-
-device_initcall(cache_sysfs_init);
-
-#endif
-- 
1.8.3.2

[PATCH 6/9] powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Anshuman Khandual <khandual@linux.vnet.ibm.com>
Cc: linuxppc-dev@lists.ozlabs.org
---
 arch/powerpc/kernel/cacheinfo.c | 813 +++++-----------------------------------
 arch/powerpc/kernel/cacheinfo.h |   8 -
 arch/powerpc/kernel/sysfs.c     |  12 +-
 3 files changed, 91 insertions(+), 742 deletions(-)
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h

diff --git a/arch/powerpc/kernel/cacheinfo.c b/arch/powerpc/kernel/cacheinfo.c
index 40198d5..b871c24 100644
--- a/arch/powerpc/kernel/cacheinfo.c
+++ b/arch/powerpc/kernel/cacheinfo.c
@@ -10,38 +10,10 @@
  * 2 as published by the Free Software Foundation.
  */
 
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
-#include <linux/cpumask.h>
 #include <linux/kernel.h>
-#include <linux/kobject.h>
-#include <linux/list.h>
-#include <linux/notifier.h>
 #include <linux/of.h>
-#include <linux/percpu.h>
-#include <linux/slab.h>
-#include <asm/prom.h>
-
-#include "cacheinfo.h"
-
-/* per-cpu object for tracking:
- * - a "cache" kobject for the top-level directory
- * - a list of "index" objects representing the cpu's local cache hierarchy
- */
-struct cache_dir {
-	struct kobject *kobj; /* bare (not embedded) kobject for cache
-			       * directory */
-	struct cache_index_dir *index; /* list of index objects */
-};
-
-/* "index" object: each cpu's cache directory has an index
- * subdirectory corresponding to a cache object associated with the
- * cpu.  This object's lifetime is managed via the embedded kobject.
- */
-struct cache_index_dir {
-	struct kobject kobj;
-	struct cache_index_dir *next; /* next index in parent directory */
-	struct cache *cache;
-};
 
 /* Template for determining which OF properties to query for a given
  * cache type */
@@ -60,11 +32,6 @@ struct cache_type_info {
 	const char *nr_sets_prop;
 };
 
-/* These are used to index the cache_type_info array. */
-#define CACHE_TYPE_UNIFIED     0
-#define CACHE_TYPE_INSTRUCTION 1
-#define CACHE_TYPE_DATA        2
-
 static const struct cache_type_info cache_type_info[] = {
 	{
 		/* PowerPC Processor binding says the [di]-cache-*
@@ -92,231 +59,83 @@ static const struct cache_type_info cache_type_info[] = {
 	},
 };
 
-/* Cache object: each instance of this corresponds to a distinct cache
- * in the system.  There are separate objects for Harvard caches: one
- * each for instruction and data, and each refers to the same OF node.
- * The refcount of the OF node is elevated for the lifetime of the
- * cache object.  A cache object is released when its shared_cpu_map
- * is cleared (see cache_cpu_clear).
- *
- * A cache object is on two lists: an unsorted global list
- * (cache_list) of cache objects; and a singly-linked list
- * representing the local cache hierarchy, which is ordered by level
- * (e.g. L1d -> L1i -> L2 -> L3).
- */
-struct cache {
-	struct device_node *ofnode;    /* OF node for this cache, may be cpu */
-	struct cpumask shared_cpu_map; /* online CPUs using this cache */
-	int type;                      /* split cache disambiguation */
-	int level;                     /* level not explicit in device tree */
-	struct list_head list;         /* global list of cache objects */
-	struct cache *next_local;      /* next cache of >= level */
-};
-
-static DEFINE_PER_CPU(struct cache_dir *, cache_dir_pcpu);
-
-/* traversal/modification of this list occurs only at cpu hotplug time;
- * access is serialized by cpu hotplug locking
- */
-static LIST_HEAD(cache_list);
-
-static struct cache_index_dir *kobj_to_cache_index_dir(struct kobject *k)
-{
-	return container_of(k, struct cache_index_dir, kobj);
-}
-
-static const char *cache_type_string(const struct cache *cache)
+static inline int get_cacheinfo_idx(enum cache_type type)
 {
-	return cache_type_info[cache->type].name;
-}
-
-static void cache_init(struct cache *cache, int type, int level,
-		       struct device_node *ofnode)
-{
-	cache->type = type;
-	cache->level = level;
-	cache->ofnode = of_node_get(ofnode);
-	INIT_LIST_HEAD(&cache->list);
-	list_add(&cache->list, &cache_list);
-}
-
-static struct cache *new_cache(int type, int level, struct device_node *ofnode)
-{
-	struct cache *cache;
-
-	cache = kzalloc(sizeof(*cache), GFP_KERNEL);
-	if (cache)
-		cache_init(cache, type, level, ofnode);
-
-	return cache;
-}
-
-static void release_cache_debugcheck(struct cache *cache)
-{
-	struct cache *iter;
-
-	list_for_each_entry(iter, &cache_list, list)
-		WARN_ONCE(iter->next_local == cache,
-			  "cache for %s(%s) refers to cache for %s(%s)\n",
-			  iter->ofnode->full_name,
-			  cache_type_string(iter),
-			  cache->ofnode->full_name,
-			  cache_type_string(cache));
-}
-
-static void release_cache(struct cache *cache)
-{
-	if (!cache)
-		return;
-
-	pr_debug("freeing L%d %s cache for %s\n", cache->level,
-		 cache_type_string(cache), cache->ofnode->full_name);
-
-	release_cache_debugcheck(cache);
-	list_del(&cache->list);
-	of_node_put(cache->ofnode);
-	kfree(cache);
-}
-
-static void cache_cpu_set(struct cache *cache, int cpu)
-{
-	struct cache *next = cache;
-
-	while (next) {
-		WARN_ONCE(cpumask_test_cpu(cpu, &next->shared_cpu_map),
-			  "CPU %i already accounted in %s(%s)\n",
-			  cpu, next->ofnode->full_name,
-			  cache_type_string(next));
-		cpumask_set_cpu(cpu, &next->shared_cpu_map);
-		next = next->next_local;
-	}
+	if (type == CACHE_TYPE_UNIFIED)
+		return 0;
+	else
+		return type;
 }
 
-static int cache_size(const struct cache *cache, unsigned int *ret)
+static void cache_size(struct cacheinfo *this_leaf)
 {
 	const char *propname;
 	const __be32 *cache_size;
+	int ct_idx;
 
-	propname = cache_type_info[cache->type].size_prop;
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	propname = cache_type_info[ct_idx].size_prop;
 
-	cache_size = of_get_property(cache->ofnode, propname, NULL);
+	cache_size = of_get_property(this_leaf->of_node, propname, NULL);
 	if (!cache_size)
-		return -ENODEV;
-
-	*ret = of_read_number(cache_size, 1);
-	return 0;
-}
-
-static int cache_size_kb(const struct cache *cache, unsigned int *ret)
-{
-	unsigned int size;
-
-	if (cache_size(cache, &size))
-		return -ENODEV;
-
-	*ret = size / 1024;
-	return 0;
+		this_leaf->size = 0;
+	else
+		this_leaf->size = of_read_number(cache_size, 1);
 }
 
 /* not cache_line_size() because that's a macro in include/linux/cache.h */
-static int cache_get_line_size(const struct cache *cache, unsigned int *ret)
+static void cache_get_line_size(struct cacheinfo *this_leaf)
 {
 	const __be32 *line_size;
-	int i, lim;
+	int i, lim, ct_idx;
 
-	lim = ARRAY_SIZE(cache_type_info[cache->type].line_size_props);
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	lim = ARRAY_SIZE(cache_type_info[ct_idx].line_size_props);
 
 	for (i = 0; i < lim; i++) {
 		const char *propname;
 
-		propname = cache_type_info[cache->type].line_size_props[i];
-		line_size = of_get_property(cache->ofnode, propname, NULL);
+		propname = cache_type_info[ct_idx].line_size_props[i];
+		line_size = of_get_property(this_leaf->of_node, propname, NULL);
 		if (line_size)
 			break;
 	}
 
 	if (!line_size)
-		return -ENODEV;
-
-	*ret = of_read_number(line_size, 1);
-	return 0;
+		this_leaf->coherency_line_size = 0;
+	else
+		this_leaf->coherency_line_size = of_read_number(line_size, 1);
 }
 
-static int cache_nr_sets(const struct cache *cache, unsigned int *ret)
+static void cache_nr_sets(struct cacheinfo *this_leaf)
 {
 	const char *propname;
 	const __be32 *nr_sets;
+	int ct_idx;
 
-	propname = cache_type_info[cache->type].nr_sets_prop;
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	propname = cache_type_info[ct_idx].nr_sets_prop;
 
-	nr_sets = of_get_property(cache->ofnode, propname, NULL);
+	nr_sets = of_get_property(this_leaf->of_node, propname, NULL);
 	if (!nr_sets)
-		return -ENODEV;
-
-	*ret = of_read_number(nr_sets, 1);
-	return 0;
+		this_leaf->number_of_sets = 0;
+	else
+		this_leaf->number_of_sets = of_read_number(nr_sets, 1);
 }
 
-static int cache_associativity(const struct cache *cache, unsigned int *ret)
+static void cache_associativity(struct cacheinfo *this_leaf)
 {
-	unsigned int line_size;
-	unsigned int nr_sets;
-	unsigned int size;
-
-	if (cache_nr_sets(cache, &nr_sets))
-		goto err;
+	unsigned int line_size = this_leaf->coherency_line_size;
+	unsigned int nr_sets = this_leaf->number_of_sets;
+	unsigned int size = this_leaf->size;
 
 	/* If the cache is fully associative, there is no need to
 	 * check the other properties.
 	 */
-	if (nr_sets == 1) {
-		*ret = 0;
-		return 0;
-	}
-
-	if (cache_get_line_size(cache, &line_size))
-		goto err;
-	if (cache_size(cache, &size))
-		goto err;
-
-	if (!(nr_sets > 0 && size > 0 && line_size > 0))
-		goto err;
-
-	*ret = (size / nr_sets) / line_size;
-	return 0;
-err:
-	return -ENODEV;
-}
-
-/* helper for dealing with split caches */
-static struct cache *cache_find_first_sibling(struct cache *cache)
-{
-	struct cache *iter;
-
-	if (cache->type == CACHE_TYPE_UNIFIED)
-		return cache;
-
-	list_for_each_entry(iter, &cache_list, list)
-		if (iter->ofnode == cache->ofnode && iter->next_local == cache)
-			return iter;
-
-	return cache;
-}
-
-/* return the first cache on a local list matching node */
-static struct cache *cache_lookup_by_node(const struct device_node *node)
-{
-	struct cache *cache = NULL;
-	struct cache *iter;
-
-	list_for_each_entry(iter, &cache_list, list) {
-		if (iter->ofnode != node)
-			continue;
-		cache = cache_find_first_sibling(iter);
-		break;
-	}
-
-	return cache;
+	if ((nr_sets == 1) || !(nr_sets > 0 && size > 0 && line_size > 0))
+		this_leaf->ways_of_associativity = 0;
+	else
+		this_leaf->ways_of_associativity = (size / nr_sets) / line_size;
 }
 
 static bool cache_node_is_unified(const struct device_node *np)
@@ -324,526 +143,74 @@ static bool cache_node_is_unified(const struct device_node *np)
 	return of_get_property(np, "cache-unified", NULL);
 }
 
-static struct cache *cache_do_one_devnode_unified(struct device_node *node,
-						  int level)
-{
-	struct cache *cache;
-
-	pr_debug("creating L%d ucache for %s\n", level, node->full_name);
-
-	cache = new_cache(CACHE_TYPE_UNIFIED, level, node);
-
-	return cache;
-}
-
-static struct cache *cache_do_one_devnode_split(struct device_node *node,
-						int level)
-{
-	struct cache *dcache, *icache;
-
-	pr_debug("creating L%d dcache and icache for %s\n", level,
-		 node->full_name);
-
-	dcache = new_cache(CACHE_TYPE_DATA, level, node);
-	icache = new_cache(CACHE_TYPE_INSTRUCTION, level, node);
-
-	if (!dcache || !icache)
-		goto err;
-
-	dcache->next_local = icache;
-
-	return dcache;
-err:
-	release_cache(dcache);
-	release_cache(icache);
-	return NULL;
-}
-
-static struct cache *cache_do_one_devnode(struct device_node *node, int level)
-{
-	struct cache *cache;
-
-	if (cache_node_is_unified(node))
-		cache = cache_do_one_devnode_unified(node, level);
-	else
-		cache = cache_do_one_devnode_split(node, level);
-
-	return cache;
-}
-
-static struct cache *cache_lookup_or_instantiate(struct device_node *node,
-						 int level)
-{
-	struct cache *cache;
-
-	cache = cache_lookup_by_node(node);
-
-	WARN_ONCE(cache && cache->level != level,
-		  "cache level mismatch on lookup (got %d, expected %d)\n",
-		  cache->level, level);
-
-	if (!cache)
-		cache = cache_do_one_devnode(node, level);
-
-	return cache;
-}
-
-static void link_cache_lists(struct cache *smaller, struct cache *bigger)
-{
-	while (smaller->next_local) {
-		if (smaller->next_local == bigger)
-			return; /* already linked */
-		smaller = smaller->next_local;
-	}
-
-	smaller->next_local = bigger;
-}
-
-static void do_subsidiary_caches_debugcheck(struct cache *cache)
-{
-	WARN_ON_ONCE(cache->level != 1);
-	WARN_ON_ONCE(strcmp(cache->ofnode->type, "cpu"));
-}
-
-static void do_subsidiary_caches(struct cache *cache)
-{
-	struct device_node *subcache_node;
-	int level = cache->level;
-
-	do_subsidiary_caches_debugcheck(cache);
-
-	while ((subcache_node = of_find_next_cache_node(cache->ofnode))) {
-		struct cache *subcache;
-
-		level++;
-		subcache = cache_lookup_or_instantiate(subcache_node, level);
-		of_node_put(subcache_node);
-		if (!subcache)
-			break;
-
-		link_cache_lists(cache, subcache);
-		cache = subcache;
-	}
-}
-
-static struct cache *cache_chain_instantiate(unsigned int cpu_id)
-{
-	struct device_node *cpu_node;
-	struct cache *cpu_cache = NULL;
-
-	pr_debug("creating cache object(s) for CPU %i\n", cpu_id);
-
-	cpu_node = of_get_cpu_node(cpu_id, NULL);
-	WARN_ONCE(!cpu_node, "no OF node found for CPU %i\n", cpu_id);
-	if (!cpu_node)
-		goto out;
-
-	cpu_cache = cache_lookup_or_instantiate(cpu_node, 1);
-	if (!cpu_cache)
-		goto out;
-
-	do_subsidiary_caches(cpu_cache);
-
-	cache_cpu_set(cpu_cache, cpu_id);
-out:
-	of_node_put(cpu_node);
-
-	return cpu_cache;
-}
-
-static struct cache_dir *cacheinfo_create_cache_dir(unsigned int cpu_id)
-{
-	struct cache_dir *cache_dir;
-	struct device *dev;
-	struct kobject *kobj = NULL;
-
-	dev = get_cpu_device(cpu_id);
-	WARN_ONCE(!dev, "no dev for CPU %i\n", cpu_id);
-	if (!dev)
-		goto err;
-
-	kobj = kobject_create_and_add("cache", &dev->kobj);
-	if (!kobj)
-		goto err;
-
-	cache_dir = kzalloc(sizeof(*cache_dir), GFP_KERNEL);
-	if (!cache_dir)
-		goto err;
-
-	cache_dir->kobj = kobj;
-
-	WARN_ON_ONCE(per_cpu(cache_dir_pcpu, cpu_id) != NULL);
-
-	per_cpu(cache_dir_pcpu, cpu_id) = cache_dir;
-
-	return cache_dir;
-err:
-	kobject_put(kobj);
-	return NULL;
-}
-
-static void cache_index_release(struct kobject *kobj)
-{
-	struct cache_index_dir *index;
-
-	index = kobj_to_cache_index_dir(kobj);
-
-	pr_debug("freeing index directory for L%d %s cache\n",
-		 index->cache->level, cache_type_string(index->cache));
-
-	kfree(index);
-}
-
-static ssize_t cache_index_show(struct kobject *k, struct attribute *attr, char *buf)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
 {
-	struct kobj_attribute *kobj_attr;
-
-	kobj_attr = container_of(attr, struct kobj_attribute, attr);
-
-	return kobj_attr->show(k, kobj_attr, buf);
+	this_leaf->level = level;
+	this_leaf->type = type;
+	cache_size(this_leaf);
+	cache_get_line_size(this_leaf);
+	cache_nr_sets(this_leaf);
+	cache_associativity(this_leaf);
 }
 
-static struct cache *index_kobj_to_cache(struct kobject *k)
+int init_cache_level(unsigned int cpu)
 {
-	struct cache_index_dir *index;
+	struct device_node *np;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int level = 0, leaves = 0;
 
-	index = kobj_to_cache_index_dir(k);
-
-	return index->cache;
-}
-
-static ssize_t size_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int size_kb;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_size_kb(cache, &size_kb))
+	if (!cpu_dev) {
+		pr_err("No cpu device for CPU %d\n", cpu);
 		return -ENODEV;
-
-	return sprintf(buf, "%uK\n", size_kb);
-}
-
-static struct kobj_attribute cache_size_attr =
-	__ATTR(size, 0444, size_show, NULL);
-
-
-static ssize_t line_size_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int line_size;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_get_line_size(cache, &line_size))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", line_size);
-}
-
-static struct kobj_attribute cache_line_size_attr =
-	__ATTR(coherency_line_size, 0444, line_size_show, NULL);
-
-static ssize_t nr_sets_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int nr_sets;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_nr_sets(cache, &nr_sets))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", nr_sets);
-}
-
-static struct kobj_attribute cache_nr_sets_attr =
-	__ATTR(number_of_sets, 0444, nr_sets_show, NULL);
-
-static ssize_t associativity_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int associativity;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_associativity(cache, &associativity))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", associativity);
-}
-
-static struct kobj_attribute cache_assoc_attr =
-	__ATTR(ways_of_associativity, 0444, associativity_show, NULL);
-
-static ssize_t type_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	return sprintf(buf, "%s\n", cache_type_string(cache));
-}
-
-static struct kobj_attribute cache_type_attr =
-	__ATTR(type, 0444, type_show, NULL);
-
-static ssize_t level_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache_index_dir *index;
-	struct cache *cache;
-
-	index = kobj_to_cache_index_dir(k);
-	cache = index->cache;
-
-	return sprintf(buf, "%d\n", cache->level);
-}
-
-static struct kobj_attribute cache_level_attr =
-	__ATTR(level, 0444, level_show, NULL);
-
-static ssize_t shared_cpu_map_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache_index_dir *index;
-	struct cache *cache;
-	int len;
-	int n = 0;
-
-	index = kobj_to_cache_index_dir(k);
-	cache = index->cache;
-	len = PAGE_SIZE - 2;
-
-	if (len > 1) {
-		n = cpumask_scnprintf(buf, len, &cache->shared_cpu_map);
-		buf[n++] = '\n';
-		buf[n] = '\0';
 	}
-	return n;
-}
-
-static struct kobj_attribute cache_shared_cpu_map_attr =
-	__ATTR(shared_cpu_map, 0444, shared_cpu_map_show, NULL);
-
-/* Attributes which should always be created -- the kobject/sysfs core
- * does this automatically via kobj_type->default_attrs.  This is the
- * minimum data required to uniquely identify a cache.
- */
-static struct attribute *cache_index_default_attrs[] = {
-	&cache_type_attr.attr,
-	&cache_level_attr.attr,
-	&cache_shared_cpu_map_attr.attr,
-	NULL,
-};
-
-/* Attributes which should be created if the cache device node has the
- * right properties -- see cacheinfo_create_index_opt_attrs
- */
-static struct kobj_attribute *cache_index_opt_attrs[] = {
-	&cache_size_attr,
-	&cache_line_size_attr,
-	&cache_nr_sets_attr,
-	&cache_assoc_attr,
-};
-
-static const struct sysfs_ops cache_index_ops = {
-	.show = cache_index_show,
-};
-
-static struct kobj_type cache_index_type = {
-	.release = cache_index_release,
-	.sysfs_ops = &cache_index_ops,
-	.default_attrs = cache_index_default_attrs,
-};
-
-static void cacheinfo_create_index_opt_attrs(struct cache_index_dir *dir)
-{
-	const char *cache_name;
-	const char *cache_type;
-	struct cache *cache;
-	char *buf;
-	int i;
-
-	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
-	if (!buf)
-		return;
-
-	cache = dir->cache;
-	cache_name = cache->ofnode->full_name;
-	cache_type = cache_type_string(cache);
-
-	/* We don't want to create an attribute that can't provide a
-	 * meaningful value.  Check the return value of each optional
-	 * attribute's ->show method before registering the
-	 * attribute.
-	 */
-	for (i = 0; i < ARRAY_SIZE(cache_index_opt_attrs); i++) {
-		struct kobj_attribute *attr;
-		ssize_t rc;
-
-		attr = cache_index_opt_attrs[i];
-
-		rc = attr->show(&dir->kobj, attr, buf);
-		if (rc <= 0) {
-			pr_debug("not creating %s attribute for "
-				 "%s(%s) (rc = %zd)\n",
-				 attr->attr.name, cache_name,
-				 cache_type, rc);
-			continue;
-		}
-		if (sysfs_create_file(&dir->kobj, &attr->attr))
-			pr_debug("could not create %s attribute for %s(%s)\n",
-				 attr->attr.name, cache_name, cache_type);
+	np = cpu_dev->of_node;
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
 	}
 
-	kfree(buf);
-}
-
-static void cacheinfo_create_index_dir(struct cache *cache, int index,
-				       struct cache_dir *cache_dir)
-{
-	struct cache_index_dir *index_dir;
-	int rc;
-
-	index_dir = kzalloc(sizeof(*index_dir), GFP_KERNEL);
-	if (!index_dir)
-		goto err;
-
-	index_dir->cache = cache;
-
-	rc = kobject_init_and_add(&index_dir->kobj, &cache_index_type,
-				  cache_dir->kobj, "index%d", index);
-	if (rc)
-		goto err;
-
-	index_dir->next = cache_dir->index;
-	cache_dir->index = index_dir;
-
-	cacheinfo_create_index_opt_attrs(index_dir);
-
-	return;
-err:
-	kfree(index_dir);
-}
-
-static void cacheinfo_sysfs_populate(unsigned int cpu_id,
-				     struct cache *cache_list)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-	int index = 0;
-
-	cache_dir = cacheinfo_create_cache_dir(cpu_id);
-	if (!cache_dir)
-		return;
-
-	cache = cache_list;
-	while (cache) {
-		cacheinfo_create_index_dir(cache, index, cache_dir);
-		index++;
-		cache = cache->next_local;
+	while (np) {
+		leaves += cache_node_is_unified(np) ? 1 : 2;
+		level++;
+		of_node_put(np);
+		np = of_find_next_cache_node(np);
 	}
-}
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
 
-void cacheinfo_cpu_online(unsigned int cpu_id)
-{
-	struct cache *cache;
-
-	cache = cache_chain_instantiate(cpu_id);
-	if (!cache)
-		return;
-
-	cacheinfo_sysfs_populate(cpu_id, cache);
-}
-
-/* functions needed to remove cache entry for cpu offline or suspend/resume */
-
-#if (defined(CONFIG_PPC_PSERIES) && defined(CONFIG_SUSPEND)) || \
-    defined(CONFIG_HOTPLUG_CPU)
-
-static struct cache *cache_lookup_by_cpu(unsigned int cpu_id)
-{
-	struct device_node *cpu_node;
-	struct cache *cache;
-
-	cpu_node = of_get_cpu_node(cpu_id, NULL);
-	WARN_ONCE(!cpu_node, "no OF node found for CPU %i\n", cpu_id);
-	if (!cpu_node)
-		return NULL;
-
-	cache = cache_lookup_by_node(cpu_node);
-	of_node_put(cpu_node);
-
-	return cache;
+	return 0;
 }
 
-static void remove_index_dirs(struct cache_dir *cache_dir)
+int populate_cache_leaves(unsigned int cpu)
 {
-	struct cache_index_dir *index;
-
-	index = cache_dir->index;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct device_node *np;
+	unsigned int level, idx;
 
-	while (index) {
-		struct cache_index_dir *next;
-
-		next = index->next;
-		kobject_put(&index->kobj);
-		index = next;
+	np = of_node_get(cpu_dev->of_node);
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
 	}
-}
-
-static void remove_cache_dir(struct cache_dir *cache_dir)
-{
-	remove_index_dirs(cache_dir);
 
-	/* Remove cache dir from sysfs */
-	kobject_del(cache_dir->kobj);
-
-	kobject_put(cache_dir->kobj);
-
-	kfree(cache_dir);
-}
-
-static void cache_cpu_clear(struct cache *cache, int cpu)
-{
-	while (cache) {
-		struct cache *next = cache->next_local;
-
-		WARN_ONCE(!cpumask_test_cpu(cpu, &cache->shared_cpu_map),
-			  "CPU %i not accounted in %s(%s)\n",
-			  cpu, cache->ofnode->full_name,
-			  cache_type_string(cache));
-
-		cpumask_clear_cpu(cpu, &cache->shared_cpu_map);
-
-		/* Release the cache object if all the cpus using it
-		 * are offline */
-		if (cpumask_empty(&cache->shared_cpu_map))
-			release_cache(cache);
-
-		cache = next;
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		this_leaf->of_node = np;
+		if (cache_node_is_unified(np)) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_UNIFIED, level);
+		} else {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		}
+		np = of_find_next_cache_node(np);
 	}
+	return 0;
 }
 
-void cacheinfo_cpu_offline(unsigned int cpu_id)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-
-	/* Prevent userspace from seeing inconsistent state - remove
-	 * the sysfs hierarchy first */
-	cache_dir = per_cpu(cache_dir_pcpu, cpu_id);
-
-	/* careful, sysfs population may have failed */
-	if (cache_dir)
-		remove_cache_dir(cache_dir);
-
-	per_cpu(cache_dir_pcpu, cpu_id) = NULL;
-
-	/* clear the CPU's bit in its cache chain, possibly freeing
-	 * cache objects */
-	cache = cache_lookup_by_cpu(cpu_id);
-	if (cache)
-		cache_cpu_clear(cache, cpu_id);
-}
-#endif /* (CONFIG_PPC_PSERIES && CONFIG_SUSPEND) || CONFIG_HOTPLUG_CPU */
diff --git a/arch/powerpc/kernel/cacheinfo.h b/arch/powerpc/kernel/cacheinfo.h
deleted file mode 100644
index a7b74d3..0000000
--- a/arch/powerpc/kernel/cacheinfo.h
+++ /dev/null
@@ -1,8 +0,0 @@
-#ifndef _PPC_CACHEINFO_H
-#define _PPC_CACHEINFO_H
-
-/* These are just hooks for sysfs.c to use. */
-extern void cacheinfo_cpu_online(unsigned int cpu_id);
-extern void cacheinfo_cpu_offline(unsigned int cpu_id);
-
-#endif /* _PPC_CACHEINFO_H */
diff --git a/arch/powerpc/kernel/sysfs.c b/arch/powerpc/kernel/sysfs.c
index 67fd2fd..6e9c5a8 100644
--- a/arch/powerpc/kernel/sysfs.c
+++ b/arch/powerpc/kernel/sysfs.c
@@ -19,8 +19,6 @@
 #include <asm/pmc.h>
 #include <asm/firmware.h>
 
-#include "cacheinfo.h"
-
 #ifdef CONFIG_PPC64
 #include <asm/paca.h>
 #include <asm/lppaca.h>
@@ -743,7 +741,6 @@ static void register_cpu_online(unsigned int cpu)
 		device_create_file(s, &dev_attr_altivec_idle_wait_time);
 	}
 #endif
-	cacheinfo_cpu_online(cpu);
 }
 
 #ifdef CONFIG_HOTPLUG_CPU
@@ -824,7 +821,6 @@ static void unregister_cpu_online(unsigned int cpu)
 		device_remove_file(s, &dev_attr_altivec_idle_wait_time);
 	}
 #endif
-	cacheinfo_cpu_offline(cpu);
 }
 
 #ifdef CONFIG_ARCH_CPU_PROBE_RELEASE
@@ -988,8 +984,7 @@ static int __init topology_init(void)
 	int cpu;
 
 	register_nodes();
-
-	cpu_notifier_register_begin();
+	register_cpu_notifier(&sysfs_cpu_nb);
 
 	for_each_possible_cpu(cpu) {
 		struct cpu *c = &per_cpu(cpu_devices, cpu);
@@ -1013,11 +1008,6 @@ static int __init topology_init(void)
 		if (cpu_online(cpu))
 			register_cpu_online(cpu);
 	}
-
-	__register_cpu_notifier(&sysfs_cpu_nb);
-
-	cpu_notifier_register_done();
-
 #ifdef CONFIG_PPC64
 	sysfs_create_dscr_default();
 #endif /* CONFIG_PPC64 */
-- 
1.8.3.2

[PATCH 6/9] powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Anshuman Khandual <khandual@linux.vnet.ibm.com>
Cc: linuxppc-dev@lists.ozlabs.org
---
 arch/powerpc/kernel/cacheinfo.c | 813 +++++-----------------------------------
 arch/powerpc/kernel/cacheinfo.h |   8 -
 arch/powerpc/kernel/sysfs.c     |  12 +-
 3 files changed, 91 insertions(+), 742 deletions(-)
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h

diff --git a/arch/powerpc/kernel/cacheinfo.c b/arch/powerpc/kernel/cacheinfo.c
index 40198d5..b871c24 100644
--- a/arch/powerpc/kernel/cacheinfo.c
+++ b/arch/powerpc/kernel/cacheinfo.c
@@ -10,38 +10,10 @@
  * 2 as published by the Free Software Foundation.
  */
 
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
-#include <linux/cpumask.h>
 #include <linux/kernel.h>
-#include <linux/kobject.h>
-#include <linux/list.h>
-#include <linux/notifier.h>
 #include <linux/of.h>
-#include <linux/percpu.h>
-#include <linux/slab.h>
-#include <asm/prom.h>
-
-#include "cacheinfo.h"
-
-/* per-cpu object for tracking:
- * - a "cache" kobject for the top-level directory
- * - a list of "index" objects representing the cpu's local cache hierarchy
- */
-struct cache_dir {
-	struct kobject *kobj; /* bare (not embedded) kobject for cache
-			       * directory */
-	struct cache_index_dir *index; /* list of index objects */
-};
-
-/* "index" object: each cpu's cache directory has an index
- * subdirectory corresponding to a cache object associated with the
- * cpu.  This object's lifetime is managed via the embedded kobject.
- */
-struct cache_index_dir {
-	struct kobject kobj;
-	struct cache_index_dir *next; /* next index in parent directory */
-	struct cache *cache;
-};
 
 /* Template for determining which OF properties to query for a given
  * cache type */
@@ -60,11 +32,6 @@ struct cache_type_info {
 	const char *nr_sets_prop;
 };
 
-/* These are used to index the cache_type_info array. */
-#define CACHE_TYPE_UNIFIED     0
-#define CACHE_TYPE_INSTRUCTION 1
-#define CACHE_TYPE_DATA        2
-
 static const struct cache_type_info cache_type_info[] = {
 	{
 		/* PowerPC Processor binding says the [di]-cache-*
@@ -92,231 +59,83 @@ static const struct cache_type_info cache_type_info[] = {
 	},
 };
 
-/* Cache object: each instance of this corresponds to a distinct cache
- * in the system.  There are separate objects for Harvard caches: one
- * each for instruction and data, and each refers to the same OF node.
- * The refcount of the OF node is elevated for the lifetime of the
- * cache object.  A cache object is released when its shared_cpu_map
- * is cleared (see cache_cpu_clear).
- *
- * A cache object is on two lists: an unsorted global list
- * (cache_list) of cache objects; and a singly-linked list
- * representing the local cache hierarchy, which is ordered by level
- * (e.g. L1d -> L1i -> L2 -> L3).
- */
-struct cache {
-	struct device_node *ofnode;    /* OF node for this cache, may be cpu */
-	struct cpumask shared_cpu_map; /* online CPUs using this cache */
-	int type;                      /* split cache disambiguation */
-	int level;                     /* level not explicit in device tree */
-	struct list_head list;         /* global list of cache objects */
-	struct cache *next_local;      /* next cache of >= level */
-};
-
-static DEFINE_PER_CPU(struct cache_dir *, cache_dir_pcpu);
-
-/* traversal/modification of this list occurs only at cpu hotplug time;
- * access is serialized by cpu hotplug locking
- */
-static LIST_HEAD(cache_list);
-
-static struct cache_index_dir *kobj_to_cache_index_dir(struct kobject *k)
-{
-	return container_of(k, struct cache_index_dir, kobj);
-}
-
-static const char *cache_type_string(const struct cache *cache)
+static inline int get_cacheinfo_idx(enum cache_type type)
 {
-	return cache_type_info[cache->type].name;
-}
-
-static void cache_init(struct cache *cache, int type, int level,
-		       struct device_node *ofnode)
-{
-	cache->type = type;
-	cache->level = level;
-	cache->ofnode = of_node_get(ofnode);
-	INIT_LIST_HEAD(&cache->list);
-	list_add(&cache->list, &cache_list);
-}
-
-static struct cache *new_cache(int type, int level, struct device_node *ofnode)
-{
-	struct cache *cache;
-
-	cache = kzalloc(sizeof(*cache), GFP_KERNEL);
-	if (cache)
-		cache_init(cache, type, level, ofnode);
-
-	return cache;
-}
-
-static void release_cache_debugcheck(struct cache *cache)
-{
-	struct cache *iter;
-
-	list_for_each_entry(iter, &cache_list, list)
-		WARN_ONCE(iter->next_local == cache,
-			  "cache for %s(%s) refers to cache for %s(%s)\n",
-			  iter->ofnode->full_name,
-			  cache_type_string(iter),
-			  cache->ofnode->full_name,
-			  cache_type_string(cache));
-}
-
-static void release_cache(struct cache *cache)
-{
-	if (!cache)
-		return;
-
-	pr_debug("freeing L%d %s cache for %s\n", cache->level,
-		 cache_type_string(cache), cache->ofnode->full_name);
-
-	release_cache_debugcheck(cache);
-	list_del(&cache->list);
-	of_node_put(cache->ofnode);
-	kfree(cache);
-}
-
-static void cache_cpu_set(struct cache *cache, int cpu)
-{
-	struct cache *next = cache;
-
-	while (next) {
-		WARN_ONCE(cpumask_test_cpu(cpu, &next->shared_cpu_map),
-			  "CPU %i already accounted in %s(%s)\n",
-			  cpu, next->ofnode->full_name,
-			  cache_type_string(next));
-		cpumask_set_cpu(cpu, &next->shared_cpu_map);
-		next = next->next_local;
-	}
+	if (type == CACHE_TYPE_UNIFIED)
+		return 0;
+	else
+		return type;
 }
 
-static int cache_size(const struct cache *cache, unsigned int *ret)
+static void cache_size(struct cacheinfo *this_leaf)
 {
 	const char *propname;
 	const __be32 *cache_size;
+	int ct_idx;
 
-	propname = cache_type_info[cache->type].size_prop;
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	propname = cache_type_info[ct_idx].size_prop;
 
-	cache_size = of_get_property(cache->ofnode, propname, NULL);
+	cache_size = of_get_property(this_leaf->of_node, propname, NULL);
 	if (!cache_size)
-		return -ENODEV;
-
-	*ret = of_read_number(cache_size, 1);
-	return 0;
-}
-
-static int cache_size_kb(const struct cache *cache, unsigned int *ret)
-{
-	unsigned int size;
-
-	if (cache_size(cache, &size))
-		return -ENODEV;
-
-	*ret = size / 1024;
-	return 0;
+		this_leaf->size = 0;
+	else
+		this_leaf->size = of_read_number(cache_size, 1);
 }
 
 /* not cache_line_size() because that's a macro in include/linux/cache.h */
-static int cache_get_line_size(const struct cache *cache, unsigned int *ret)
+static void cache_get_line_size(struct cacheinfo *this_leaf)
 {
 	const __be32 *line_size;
-	int i, lim;
+	int i, lim, ct_idx;
 
-	lim = ARRAY_SIZE(cache_type_info[cache->type].line_size_props);
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	lim = ARRAY_SIZE(cache_type_info[ct_idx].line_size_props);
 
 	for (i = 0; i < lim; i++) {
 		const char *propname;
 
-		propname = cache_type_info[cache->type].line_size_props[i];
-		line_size = of_get_property(cache->ofnode, propname, NULL);
+		propname = cache_type_info[ct_idx].line_size_props[i];
+		line_size = of_get_property(this_leaf->of_node, propname, NULL);
 		if (line_size)
 			break;
 	}
 
 	if (!line_size)
-		return -ENODEV;
-
-	*ret = of_read_number(line_size, 1);
-	return 0;
+		this_leaf->coherency_line_size = 0;
+	else
+		this_leaf->coherency_line_size = of_read_number(line_size, 1);
 }
 
-static int cache_nr_sets(const struct cache *cache, unsigned int *ret)
+static void cache_nr_sets(struct cacheinfo *this_leaf)
 {
 	const char *propname;
 	const __be32 *nr_sets;
+	int ct_idx;
 
-	propname = cache_type_info[cache->type].nr_sets_prop;
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	propname = cache_type_info[ct_idx].nr_sets_prop;
 
-	nr_sets = of_get_property(cache->ofnode, propname, NULL);
+	nr_sets = of_get_property(this_leaf->of_node, propname, NULL);
 	if (!nr_sets)
-		return -ENODEV;
-
-	*ret = of_read_number(nr_sets, 1);
-	return 0;
+		this_leaf->number_of_sets = 0;
+	else
+		this_leaf->number_of_sets = of_read_number(nr_sets, 1);
 }
 
-static int cache_associativity(const struct cache *cache, unsigned int *ret)
+static void cache_associativity(struct cacheinfo *this_leaf)
 {
-	unsigned int line_size;
-	unsigned int nr_sets;
-	unsigned int size;
-
-	if (cache_nr_sets(cache, &nr_sets))
-		goto err;
+	unsigned int line_size = this_leaf->coherency_line_size;
+	unsigned int nr_sets = this_leaf->number_of_sets;
+	unsigned int size = this_leaf->size;
 
 	/* If the cache is fully associative, there is no need to
 	 * check the other properties.
 	 */
-	if (nr_sets == 1) {
-		*ret = 0;
-		return 0;
-	}
-
-	if (cache_get_line_size(cache, &line_size))
-		goto err;
-	if (cache_size(cache, &size))
-		goto err;
-
-	if (!(nr_sets > 0 && size > 0 && line_size > 0))
-		goto err;
-
-	*ret = (size / nr_sets) / line_size;
-	return 0;
-err:
-	return -ENODEV;
-}
-
-/* helper for dealing with split caches */
-static struct cache *cache_find_first_sibling(struct cache *cache)
-{
-	struct cache *iter;
-
-	if (cache->type == CACHE_TYPE_UNIFIED)
-		return cache;
-
-	list_for_each_entry(iter, &cache_list, list)
-		if (iter->ofnode == cache->ofnode && iter->next_local == cache)
-			return iter;
-
-	return cache;
-}
-
-/* return the first cache on a local list matching node */
-static struct cache *cache_lookup_by_node(const struct device_node *node)
-{
-	struct cache *cache = NULL;
-	struct cache *iter;
-
-	list_for_each_entry(iter, &cache_list, list) {
-		if (iter->ofnode != node)
-			continue;
-		cache = cache_find_first_sibling(iter);
-		break;
-	}
-
-	return cache;
+	if ((nr_sets == 1) || !(nr_sets > 0 && size > 0 && line_size > 0))
+		this_leaf->ways_of_associativity = 0;
+	else
+		this_leaf->ways_of_associativity = (size / nr_sets) / line_size;
 }
 
 static bool cache_node_is_unified(const struct device_node *np)
@@ -324,526 +143,74 @@ static bool cache_node_is_unified(const struct device_node *np)
 	return of_get_property(np, "cache-unified", NULL);
 }
 
-static struct cache *cache_do_one_devnode_unified(struct device_node *node,
-						  int level)
-{
-	struct cache *cache;
-
-	pr_debug("creating L%d ucache for %s\n", level, node->full_name);
-
-	cache = new_cache(CACHE_TYPE_UNIFIED, level, node);
-
-	return cache;
-}
-
-static struct cache *cache_do_one_devnode_split(struct device_node *node,
-						int level)
-{
-	struct cache *dcache, *icache;
-
-	pr_debug("creating L%d dcache and icache for %s\n", level,
-		 node->full_name);
-
-	dcache = new_cache(CACHE_TYPE_DATA, level, node);
-	icache = new_cache(CACHE_TYPE_INSTRUCTION, level, node);
-
-	if (!dcache || !icache)
-		goto err;
-
-	dcache->next_local = icache;
-
-	return dcache;
-err:
-	release_cache(dcache);
-	release_cache(icache);
-	return NULL;
-}
-
-static struct cache *cache_do_one_devnode(struct device_node *node, int level)
-{
-	struct cache *cache;
-
-	if (cache_node_is_unified(node))
-		cache = cache_do_one_devnode_unified(node, level);
-	else
-		cache = cache_do_one_devnode_split(node, level);
-
-	return cache;
-}
-
-static struct cache *cache_lookup_or_instantiate(struct device_node *node,
-						 int level)
-{
-	struct cache *cache;
-
-	cache = cache_lookup_by_node(node);
-
-	WARN_ONCE(cache && cache->level != level,
-		  "cache level mismatch on lookup (got %d, expected %d)\n",
-		  cache->level, level);
-
-	if (!cache)
-		cache = cache_do_one_devnode(node, level);
-
-	return cache;
-}
-
-static void link_cache_lists(struct cache *smaller, struct cache *bigger)
-{
-	while (smaller->next_local) {
-		if (smaller->next_local == bigger)
-			return; /* already linked */
-		smaller = smaller->next_local;
-	}
-
-	smaller->next_local = bigger;
-}
-
-static void do_subsidiary_caches_debugcheck(struct cache *cache)
-{
-	WARN_ON_ONCE(cache->level != 1);
-	WARN_ON_ONCE(strcmp(cache->ofnode->type, "cpu"));
-}
-
-static void do_subsidiary_caches(struct cache *cache)
-{
-	struct device_node *subcache_node;
-	int level = cache->level;
-
-	do_subsidiary_caches_debugcheck(cache);
-
-	while ((subcache_node = of_find_next_cache_node(cache->ofnode))) {
-		struct cache *subcache;
-
-		level++;
-		subcache = cache_lookup_or_instantiate(subcache_node, level);
-		of_node_put(subcache_node);
-		if (!subcache)
-			break;
-
-		link_cache_lists(cache, subcache);
-		cache = subcache;
-	}
-}
-
-static struct cache *cache_chain_instantiate(unsigned int cpu_id)
-{
-	struct device_node *cpu_node;
-	struct cache *cpu_cache = NULL;
-
-	pr_debug("creating cache object(s) for CPU %i\n", cpu_id);
-
-	cpu_node = of_get_cpu_node(cpu_id, NULL);
-	WARN_ONCE(!cpu_node, "no OF node found for CPU %i\n", cpu_id);
-	if (!cpu_node)
-		goto out;
-
-	cpu_cache = cache_lookup_or_instantiate(cpu_node, 1);
-	if (!cpu_cache)
-		goto out;
-
-	do_subsidiary_caches(cpu_cache);
-
-	cache_cpu_set(cpu_cache, cpu_id);
-out:
-	of_node_put(cpu_node);
-
-	return cpu_cache;
-}
-
-static struct cache_dir *cacheinfo_create_cache_dir(unsigned int cpu_id)
-{
-	struct cache_dir *cache_dir;
-	struct device *dev;
-	struct kobject *kobj = NULL;
-
-	dev = get_cpu_device(cpu_id);
-	WARN_ONCE(!dev, "no dev for CPU %i\n", cpu_id);
-	if (!dev)
-		goto err;
-
-	kobj = kobject_create_and_add("cache", &dev->kobj);
-	if (!kobj)
-		goto err;
-
-	cache_dir = kzalloc(sizeof(*cache_dir), GFP_KERNEL);
-	if (!cache_dir)
-		goto err;
-
-	cache_dir->kobj = kobj;
-
-	WARN_ON_ONCE(per_cpu(cache_dir_pcpu, cpu_id) != NULL);
-
-	per_cpu(cache_dir_pcpu, cpu_id) = cache_dir;
-
-	return cache_dir;
-err:
-	kobject_put(kobj);
-	return NULL;
-}
-
-static void cache_index_release(struct kobject *kobj)
-{
-	struct cache_index_dir *index;
-
-	index = kobj_to_cache_index_dir(kobj);
-
-	pr_debug("freeing index directory for L%d %s cache\n",
-		 index->cache->level, cache_type_string(index->cache));
-
-	kfree(index);
-}
-
-static ssize_t cache_index_show(struct kobject *k, struct attribute *attr, char *buf)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
 {
-	struct kobj_attribute *kobj_attr;
-
-	kobj_attr = container_of(attr, struct kobj_attribute, attr);
-
-	return kobj_attr->show(k, kobj_attr, buf);
+	this_leaf->level = level;
+	this_leaf->type = type;
+	cache_size(this_leaf);
+	cache_get_line_size(this_leaf);
+	cache_nr_sets(this_leaf);
+	cache_associativity(this_leaf);
 }
 
-static struct cache *index_kobj_to_cache(struct kobject *k)
+int init_cache_level(unsigned int cpu)
 {
-	struct cache_index_dir *index;
+	struct device_node *np;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int level = 0, leaves = 0;
 
-	index = kobj_to_cache_index_dir(k);
-
-	return index->cache;
-}
-
-static ssize_t size_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int size_kb;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_size_kb(cache, &size_kb))
+	if (!cpu_dev) {
+		pr_err("No cpu device for CPU %d\n", cpu);
 		return -ENODEV;
-
-	return sprintf(buf, "%uK\n", size_kb);
-}
-
-static struct kobj_attribute cache_size_attr =
-	__ATTR(size, 0444, size_show, NULL);
-
-
-static ssize_t line_size_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int line_size;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_get_line_size(cache, &line_size))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", line_size);
-}
-
-static struct kobj_attribute cache_line_size_attr =
-	__ATTR(coherency_line_size, 0444, line_size_show, NULL);
-
-static ssize_t nr_sets_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int nr_sets;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_nr_sets(cache, &nr_sets))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", nr_sets);
-}
-
-static struct kobj_attribute cache_nr_sets_attr =
-	__ATTR(number_of_sets, 0444, nr_sets_show, NULL);
-
-static ssize_t associativity_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int associativity;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_associativity(cache, &associativity))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", associativity);
-}
-
-static struct kobj_attribute cache_assoc_attr =
-	__ATTR(ways_of_associativity, 0444, associativity_show, NULL);
-
-static ssize_t type_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	return sprintf(buf, "%s\n", cache_type_string(cache));
-}
-
-static struct kobj_attribute cache_type_attr =
-	__ATTR(type, 0444, type_show, NULL);
-
-static ssize_t level_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache_index_dir *index;
-	struct cache *cache;
-
-	index = kobj_to_cache_index_dir(k);
-	cache = index->cache;
-
-	return sprintf(buf, "%d\n", cache->level);
-}
-
-static struct kobj_attribute cache_level_attr =
-	__ATTR(level, 0444, level_show, NULL);
-
-static ssize_t shared_cpu_map_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache_index_dir *index;
-	struct cache *cache;
-	int len;
-	int n = 0;
-
-	index = kobj_to_cache_index_dir(k);
-	cache = index->cache;
-	len = PAGE_SIZE - 2;
-
-	if (len > 1) {
-		n = cpumask_scnprintf(buf, len, &cache->shared_cpu_map);
-		buf[n++] = '\n';
-		buf[n] = '\0';
 	}
-	return n;
-}
-
-static struct kobj_attribute cache_shared_cpu_map_attr =
-	__ATTR(shared_cpu_map, 0444, shared_cpu_map_show, NULL);
-
-/* Attributes which should always be created -- the kobject/sysfs core
- * does this automatically via kobj_type->default_attrs.  This is the
- * minimum data required to uniquely identify a cache.
- */
-static struct attribute *cache_index_default_attrs[] = {
-	&cache_type_attr.attr,
-	&cache_level_attr.attr,
-	&cache_shared_cpu_map_attr.attr,
-	NULL,
-};
-
-/* Attributes which should be created if the cache device node has the
- * right properties -- see cacheinfo_create_index_opt_attrs
- */
-static struct kobj_attribute *cache_index_opt_attrs[] = {
-	&cache_size_attr,
-	&cache_line_size_attr,
-	&cache_nr_sets_attr,
-	&cache_assoc_attr,
-};
-
-static const struct sysfs_ops cache_index_ops = {
-	.show = cache_index_show,
-};
-
-static struct kobj_type cache_index_type = {
-	.release = cache_index_release,
-	.sysfs_ops = &cache_index_ops,
-	.default_attrs = cache_index_default_attrs,
-};
-
-static void cacheinfo_create_index_opt_attrs(struct cache_index_dir *dir)
-{
-	const char *cache_name;
-	const char *cache_type;
-	struct cache *cache;
-	char *buf;
-	int i;
-
-	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
-	if (!buf)
-		return;
-
-	cache = dir->cache;
-	cache_name = cache->ofnode->full_name;
-	cache_type = cache_type_string(cache);
-
-	/* We don't want to create an attribute that can't provide a
-	 * meaningful value.  Check the return value of each optional
-	 * attribute's ->show method before registering the
-	 * attribute.
-	 */
-	for (i = 0; i < ARRAY_SIZE(cache_index_opt_attrs); i++) {
-		struct kobj_attribute *attr;
-		ssize_t rc;
-
-		attr = cache_index_opt_attrs[i];
-
-		rc = attr->show(&dir->kobj, attr, buf);
-		if (rc <= 0) {
-			pr_debug("not creating %s attribute for "
-				 "%s(%s) (rc = %zd)\n",
-				 attr->attr.name, cache_name,
-				 cache_type, rc);
-			continue;
-		}
-		if (sysfs_create_file(&dir->kobj, &attr->attr))
-			pr_debug("could not create %s attribute for %s(%s)\n",
-				 attr->attr.name, cache_name, cache_type);
+	np = cpu_dev->of_node;
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
 	}
 
-	kfree(buf);
-}
-
-static void cacheinfo_create_index_dir(struct cache *cache, int index,
-				       struct cache_dir *cache_dir)
-{
-	struct cache_index_dir *index_dir;
-	int rc;
-
-	index_dir = kzalloc(sizeof(*index_dir), GFP_KERNEL);
-	if (!index_dir)
-		goto err;
-
-	index_dir->cache = cache;
-
-	rc = kobject_init_and_add(&index_dir->kobj, &cache_index_type,
-				  cache_dir->kobj, "index%d", index);
-	if (rc)
-		goto err;
-
-	index_dir->next = cache_dir->index;
-	cache_dir->index = index_dir;
-
-	cacheinfo_create_index_opt_attrs(index_dir);
-
-	return;
-err:
-	kfree(index_dir);
-}
-
-static void cacheinfo_sysfs_populate(unsigned int cpu_id,
-				     struct cache *cache_list)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-	int index = 0;
-
-	cache_dir = cacheinfo_create_cache_dir(cpu_id);
-	if (!cache_dir)
-		return;
-
-	cache = cache_list;
-	while (cache) {
-		cacheinfo_create_index_dir(cache, index, cache_dir);
-		index++;
-		cache = cache->next_local;
+	while (np) {
+		leaves += cache_node_is_unified(np) ? 1 : 2;
+		level++;
+		of_node_put(np);
+		np = of_find_next_cache_node(np);
 	}
-}
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
 
-void cacheinfo_cpu_online(unsigned int cpu_id)
-{
-	struct cache *cache;
-
-	cache = cache_chain_instantiate(cpu_id);
-	if (!cache)
-		return;
-
-	cacheinfo_sysfs_populate(cpu_id, cache);
-}
-
-/* functions needed to remove cache entry for cpu offline or suspend/resume */
-
-#if (defined(CONFIG_PPC_PSERIES) && defined(CONFIG_SUSPEND)) || \
-    defined(CONFIG_HOTPLUG_CPU)
-
-static struct cache *cache_lookup_by_cpu(unsigned int cpu_id)
-{
-	struct device_node *cpu_node;
-	struct cache *cache;
-
-	cpu_node = of_get_cpu_node(cpu_id, NULL);
-	WARN_ONCE(!cpu_node, "no OF node found for CPU %i\n", cpu_id);
-	if (!cpu_node)
-		return NULL;
-
-	cache = cache_lookup_by_node(cpu_node);
-	of_node_put(cpu_node);
-
-	return cache;
+	return 0;
 }
 
-static void remove_index_dirs(struct cache_dir *cache_dir)
+int populate_cache_leaves(unsigned int cpu)
 {
-	struct cache_index_dir *index;
-
-	index = cache_dir->index;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct device_node *np;
+	unsigned int level, idx;
 
-	while (index) {
-		struct cache_index_dir *next;
-
-		next = index->next;
-		kobject_put(&index->kobj);
-		index = next;
+	np = of_node_get(cpu_dev->of_node);
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
 	}
-}
-
-static void remove_cache_dir(struct cache_dir *cache_dir)
-{
-	remove_index_dirs(cache_dir);
 
-	/* Remove cache dir from sysfs */
-	kobject_del(cache_dir->kobj);
-
-	kobject_put(cache_dir->kobj);
-
-	kfree(cache_dir);
-}
-
-static void cache_cpu_clear(struct cache *cache, int cpu)
-{
-	while (cache) {
-		struct cache *next = cache->next_local;
-
-		WARN_ONCE(!cpumask_test_cpu(cpu, &cache->shared_cpu_map),
-			  "CPU %i not accounted in %s(%s)\n",
-			  cpu, cache->ofnode->full_name,
-			  cache_type_string(cache));
-
-		cpumask_clear_cpu(cpu, &cache->shared_cpu_map);
-
-		/* Release the cache object if all the cpus using it
-		 * are offline */
-		if (cpumask_empty(&cache->shared_cpu_map))
-			release_cache(cache);
-
-		cache = next;
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		this_leaf->of_node = np;
+		if (cache_node_is_unified(np)) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_UNIFIED, level);
+		} else {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		}
+		np = of_find_next_cache_node(np);
 	}
+	return 0;
 }
 
-void cacheinfo_cpu_offline(unsigned int cpu_id)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-
-	/* Prevent userspace from seeing inconsistent state - remove
-	 * the sysfs hierarchy first */
-	cache_dir = per_cpu(cache_dir_pcpu, cpu_id);
-
-	/* careful, sysfs population may have failed */
-	if (cache_dir)
-		remove_cache_dir(cache_dir);
-
-	per_cpu(cache_dir_pcpu, cpu_id) = NULL;
-
-	/* clear the CPU's bit in its cache chain, possibly freeing
-	 * cache objects */
-	cache = cache_lookup_by_cpu(cpu_id);
-	if (cache)
-		cache_cpu_clear(cache, cpu_id);
-}
-#endif /* (CONFIG_PPC_PSERIES && CONFIG_SUSPEND) || CONFIG_HOTPLUG_CPU */
diff --git a/arch/powerpc/kernel/cacheinfo.h b/arch/powerpc/kernel/cacheinfo.h
deleted file mode 100644
index a7b74d3..0000000
--- a/arch/powerpc/kernel/cacheinfo.h
+++ /dev/null
@@ -1,8 +0,0 @@
-#ifndef _PPC_CACHEINFO_H
-#define _PPC_CACHEINFO_H
-
-/* These are just hooks for sysfs.c to use. */
-extern void cacheinfo_cpu_online(unsigned int cpu_id);
-extern void cacheinfo_cpu_offline(unsigned int cpu_id);
-
-#endif /* _PPC_CACHEINFO_H */
diff --git a/arch/powerpc/kernel/sysfs.c b/arch/powerpc/kernel/sysfs.c
index 67fd2fd..6e9c5a8 100644
--- a/arch/powerpc/kernel/sysfs.c
+++ b/arch/powerpc/kernel/sysfs.c
@@ -19,8 +19,6 @@
 #include <asm/pmc.h>
 #include <asm/firmware.h>
 
-#include "cacheinfo.h"
-
 #ifdef CONFIG_PPC64
 #include <asm/paca.h>
 #include <asm/lppaca.h>
@@ -743,7 +741,6 @@ static void register_cpu_online(unsigned int cpu)
 		device_create_file(s, &dev_attr_altivec_idle_wait_time);
 	}
 #endif
-	cacheinfo_cpu_online(cpu);
 }
 
 #ifdef CONFIG_HOTPLUG_CPU
@@ -824,7 +821,6 @@ static void unregister_cpu_online(unsigned int cpu)
 		device_remove_file(s, &dev_attr_altivec_idle_wait_time);
 	}
 #endif
-	cacheinfo_cpu_offline(cpu);
 }
 
 #ifdef CONFIG_ARCH_CPU_PROBE_RELEASE
@@ -988,8 +984,7 @@ static int __init topology_init(void)
 	int cpu;
 
 	register_nodes();
-
-	cpu_notifier_register_begin();
+	register_cpu_notifier(&sysfs_cpu_nb);
 
 	for_each_possible_cpu(cpu) {
 		struct cpu *c = &per_cpu(cpu_devices, cpu);
@@ -1013,11 +1008,6 @@ static int __init topology_init(void)
 		if (cpu_online(cpu))
 			register_cpu_online(cpu);
 	}
-
-	__register_cpu_notifier(&sysfs_cpu_nb);
-
-	cpu_notifier_register_done();
-
 #ifdef CONFIG_PPC64
 	sysfs_create_dscr_default();
 #endif /* CONFIG_PPC64 */
-- 
1.8.3.2

[PATCH 7/9] ARM64: kernel: add support for cpu cache information

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds support for cacheinfo on ARM64.

On ARMv8, the cache hierarchy can be identified through Cache Level ID
(CLIDR) register while the cache geometry is provided by Cache Size ID
(CCSIDR) register.

Since the architecture doesn't provide any way of detecting the cpus
sharing particular cache, device tree is used for the same purpose.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
---
 arch/arm64/kernel/Makefile    |   3 +-
 arch/arm64/kernel/cacheinfo.c | 135 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 137 insertions(+), 1 deletion(-)
 create mode 100644 arch/arm64/kernel/cacheinfo.c

diff --git a/arch/arm64/kernel/Makefile b/arch/arm64/kernel/Makefile
index cdaedad..754a3d0 100644
--- a/arch/arm64/kernel/Makefile
+++ b/arch/arm64/kernel/Makefile
@@ -15,7 +15,8 @@ CFLAGS_REMOVE_return_address.o = -pg
 arm64-obj-y		:= cputable.o debug-monitors.o entry.o irq.o fpsimd.o	\
 			   entry-fpsimd.o process.o ptrace.o setup.o signal.o	\
 			   sys.o stacktrace.o time.o traps.o io.o vdso.o	\
-			   hyp-stub.o psci.o cpu_ops.o insn.o return_address.o
+			   hyp-stub.o psci.o cpu_ops.o insn.o return_address.o	\
+			   cacheinfo.o
 
 arm64-obj-$(CONFIG_COMPAT)		+= sys32.o kuser32.o signal32.o 	\
 					   sys_compat.o
diff --git a/arch/arm64/kernel/cacheinfo.c b/arch/arm64/kernel/cacheinfo.c
new file mode 100644
index 0000000..a6f81cc
--- /dev/null
+++ b/arch/arm64/kernel/cacheinfo.c
@@ -0,0 +1,135 @@
+/*
+ *  ARM64 cacheinfo support
+ *
+ *  Copyright (C) 2014 ARM Ltd.
+ *  All Rights Reserved
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/cpu.h>
+#include <linux/compiler.h>
+#include <linux/of.h>
+
+#include <asm/processor.h>
+
+#define MAX_CACHE_LEVEL			7	/* Max 7 level supported */
+/* Ctypen, bits[3(n - 1) + 2 : 3(n - 1)], for n = 1 to 7 */
+#define CLIDR_CTYPE_SHIFT(level)	(3 * (level - 1))
+#define CLIDR_CTYPE_MASK(level)		(7 << CLIDR_CTYPE_SHIFT(level))
+#define CLIDR_CTYPE(clidr, level)	\
+	(((clidr) & CLIDR_CTYPE_MASK(level)) >> CLIDR_CTYPE_SHIFT(level))
+
+static inline enum cache_type get_cache_type(int level)
+{
+	unsigned int clidr;
+
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	asm volatile ("mrs     %0, clidr_el1" : "=r" (clidr));
+	return CLIDR_CTYPE(clidr, level);
+}
+
+/*
+ * NumSets, bits[27:13] - (Number of sets in cache) - 1
+ * Associativity, bits[12:3] - (Associativity of cache) - 1
+ * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
+ */
+#define CCSIDR_WRITE_THROUGH	BIT(31)
+#define CCSIDR_WRITE_BACK	BIT(30)
+#define CCSIDR_READ_ALLOCATE	BIT(29)
+#define CCSIDR_WRITE_ALLOCATE	BIT(28)
+#define CCSIDR_LINESIZE_MASK	0x7
+#define CCSIDR_ASSOCIAT_SHIFT	3
+#define CCSIDR_ASSOCIAT_MASK	0x3FF
+#define CCSIDR_NUMSETS_SHIFT	13
+#define CCSIDR_NUMSETS_MASK	0x7FF
+
+/*
+ * Which cache CCSIDR represents depends on CSSELR value
+ * Make sure no one else changes CSSELR during this
+ * smp_call_function_single prevents preemption for us
+ */
+static inline u32 get_ccsidr(u32 csselr)
+{
+	u32 ccsidr;
+
+	/* Put value into CSSELR */
+	asm volatile("msr csselr_el1, %x0" : : "r" (csselr));
+	isb();
+	/* Read result out of CCSIDR */
+	asm volatile("mrs %0, ccsidr_el1" : "=r" (ccsidr));
+
+	return ccsidr;
+}
+
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
+{
+	bool is_instr_cache = type & CACHE_TYPE_INST;
+	u32 tmp = get_ccsidr((level - 1) << 1 | is_instr_cache);
+
+	this_leaf->level = level;
+	this_leaf->type = type;
+	this_leaf->coherency_line_size =
+	    (1 << ((tmp & CCSIDR_LINESIZE_MASK) + 2)) * 4;
+	this_leaf->number_of_sets =
+	    ((tmp >> CCSIDR_NUMSETS_SHIFT) & CCSIDR_NUMSETS_MASK) + 1;
+	this_leaf->ways_of_associativity =
+	    ((tmp >> CCSIDR_ASSOCIAT_SHIFT) & CCSIDR_ASSOCIAT_MASK) + 1;
+	this_leaf->size = this_leaf->number_of_sets *
+	    this_leaf->coherency_line_size * this_leaf->ways_of_associativity;
+	this_leaf->attributes =
+		((tmp & CCSIDR_WRITE_THROUGH) ? CACHE_WRITE_THROUGH : 0) |
+		((tmp & CCSIDR_WRITE_BACK) ? CACHE_WRITE_BACK : 0) |
+		((tmp & CCSIDR_READ_ALLOCATE) ? CACHE_READ_ALLOCATE : 0) |
+		((tmp & CCSIDR_WRITE_ALLOCATE) ? CACHE_WRITE_ALLOCATE : 0);
+}
+
+int init_cache_level(unsigned int cpu)
+{
+	unsigned int ctype, level = 1, leaves = 0;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	if (!this_cpu_ci)
+		return -EINVAL;
+
+	do {
+		ctype = get_cache_type(level);
+		if (ctype == CACHE_TYPE_NOCACHE)
+			break;
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	} while (++level <= MAX_CACHE_LEVEL);
+
+	this_cpu_ci->num_levels = level - 1;
+	this_cpu_ci->num_leaves = leaves;
+	return 0;
+}
+
+int populate_cache_leaves(unsigned int cpu)
+{
+	unsigned int level, idx;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		type = get_cache_type(level);
+		if (type == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, type, level);
+		}
+	}
+	return 0;
+}
-- 
1.8.3.2

[PATCH 7/9] ARM64: kernel: add support for cpu cache information

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds support for cacheinfo on ARM64.

On ARMv8, the cache hierarchy can be identified through Cache Level ID
(CLIDR) register while the cache geometry is provided by Cache Size ID
(CCSIDR) register.

Since the architecture doesn't provide any way of detecting the cpus
sharing particular cache, device tree is used for the same purpose.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel at lists.infradead.org
---
 arch/arm64/kernel/Makefile    |   3 +-
 arch/arm64/kernel/cacheinfo.c | 135 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 137 insertions(+), 1 deletion(-)
 create mode 100644 arch/arm64/kernel/cacheinfo.c

diff --git a/arch/arm64/kernel/Makefile b/arch/arm64/kernel/Makefile
index cdaedad..754a3d0 100644
--- a/arch/arm64/kernel/Makefile
+++ b/arch/arm64/kernel/Makefile
@@ -15,7 +15,8 @@ CFLAGS_REMOVE_return_address.o = -pg
 arm64-obj-y		:= cputable.o debug-monitors.o entry.o irq.o fpsimd.o	\
 			   entry-fpsimd.o process.o ptrace.o setup.o signal.o	\
 			   sys.o stacktrace.o time.o traps.o io.o vdso.o	\
-			   hyp-stub.o psci.o cpu_ops.o insn.o return_address.o
+			   hyp-stub.o psci.o cpu_ops.o insn.o return_address.o	\
+			   cacheinfo.o
 
 arm64-obj-$(CONFIG_COMPAT)		+= sys32.o kuser32.o signal32.o 	\
 					   sys_compat.o
diff --git a/arch/arm64/kernel/cacheinfo.c b/arch/arm64/kernel/cacheinfo.c
new file mode 100644
index 0000000..a6f81cc
--- /dev/null
+++ b/arch/arm64/kernel/cacheinfo.c
@@ -0,0 +1,135 @@
+/*
+ *  ARM64 cacheinfo support
+ *
+ *  Copyright (C) 2014 ARM Ltd.
+ *  All Rights Reserved
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/cpu.h>
+#include <linux/compiler.h>
+#include <linux/of.h>
+
+#include <asm/processor.h>
+
+#define MAX_CACHE_LEVEL			7	/* Max 7 level supported */
+/* Ctypen, bits[3(n - 1) + 2 : 3(n - 1)], for n = 1 to 7 */
+#define CLIDR_CTYPE_SHIFT(level)	(3 * (level - 1))
+#define CLIDR_CTYPE_MASK(level)		(7 << CLIDR_CTYPE_SHIFT(level))
+#define CLIDR_CTYPE(clidr, level)	\
+	(((clidr) & CLIDR_CTYPE_MASK(level)) >> CLIDR_CTYPE_SHIFT(level))
+
+static inline enum cache_type get_cache_type(int level)
+{
+	unsigned int clidr;
+
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	asm volatile ("mrs     %0, clidr_el1" : "=r" (clidr));
+	return CLIDR_CTYPE(clidr, level);
+}
+
+/*
+ * NumSets, bits[27:13] - (Number of sets in cache) - 1
+ * Associativity, bits[12:3] - (Associativity of cache) - 1
+ * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
+ */
+#define CCSIDR_WRITE_THROUGH	BIT(31)
+#define CCSIDR_WRITE_BACK	BIT(30)
+#define CCSIDR_READ_ALLOCATE	BIT(29)
+#define CCSIDR_WRITE_ALLOCATE	BIT(28)
+#define CCSIDR_LINESIZE_MASK	0x7
+#define CCSIDR_ASSOCIAT_SHIFT	3
+#define CCSIDR_ASSOCIAT_MASK	0x3FF
+#define CCSIDR_NUMSETS_SHIFT	13
+#define CCSIDR_NUMSETS_MASK	0x7FF
+
+/*
+ * Which cache CCSIDR represents depends on CSSELR value
+ * Make sure no one else changes CSSELR during this
+ * smp_call_function_single prevents preemption for us
+ */
+static inline u32 get_ccsidr(u32 csselr)
+{
+	u32 ccsidr;
+
+	/* Put value into CSSELR */
+	asm volatile("msr csselr_el1, %x0" : : "r" (csselr));
+	isb();
+	/* Read result out of CCSIDR */
+	asm volatile("mrs %0, ccsidr_el1" : "=r" (ccsidr));
+
+	return ccsidr;
+}
+
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
+{
+	bool is_instr_cache = type & CACHE_TYPE_INST;
+	u32 tmp = get_ccsidr((level - 1) << 1 | is_instr_cache);
+
+	this_leaf->level = level;
+	this_leaf->type = type;
+	this_leaf->coherency_line_size =
+	    (1 << ((tmp & CCSIDR_LINESIZE_MASK) + 2)) * 4;
+	this_leaf->number_of_sets =
+	    ((tmp >> CCSIDR_NUMSETS_SHIFT) & CCSIDR_NUMSETS_MASK) + 1;
+	this_leaf->ways_of_associativity =
+	    ((tmp >> CCSIDR_ASSOCIAT_SHIFT) & CCSIDR_ASSOCIAT_MASK) + 1;
+	this_leaf->size = this_leaf->number_of_sets *
+	    this_leaf->coherency_line_size * this_leaf->ways_of_associativity;
+	this_leaf->attributes =
+		((tmp & CCSIDR_WRITE_THROUGH) ? CACHE_WRITE_THROUGH : 0) |
+		((tmp & CCSIDR_WRITE_BACK) ? CACHE_WRITE_BACK : 0) |
+		((tmp & CCSIDR_READ_ALLOCATE) ? CACHE_READ_ALLOCATE : 0) |
+		((tmp & CCSIDR_WRITE_ALLOCATE) ? CACHE_WRITE_ALLOCATE : 0);
+}
+
+int init_cache_level(unsigned int cpu)
+{
+	unsigned int ctype, level = 1, leaves = 0;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	if (!this_cpu_ci)
+		return -EINVAL;
+
+	do {
+		ctype = get_cache_type(level);
+		if (ctype == CACHE_TYPE_NOCACHE)
+			break;
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	} while (++level <= MAX_CACHE_LEVEL);
+
+	this_cpu_ci->num_levels = level - 1;
+	this_cpu_ci->num_leaves = leaves;
+	return 0;
+}
+
+int populate_cache_leaves(unsigned int cpu)
+{
+	unsigned int level, idx;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		type = get_cache_type(level);
+		if (type == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, type, level);
+		}
+	}
+	return 0;
+}
-- 
1.8.3.2

Re: [PATCH 7/9] ARM64: kernel: add support for cpu cache information

[Mark Rutland]

Hi Sudeep,

On Wed, Jun 25, 2014 at 06:30:42PM +0100, Sudeep Holla wrote:
> From: Sudeep Holla <sudeep.holla@arm.com>
> 
> This patch adds support for cacheinfo on ARM64.
> 
> On ARMv8, the cache hierarchy can be identified through Cache Level ID
> (CLIDR) register while the cache geometry is provided by Cache Size ID
> (CCSIDR) register.
> 
> Since the architecture doesn't provide any way of detecting the cpus
> sharing particular cache, device tree is used for the same purpose.
> 
> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
> Cc: Catalin Marinas <catalin.marinas@arm.com>
> Cc: Will Deacon <will.deacon@arm.com>
> Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
> Cc: linux-arm-kernel@lists.infradead.org
> ---
>  arch/arm64/kernel/Makefile    |   3 +-
>  arch/arm64/kernel/cacheinfo.c | 135 ++++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 137 insertions(+), 1 deletion(-)
>  create mode 100644 arch/arm64/kernel/cacheinfo.c

[...]

> +static inline enum cache_type get_cache_type(int level)
> +{
> +	unsigned int clidr;
> +
> +	if (level > MAX_CACHE_LEVEL)
> +		return CACHE_TYPE_NOCACHE;
> +	asm volatile ("mrs     %0, clidr_el1" : "=r" (clidr));

Can't that allocate a w register?

You can make clidr a u64 to avoid that.

> +	return CLIDR_CTYPE(clidr, level);
> +}
> +
> +/*
> + * NumSets, bits[27:13] - (Number of sets in cache) - 1
> + * Associativity, bits[12:3] - (Associativity of cache) - 1
> + * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
> + */
> +#define CCSIDR_WRITE_THROUGH	BIT(31)
> +#define CCSIDR_WRITE_BACK	BIT(30)
> +#define CCSIDR_READ_ALLOCATE	BIT(29)
> +#define CCSIDR_WRITE_ALLOCATE	BIT(28)
> +#define CCSIDR_LINESIZE_MASK	0x7
> +#define CCSIDR_ASSOCIAT_SHIFT	3
> +#define CCSIDR_ASSOCIAT_MASK	0x3FF

ASSOCIAT doesn't quite roll off of the tongue...

> +#define CCSIDR_NUMSETS_SHIFT	13
> +#define CCSIDR_NUMSETS_MASK	0x7FF
> +
> +/*
> + * Which cache CCSIDR represents depends on CSSELR value
> + * Make sure no one else changes CSSELR during this
> + * smp_call_function_single prevents preemption for us
> + */
> +static inline u32 get_ccsidr(u32 csselr)
> +{
> +	u32 ccsidr;
> +
> +	/* Put value into CSSELR */
> +	asm volatile("msr csselr_el1, %x0" : : "r" (csselr));

This looks a little dodgy. I think GCC can leave the upper 32 bits in a
random state. Why not cast csselr to a u64 here?

> +	isb();
> +	/* Read result out of CCSIDR */
> +	asm volatile("mrs %0, ccsidr_el1" : "=r" (ccsidr));
> +
> +	return ccsidr;

Similarly it might make sense to make the temporary variable a u64.

[...]

> +int init_cache_level(unsigned int cpu)
> +{
> +	unsigned int ctype, level = 1, leaves = 0;
> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
> +
> +	if (!this_cpu_ci)
> +		return -EINVAL;
> +
> +	do {
> +		ctype = get_cache_type(level);
> +		if (ctype == CACHE_TYPE_NOCACHE)
> +			break;
> +		/* Separate instruction and data caches */
> +		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
> +	} while (++level <= MAX_CACHE_LEVEL);

I think this would be clearer with:

for (level = 1; level <= MAX_CACHE_LEVEL; level++)

We do something like that in populate_cache_leaves below.

> +
> +	this_cpu_ci->num_levels = level - 1;
> +	this_cpu_ci->num_leaves = leaves;
> +	return 0;
> +}
> +
> +int populate_cache_leaves(unsigned int cpu)
> +{
> +	unsigned int level, idx;
> +	enum cache_type type;
> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
> +	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
> +
> +	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
> +	     idx < this_cpu_ci->num_leaves; idx++, level++) {
> +		if (!this_leaf)
> +			return -EINVAL;
> +
> +		type = get_cache_type(level);
> +		if (type == CACHE_TYPE_SEPARATE) {
> +			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
> +			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
> +		} else {
> +			ci_leaf_init(this_leaf++, type, level);
> +		}
> +	}
> +	return 0;
> +}

Cheers,
Mark.

[PATCH 7/9] ARM64: kernel: add support for cpu cache information

[Mark Rutland]

Hi Sudeep,

On Wed, Jun 25, 2014 at 06:30:42PM +0100, Sudeep Holla wrote:
> From: Sudeep Holla <sudeep.holla@arm.com>
> 
> This patch adds support for cacheinfo on ARM64.
> 
> On ARMv8, the cache hierarchy can be identified through Cache Level ID
> (CLIDR) register while the cache geometry is provided by Cache Size ID
> (CCSIDR) register.
> 
> Since the architecture doesn't provide any way of detecting the cpus
> sharing particular cache, device tree is used for the same purpose.
> 
> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
> Cc: Catalin Marinas <catalin.marinas@arm.com>
> Cc: Will Deacon <will.deacon@arm.com>
> Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
> Cc: linux-arm-kernel at lists.infradead.org
> ---
>  arch/arm64/kernel/Makefile    |   3 +-
>  arch/arm64/kernel/cacheinfo.c | 135 ++++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 137 insertions(+), 1 deletion(-)
>  create mode 100644 arch/arm64/kernel/cacheinfo.c

[...]

> +static inline enum cache_type get_cache_type(int level)
> +{
> +	unsigned int clidr;
> +
> +	if (level > MAX_CACHE_LEVEL)
> +		return CACHE_TYPE_NOCACHE;
> +	asm volatile ("mrs     %0, clidr_el1" : "=r" (clidr));

Can't that allocate a w register?

You can make clidr a u64 to avoid that.

> +	return CLIDR_CTYPE(clidr, level);
> +}
> +
> +/*
> + * NumSets, bits[27:13] - (Number of sets in cache) - 1
> + * Associativity, bits[12:3] - (Associativity of cache) - 1
> + * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
> + */
> +#define CCSIDR_WRITE_THROUGH	BIT(31)
> +#define CCSIDR_WRITE_BACK	BIT(30)
> +#define CCSIDR_READ_ALLOCATE	BIT(29)
> +#define CCSIDR_WRITE_ALLOCATE	BIT(28)
> +#define CCSIDR_LINESIZE_MASK	0x7
> +#define CCSIDR_ASSOCIAT_SHIFT	3
> +#define CCSIDR_ASSOCIAT_MASK	0x3FF

ASSOCIAT doesn't quite roll off of the tongue...

> +#define CCSIDR_NUMSETS_SHIFT	13
> +#define CCSIDR_NUMSETS_MASK	0x7FF
> +
> +/*
> + * Which cache CCSIDR represents depends on CSSELR value
> + * Make sure no one else changes CSSELR during this
> + * smp_call_function_single prevents preemption for us
> + */
> +static inline u32 get_ccsidr(u32 csselr)
> +{
> +	u32 ccsidr;
> +
> +	/* Put value into CSSELR */
> +	asm volatile("msr csselr_el1, %x0" : : "r" (csselr));

This looks a little dodgy. I think GCC can leave the upper 32 bits in a
random state. Why not cast csselr to a u64 here?

> +	isb();
> +	/* Read result out of CCSIDR */
> +	asm volatile("mrs %0, ccsidr_el1" : "=r" (ccsidr));
> +
> +	return ccsidr;

Similarly it might make sense to make the temporary variable a u64.

[...]

> +int init_cache_level(unsigned int cpu)
> +{
> +	unsigned int ctype, level = 1, leaves = 0;
> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
> +
> +	if (!this_cpu_ci)
> +		return -EINVAL;
> +
> +	do {
> +		ctype = get_cache_type(level);
> +		if (ctype == CACHE_TYPE_NOCACHE)
> +			break;
> +		/* Separate instruction and data caches */
> +		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
> +	} while (++level <= MAX_CACHE_LEVEL);

I think this would be clearer with:

for (level = 1; level <= MAX_CACHE_LEVEL; level++)

We do something like that in populate_cache_leaves below.

> +
> +	this_cpu_ci->num_levels = level - 1;
> +	this_cpu_ci->num_leaves = leaves;
> +	return 0;
> +}
> +
> +int populate_cache_leaves(unsigned int cpu)
> +{
> +	unsigned int level, idx;
> +	enum cache_type type;
> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
> +	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
> +
> +	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
> +	     idx < this_cpu_ci->num_leaves; idx++, level++) {
> +		if (!this_leaf)
> +			return -EINVAL;
> +
> +		type = get_cache_type(level);
> +		if (type == CACHE_TYPE_SEPARATE) {
> +			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
> +			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
> +		} else {
> +			ci_leaf_init(this_leaf++, type, level);
> +		}
> +	}
> +	return 0;
> +}

Cheers,
Mark.

Re: [PATCH 7/9] ARM64: kernel: add support for cpu cache information

[Sudeep Holla]

Hi Mark,

Thanks for the review.

On 27/06/14 11:36, Mark Rutland wrote:
> Hi Sudeep,
>
> On Wed, Jun 25, 2014 at 06:30:42PM +0100, Sudeep Holla wrote:
>> From: Sudeep Holla <sudeep.holla@arm.com>
>>
>> This patch adds support for cacheinfo on ARM64.
>>
>> On ARMv8, the cache hierarchy can be identified through Cache Level ID
>> (CLIDR) register while the cache geometry is provided by Cache Size ID
>> (CCSIDR) register.
>>
>> Since the architecture doesn't provide any way of detecting the cpus
>> sharing particular cache, device tree is used for the same purpose.
>>
>> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
>> Cc: Catalin Marinas <catalin.marinas@arm.com>
>> Cc: Will Deacon <will.deacon@arm.com>
>> Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
>> Cc: linux-arm-kernel@lists.infradead.org
>> ---
>>   arch/arm64/kernel/Makefile    |   3 +-
>>   arch/arm64/kernel/cacheinfo.c | 135 ++++++++++++++++++++++++++++++++++++++++++
>>   2 files changed, 137 insertions(+), 1 deletion(-)
>>   create mode 100644 arch/arm64/kernel/cacheinfo.c
>
> [...]
>
>> +static inline enum cache_type get_cache_type(int level)
>> +{
>> +	unsigned int clidr;
>> +
>> +	if (level > MAX_CACHE_LEVEL)
>> +		return CACHE_TYPE_NOCACHE;
>> +	asm volatile ("mrs     %0, clidr_el1" : "=r" (clidr));
>
> Can't that allocate a w register?
>

That should be fine, as all of these cache info registers are 32-bit.

> You can make clidr a u64 to avoid that.
>

What would be the preference ?
Using w registers for all these cache registers or using u64 with x registers?

>> +	return CLIDR_CTYPE(clidr, level);
>> +}
>> +
>> +/*
>> + * NumSets, bits[27:13] - (Number of sets in cache) - 1
>> + * Associativity, bits[12:3] - (Associativity of cache) - 1
>> + * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
>> + */
>> +#define CCSIDR_WRITE_THROUGH	BIT(31)
>> +#define CCSIDR_WRITE_BACK	BIT(30)
>> +#define CCSIDR_READ_ALLOCATE	BIT(29)
>> +#define CCSIDR_WRITE_ALLOCATE	BIT(28)
>> +#define CCSIDR_LINESIZE_MASK	0x7
>> +#define CCSIDR_ASSOCIAT_SHIFT	3
>> +#define CCSIDR_ASSOCIAT_MASK	0x3FF
>
> ASSOCIAT doesn't quite roll off of the tongue...
>

I have no idea why I chose that incomplete name :(

>> +#define CCSIDR_NUMSETS_SHIFT	13
>> +#define CCSIDR_NUMSETS_MASK	0x7FF
>> +
>> +/*
>> + * Which cache CCSIDR represents depends on CSSELR value
>> + * Make sure no one else changes CSSELR during this
>> + * smp_call_function_single prevents preemption for us
>> + */
>> +static inline u32 get_ccsidr(u32 csselr)
>> +{
>> +	u32 ccsidr;
>> +
>> +	/* Put value into CSSELR */
>> +	asm volatile("msr csselr_el1, %x0" : : "r" (csselr));
>
> This looks a little dodgy. I think GCC can leave the upper 32 bits in a
> random state. Why not cast csselr to a u64 here?
>
>> +	isb();
>> +	/* Read result out of CCSIDR */
>> +	asm volatile("mrs %0, ccsidr_el1" : "=r" (ccsidr));
>> +
>> +	return ccsidr;
>
> Similarly it might make sense to make the temporary variable a u64.
>
> [...]
>
>> +int init_cache_level(unsigned int cpu)
>> +{
>> +	unsigned int ctype, level = 1, leaves = 0;
>> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
>> +
>> +	if (!this_cpu_ci)
>> +		return -EINVAL;
>> +
>> +	do {
>> +		ctype = get_cache_type(level);
>> +		if (ctype == CACHE_TYPE_NOCACHE)
>> +			break;
>> +		/* Separate instruction and data caches */
>> +		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
>> +	} while (++level <= MAX_CACHE_LEVEL);
>
> I think this would be clearer with:
>
> for (level = 1; level <= MAX_CACHE_LEVEL; level++)
>
> We do something like that in populate_cache_leaves below.
>

Right will change it.

[PATCH 7/9] ARM64: kernel: add support for cpu cache information

[Sudeep Holla]

Hi Mark,

Thanks for the review.

On 27/06/14 11:36, Mark Rutland wrote:
> Hi Sudeep,
>
> On Wed, Jun 25, 2014 at 06:30:42PM +0100, Sudeep Holla wrote:
>> From: Sudeep Holla <sudeep.holla@arm.com>
>>
>> This patch adds support for cacheinfo on ARM64.
>>
>> On ARMv8, the cache hierarchy can be identified through Cache Level ID
>> (CLIDR) register while the cache geometry is provided by Cache Size ID
>> (CCSIDR) register.
>>
>> Since the architecture doesn't provide any way of detecting the cpus
>> sharing particular cache, device tree is used for the same purpose.
>>
>> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
>> Cc: Catalin Marinas <catalin.marinas@arm.com>
>> Cc: Will Deacon <will.deacon@arm.com>
>> Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
>> Cc: linux-arm-kernel at lists.infradead.org
>> ---
>>   arch/arm64/kernel/Makefile    |   3 +-
>>   arch/arm64/kernel/cacheinfo.c | 135 ++++++++++++++++++++++++++++++++++++++++++
>>   2 files changed, 137 insertions(+), 1 deletion(-)
>>   create mode 100644 arch/arm64/kernel/cacheinfo.c
>
> [...]
>
>> +static inline enum cache_type get_cache_type(int level)
>> +{
>> +	unsigned int clidr;
>> +
>> +	if (level > MAX_CACHE_LEVEL)
>> +		return CACHE_TYPE_NOCACHE;
>> +	asm volatile ("mrs     %0, clidr_el1" : "=r" (clidr));
>
> Can't that allocate a w register?
>

That should be fine, as all of these cache info registers are 32-bit.

> You can make clidr a u64 to avoid that.
>

What would be the preference ?
Using w registers for all these cache registers or using u64 with x registers?

>> +	return CLIDR_CTYPE(clidr, level);
>> +}
>> +
>> +/*
>> + * NumSets, bits[27:13] - (Number of sets in cache) - 1
>> + * Associativity, bits[12:3] - (Associativity of cache) - 1
>> + * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
>> + */
>> +#define CCSIDR_WRITE_THROUGH	BIT(31)
>> +#define CCSIDR_WRITE_BACK	BIT(30)
>> +#define CCSIDR_READ_ALLOCATE	BIT(29)
>> +#define CCSIDR_WRITE_ALLOCATE	BIT(28)
>> +#define CCSIDR_LINESIZE_MASK	0x7
>> +#define CCSIDR_ASSOCIAT_SHIFT	3
>> +#define CCSIDR_ASSOCIAT_MASK	0x3FF
>
> ASSOCIAT doesn't quite roll off of the tongue...
>

I have no idea why I chose that incomplete name :(

>> +#define CCSIDR_NUMSETS_SHIFT	13
>> +#define CCSIDR_NUMSETS_MASK	0x7FF
>> +
>> +/*
>> + * Which cache CCSIDR represents depends on CSSELR value
>> + * Make sure no one else changes CSSELR during this
>> + * smp_call_function_single prevents preemption for us
>> + */
>> +static inline u32 get_ccsidr(u32 csselr)
>> +{
>> +	u32 ccsidr;
>> +
>> +	/* Put value into CSSELR */
>> +	asm volatile("msr csselr_el1, %x0" : : "r" (csselr));
>
> This looks a little dodgy. I think GCC can leave the upper 32 bits in a
> random state. Why not cast csselr to a u64 here?
>
>> +	isb();
>> +	/* Read result out of CCSIDR */
>> +	asm volatile("mrs %0, ccsidr_el1" : "=r" (ccsidr));
>> +
>> +	return ccsidr;
>
> Similarly it might make sense to make the temporary variable a u64.
>
> [...]
>
>> +int init_cache_level(unsigned int cpu)
>> +{
>> +	unsigned int ctype, level = 1, leaves = 0;
>> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
>> +
>> +	if (!this_cpu_ci)
>> +		return -EINVAL;
>> +
>> +	do {
>> +		ctype = get_cache_type(level);
>> +		if (ctype == CACHE_TYPE_NOCACHE)
>> +			break;
>> +		/* Separate instruction and data caches */
>> +		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
>> +	} while (++level <= MAX_CACHE_LEVEL);
>
> I think this would be clearer with:
>
> for (level = 1; level <= MAX_CACHE_LEVEL; level++)
>
> We do something like that in populate_cache_leaves below.
>

Right will change it.

Re: [PATCH 7/9] ARM64: kernel: add support for cpu cache information

[Mark Rutland]

On Fri, Jun 27, 2014 at 12:22:17PM +0100, Sudeep Holla wrote:
> Hi Mark,
> 
> Thanks for the review.
> 
> On 27/06/14 11:36, Mark Rutland wrote:
> > Hi Sudeep,
> >
> > On Wed, Jun 25, 2014 at 06:30:42PM +0100, Sudeep Holla wrote:
> >> From: Sudeep Holla <sudeep.holla@arm.com>
> >>
> >> This patch adds support for cacheinfo on ARM64.
> >>
> >> On ARMv8, the cache hierarchy can be identified through Cache Level ID
> >> (CLIDR) register while the cache geometry is provided by Cache Size ID
> >> (CCSIDR) register.
> >>
> >> Since the architecture doesn't provide any way of detecting the cpus
> >> sharing particular cache, device tree is used for the same purpose.
> >>
> >> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
> >> Cc: Catalin Marinas <catalin.marinas@arm.com>
> >> Cc: Will Deacon <will.deacon@arm.com>
> >> Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
> >> Cc: linux-arm-kernel@lists.infradead.org
> >> ---
> >>   arch/arm64/kernel/Makefile    |   3 +-
> >>   arch/arm64/kernel/cacheinfo.c | 135 ++++++++++++++++++++++++++++++++++++++++++
> >>   2 files changed, 137 insertions(+), 1 deletion(-)
> >>   create mode 100644 arch/arm64/kernel/cacheinfo.c
> >
> > [...]
> >
> >> +static inline enum cache_type get_cache_type(int level)
> >> +{
> >> +	unsigned int clidr;
> >> +
> >> +	if (level > MAX_CACHE_LEVEL)
> >> +		return CACHE_TYPE_NOCACHE;
> >> +	asm volatile ("mrs     %0, clidr_el1" : "=r" (clidr));
> >
> > Can't that allocate a w register?
> >
> 
> That should be fine, as all of these cache info registers are 32-bit.

In A64 mrs/msr only works for x registers, and gas will barf for w registers:

[mark@leverpostej:~]% echo "mrs x0, clidr_el1" | aarch64-linux-gnu-as - 
[mark@leverpostej:~]% echo "mrs w0, clidr_el1" | aarch64-linux-gnu-as - 
{standard input}: Assembler messages:
{standard input}:1: Error: operand mismatch -- `mrs w0,clidr_el1'
[mark@leverpostej:~]% echo "msr clidr_el1, x0" | aarch64-linux-gnu-as - 
[mark@leverpostej:~]% echo "msr clidr_el1, w0" | aarch64-linux-gnu-as - 
{standard input}: Assembler messages:
{standard input}:1: Error: operand mismatch -- `msr clidr_el1,w0'
[mark@leverpostej:~]% 

> > You can make clidr a u64 to avoid that.
> >
> 
> What would be the preference ?
> Using w registers for all these cache registers or using u64 with x registers?

You must use x registers.

To prevent GCC from making the assumption that the upper 32-bits are
irrelevant, it's better to cast to a u64 than use %xN in the asm.

> >> +	return CLIDR_CTYPE(clidr, level);
> >> +}
> >> +
> >> +/*
> >> + * NumSets, bits[27:13] - (Number of sets in cache) - 1
> >> + * Associativity, bits[12:3] - (Associativity of cache) - 1
> >> + * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
> >> + */
> >> +#define CCSIDR_WRITE_THROUGH	BIT(31)
> >> +#define CCSIDR_WRITE_BACK	BIT(30)
> >> +#define CCSIDR_READ_ALLOCATE	BIT(29)
> >> +#define CCSIDR_WRITE_ALLOCATE	BIT(28)
> >> +#define CCSIDR_LINESIZE_MASK	0x7
> >> +#define CCSIDR_ASSOCIAT_SHIFT	3
> >> +#define CCSIDR_ASSOCIAT_MASK	0x3FF
> >
> > ASSOCIAT doesn't quite roll off of the tongue...
> >
> 
> I have no idea why I chose that incomplete name :(

At least we can fix it :)

Cheers,
Mark.

[PATCH 7/9] ARM64: kernel: add support for cpu cache information

[Mark Rutland]

On Fri, Jun 27, 2014 at 12:22:17PM +0100, Sudeep Holla wrote:
> Hi Mark,
> 
> Thanks for the review.
> 
> On 27/06/14 11:36, Mark Rutland wrote:
> > Hi Sudeep,
> >
> > On Wed, Jun 25, 2014 at 06:30:42PM +0100, Sudeep Holla wrote:
> >> From: Sudeep Holla <sudeep.holla@arm.com>
> >>
> >> This patch adds support for cacheinfo on ARM64.
> >>
> >> On ARMv8, the cache hierarchy can be identified through Cache Level ID
> >> (CLIDR) register while the cache geometry is provided by Cache Size ID
> >> (CCSIDR) register.
> >>
> >> Since the architecture doesn't provide any way of detecting the cpus
> >> sharing particular cache, device tree is used for the same purpose.
> >>
> >> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
> >> Cc: Catalin Marinas <catalin.marinas@arm.com>
> >> Cc: Will Deacon <will.deacon@arm.com>
> >> Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
> >> Cc: linux-arm-kernel at lists.infradead.org
> >> ---
> >>   arch/arm64/kernel/Makefile    |   3 +-
> >>   arch/arm64/kernel/cacheinfo.c | 135 ++++++++++++++++++++++++++++++++++++++++++
> >>   2 files changed, 137 insertions(+), 1 deletion(-)
> >>   create mode 100644 arch/arm64/kernel/cacheinfo.c
> >
> > [...]
> >
> >> +static inline enum cache_type get_cache_type(int level)
> >> +{
> >> +	unsigned int clidr;
> >> +
> >> +	if (level > MAX_CACHE_LEVEL)
> >> +		return CACHE_TYPE_NOCACHE;
> >> +	asm volatile ("mrs     %0, clidr_el1" : "=r" (clidr));
> >
> > Can't that allocate a w register?
> >
> 
> That should be fine, as all of these cache info registers are 32-bit.

In A64 mrs/msr only works for x registers, and gas will barf for w registers:

[mark at leverpostej:~]% echo "mrs x0, clidr_el1" | aarch64-linux-gnu-as - 
[mark at leverpostej:~]% echo "mrs w0, clidr_el1" | aarch64-linux-gnu-as - 
{standard input}: Assembler messages:
{standard input}:1: Error: operand mismatch -- `mrs w0,clidr_el1'
[mark at leverpostej:~]% echo "msr clidr_el1, x0" | aarch64-linux-gnu-as - 
[mark at leverpostej:~]% echo "msr clidr_el1, w0" | aarch64-linux-gnu-as - 
{standard input}: Assembler messages:
{standard input}:1: Error: operand mismatch -- `msr clidr_el1,w0'
[mark at leverpostej:~]% 

> > You can make clidr a u64 to avoid that.
> >
> 
> What would be the preference ?
> Using w registers for all these cache registers or using u64 with x registers?

You must use x registers.

To prevent GCC from making the assumption that the upper 32-bits are
irrelevant, it's better to cast to a u64 than use %xN in the asm.

> >> +	return CLIDR_CTYPE(clidr, level);
> >> +}
> >> +
> >> +/*
> >> + * NumSets, bits[27:13] - (Number of sets in cache) - 1
> >> + * Associativity, bits[12:3] - (Associativity of cache) - 1
> >> + * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
> >> + */
> >> +#define CCSIDR_WRITE_THROUGH	BIT(31)
> >> +#define CCSIDR_WRITE_BACK	BIT(30)
> >> +#define CCSIDR_READ_ALLOCATE	BIT(29)
> >> +#define CCSIDR_WRITE_ALLOCATE	BIT(28)
> >> +#define CCSIDR_LINESIZE_MASK	0x7
> >> +#define CCSIDR_ASSOCIAT_SHIFT	3
> >> +#define CCSIDR_ASSOCIAT_MASK	0x3FF
> >
> > ASSOCIAT doesn't quite roll off of the tongue...
> >
> 
> I have no idea why I chose that incomplete name :(

At least we can fix it :)

Cheers,
Mark.

[PATCH 8/9] ARM: kernel: add support for cpu cache information

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds support for cacheinfo on ARM platforms.

On ARMv7, the cache hierarchy can be identified through Cache Level ID
register(CLIDR) while the cache geometry is provided by Cache Size ID
register(CCSIDR).

On architecture versions before ARMv7, CLIDR and CCSIDR is not
implemented. The cache type register(CTR) provides both cache hierarchy
and geometry if implemented. For implementations that doesn't support
CTR, we need to list the probable value of CTR if it was implemented
along with the cpuid for the sake of simplicity to handle them.

Since the architecture doesn't provide any way of detecting the cpus
sharing particular cache, device tree is used fo the same purpose.
On non-DT platforms, first level caches are per-cpu while higher level
caches are assumed system-wide.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Nicolas Pitre <nicolas.pitre@linaro.org>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
---
 arch/arm/kernel/Makefile    |   1 +
 arch/arm/kernel/cacheinfo.c | 229 ++++++++++++++++++++++++++++++++++++++++++++
 arch/arm/mm/Kconfig         |  13 +++
 3 files changed, 243 insertions(+)
 create mode 100644 arch/arm/kernel/cacheinfo.c

diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile
index 38ddd9f..2c5ff0e 100644
--- a/arch/arm/kernel/Makefile
+++ b/arch/arm/kernel/Makefile
@@ -29,6 +29,7 @@ obj-y		+= entry-v7m.o v7m.o
 else
 obj-y		+= entry-armv.o
 endif
+obj-$(CONFIG_CPU_HAS_CACHE) += cacheinfo.o
 
 obj-$(CONFIG_OC_ETM)		+= etm.o
 obj-$(CONFIG_CPU_IDLE)		+= cpuidle.o
diff --git a/arch/arm/kernel/cacheinfo.c b/arch/arm/kernel/cacheinfo.c
new file mode 100644
index 0000000..ab70993
--- /dev/null
+++ b/arch/arm/kernel/cacheinfo.c
@@ -0,0 +1,229 @@
+/*
+ *  ARM cacheinfo support
+ *
+ *  Copyright (C) 2014 ARM Ltd.
+ *  All Rights Reserved
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/cpu.h>
+#include <linux/compiler.h>
+#include <linux/of.h>
+
+#include <asm/cputype.h>
+#include <asm/processor.h>
+
+#if __LINUX_ARM_ARCH__ < 7 /* pre ARMv7 */
+
+#define MAX_CACHE_LEVEL		1	/* Only 1 level supported */
+#define CTR_CTYPE_SHIFT		24
+#define CTR_CTYPE_MASK		(1 << CTR_CTYPE_SHIFT)
+
+struct ctr_info {
+	unsigned int cpuid_id;
+	unsigned int ctr;
+};
+
+static struct ctr_info cache_ctr_list[] = {
+};
+
+static int get_unimplemented_ctr(unsigned int *ctr)
+{
+	int i, cpuid_id = read_cpuid_id();
+
+	for (i = 0; i < ARRAY_SIZE(cache_ctr_list); i++)
+		if (cache_ctr_list[i].cpuid_id == cpuid_id) {
+			*ctr = cache_ctr_list[i].ctr;
+			return 0;
+		}
+	return -ENOENT;
+}
+
+static unsigned int get_ctr(void)
+{
+	unsigned int ctr;
+
+	if (get_unimplemented_ctr(&ctr))
+		asm volatile ("mrc p15, 0, %0, c0, c0, 1" : "=r" (ctr));
+	return ctr;
+}
+
+static enum cache_type get_cache_type(int level)
+{
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	return get_ctr() & CTR_CTYPE_MASK ?
+		CACHE_TYPE_SEPARATE : CACHE_TYPE_UNIFIED;
+}
+
+/*
+ *  +---------------------------------+
+ *  | 9  8  7  6 | 5  4  3 | 2 | 1  0 |
+ *  +---------------------------------+
+ *  |    size    |  assoc  | m |  len |
+ *  +---------------------------------+
+ * linelen        = 1 << (len + 3)
+ * multiplier     = 2 + m
+ * nsets          = 1 << (size + 6 - assoc - len)
+ * associativity  = multiplier << (assoc - 1)
+ * cache_size     = multiplier << (size + 8)
+ */
+#define CTR_LINESIZE_MASK	0x3
+#define CTR_MULTIPLIER_SHIFT	2
+#define CTR_MULTIPLIER_MASK	0x1
+#define CTR_ASSOCIAT_SHIFT	3
+#define CTR_ASSOCIAT_MASK	0x7
+#define CTR_SIZE_SHIFT		6
+#define CTR_SIZE_MASK		0xF
+#define CTR_DCACHE_SHIFT	12
+
+static void __ci_leaf_init(enum cache_type type, struct cacheinfo *this_leaf)
+{
+	unsigned int size, multiplier, assoc, len, tmp = get_ctr();
+
+	if (type == CACHE_TYPE_DATA)
+		tmp >>= CTR_DCACHE_SHIFT;
+
+	len = tmp & CTR_LINESIZE_MASK;
+	size = (tmp >> CTR_SIZE_SHIFT) & CTR_SIZE_MASK;
+	assoc = (tmp >> CTR_ASSOCIAT_SHIFT) & CTR_ASSOCIAT_MASK;
+	multiplier = ((tmp >> CTR_MULTIPLIER_SHIFT) & CTR_MULTIPLIER_MASK) + 2;
+
+	this_leaf->type = type;
+	this_leaf->coherency_line_size = 1 << (len + 3);
+	this_leaf->number_of_sets = 1 << (size + 6 - assoc - len);
+	this_leaf->ways_of_associativity = multiplier << (assoc - 1);
+	this_leaf->size = multiplier << (size + 8);
+}
+
+#else /* ARMv7 */
+
+#define MAX_CACHE_LEVEL			7	/* Max 7 level supported */
+/* Ctypen, bits[3(n - 1) + 2 : 3(n - 1)], for n = 1 to 7 */
+#define CLIDR_CTYPE_SHIFT(level)	(3 * (level - 1))
+#define CLIDR_CTYPE_MASK(level)		(7 << CLIDR_CTYPE_SHIFT(level))
+#define CLIDR_CTYPE(clidr, level)	\
+	(((clidr) & CLIDR_CTYPE_MASK(level)) >> CLIDR_CTYPE_SHIFT(level))
+
+static inline enum cache_type get_cache_type(int level)
+{
+	unsigned int clidr;
+
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	asm volatile ("mrc p15, 1, %0, c0, c0, 1" : "=r" (clidr));
+	return CLIDR_CTYPE(clidr, level);
+}
+
+/*
+ * NumSets, bits[27:13] - (Number of sets in cache) - 1
+ * Associativity, bits[12:3] - (Associativity of cache) - 1
+ * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
+ */
+#define CCSIDR_WRITE_THROUGH	BIT(31)
+#define CCSIDR_WRITE_BACK	BIT(30)
+#define CCSIDR_READ_ALLOCATE	BIT(29)
+#define CCSIDR_WRITE_ALLOCATE	BIT(28)
+#define CCSIDR_LINESIZE_MASK	0x7
+#define CCSIDR_ASSOCIAT_SHIFT	3
+#define CCSIDR_ASSOCIAT_MASK	0x3FF
+#define CCSIDR_NUMSETS_SHIFT	13
+#define CCSIDR_NUMSETS_MASK	0x7FF
+
+/*
+ * Which cache CCSIDR represents depends on CSSELR value
+ * Make sure no one else changes CSSELR during this
+ * smp_call_function_single prevents preemption for us
+ */
+static inline u32 get_ccsidr(u32 csselr)
+{
+	u32 ccsidr;
+
+	/* Put value into CSSELR */
+	asm volatile ("mcr p15, 2, %0, c0, c0, 0" : : "r" (csselr));
+	isb();
+	/* Read result out of CCSIDR */
+	asm volatile ("mrc p15, 1, %0, c0, c0, 0" : "=r" (ccsidr));
+
+	return ccsidr;
+}
+
+static void __ci_leaf_init(enum cache_type type, struct cacheinfo *this_leaf)
+{
+	bool is_instr_cache = type & CACHE_TYPE_INST;
+	u32 tmp = get_ccsidr((this_leaf->level - 1) << 1 | is_instr_cache);
+
+	this_leaf->type = type;
+	this_leaf->coherency_line_size =
+	    (1 << ((tmp & CCSIDR_LINESIZE_MASK) + 2)) * 4;
+	this_leaf->number_of_sets =
+	    ((tmp >> CCSIDR_NUMSETS_SHIFT) & CCSIDR_NUMSETS_MASK) + 1;
+	this_leaf->ways_of_associativity =
+	    ((tmp >> CCSIDR_ASSOCIAT_SHIFT) & CCSIDR_ASSOCIAT_MASK) + 1;
+	this_leaf->size = this_leaf->number_of_sets *
+	    this_leaf->coherency_line_size * this_leaf->ways_of_associativity;
+	this_leaf->attributes =
+		((tmp & CCSIDR_WRITE_THROUGH) ? CACHE_WRITE_THROUGH : 0) |
+		((tmp & CCSIDR_WRITE_BACK) ? CACHE_WRITE_BACK : 0) |
+		((tmp & CCSIDR_READ_ALLOCATE) ? CACHE_READ_ALLOCATE : 0) |
+		((tmp & CCSIDR_WRITE_ALLOCATE) ? CACHE_WRITE_ALLOCATE : 0);
+}
+
+#endif
+
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
+{
+	this_leaf->level = level;
+	__ci_leaf_init(type, this_leaf);
+}
+
+int init_cache_level(unsigned int cpu)
+{
+	unsigned int ctype, level = 1, leaves = 0;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	if (!this_cpu_ci)
+		return -EINVAL;
+
+	do {
+		ctype = get_cache_type(level);
+		if (ctype == CACHE_TYPE_NOCACHE)
+			break;
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	} while (++level <= MAX_CACHE_LEVEL);
+
+	this_cpu_ci->num_levels = level - 1;
+	this_cpu_ci->num_leaves = leaves;
+
+	return 0;
+}
+
+int populate_cache_leaves(unsigned int cpu)
+{
+	unsigned int level, idx;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		type = get_cache_type(level);
+		if (type == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, type, level);
+		}
+	}
+	return 0;
+}
diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig
index eda0dd0..fac8646 100644
--- a/arch/arm/mm/Kconfig
+++ b/arch/arm/mm/Kconfig
@@ -494,30 +494,42 @@ config CPU_PABRT_V7
 # The cache model
 config CPU_CACHE_V4
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V4WT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V4WB
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V6
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V7
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_NOP
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_VIVT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_VIPT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_FA
 	bool
+	select CPU_HAS_CACHE
+
+config CPU_HAS_CACHE
+	bool
 
 if MMU
 # The copy-page model
@@ -845,6 +857,7 @@ config DMA_CACHE_RWFO
 
 config OUTER_CACHE
 	bool
+	select CPU_HAS_CACHE
 
 config OUTER_CACHE_SYNC
 	bool
-- 
1.8.3.2

[PATCH 8/9] ARM: kernel: add support for cpu cache information

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds support for cacheinfo on ARM platforms.

On ARMv7, the cache hierarchy can be identified through Cache Level ID
register(CLIDR) while the cache geometry is provided by Cache Size ID
register(CCSIDR).

On architecture versions before ARMv7, CLIDR and CCSIDR is not
implemented. The cache type register(CTR) provides both cache hierarchy
and geometry if implemented. For implementations that doesn't support
CTR, we need to list the probable value of CTR if it was implemented
along with the cpuid for the sake of simplicity to handle them.

Since the architecture doesn't provide any way of detecting the cpus
sharing particular cache, device tree is used fo the same purpose.
On non-DT platforms, first level caches are per-cpu while higher level
caches are assumed system-wide.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Nicolas Pitre <nicolas.pitre@linaro.org>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel at lists.infradead.org
---
 arch/arm/kernel/Makefile    |   1 +
 arch/arm/kernel/cacheinfo.c | 229 ++++++++++++++++++++++++++++++++++++++++++++
 arch/arm/mm/Kconfig         |  13 +++
 3 files changed, 243 insertions(+)
 create mode 100644 arch/arm/kernel/cacheinfo.c

diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile
index 38ddd9f..2c5ff0e 100644
--- a/arch/arm/kernel/Makefile
+++ b/arch/arm/kernel/Makefile
@@ -29,6 +29,7 @@ obj-y		+= entry-v7m.o v7m.o
 else
 obj-y		+= entry-armv.o
 endif
+obj-$(CONFIG_CPU_HAS_CACHE) += cacheinfo.o
 
 obj-$(CONFIG_OC_ETM)		+= etm.o
 obj-$(CONFIG_CPU_IDLE)		+= cpuidle.o
diff --git a/arch/arm/kernel/cacheinfo.c b/arch/arm/kernel/cacheinfo.c
new file mode 100644
index 0000000..ab70993
--- /dev/null
+++ b/arch/arm/kernel/cacheinfo.c
@@ -0,0 +1,229 @@
+/*
+ *  ARM cacheinfo support
+ *
+ *  Copyright (C) 2014 ARM Ltd.
+ *  All Rights Reserved
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/cpu.h>
+#include <linux/compiler.h>
+#include <linux/of.h>
+
+#include <asm/cputype.h>
+#include <asm/processor.h>
+
+#if __LINUX_ARM_ARCH__ < 7 /* pre ARMv7 */
+
+#define MAX_CACHE_LEVEL		1	/* Only 1 level supported */
+#define CTR_CTYPE_SHIFT		24
+#define CTR_CTYPE_MASK		(1 << CTR_CTYPE_SHIFT)
+
+struct ctr_info {
+	unsigned int cpuid_id;
+	unsigned int ctr;
+};
+
+static struct ctr_info cache_ctr_list[] = {
+};
+
+static int get_unimplemented_ctr(unsigned int *ctr)
+{
+	int i, cpuid_id = read_cpuid_id();
+
+	for (i = 0; i < ARRAY_SIZE(cache_ctr_list); i++)
+		if (cache_ctr_list[i].cpuid_id == cpuid_id) {
+			*ctr = cache_ctr_list[i].ctr;
+			return 0;
+		}
+	return -ENOENT;
+}
+
+static unsigned int get_ctr(void)
+{
+	unsigned int ctr;
+
+	if (get_unimplemented_ctr(&ctr))
+		asm volatile ("mrc p15, 0, %0, c0, c0, 1" : "=r" (ctr));
+	return ctr;
+}
+
+static enum cache_type get_cache_type(int level)
+{
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	return get_ctr() & CTR_CTYPE_MASK ?
+		CACHE_TYPE_SEPARATE : CACHE_TYPE_UNIFIED;
+}
+
+/*
+ *  +---------------------------------+
+ *  | 9  8  7  6 | 5  4  3 | 2 | 1  0 |
+ *  +---------------------------------+
+ *  |    size    |  assoc  | m |  len |
+ *  +---------------------------------+
+ * linelen        = 1 << (len + 3)
+ * multiplier     = 2 + m
+ * nsets          = 1 << (size + 6 - assoc - len)
+ * associativity  = multiplier << (assoc - 1)
+ * cache_size     = multiplier << (size + 8)
+ */
+#define CTR_LINESIZE_MASK	0x3
+#define CTR_MULTIPLIER_SHIFT	2
+#define CTR_MULTIPLIER_MASK	0x1
+#define CTR_ASSOCIAT_SHIFT	3
+#define CTR_ASSOCIAT_MASK	0x7
+#define CTR_SIZE_SHIFT		6
+#define CTR_SIZE_MASK		0xF
+#define CTR_DCACHE_SHIFT	12
+
+static void __ci_leaf_init(enum cache_type type, struct cacheinfo *this_leaf)
+{
+	unsigned int size, multiplier, assoc, len, tmp = get_ctr();
+
+	if (type == CACHE_TYPE_DATA)
+		tmp >>= CTR_DCACHE_SHIFT;
+
+	len = tmp & CTR_LINESIZE_MASK;
+	size = (tmp >> CTR_SIZE_SHIFT) & CTR_SIZE_MASK;
+	assoc = (tmp >> CTR_ASSOCIAT_SHIFT) & CTR_ASSOCIAT_MASK;
+	multiplier = ((tmp >> CTR_MULTIPLIER_SHIFT) & CTR_MULTIPLIER_MASK) + 2;
+
+	this_leaf->type = type;
+	this_leaf->coherency_line_size = 1 << (len + 3);
+	this_leaf->number_of_sets = 1 << (size + 6 - assoc - len);
+	this_leaf->ways_of_associativity = multiplier << (assoc - 1);
+	this_leaf->size = multiplier << (size + 8);
+}
+
+#else /* ARMv7 */
+
+#define MAX_CACHE_LEVEL			7	/* Max 7 level supported */
+/* Ctypen, bits[3(n - 1) + 2 : 3(n - 1)], for n = 1 to 7 */
+#define CLIDR_CTYPE_SHIFT(level)	(3 * (level - 1))
+#define CLIDR_CTYPE_MASK(level)		(7 << CLIDR_CTYPE_SHIFT(level))
+#define CLIDR_CTYPE(clidr, level)	\
+	(((clidr) & CLIDR_CTYPE_MASK(level)) >> CLIDR_CTYPE_SHIFT(level))
+
+static inline enum cache_type get_cache_type(int level)
+{
+	unsigned int clidr;
+
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	asm volatile ("mrc p15, 1, %0, c0, c0, 1" : "=r" (clidr));
+	return CLIDR_CTYPE(clidr, level);
+}
+
+/*
+ * NumSets, bits[27:13] - (Number of sets in cache) - 1
+ * Associativity, bits[12:3] - (Associativity of cache) - 1
+ * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
+ */
+#define CCSIDR_WRITE_THROUGH	BIT(31)
+#define CCSIDR_WRITE_BACK	BIT(30)
+#define CCSIDR_READ_ALLOCATE	BIT(29)
+#define CCSIDR_WRITE_ALLOCATE	BIT(28)
+#define CCSIDR_LINESIZE_MASK	0x7
+#define CCSIDR_ASSOCIAT_SHIFT	3
+#define CCSIDR_ASSOCIAT_MASK	0x3FF
+#define CCSIDR_NUMSETS_SHIFT	13
+#define CCSIDR_NUMSETS_MASK	0x7FF
+
+/*
+ * Which cache CCSIDR represents depends on CSSELR value
+ * Make sure no one else changes CSSELR during this
+ * smp_call_function_single prevents preemption for us
+ */
+static inline u32 get_ccsidr(u32 csselr)
+{
+	u32 ccsidr;
+
+	/* Put value into CSSELR */
+	asm volatile ("mcr p15, 2, %0, c0, c0, 0" : : "r" (csselr));
+	isb();
+	/* Read result out of CCSIDR */
+	asm volatile ("mrc p15, 1, %0, c0, c0, 0" : "=r" (ccsidr));
+
+	return ccsidr;
+}
+
+static void __ci_leaf_init(enum cache_type type, struct cacheinfo *this_leaf)
+{
+	bool is_instr_cache = type & CACHE_TYPE_INST;
+	u32 tmp = get_ccsidr((this_leaf->level - 1) << 1 | is_instr_cache);
+
+	this_leaf->type = type;
+	this_leaf->coherency_line_size =
+	    (1 << ((tmp & CCSIDR_LINESIZE_MASK) + 2)) * 4;
+	this_leaf->number_of_sets =
+	    ((tmp >> CCSIDR_NUMSETS_SHIFT) & CCSIDR_NUMSETS_MASK) + 1;
+	this_leaf->ways_of_associativity =
+	    ((tmp >> CCSIDR_ASSOCIAT_SHIFT) & CCSIDR_ASSOCIAT_MASK) + 1;
+	this_leaf->size = this_leaf->number_of_sets *
+	    this_leaf->coherency_line_size * this_leaf->ways_of_associativity;
+	this_leaf->attributes =
+		((tmp & CCSIDR_WRITE_THROUGH) ? CACHE_WRITE_THROUGH : 0) |
+		((tmp & CCSIDR_WRITE_BACK) ? CACHE_WRITE_BACK : 0) |
+		((tmp & CCSIDR_READ_ALLOCATE) ? CACHE_READ_ALLOCATE : 0) |
+		((tmp & CCSIDR_WRITE_ALLOCATE) ? CACHE_WRITE_ALLOCATE : 0);
+}
+
+#endif
+
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
+{
+	this_leaf->level = level;
+	__ci_leaf_init(type, this_leaf);
+}
+
+int init_cache_level(unsigned int cpu)
+{
+	unsigned int ctype, level = 1, leaves = 0;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	if (!this_cpu_ci)
+		return -EINVAL;
+
+	do {
+		ctype = get_cache_type(level);
+		if (ctype == CACHE_TYPE_NOCACHE)
+			break;
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	} while (++level <= MAX_CACHE_LEVEL);
+
+	this_cpu_ci->num_levels = level - 1;
+	this_cpu_ci->num_leaves = leaves;
+
+	return 0;
+}
+
+int populate_cache_leaves(unsigned int cpu)
+{
+	unsigned int level, idx;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		type = get_cache_type(level);
+		if (type == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, type, level);
+		}
+	}
+	return 0;
+}
diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig
index eda0dd0..fac8646 100644
--- a/arch/arm/mm/Kconfig
+++ b/arch/arm/mm/Kconfig
@@ -494,30 +494,42 @@ config CPU_PABRT_V7
 # The cache model
 config CPU_CACHE_V4
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V4WT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V4WB
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V6
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V7
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_NOP
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_VIVT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_VIPT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_FA
 	bool
+	select CPU_HAS_CACHE
+
+config CPU_HAS_CACHE
+	bool
 
 if MMU
 # The copy-page model
@@ -845,6 +857,7 @@ config DMA_CACHE_RWFO
 
 config OUTER_CACHE
 	bool
+	select CPU_HAS_CACHE
 
 config OUTER_CACHE_SYNC
 	bool
-- 
1.8.3.2

Re: [PATCH 8/9] ARM: kernel: add support for cpu cache information

[Russell King - ARM Linux]

On Wed, Jun 25, 2014 at 06:30:43PM +0100, Sudeep Holla wrote:
> diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile
> index 38ddd9f..2c5ff0e 100644
> --- a/arch/arm/kernel/Makefile
> +++ b/arch/arm/kernel/Makefile
> @@ -29,6 +29,7 @@ obj-y		+= entry-v7m.o v7m.o
>  else
>  obj-y		+= entry-armv.o
>  endif
> +obj-$(CONFIG_CPU_HAS_CACHE) += cacheinfo.o
>  
>  obj-$(CONFIG_OC_ETM)		+= etm.o
>  obj-$(CONFIG_CPU_IDLE)		+= cpuidle.o
> diff --git a/arch/arm/kernel/cacheinfo.c b/arch/arm/kernel/cacheinfo.c
> new file mode 100644
> index 0000000..ab70993
> --- /dev/null
> +++ b/arch/arm/kernel/cacheinfo.c
> @@ -0,0 +1,229 @@
> +/*
> + *  ARM cacheinfo support
> + *
> + *  Copyright (C) 2014 ARM Ltd.
> + *  All Rights Reserved
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 as
> + * published by the Free Software Foundation.
> + */
> +
> +#include <linux/bitops.h>
> +#include <linux/cacheinfo.h>
> +#include <linux/cpu.h>
> +#include <linux/compiler.h>
> +#include <linux/of.h>
> +
> +#include <asm/cputype.h>
> +#include <asm/processor.h>
> +
> +#if __LINUX_ARM_ARCH__ < 7 /* pre ARMv7 */

__LINUX_ARM_ARCH__ defines the minimum architecture version we are building
for - we may support later versions than the architecture version denoted
by this symbol.  It does not define which CPUs we are building for.  Are
you sure that this is correct here?  What if we build a kernel supporting
both v6 + v7, as the OMAP guys do?

> +
> +#define MAX_CACHE_LEVEL		1	/* Only 1 level supported */
> +#define CTR_CTYPE_SHIFT		24
> +#define CTR_CTYPE_MASK		(1 << CTR_CTYPE_SHIFT)
> +
> +struct ctr_info {
> +	unsigned int cpuid_id;
> +	unsigned int ctr;
> +};
> +
> +static struct ctr_info cache_ctr_list[] = {
> +};

This list needs to be populated.  Early CPUs (such as StrongARM) do not
have the CTR register.

> +static int get_unimplemented_ctr(unsigned int *ctr)
> +{
> +	int i, cpuid_id = read_cpuid_id();
> +
> +	for (i = 0; i < ARRAY_SIZE(cache_ctr_list); i++)
> +		if (cache_ctr_list[i].cpuid_id == cpuid_id) {
> +			*ctr = cache_ctr_list[i].ctr;
> +			return 0;
> +		}
> +	return -ENOENT;
> +}
> +
> +static unsigned int get_ctr(void)
> +{
> +	unsigned int ctr;
> +
> +	if (get_unimplemented_ctr(&ctr))
> +		asm volatile ("mrc p15, 0, %0, c0, c0, 1" : "=r" (ctr));

read_cpuid_cachetype() ?

-- 
FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly
improving, and getting towards what was expected from it.

[PATCH 8/9] ARM: kernel: add support for cpu cache information

[Russell King - ARM Linux]

On Wed, Jun 25, 2014 at 06:30:43PM +0100, Sudeep Holla wrote:
> diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile
> index 38ddd9f..2c5ff0e 100644
> --- a/arch/arm/kernel/Makefile
> +++ b/arch/arm/kernel/Makefile
> @@ -29,6 +29,7 @@ obj-y		+= entry-v7m.o v7m.o
>  else
>  obj-y		+= entry-armv.o
>  endif
> +obj-$(CONFIG_CPU_HAS_CACHE) += cacheinfo.o
>  
>  obj-$(CONFIG_OC_ETM)		+= etm.o
>  obj-$(CONFIG_CPU_IDLE)		+= cpuidle.o
> diff --git a/arch/arm/kernel/cacheinfo.c b/arch/arm/kernel/cacheinfo.c
> new file mode 100644
> index 0000000..ab70993
> --- /dev/null
> +++ b/arch/arm/kernel/cacheinfo.c
> @@ -0,0 +1,229 @@
> +/*
> + *  ARM cacheinfo support
> + *
> + *  Copyright (C) 2014 ARM Ltd.
> + *  All Rights Reserved
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 as
> + * published by the Free Software Foundation.
> + */
> +
> +#include <linux/bitops.h>
> +#include <linux/cacheinfo.h>
> +#include <linux/cpu.h>
> +#include <linux/compiler.h>
> +#include <linux/of.h>
> +
> +#include <asm/cputype.h>
> +#include <asm/processor.h>
> +
> +#if __LINUX_ARM_ARCH__ < 7 /* pre ARMv7 */

__LINUX_ARM_ARCH__ defines the minimum architecture version we are building
for - we may support later versions than the architecture version denoted
by this symbol.  It does not define which CPUs we are building for.  Are
you sure that this is correct here?  What if we build a kernel supporting
both v6 + v7, as the OMAP guys do?

> +
> +#define MAX_CACHE_LEVEL		1	/* Only 1 level supported */
> +#define CTR_CTYPE_SHIFT		24
> +#define CTR_CTYPE_MASK		(1 << CTR_CTYPE_SHIFT)
> +
> +struct ctr_info {
> +	unsigned int cpuid_id;
> +	unsigned int ctr;
> +};
> +
> +static struct ctr_info cache_ctr_list[] = {
> +};

This list needs to be populated.  Early CPUs (such as StrongARM) do not
have the CTR register.

> +static int get_unimplemented_ctr(unsigned int *ctr)
> +{
> +	int i, cpuid_id = read_cpuid_id();
> +
> +	for (i = 0; i < ARRAY_SIZE(cache_ctr_list); i++)
> +		if (cache_ctr_list[i].cpuid_id == cpuid_id) {
> +			*ctr = cache_ctr_list[i].ctr;
> +			return 0;
> +		}
> +	return -ENOENT;
> +}
> +
> +static unsigned int get_ctr(void)
> +{
> +	unsigned int ctr;
> +
> +	if (get_unimplemented_ctr(&ctr))
> +		asm volatile ("mrc p15, 0, %0, c0, c0, 1" : "=r" (ctr));

read_cpuid_cachetype() ?

-- 
FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly
improving, and getting towards what was expected from it.

Re: [PATCH 8/9] ARM: kernel: add support for cpu cache information

[Sudeep Holla]

Hi Russell,

Thanks for the reviews.

On 25/06/14 23:33, Russell King - ARM Linux wrote:
> On Wed, Jun 25, 2014 at 06:30:43PM +0100, Sudeep Holla wrote:
[...]
>> +
>> +#include <linux/bitops.h>
>> +#include <linux/cacheinfo.h>
>> +#include <linux/cpu.h>
>> +#include <linux/compiler.h>
>> +#include <linux/of.h>
>> +
>> +#include <asm/cputype.h>
>> +#include <asm/processor.h>
>> +
>> +#if __LINUX_ARM_ARCH__ < 7 /* pre ARMv7 */
>
> __LINUX_ARM_ARCH__ defines the minimum architecture version we are building
> for - we may support later versions than the architecture version denoted
> by this symbol.  It does not define which CPUs we are building for.  Are
> you sure that this is correct here?  What if we build a kernel supporting
> both v6 + v7, as the OMAP guys do?
>

You are right, I have not considered v6 + v7, I will use cpu_architecture and
make it runtime.

>> +
>> +#define MAX_CACHE_LEVEL		1	/* Only 1 level supported */
>> +#define CTR_CTYPE_SHIFT		24
>> +#define CTR_CTYPE_MASK		(1 << CTR_CTYPE_SHIFT)
>> +
>> +struct ctr_info {
>> +	unsigned int cpuid_id;
>> +	unsigned int ctr;
>> +};
>> +
>> +static struct ctr_info cache_ctr_list[] = {
>> +};
>
> This list needs to be populated.  Early CPUs (such as StrongARM) do not
> have the CTR register.
>

Right, since I didn't have the list left it empty. I will compile the list
soon but I need your help. The list of StrongARM I can come up is:
1. SA-110
2. SA-1100
3. SA-1110
4. SA-1500 (grep didn't show this in kernel, not sure if it's supported)

I also have to find all other ARMv4 implementations not having CTR.

>> +static int get_unimplemented_ctr(unsigned int *ctr)
>> +{
>> +	int i, cpuid_id = read_cpuid_id();
>> +
>> +	for (i = 0; i < ARRAY_SIZE(cache_ctr_list); i++)
>> +		if (cache_ctr_list[i].cpuid_id == cpuid_id) {
>> +			*ctr = cache_ctr_list[i].ctr;
>> +			return 0;
>> +		}
>> +	return -ENOENT;
>> +}
>> +
>> +static unsigned int get_ctr(void)
>> +{
>> +	unsigned int ctr;
>> +
>> +	if (get_unimplemented_ctr(&ctr))
>> +		asm volatile ("mrc p15, 0, %0, c0, c0, 1" : "=r" (ctr));
>
> read_cpuid_cachetype() ?
>
Ah, I missed to see that, will use it.

Regards,
Sudeep

[PATCH 8/9] ARM: kernel: add support for cpu cache information

[Sudeep Holla]

Hi Russell,

Thanks for the reviews.

On 25/06/14 23:33, Russell King - ARM Linux wrote:
> On Wed, Jun 25, 2014 at 06:30:43PM +0100, Sudeep Holla wrote:
[...]
>> +
>> +#include <linux/bitops.h>
>> +#include <linux/cacheinfo.h>
>> +#include <linux/cpu.h>
>> +#include <linux/compiler.h>
>> +#include <linux/of.h>
>> +
>> +#include <asm/cputype.h>
>> +#include <asm/processor.h>
>> +
>> +#if __LINUX_ARM_ARCH__ < 7 /* pre ARMv7 */
>
> __LINUX_ARM_ARCH__ defines the minimum architecture version we are building
> for - we may support later versions than the architecture version denoted
> by this symbol.  It does not define which CPUs we are building for.  Are
> you sure that this is correct here?  What if we build a kernel supporting
> both v6 + v7, as the OMAP guys do?
>

You are right, I have not considered v6 + v7, I will use cpu_architecture and
make it runtime.

>> +
>> +#define MAX_CACHE_LEVEL		1	/* Only 1 level supported */
>> +#define CTR_CTYPE_SHIFT		24
>> +#define CTR_CTYPE_MASK		(1 << CTR_CTYPE_SHIFT)
>> +
>> +struct ctr_info {
>> +	unsigned int cpuid_id;
>> +	unsigned int ctr;
>> +};
>> +
>> +static struct ctr_info cache_ctr_list[] = {
>> +};
>
> This list needs to be populated.  Early CPUs (such as StrongARM) do not
> have the CTR register.
>

Right, since I didn't have the list left it empty. I will compile the list
soon but I need your help. The list of StrongARM I can come up is:
1. SA-110
2. SA-1100
3. SA-1110
4. SA-1500 (grep didn't show this in kernel, not sure if it's supported)

I also have to find all other ARMv4 implementations not having CTR.

>> +static int get_unimplemented_ctr(unsigned int *ctr)
>> +{
>> +	int i, cpuid_id = read_cpuid_id();
>> +
>> +	for (i = 0; i < ARRAY_SIZE(cache_ctr_list); i++)
>> +		if (cache_ctr_list[i].cpuid_id == cpuid_id) {
>> +			*ctr = cache_ctr_list[i].ctr;
>> +			return 0;
>> +		}
>> +	return -ENOENT;
>> +}
>> +
>> +static unsigned int get_ctr(void)
>> +{
>> +	unsigned int ctr;
>> +
>> +	if (get_unimplemented_ctr(&ctr))
>> +		asm volatile ("mrc p15, 0, %0, c0, c0, 1" : "=r" (ctr));
>
> read_cpuid_cachetype() ?
>
Ah, I missed to see that, will use it.

Regards,
Sudeep

Re: [PATCH 8/9] ARM: kernel: add support for cpu cache information

[Stephen Boyd]

On 06/25/14 10:30, Sudeep Holla wrote:
> +
> +/*
> + * Which cache CCSIDR represents depends on CSSELR value
> + * Make sure no one else changes CSSELR during this
> + * smp_call_function_single prevents preemption for us
> + */

Where's the smp_call_function_single() or preemption disable happening?

> +static inline u32 get_ccsidr(u32 csselr)
> +{
> +	u32 ccsidr;
> +
> +	/* Put value into CSSELR */
> +	asm volatile ("mcr p15, 2, %0, c0, c0, 0" : : "r" (csselr));
> +	isb();
> +	/* Read result out of CCSIDR */
> +	asm volatile ("mrc p15, 1, %0, c0, c0, 0" : "=r" (ccsidr));
> +
> +	return ccsidr;
> +}
> +

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation

[PATCH 8/9] ARM: kernel: add support for cpu cache information

[Stephen Boyd]

On 06/25/14 10:30, Sudeep Holla wrote:
> +
> +/*
> + * Which cache CCSIDR represents depends on CSSELR value
> + * Make sure no one else changes CSSELR during this
> + * smp_call_function_single prevents preemption for us
> + */

Where's the smp_call_function_single() or preemption disable happening?

> +static inline u32 get_ccsidr(u32 csselr)
> +{
> +	u32 ccsidr;
> +
> +	/* Put value into CSSELR */
> +	asm volatile ("mcr p15, 2, %0, c0, c0, 0" : : "r" (csselr));
> +	isb();
> +	/* Read result out of CCSIDR */
> +	asm volatile ("mrc p15, 1, %0, c0, c0, 0" : "=r" (ccsidr));
> +
> +	return ccsidr;
> +}
> +

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation

Re: [PATCH 8/9] ARM: kernel: add support for cpu cache information

[Sudeep Holla]

Hi Stephen,

On 26/06/14 01:19, Stephen Boyd wrote:
> On 06/25/14 10:30, Sudeep Holla wrote:
>> +
>> +/*
>> + * Which cache CCSIDR represents depends on CSSELR value
>> + * Make sure no one else changes CSSELR during this
>> + * smp_call_function_single prevents preemption for us
>> + */
>
> Where's the smp_call_function_single() or preemption disable happening?
>

init_cache_level is called using smp_call_function_single in
drivers/base/cacheinfo.c(PATCH 2/9)

Regards,
Sudeep

[PATCH 8/9] ARM: kernel: add support for cpu cache information

[Sudeep Holla]

Hi Stephen,

On 26/06/14 01:19, Stephen Boyd wrote:
> On 06/25/14 10:30, Sudeep Holla wrote:
>> +
>> +/*
>> + * Which cache CCSIDR represents depends on CSSELR value
>> + * Make sure no one else changes CSSELR during this
>> + * smp_call_function_single prevents preemption for us
>> + */
>
> Where's the smp_call_function_single() or preemption disable happening?
>

init_cache_level is called using smp_call_function_single in
drivers/base/cacheinfo.c(PATCH 2/9)

Regards,
Sudeep

Re: [PATCH 8/9] ARM: kernel: add support for cpu cache information

[Stephen Boyd]

On 06/26/14 04:36, Sudeep Holla wrote:
> Hi Stephen,
>
> On 26/06/14 01:19, Stephen Boyd wrote:
>> On 06/25/14 10:30, Sudeep Holla wrote:
>>> +
>>> +/*
>>> + * Which cache CCSIDR represents depends on CSSELR value
>>> + * Make sure no one else changes CSSELR during this
>>> + * smp_call_function_single prevents preemption for us
>>> + */
>>
>> Where's the smp_call_function_single() or preemption disable happening?
>>
>
> init_cache_level is called using smp_call_function_single in
> drivers/base/cacheinfo.c(PATCH 2/9)

Oh that's unexpected. Do other architectures require the use of
smp_call_function_single() to read their cache information? It seems
like an ARM architecture specific detail that has been pushed up into
the generic layer.

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation

[PATCH 8/9] ARM: kernel: add support for cpu cache information

[Stephen Boyd]

On 06/26/14 04:36, Sudeep Holla wrote:
> Hi Stephen,
>
> On 26/06/14 01:19, Stephen Boyd wrote:
>> On 06/25/14 10:30, Sudeep Holla wrote:
>>> +
>>> +/*
>>> + * Which cache CCSIDR represents depends on CSSELR value
>>> + * Make sure no one else changes CSSELR during this
>>> + * smp_call_function_single prevents preemption for us
>>> + */
>>
>> Where's the smp_call_function_single() or preemption disable happening?
>>
>
> init_cache_level is called using smp_call_function_single in
> drivers/base/cacheinfo.c(PATCH 2/9)

Oh that's unexpected. Do other architectures require the use of
smp_call_function_single() to read their cache information? It seems
like an ARM architecture specific detail that has been pushed up into
the generic layer.

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation

Re: [PATCH 8/9] ARM: kernel: add support for cpu cache information

[Sudeep Holla]

Hi,

On 26/06/14 19:45, Stephen Boyd wrote:
> On 06/26/14 04:36, Sudeep Holla wrote:
>> Hi Stephen,
>>
>> On 26/06/14 01:19, Stephen Boyd wrote:
>>> On 06/25/14 10:30, Sudeep Holla wrote:
>>>> +
>>>> +/*
>>>> + * Which cache CCSIDR represents depends on CSSELR value
>>>> + * Make sure no one else changes CSSELR during this
>>>> + * smp_call_function_single prevents preemption for us
>>>> + */
>>>
>>> Where's the smp_call_function_single() or preemption disable happening?
>>>
>>
>> init_cache_level is called using smp_call_function_single in
>> drivers/base/cacheinfo.c(PATCH 2/9)
>
> Oh that's unexpected. Do other architectures require the use of
> smp_call_function_single() to read their cache information? It seems
> like an ARM architecture specific detail that has been pushed up into
> the generic layer.
>

Right, since I started with x86 as reference and it requires it, I missed to
see others. So x86,ARM{32,64} requires it while ppc,s390 and ia64 doesn't.
I see how to fix that. Thanks for spotting it.

Regards,
Sudeep

[PATCH 8/9] ARM: kernel: add support for cpu cache information

[Sudeep Holla]

Hi,

On 26/06/14 19:45, Stephen Boyd wrote:
> On 06/26/14 04:36, Sudeep Holla wrote:
>> Hi Stephen,
>>
>> On 26/06/14 01:19, Stephen Boyd wrote:
>>> On 06/25/14 10:30, Sudeep Holla wrote:
>>>> +
>>>> +/*
>>>> + * Which cache CCSIDR represents depends on CSSELR value
>>>> + * Make sure no one else changes CSSELR during this
>>>> + * smp_call_function_single prevents preemption for us
>>>> + */
>>>
>>> Where's the smp_call_function_single() or preemption disable happening?
>>>
>>
>> init_cache_level is called using smp_call_function_single in
>> drivers/base/cacheinfo.c(PATCH 2/9)
>
> Oh that's unexpected. Do other architectures require the use of
> smp_call_function_single() to read their cache information? It seems
> like an ARM architecture specific detail that has been pushed up into
> the generic layer.
>

Right, since I started with x86 as reference and it requires it, I missed to
see others. So x86,ARM{32,64} requires it while ppc,s390 and ia64 doesn't.
I see how to fix that. Thanks for spotting it.

Regards,
Sudeep

[PATCH 9/9] ARM: kernel: add outer cache support for cacheinfo implementation

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

In order to support outer cache in the cacheinfo infrastructure, a new
function 'get_info' is added to outer_cache_fns. This function is used
to get the outer cache information namely: line size, number of ways of
associativity and number of sets.

This patch adds 'get_info' supports to all L2 cache implementations on
ARM except Marvell's Feroceon L2 cache.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
---
 arch/arm/include/asm/outercache.h | 13 +++++++++++++
 arch/arm/kernel/cacheinfo.c       | 22 +++++++++++++++++++++-
 arch/arm/mm/cache-l2x0.c          | 10 ++++++++++
 arch/arm/mm/cache-tauros2.c       | 34 ++++++++++++++++++++++++++++++++++
 arch/arm/mm/cache-xsc3l2.c        | 15 +++++++++++++++
 5 files changed, 93 insertions(+), 1 deletion(-)

diff --git a/arch/arm/include/asm/outercache.h b/arch/arm/include/asm/outercache.h
index 891a56b..991cf63 100644
--- a/arch/arm/include/asm/outercache.h
+++ b/arch/arm/include/asm/outercache.h
@@ -23,7 +23,14 @@
 
 #include <linux/types.h>
 
+struct outer_cache_info {
+	unsigned int num_ways;
+	unsigned int num_sets;
+	unsigned int line_size;
+};
+
 struct outer_cache_fns {
+	void (*get_info)(struct outer_cache_info *info);
 	void (*inv_range)(unsigned long, unsigned long);
 	void (*clean_range)(unsigned long, unsigned long);
 	void (*flush_range)(unsigned long, unsigned long);
@@ -112,6 +119,11 @@ static inline void outer_resume(void)
 		outer_cache.resume();
 }
 
+static inline void outer_get_info(struct outer_cache_info *info)
+{
+	if (outer_cache.get_info)
+		outer_cache.get_info(info);
+}
 #else
 
 static inline void outer_inv_range(phys_addr_t start, phys_addr_t end)
@@ -123,6 +135,7 @@ static inline void outer_flush_range(phys_addr_t start, phys_addr_t end)
 static inline void outer_flush_all(void) { }
 static inline void outer_disable(void) { }
 static inline void outer_resume(void) { }
+static inline void outer_get_info(struct outer_cache_info *info) { }
 
 #endif
 
diff --git a/arch/arm/kernel/cacheinfo.c b/arch/arm/kernel/cacheinfo.c
index ab70993..88b552b 100644
--- a/arch/arm/kernel/cacheinfo.c
+++ b/arch/arm/kernel/cacheinfo.c
@@ -16,6 +16,7 @@
 #include <linux/of.h>
 
 #include <asm/cputype.h>
+#include <asm/outercache.h>
 #include <asm/processor.h>
 
 #if __LINUX_ARM_ARCH__ < 7 /* pre ARMv7 */
@@ -176,11 +177,27 @@ static void __ci_leaf_init(enum cache_type type, struct cacheinfo *this_leaf)
 
 #endif
 
+static void __outer_ci_leaf_init(struct cacheinfo *this_leaf)
+{
+	struct outer_cache_info info;
+
+	outer_get_info(&info);
+
+	this_leaf->type = CACHE_TYPE_UNIFIED;/* record it as Unified */
+	this_leaf->ways_of_associativity = info.num_ways;
+	this_leaf->number_of_sets = info.num_sets;
+	this_leaf->coherency_line_size = info.line_size;
+	this_leaf->size = info.num_ways * info.num_sets * info.line_size;
+}
+
 static void ci_leaf_init(struct cacheinfo *this_leaf,
 			 enum cache_type type, unsigned int level)
 {
 	this_leaf->level = level;
-	__ci_leaf_init(type, this_leaf);
+	if (type == CACHE_TYPE_NOCACHE)	/* must be outer cache */
+		__outer_ci_leaf_init(this_leaf);
+	else
+		__ci_leaf_init(type, this_leaf);
 }
 
 int init_cache_level(unsigned int cpu)
@@ -202,6 +219,9 @@ int init_cache_level(unsigned int cpu)
 	this_cpu_ci->num_levels = level - 1;
 	this_cpu_ci->num_leaves = leaves;
 
+	if (IS_ENABLED(CONFIG_OUTER_CACHE) && outer_cache.get_info)
+		this_cpu_ci->num_leaves++, this_cpu_ci->num_levels++;
+
 	return 0;
 }
 
diff --git a/arch/arm/mm/cache-l2x0.c b/arch/arm/mm/cache-l2x0.c
index efc5cab..30ca151 100644
--- a/arch/arm/mm/cache-l2x0.c
+++ b/arch/arm/mm/cache-l2x0.c
@@ -105,6 +105,15 @@ static inline void l2c_unlock(void __iomem *base, unsigned num)
 	}
 }
 
+static void l2x0_getinfo(struct outer_cache_info *info)
+{
+	if (!info)
+		return;
+	info->num_ways = get_count_order(l2x0_way_mask);
+	info->line_size = CACHE_LINE_SIZE;
+	info->num_sets = l2x0_size / (info->num_ways * CACHE_LINE_SIZE);
+}
+
 /*
  * Enable the L2 cache controller.  This function must only be
  * called when the cache controller is known to be disabled.
@@ -894,6 +903,7 @@ static void __init __l2c_init(const struct l2c_init_data *data,
 		data->enable(l2x0_base, aux, data->num_lock);
 
 	outer_cache = fns;
+	outer_cache.get_info = l2x0_getinfo;
 
 	/*
 	 * It is strange to save the register state before initialisation,
diff --git a/arch/arm/mm/cache-tauros2.c b/arch/arm/mm/cache-tauros2.c
index b273739..8708684 100644
--- a/arch/arm/mm/cache-tauros2.c
+++ b/arch/arm/mm/cache-tauros2.c
@@ -60,6 +60,7 @@ static inline void tauros2_inv_pa(unsigned long addr)
  * noninclusive.
  */
 #define CACHE_LINE_SIZE		32
+#define CACHE_LINE_SHIFT	5
 
 static void tauros2_inv_range(unsigned long start, unsigned long end)
 {
@@ -131,6 +132,38 @@ static void tauros2_resume(void)
 	"mcr	p15, 0, %0, c1, c0, 0 @Enable L2 Cache\n\t"
 	: : "r" (0x0));
 }
+
+/*
+ *  +----------------------------------------+
+ *  | 11 10 9  8 | 7  6  5  4  3 | 2 |  1  0 |
+ *  +----------------------------------------+
+ *  |  way size  | associativity | - |line_sz|
+ *  +----------------------------------------+
+ */
+#define L2CTR_ASSOCIAT_SHIFT	3
+#define L2CTR_ASSOCIAT_MASK	0x1F
+#define L2CTR_WAYSIZE_SHIFT	8
+#define L2CTR_WAYSIZE_MASK	0xF
+#define CACHE_WAY_PER_SET(l2ctr)	\
+	(((l2_ctr) >> L2CTR_ASSOCIAT_SHIFT) & L2CTR_ASSOCIAT_MASK)
+#define CACHE_WAY_SIZE(l2ctr)		\
+	(8192 << (((l2ctr) >> L2CTR_WAYSIZE_SHIFT) & L2CTR_WAYSIZE_MASK))
+#define CACHE_SET_SIZE(l2ctr)	(CACHE_WAY_SIZE(l2ctr) >> CACHE_LINE_SHIFT)
+
+static void tauros2_getinfo(struct outer_cache_info *info)
+{
+	unsigned int l2_ctr;
+
+	if (!info)
+		return;
+
+	__asm__("mrc p15, 1, %0, c0, c0, 1" : "=r" (l2_ctr));
+
+	info->line_size = CACHE_LINE_SIZE;
+	info->num_ways = CACHE_WAY_PER_SET(l2_ctr);
+	info->num_sets = CACHE_SET_SIZE(l2_ctr);
+}
+
 #endif
 
 static inline u32 __init read_extra_features(void)
@@ -226,6 +259,7 @@ static void __init tauros2_internal_init(unsigned int features)
 		outer_cache.flush_range = tauros2_flush_range;
 		outer_cache.disable = tauros2_disable;
 		outer_cache.resume = tauros2_resume;
+		outer_cache.get_info = tauros2_getinfo;
 	}
 #endif
 
diff --git a/arch/arm/mm/cache-xsc3l2.c b/arch/arm/mm/cache-xsc3l2.c
index 6c3edeb..353c642 100644
--- a/arch/arm/mm/cache-xsc3l2.c
+++ b/arch/arm/mm/cache-xsc3l2.c
@@ -201,6 +201,20 @@ static void xsc3_l2_flush_range(unsigned long start, unsigned long end)
 	dsb();
 }
 
+static void xsc3_l2_getinfo(struct outer_cache_info *info)
+{
+	unsigned long l2ctype;
+
+	if (!info)
+		return;
+
+	__asm__("mrc p15, 1, %0, c0, c0, 1" : "=r" (l2ctype));
+
+	info->num_ways = CACHE_WAY_PER_SET;
+	info->line_size = CACHE_LINE_SIZE;
+	info->num_sets = CACHE_SET_SIZE(l2ctype);
+}
+
 static int __init xsc3_l2_init(void)
 {
 	if (!cpu_is_xsc3() || !xsc3_l2_present())
@@ -213,6 +227,7 @@ static int __init xsc3_l2_init(void)
 		outer_cache.inv_range = xsc3_l2_inv_range;
 		outer_cache.clean_range = xsc3_l2_clean_range;
 		outer_cache.flush_range = xsc3_l2_flush_range;
+		outer_cache.get_info    = xsc3_l2_getinfo;
 	}
 
 	return 0;
-- 
1.8.3.2

[PATCH 9/9] ARM: kernel: add outer cache support for cacheinfo implementation

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

In order to support outer cache in the cacheinfo infrastructure, a new
function 'get_info' is added to outer_cache_fns. This function is used
to get the outer cache information namely: line size, number of ways of
associativity and number of sets.

This patch adds 'get_info' supports to all L2 cache implementations on
ARM except Marvell's Feroceon L2 cache.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel at lists.infradead.org
---
 arch/arm/include/asm/outercache.h | 13 +++++++++++++
 arch/arm/kernel/cacheinfo.c       | 22 +++++++++++++++++++++-
 arch/arm/mm/cache-l2x0.c          | 10 ++++++++++
 arch/arm/mm/cache-tauros2.c       | 34 ++++++++++++++++++++++++++++++++++
 arch/arm/mm/cache-xsc3l2.c        | 15 +++++++++++++++
 5 files changed, 93 insertions(+), 1 deletion(-)

diff --git a/arch/arm/include/asm/outercache.h b/arch/arm/include/asm/outercache.h
index 891a56b..991cf63 100644
--- a/arch/arm/include/asm/outercache.h
+++ b/arch/arm/include/asm/outercache.h
@@ -23,7 +23,14 @@
 
 #include <linux/types.h>
 
+struct outer_cache_info {
+	unsigned int num_ways;
+	unsigned int num_sets;
+	unsigned int line_size;
+};
+
 struct outer_cache_fns {
+	void (*get_info)(struct outer_cache_info *info);
 	void (*inv_range)(unsigned long, unsigned long);
 	void (*clean_range)(unsigned long, unsigned long);
 	void (*flush_range)(unsigned long, unsigned long);
@@ -112,6 +119,11 @@ static inline void outer_resume(void)
 		outer_cache.resume();
 }
 
+static inline void outer_get_info(struct outer_cache_info *info)
+{
+	if (outer_cache.get_info)
+		outer_cache.get_info(info);
+}
 #else
 
 static inline void outer_inv_range(phys_addr_t start, phys_addr_t end)
@@ -123,6 +135,7 @@ static inline void outer_flush_range(phys_addr_t start, phys_addr_t end)
 static inline void outer_flush_all(void) { }
 static inline void outer_disable(void) { }
 static inline void outer_resume(void) { }
+static inline void outer_get_info(struct outer_cache_info *info) { }
 
 #endif
 
diff --git a/arch/arm/kernel/cacheinfo.c b/arch/arm/kernel/cacheinfo.c
index ab70993..88b552b 100644
--- a/arch/arm/kernel/cacheinfo.c
+++ b/arch/arm/kernel/cacheinfo.c
@@ -16,6 +16,7 @@
 #include <linux/of.h>
 
 #include <asm/cputype.h>
+#include <asm/outercache.h>
 #include <asm/processor.h>
 
 #if __LINUX_ARM_ARCH__ < 7 /* pre ARMv7 */
@@ -176,11 +177,27 @@ static void __ci_leaf_init(enum cache_type type, struct cacheinfo *this_leaf)
 
 #endif
 
+static void __outer_ci_leaf_init(struct cacheinfo *this_leaf)
+{
+	struct outer_cache_info info;
+
+	outer_get_info(&info);
+
+	this_leaf->type = CACHE_TYPE_UNIFIED;/* record it as Unified */
+	this_leaf->ways_of_associativity = info.num_ways;
+	this_leaf->number_of_sets = info.num_sets;
+	this_leaf->coherency_line_size = info.line_size;
+	this_leaf->size = info.num_ways * info.num_sets * info.line_size;
+}
+
 static void ci_leaf_init(struct cacheinfo *this_leaf,
 			 enum cache_type type, unsigned int level)
 {
 	this_leaf->level = level;
-	__ci_leaf_init(type, this_leaf);
+	if (type == CACHE_TYPE_NOCACHE)	/* must be outer cache */
+		__outer_ci_leaf_init(this_leaf);
+	else
+		__ci_leaf_init(type, this_leaf);
 }
 
 int init_cache_level(unsigned int cpu)
@@ -202,6 +219,9 @@ int init_cache_level(unsigned int cpu)
 	this_cpu_ci->num_levels = level - 1;
 	this_cpu_ci->num_leaves = leaves;
 
+	if (IS_ENABLED(CONFIG_OUTER_CACHE) && outer_cache.get_info)
+		this_cpu_ci->num_leaves++, this_cpu_ci->num_levels++;
+
 	return 0;
 }
 
diff --git a/arch/arm/mm/cache-l2x0.c b/arch/arm/mm/cache-l2x0.c
index efc5cab..30ca151 100644
--- a/arch/arm/mm/cache-l2x0.c
+++ b/arch/arm/mm/cache-l2x0.c
@@ -105,6 +105,15 @@ static inline void l2c_unlock(void __iomem *base, unsigned num)
 	}
 }
 
+static void l2x0_getinfo(struct outer_cache_info *info)
+{
+	if (!info)
+		return;
+	info->num_ways = get_count_order(l2x0_way_mask);
+	info->line_size = CACHE_LINE_SIZE;
+	info->num_sets = l2x0_size / (info->num_ways * CACHE_LINE_SIZE);
+}
+
 /*
  * Enable the L2 cache controller.  This function must only be
  * called when the cache controller is known to be disabled.
@@ -894,6 +903,7 @@ static void __init __l2c_init(const struct l2c_init_data *data,
 		data->enable(l2x0_base, aux, data->num_lock);
 
 	outer_cache = fns;
+	outer_cache.get_info = l2x0_getinfo;
 
 	/*
 	 * It is strange to save the register state before initialisation,
diff --git a/arch/arm/mm/cache-tauros2.c b/arch/arm/mm/cache-tauros2.c
index b273739..8708684 100644
--- a/arch/arm/mm/cache-tauros2.c
+++ b/arch/arm/mm/cache-tauros2.c
@@ -60,6 +60,7 @@ static inline void tauros2_inv_pa(unsigned long addr)
  * noninclusive.
  */
 #define CACHE_LINE_SIZE		32
+#define CACHE_LINE_SHIFT	5
 
 static void tauros2_inv_range(unsigned long start, unsigned long end)
 {
@@ -131,6 +132,38 @@ static void tauros2_resume(void)
 	"mcr	p15, 0, %0, c1, c0, 0 @Enable L2 Cache\n\t"
 	: : "r" (0x0));
 }
+
+/*
+ *  +----------------------------------------+
+ *  | 11 10 9  8 | 7  6  5  4  3 | 2 |  1  0 |
+ *  +----------------------------------------+
+ *  |  way size  | associativity | - |line_sz|
+ *  +----------------------------------------+
+ */
+#define L2CTR_ASSOCIAT_SHIFT	3
+#define L2CTR_ASSOCIAT_MASK	0x1F
+#define L2CTR_WAYSIZE_SHIFT	8
+#define L2CTR_WAYSIZE_MASK	0xF
+#define CACHE_WAY_PER_SET(l2ctr)	\
+	(((l2_ctr) >> L2CTR_ASSOCIAT_SHIFT) & L2CTR_ASSOCIAT_MASK)
+#define CACHE_WAY_SIZE(l2ctr)		\
+	(8192 << (((l2ctr) >> L2CTR_WAYSIZE_SHIFT) & L2CTR_WAYSIZE_MASK))
+#define CACHE_SET_SIZE(l2ctr)	(CACHE_WAY_SIZE(l2ctr) >> CACHE_LINE_SHIFT)
+
+static void tauros2_getinfo(struct outer_cache_info *info)
+{
+	unsigned int l2_ctr;
+
+	if (!info)
+		return;
+
+	__asm__("mrc p15, 1, %0, c0, c0, 1" : "=r" (l2_ctr));
+
+	info->line_size = CACHE_LINE_SIZE;
+	info->num_ways = CACHE_WAY_PER_SET(l2_ctr);
+	info->num_sets = CACHE_SET_SIZE(l2_ctr);
+}
+
 #endif
 
 static inline u32 __init read_extra_features(void)
@@ -226,6 +259,7 @@ static void __init tauros2_internal_init(unsigned int features)
 		outer_cache.flush_range = tauros2_flush_range;
 		outer_cache.disable = tauros2_disable;
 		outer_cache.resume = tauros2_resume;
+		outer_cache.get_info = tauros2_getinfo;
 	}
 #endif
 
diff --git a/arch/arm/mm/cache-xsc3l2.c b/arch/arm/mm/cache-xsc3l2.c
index 6c3edeb..353c642 100644
--- a/arch/arm/mm/cache-xsc3l2.c
+++ b/arch/arm/mm/cache-xsc3l2.c
@@ -201,6 +201,20 @@ static void xsc3_l2_flush_range(unsigned long start, unsigned long end)
 	dsb();
 }
 
+static void xsc3_l2_getinfo(struct outer_cache_info *info)
+{
+	unsigned long l2ctype;
+
+	if (!info)
+		return;
+
+	__asm__("mrc p15, 1, %0, c0, c0, 1" : "=r" (l2ctype));
+
+	info->num_ways = CACHE_WAY_PER_SET;
+	info->line_size = CACHE_LINE_SIZE;
+	info->num_sets = CACHE_SET_SIZE(l2ctype);
+}
+
 static int __init xsc3_l2_init(void)
 {
 	if (!cpu_is_xsc3() || !xsc3_l2_present())
@@ -213,6 +227,7 @@ static int __init xsc3_l2_init(void)
 		outer_cache.inv_range = xsc3_l2_inv_range;
 		outer_cache.clean_range = xsc3_l2_clean_range;
 		outer_cache.flush_range = xsc3_l2_flush_range;
+		outer_cache.get_info    = xsc3_l2_getinfo;
 	}
 
 	return 0;
-- 
1.8.3.2

Re: [PATCH 9/9] ARM: kernel: add outer cache support for cacheinfo implementation

[Russell King - ARM Linux]

On Wed, Jun 25, 2014 at 06:30:44PM +0100, Sudeep Holla wrote:
> diff --git a/arch/arm/mm/cache-l2x0.c b/arch/arm/mm/cache-l2x0.c
> index efc5cab..30ca151 100644
> --- a/arch/arm/mm/cache-l2x0.c
> +++ b/arch/arm/mm/cache-l2x0.c
> @@ -105,6 +105,15 @@ static inline void l2c_unlock(void __iomem *base, unsigned num)
>  	}
>  }
>  
> +static void l2x0_getinfo(struct outer_cache_info *info)
> +{
> +	if (!info)
> +		return;

Pointless NULL test.  If someone passes NULL to this function (which
you never do in this file) then we want to know about it because _that_
is a kernel bug - it is invalid to pass NULL.  Hence the kernel should
oops.

Please, don't go around adding stupid NULL tests for conditions which
should _never_ happen, instead, rely on the kernel to oops if these
invalid conditions occur.  That's why we produce a backtrace from such
events, to allow invalid conditions to be debugged and fixed.

Having stuff silently ignore in this way does not detect these bugs so
they go by unnoticed.

Take a moment to read some of the fs/ or kernel/ code, and you'll find
a lack of NULL checks in there.  That's what gives that code performance,
because it's not spending its time doing loads of useless NULL checks.

> @@ -894,6 +903,7 @@ static void __init __l2c_init(const struct l2c_init_data *data,
>  		data->enable(l2x0_base, aux, data->num_lock);
>  
>  	outer_cache = fns;
> +	outer_cache.get_info = l2x0_getinfo;

NAK.  Think about it.

-- 
FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly
improving, and getting towards what was expected from it.

[PATCH 9/9] ARM: kernel: add outer cache support for cacheinfo implementation

[Russell King - ARM Linux]

On Wed, Jun 25, 2014 at 06:30:44PM +0100, Sudeep Holla wrote:
> diff --git a/arch/arm/mm/cache-l2x0.c b/arch/arm/mm/cache-l2x0.c
> index efc5cab..30ca151 100644
> --- a/arch/arm/mm/cache-l2x0.c
> +++ b/arch/arm/mm/cache-l2x0.c
> @@ -105,6 +105,15 @@ static inline void l2c_unlock(void __iomem *base, unsigned num)
>  	}
>  }
>  
> +static void l2x0_getinfo(struct outer_cache_info *info)
> +{
> +	if (!info)
> +		return;

Pointless NULL test.  If someone passes NULL to this function (which
you never do in this file) then we want to know about it because _that_
is a kernel bug - it is invalid to pass NULL.  Hence the kernel should
oops.

Please, don't go around adding stupid NULL tests for conditions which
should _never_ happen, instead, rely on the kernel to oops if these
invalid conditions occur.  That's why we produce a backtrace from such
events, to allow invalid conditions to be debugged and fixed.

Having stuff silently ignore in this way does not detect these bugs so
they go by unnoticed.

Take a moment to read some of the fs/ or kernel/ code, and you'll find
a lack of NULL checks in there.  That's what gives that code performance,
because it's not spending its time doing loads of useless NULL checks.

> @@ -894,6 +903,7 @@ static void __init __l2c_init(const struct l2c_init_data *data,
>  		data->enable(l2x0_base, aux, data->num_lock);
>  
>  	outer_cache = fns;
> +	outer_cache.get_info = l2x0_getinfo;

NAK.  Think about it.

-- 
FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly
improving, and getting towards what was expected from it.

Re: [PATCH 9/9] ARM: kernel: add outer cache support for cacheinfo implementation

[Sudeep Holla]

Hi Russell,

On 25/06/14 23:37, Russell King - ARM Linux wrote:
> On Wed, Jun 25, 2014 at 06:30:44PM +0100, Sudeep Holla wrote:
>> diff --git a/arch/arm/mm/cache-l2x0.c b/arch/arm/mm/cache-l2x0.c
>> index efc5cab..30ca151 100644
>> --- a/arch/arm/mm/cache-l2x0.c
>> +++ b/arch/arm/mm/cache-l2x0.c
>> @@ -105,6 +105,15 @@ static inline void l2c_unlock(void __iomem *base, unsigned num)
>>   	}
>>   }
>>
>> +static void l2x0_getinfo(struct outer_cache_info *info)
>> +{
>> +	if (!info)
>> +		return;
>
> Pointless NULL test.  If someone passes NULL to this function (which
> you never do in this file) then we want to know about it because _that_
> is a kernel bug - it is invalid to pass NULL.  Hence the kernel should
> oops.
>
> Please, don't go around adding stupid NULL tests for conditions which
> should _never_ happen, instead, rely on the kernel to oops if these
> invalid conditions occur.  That's why we produce a backtrace from such
> events, to allow invalid conditions to be debugged and fixed.
>
> Having stuff silently ignore in this way does not detect these bugs so
> they go by unnoticed.
>
> Take a moment to read some of the fs/ or kernel/ code, and you'll find
> a lack of NULL checks in there.  That's what gives that code performance,
> because it's not spending its time doing loads of useless NULL checks.
>

Understood, will get rid of it.

>> @@ -894,6 +903,7 @@ static void __init __l2c_init(const struct l2c_init_data *data,
>>   		data->enable(l2x0_base, aux, data->num_lock);
>>
>>   	outer_cache = fns;
>> +	outer_cache.get_info = l2x0_getinfo;
>
> NAK.  Think about it.
>

Ah, will specify in l2c_init_data for individual implementations so that
fixups is possible if needed for get_info. Sorry for missing this.

Regards,
Sudeep

[PATCH 9/9] ARM: kernel: add outer cache support for cacheinfo implementation

[Sudeep Holla]

Hi Russell,

On 25/06/14 23:37, Russell King - ARM Linux wrote:
> On Wed, Jun 25, 2014 at 06:30:44PM +0100, Sudeep Holla wrote:
>> diff --git a/arch/arm/mm/cache-l2x0.c b/arch/arm/mm/cache-l2x0.c
>> index efc5cab..30ca151 100644
>> --- a/arch/arm/mm/cache-l2x0.c
>> +++ b/arch/arm/mm/cache-l2x0.c
>> @@ -105,6 +105,15 @@ static inline void l2c_unlock(void __iomem *base, unsigned num)
>>   	}
>>   }
>>
>> +static void l2x0_getinfo(struct outer_cache_info *info)
>> +{
>> +	if (!info)
>> +		return;
>
> Pointless NULL test.  If someone passes NULL to this function (which
> you never do in this file) then we want to know about it because _that_
> is a kernel bug - it is invalid to pass NULL.  Hence the kernel should
> oops.
>
> Please, don't go around adding stupid NULL tests for conditions which
> should _never_ happen, instead, rely on the kernel to oops if these
> invalid conditions occur.  That's why we produce a backtrace from such
> events, to allow invalid conditions to be debugged and fixed.
>
> Having stuff silently ignore in this way does not detect these bugs so
> they go by unnoticed.
>
> Take a moment to read some of the fs/ or kernel/ code, and you'll find
> a lack of NULL checks in there.  That's what gives that code performance,
> because it's not spending its time doing loads of useless NULL checks.
>

Understood, will get rid of it.

>> @@ -894,6 +903,7 @@ static void __init __l2c_init(const struct l2c_init_data *data,
>>   		data->enable(l2x0_base, aux, data->num_lock);
>>
>>   	outer_cache = fns;
>> +	outer_cache.get_info = l2x0_getinfo;
>
> NAK.  Think about it.
>

Ah, will specify in l2c_init_data for individual implementations so that
fixups is possible if needed for get_info. Sorry for missing this.

Regards,
Sudeep

[PATCH v2 0/9] drivers: cacheinfo support

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This series adds a generic cacheinfo support similar to topology. The
implementation is based on x86 cacheinfo support. Currently x86, powerpc,
ia64 and s390 have their own implementations. While adding similar support
to ARM and ARM64, here is the attempt to make it generic quite similar to
topology info support. It also adds the missing ABI documentation for
the cacheinfo sysfs which is already being used.

It moves all the existing different implementations on x86, ia64, powerpc
and s390 to use the generic cacheinfo infrastructure introduced here.
These changes on non-ARM platforms are only compile tested and tested on x86.

This series also adds support for ARM and ARM64 architectures based on
the generic support.

Since there was no objection to the idea in RFC, I am posting non-RFC
version here.

The code can be fetched from:
 git://linux-arm.org/linux-skn cacheinfo


Changes v1->v2:
- removed custom device_{add,remove}_attrs, using is_visible callback
  instead(suggested by GregKH)
- arm64: changes as per MarkR review comments
- Moved smp_call_function_single to architectures using it(arm, arm64,
  x86) (suggested by Stephen Boyd)
- arm (mostly changes as per RMK's review comments)
	- fixed to allow v7 + v6 build
	- l2 cache changes to remove extra structure
	- populated CTR for few StrongARM CPU's not implementing CTR

Previous RFCs:
[1] https://lkml.org/lkml/2014/1/8/523
[2] https://lkml.org/lkml/2014/2/7/654
[3] https://lkml.org/lkml/2014/2/19/391

Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-ia64@vger.kernel.org
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org

Sudeep Holla (9):
  drivers: base: add new class "cpu" to group cpu devices
  drivers: base: support cpu cache information interface to userspace
    via sysfs
  ia64: move cacheinfo sysfs to generic cacheinfo infrastructure
  s390: move cacheinfo sysfs to generic cacheinfo infrastructure
  x86: move cacheinfo sysfs to generic cacheinfo infrastructure
  powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure
  ARM64: kernel: add support for cpu cache information
  ARM: kernel: add support for cpu cache information
  ARM: kernel: add outer cache support for cacheinfo implementation

 Documentation/ABI/testing/sysfs-devices-system-cpu |  41 ++
 arch/arm/include/asm/outercache.h                  |   9 +
 arch/arm/kernel/Makefile                           |   1 +
 arch/arm/kernel/cacheinfo.c                        | 284 +++++++
 arch/arm/mm/Kconfig                                |  13 +
 arch/arm/mm/cache-l2x0.c                           |  35 +-
 arch/arm/mm/cache-tauros2.c                        |  35 +
 arch/arm/mm/cache-xsc3l2.c                         |  16 +
 arch/arm64/kernel/Makefile                         |   3 +-
 arch/arm64/kernel/cacheinfo.c                      | 142 ++++
 arch/ia64/kernel/topology.c                        | 401 ++--------
 arch/powerpc/kernel/cacheinfo.c                    | 813 +++------------------
 arch/powerpc/kernel/cacheinfo.h                    |   8 -
 arch/powerpc/kernel/sysfs.c                        |  12 +-
 arch/s390/kernel/cache.c                           | 388 +++-------
 arch/x86/kernel/cpu/intel_cacheinfo.c              | 680 +++++------------
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 539 ++++++++++++++
 drivers/base/core.c                                |  39 +-
 drivers/base/cpu.c                                 |   7 +
 include/linux/cacheinfo.h                          |  73 ++
 include/linux/cpu.h                                |   2 +
 22 files changed, 1660 insertions(+), 1883 deletions(-)
 create mode 100644 arch/arm/kernel/cacheinfo.c
 create mode 100644 arch/arm64/kernel/cacheinfo.c
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

-- 
1.8.3.2

[PATCH v2 0/9] drivers: cacheinfo support

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This series adds a generic cacheinfo support similar to topology. The
implementation is based on x86 cacheinfo support. Currently x86, powerpc,
ia64 and s390 have their own implementations. While adding similar support
to ARM and ARM64, here is the attempt to make it generic quite similar to
topology info support. It also adds the missing ABI documentation for
the cacheinfo sysfs which is already being used.

It moves all the existing different implementations on x86, ia64, powerpc
and s390 to use the generic cacheinfo infrastructure introduced here.
These changes on non-ARM platforms are only compile tested and tested on x86.

This series also adds support for ARM and ARM64 architectures based on
the generic support.

Since there was no objection to the idea in RFC, I am posting non-RFC
version here.

The code can be fetched from:
 git://linux-arm.org/linux-skn cacheinfo


Changes v1->v2:
- removed custom device_{add,remove}_attrs, using is_visible callback
  instead(suggested by GregKH)
- arm64: changes as per MarkR review comments
- Moved smp_call_function_single to architectures using it(arm, arm64,
  x86) (suggested by Stephen Boyd)
- arm (mostly changes as per RMK's review comments)
	- fixed to allow v7 + v6 build
	- l2 cache changes to remove extra structure
	- populated CTR for few StrongARM CPU's not implementing CTR

Previous RFCs:
[1] https://lkml.org/lkml/2014/1/8/523
[2] https://lkml.org/lkml/2014/2/7/654
[3] https://lkml.org/lkml/2014/2/19/391

Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-ia64@vger.kernel.org
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org

Sudeep Holla (9):
  drivers: base: add new class "cpu" to group cpu devices
  drivers: base: support cpu cache information interface to userspace
    via sysfs
  ia64: move cacheinfo sysfs to generic cacheinfo infrastructure
  s390: move cacheinfo sysfs to generic cacheinfo infrastructure
  x86: move cacheinfo sysfs to generic cacheinfo infrastructure
  powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure
  ARM64: kernel: add support for cpu cache information
  ARM: kernel: add support for cpu cache information
  ARM: kernel: add outer cache support for cacheinfo implementation

 Documentation/ABI/testing/sysfs-devices-system-cpu |  41 ++
 arch/arm/include/asm/outercache.h                  |   9 +
 arch/arm/kernel/Makefile                           |   1 +
 arch/arm/kernel/cacheinfo.c                        | 284 +++++++
 arch/arm/mm/Kconfig                                |  13 +
 arch/arm/mm/cache-l2x0.c                           |  35 +-
 arch/arm/mm/cache-tauros2.c                        |  35 +
 arch/arm/mm/cache-xsc3l2.c                         |  16 +
 arch/arm64/kernel/Makefile                         |   3 +-
 arch/arm64/kernel/cacheinfo.c                      | 142 ++++
 arch/ia64/kernel/topology.c                        | 401 ++--------
 arch/powerpc/kernel/cacheinfo.c                    | 813 +++------------------
 arch/powerpc/kernel/cacheinfo.h                    |   8 -
 arch/powerpc/kernel/sysfs.c                        |  12 +-
 arch/s390/kernel/cache.c                           | 388 +++-------
 arch/x86/kernel/cpu/intel_cacheinfo.c              | 680 +++++------------
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 539 ++++++++++++++
 drivers/base/core.c                                |  39 +-
 drivers/base/cpu.c                                 |   7 +
 include/linux/cacheinfo.h                          |  73 ++
 include/linux/cpu.h                                |   2 +
 22 files changed, 1660 insertions(+), 1883 deletions(-)
 create mode 100644 arch/arm/kernel/cacheinfo.c
 create mode 100644 arch/arm64/kernel/cacheinfo.c
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

-- 
1.8.3.2

[PATCH v2 0/9] drivers: cacheinfo support

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This series adds a generic cacheinfo support similar to topology. The
implementation is based on x86 cacheinfo support. Currently x86, powerpc,
ia64 and s390 have their own implementations. While adding similar support
to ARM and ARM64, here is the attempt to make it generic quite similar to
topology info support. It also adds the missing ABI documentation for
the cacheinfo sysfs which is already being used.

It moves all the existing different implementations on x86, ia64, powerpc
and s390 to use the generic cacheinfo infrastructure introduced here.
These changes on non-ARM platforms are only compile tested and tested on x86.

This series also adds support for ARM and ARM64 architectures based on
the generic support.

Since there was no objection to the idea in RFC, I am posting non-RFC
version here.

The code can be fetched from:
 git://linux-arm.org/linux-skn cacheinfo


Changes v1->v2:
- removed custom device_{add,remove}_attrs, using is_visible callback
  instead(suggested by GregKH)
- arm64: changes as per MarkR review comments
- Moved smp_call_function_single to architectures using it(arm, arm64,
  x86) (suggested by Stephen Boyd)
- arm (mostly changes as per RMK's review comments)
	- fixed to allow v7 + v6 build
	- l2 cache changes to remove extra structure
	- populated CTR for few StrongARM CPU's not implementing CTR

Previous RFCs:
[1] https://lkml.org/lkml/2014/1/8/523
[2] https://lkml.org/lkml/2014/2/7/654
[3] https://lkml.org/lkml/2014/2/19/391

Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-ia64@vger.kernel.org
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org

Sudeep Holla (9):
  drivers: base: add new class "cpu" to group cpu devices
  drivers: base: support cpu cache information interface to userspace
    via sysfs
  ia64: move cacheinfo sysfs to generic cacheinfo infrastructure
  s390: move cacheinfo sysfs to generic cacheinfo infrastructure
  x86: move cacheinfo sysfs to generic cacheinfo infrastructure
  powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure
  ARM64: kernel: add support for cpu cache information
  ARM: kernel: add support for cpu cache information
  ARM: kernel: add outer cache support for cacheinfo implementation

 Documentation/ABI/testing/sysfs-devices-system-cpu |  41 ++
 arch/arm/include/asm/outercache.h                  |   9 +
 arch/arm/kernel/Makefile                           |   1 +
 arch/arm/kernel/cacheinfo.c                        | 284 +++++++
 arch/arm/mm/Kconfig                                |  13 +
 arch/arm/mm/cache-l2x0.c                           |  35 +-
 arch/arm/mm/cache-tauros2.c                        |  35 +
 arch/arm/mm/cache-xsc3l2.c                         |  16 +
 arch/arm64/kernel/Makefile                         |   3 +-
 arch/arm64/kernel/cacheinfo.c                      | 142 ++++
 arch/ia64/kernel/topology.c                        | 401 ++--------
 arch/powerpc/kernel/cacheinfo.c                    | 813 +++------------------
 arch/powerpc/kernel/cacheinfo.h                    |   8 -
 arch/powerpc/kernel/sysfs.c                        |  12 +-
 arch/s390/kernel/cache.c                           | 388 +++-------
 arch/x86/kernel/cpu/intel_cacheinfo.c              | 680 +++++------------
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 539 ++++++++++++++
 drivers/base/core.c                                |  39 +-
 drivers/base/cpu.c                                 |   7 +
 include/linux/cacheinfo.h                          |  73 ++
 include/linux/cpu.h                                |   2 +
 22 files changed, 1660 insertions(+), 1883 deletions(-)
 create mode 100644 arch/arm/kernel/cacheinfo.c
 create mode 100644 arch/arm64/kernel/cacheinfo.c
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

-- 
1.8.3.2

[PATCH v2 1/9] drivers: base: add new class "cpu" to group cpu devices

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch creates a new class called "cpu" and assigns it to all the
cpu devices. This helps in grouping all the cpu devices and associated
child devices under the same class.

This patch also:
1. modifies the get_parent_device to return the legacy path
   (/sys/devices/system/cpu/..) for the cpu class devices to support
   existing sysfs ABI
2. avoids creating link in the class directory pointing to the device as
   there would be per-cpu instance of these devices with the same name
3. makes sure subsystem symlink continues pointing to cpu bus instead of
   cpu class for cpu devices

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 drivers/base/core.c | 39 +++++++++++++++++++++++++++++++++------
 drivers/base/cpu.c  |  7 +++++++
 include/linux/cpu.h |  2 ++
 3 files changed, 42 insertions(+), 6 deletions(-)

diff --git a/drivers/base/core.c b/drivers/base/core.c
index 20da3ad1696b..fe622e2a48d0 100644
--- a/drivers/base/core.c
+++ b/drivers/base/core.c
@@ -10,6 +10,7 @@
  *
  */
 
+#include <linux/cpu.h>
 #include <linux/device.h>
 #include <linux/err.h>
 #include <linux/init.h>
@@ -742,6 +743,12 @@ static struct kobject *get_device_parent(struct device *dev,
 			return &block_class.p->subsys.kobj;
 		}
 #endif
+		/*
+		 * if the device is in cpu class, then use the default/legacy
+		 * /sys/devices/system/cpu/.. path
+		 */
+		if (dev->class == cpu_class)
+			return &parent->kobj;
 
 		/*
 		 * If we have no parent, we live in "virtual".
@@ -808,11 +815,17 @@ static int device_add_class_symlinks(struct device *dev)
 	if (!dev->class)
 		return 0;
 
-	error = sysfs_create_link(&dev->kobj,
-				  &dev->class->p->subsys.kobj,
-				  "subsystem");
-	if (error)
-		goto out;
+	/*
+	 * the subsystem symlink in each cpu device needs to continue
+	 * pointing to cpu bus
+	 */
+	if (dev->bus != &cpu_subsys) {
+		error = sysfs_create_link(&dev->kobj,
+					  &dev->class->p->subsys.kobj,
+					  "subsystem");
+		if (error)
+			goto out;
+	}
 
 	if (dev->parent && device_is_not_partition(dev)) {
 		error = sysfs_create_link(&dev->kobj, &dev->parent->kobj,
@@ -826,6 +839,13 @@ static int device_add_class_symlinks(struct device *dev)
 	if (sysfs_deprecated && dev->class == &block_class)
 		return 0;
 #endif
+	/*
+	 * don't create a link in the cpu class directory pointing to the
+	 * device as there would be per-cpu instance of these devices with
+	 * the same name
+	 */
+	if (dev->class == cpu_class)
+		return 0;
 
 	/* link in the class directory pointing to the device */
 	error = sysfs_create_link(&dev->class->p->subsys.kobj,
@@ -851,11 +871,18 @@ static void device_remove_class_symlinks(struct device *dev)
 
 	if (dev->parent && device_is_not_partition(dev))
 		sysfs_remove_link(&dev->kobj, "device");
-	sysfs_remove_link(&dev->kobj, "subsystem");
+
+	/* if subsystem points to cpu bus, bus_remove_device will remove it */
+	if (dev->bus != &cpu_subsys)
+		sysfs_remove_link(&dev->kobj, "subsystem");
 #ifdef CONFIG_BLOCK
 	if (sysfs_deprecated && dev->class == &block_class)
 		return;
 #endif
+	/* symlinks are not created for cpu class devices, nothing to remove */
+	if (dev->class == cpu_class)
+		return;
+
 	sysfs_delete_link(&dev->class->p->subsys.kobj, &dev->kobj, dev_name(dev));
 }
 
diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c
index 006b1bc5297d..056b8854e4fb 100644
--- a/drivers/base/cpu.c
+++ b/drivers/base/cpu.c
@@ -322,6 +322,7 @@ static int cpu_uevent(struct device *dev, struct kobj_uevent_env *env)
 }
 #endif
 
+struct class *cpu_class;
 /*
  * register_cpu - Setup a sysfs device for a CPU.
  * @cpu - cpu->hotpluggable field set to 1 will generate a control file in
@@ -338,6 +339,8 @@ int register_cpu(struct cpu *cpu, int num)
 	memset(&cpu->dev, 0x00, sizeof(struct device));
 	cpu->dev.id = num;
 	cpu->dev.bus = &cpu_subsys;
+	cpu->dev.parent = cpu_subsys.dev_root;
+	cpu->dev.class = cpu_class;
 	cpu->dev.release = cpu_device_release;
 	cpu->dev.offline_disabled = !cpu->hotpluggable;
 	cpu->dev.offline = !cpu_online(num);
@@ -423,5 +426,9 @@ void __init cpu_dev_init(void)
 	if (subsys_system_register(&cpu_subsys, cpu_root_attr_groups))
 		panic("Failed to register CPU subsystem");
 
+	cpu_class = class_create(THIS_MODULE, "cpu");
+	if (IS_ERR(cpu_class))
+		panic("Failed to register CPU class");
+
 	cpu_dev_register_generic();
 }
diff --git a/include/linux/cpu.h b/include/linux/cpu.h
index 95978ad7fcdd..8c0fc9b0acad 100644
--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -39,6 +39,8 @@ extern void cpu_remove_dev_attr(struct device_attribute *attr);
 extern int cpu_add_dev_attr_group(struct attribute_group *attrs);
 extern void cpu_remove_dev_attr_group(struct attribute_group *attrs);
 
+extern struct class *cpu_class;
+
 #ifdef CONFIG_HOTPLUG_CPU
 extern void unregister_cpu(struct cpu *cpu);
 extern ssize_t arch_cpu_probe(const char *, size_t);
-- 
1.8.3.2

Re: [PATCH v2 1/9] drivers: base: add new class "cpu" to group cpu devices

[Stephen Boyd]

On 07/25/14 09:44, Sudeep Holla wrote:
> From: Sudeep Holla <sudeep.holla@arm.com>
>
> This patch creates a new class called "cpu" and assigns it to all the
> cpu devices. This helps in grouping all the cpu devices and associated
> child devices under the same class.
>
> This patch also:
> 1. modifies the get_parent_device to return the legacy path
>    (/sys/devices/system/cpu/..) for the cpu class devices to support
>    existing sysfs ABI
> 2. avoids creating link in the class directory pointing to the device as
>    there would be per-cpu instance of these devices with the same name
> 3. makes sure subsystem symlink continues pointing to cpu bus instead of
>    cpu class for cpu devices
>
> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Why can't we make a 'cache' bus that the cache devices hang off of? It
doesn't make sense to me why we would want to have cache devices living
on the cpu bus.

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation

Re: [PATCH v2 1/9] drivers: base: add new class "cpu" to group cpu devices

[Sudeep Holla]

On 25/07/14 20:09, Stephen Boyd wrote:
> On 07/25/14 09:44, Sudeep Holla wrote:
>> From: Sudeep Holla <sudeep.holla@arm.com>
>>
>> This patch creates a new class called "cpu" and assigns it to all the
>> cpu devices. This helps in grouping all the cpu devices and associated
>> child devices under the same class.
>>
>> This patch also:
>> 1. modifies the get_parent_device to return the legacy path
>>     (/sys/devices/system/cpu/..) for the cpu class devices to support
>>     existing sysfs ABI
>> 2. avoids creating link in the class directory pointing to the device as
>>     there would be per-cpu instance of these devices with the same name
>> 3. makes sure subsystem symlink continues pointing to cpu bus instead of
>>     cpu class for cpu devices
>>
>> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
>> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
>
> Why can't we make a 'cache' bus that the cache devices hang off of? It
> doesn't make sense to me why we would want to have cache devices living
> on the cpu bus.
>

Yes we can group them under cache, but to support legacy sysfs path
under which various cpu related information including caches, we need
to have a workaround as in this patch.

Instead of having this fixup for each class like cache, having them
grouped under cpu is better IMO and hence I added caches to cpu class.
It also helps to move few other cpu related subsystems using raw
kobjects to the device attribute groups and corresponding APIs.

Regards,
Sudeep

[PATCH v2 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds initial support for providing processor cache information
to userspace through sysfs interface. This is based on already existing
implementations(x86, ia64, s390 and powerpc) and hence the interface is
intended to be fully compatible.

The main purpose of this generic support is to avoid further code
duplication to support new architectures and also to unify all the existing
different implementations.

This implementation maintains the hierarchy of cache objects which reflects
the system's cache topology. Cache devices are instantiated as needed as
CPUs come online. The cache information is replicated per-cpu even if they are
shared. A per-cpu array of cache information maintained is used mainly for
sysfs-related book keeping.

It also implements the shared_cpu_map attribute, which is essential for
enabling both kernel and user-space to discover the system's overall cache
topology.

This patch also add the missing ABI documentation for the cacheinfo sysfs
interface already, which is well defined and widely used.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-doc@vger.kernel.org
---
 Documentation/ABI/testing/sysfs-devices-system-cpu |  41 ++
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 539 +++++++++++++++++++++
 include/linux/cacheinfo.h                          |  73 +++
 4 files changed, 654 insertions(+), 1 deletion(-)
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

diff --git a/Documentation/ABI/testing/sysfs-devices-system-cpu b/Documentation/ABI/testing/sysfs-devices-system-cpu
index acb9bfc89b48..832b7f2ed6d2 100644
--- a/Documentation/ABI/testing/sysfs-devices-system-cpu
+++ b/Documentation/ABI/testing/sysfs-devices-system-cpu
@@ -224,3 +224,44 @@ Description:	Parameters for the Intel P-state driver
 		frequency range.
 
 		More details can be found in Documentation/cpu-freq/intel-pstate.txt
+
+What:		/sys/devices/system/cpu/cpu*/cache/index*/<set_of_attributes_mentioned_below>
+Date:		July 2014(documented, existed before August 2008)
+Contact:	Sudeep Holla <sudeep.holla@arm.com>
+		Linux kernel mailing list <linux-kernel@vger.kernel.org>
+Description:	Parameters for the CPU cache attributes
+
+		attributes:
+			- writethrough: data is written to both the cache line
+					and to the block in the lower-level memory
+			- writeback: data is written only to the cache line and
+				     the modified cache line is written to main
+				     memory only when it is replaced
+			- writeallocate: allocate a memory location to a cache line
+					 on a cache miss because of a write
+			- readallocate: allocate a memory location to a cache line
+					on a cache miss because of a read
+
+		coherency_line_size: the minimum amount of data that gets transferred
+
+		level: the cache hierarcy in the multi-level cache configuration
+
+		number_of_sets: total number of sets in the cache, a set is a
+				collection of cache lines with the same cache index
+
+		physical_line_partition: number of physical cache line per cache tag
+
+		shared_cpu_list: the list of cpus sharing the cache
+
+		shared_cpu_map: logical cpu mask containing the list of cpus sharing
+				the cache
+
+		size: the total cache size in kB
+
+		type:
+			- instruction: cache that only holds instructions
+			- data: cache that only caches data
+			- unified: cache that holds both data and instructions
+
+		ways_of_associativity: degree of freedom in placing a particular block
+					of memory in the cache
diff --git a/drivers/base/Makefile b/drivers/base/Makefile
index 04b314e0fa51..bad2ff809bec 100644
--- a/drivers/base/Makefile
+++ b/drivers/base/Makefile
@@ -4,7 +4,7 @@ obj-y			:= component.o core.o bus.o dd.o syscore.o \
 			   driver.o class.o platform.o \
 			   cpu.o firmware.o init.o map.o devres.o \
 			   attribute_container.o transport_class.o \
-			   topology.o container.o
+			   topology.o container.o cacheinfo.o
 obj-$(CONFIG_DEVTMPFS)	+= devtmpfs.o
 obj-$(CONFIG_DMA_CMA) += dma-contiguous.o
 obj-y			+= power/
diff --git a/drivers/base/cacheinfo.c b/drivers/base/cacheinfo.c
new file mode 100644
index 000000000000..983728a919ec
--- /dev/null
+++ b/drivers/base/cacheinfo.c
@@ -0,0 +1,539 @@
+/*
+ * cacheinfo support - processor cache information via sysfs
+ *
+ * Based on arch/x86/kernel/cpu/intel_cacheinfo.c
+ * Author: Sudeep Holla <sudeep.holla@arm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/compiler.h>
+#include <linux/cpu.h>
+#include <linux/device.h>
+#include <linux/init.h>
+#include <linux/of.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/smp.h>
+#include <linux/sysfs.h>
+
+/* pointer to per cpu cacheinfo */
+static DEFINE_PER_CPU(struct cpu_cacheinfo, ci_cpu_cacheinfo);
+#define ci_cacheinfo(cpu)	(&per_cpu(ci_cpu_cacheinfo, cpu))
+#define cache_leaves(cpu)	(ci_cacheinfo(cpu)->num_leaves)
+#define per_cpu_cacheinfo(cpu)	(ci_cacheinfo(cpu)->info_list)
+
+struct cpu_cacheinfo *get_cpu_cacheinfo(unsigned int cpu)
+{
+	return ci_cacheinfo(cpu);
+}
+
+#ifdef CONFIG_OF
+static int cache_setup_of_node(unsigned int cpu)
+{
+	struct device_node *np;
+	struct cacheinfo *this_leaf;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int index = 0;
+
+	/* skip if of_node is already populated */
+	if (this_cpu_ci->info_list->of_node)
+		return 0;
+
+	if (!cpu_dev) {
+		pr_err("No cpu device for CPU %d\n", cpu);
+		return -ENODEV;
+	}
+	np = cpu_dev->of_node;
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
+	}
+
+	while (np && index < cache_leaves(cpu)) {
+		this_leaf = this_cpu_ci->info_list + index;
+		if (this_leaf->level != 1)
+			np = of_find_next_cache_node(np);
+		else
+			np = of_node_get(np);/* cpu node itself */
+		this_leaf->of_node = np;
+		index++;
+	}
+	return 0;
+}
+
+static inline bool cache_leaves_are_shared(struct cacheinfo *this_leaf,
+					   struct cacheinfo *sib_leaf)
+{
+	return sib_leaf->of_node == this_leaf->of_node;
+}
+
+static int of_cache_shared_cpu_map_setup(unsigned int cpu)
+{
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sib_leaf;
+	unsigned int index;
+	int ret;
+
+	ret = cache_setup_of_node(cpu);
+	if (ret)
+		return ret;
+
+	for (index = 0; index < cache_leaves(cpu); index++) {
+		unsigned int i;
+
+		this_leaf = this_cpu_ci->info_list + index;
+		cpumask_set_cpu(cpu, &this_leaf->shared_cpu_map);
+
+		for_each_online_cpu(i) {
+			struct cpu_cacheinfo *sib_cpu_ci = get_cpu_cacheinfo(i);
+
+			if (i == cpu || !sib_cpu_ci->info_list)
+				continue;/* skip if itself or no cacheinfo */
+			sib_leaf = sib_cpu_ci->info_list + index;
+			if (cache_leaves_are_shared(this_leaf, sib_leaf)) {
+				cpumask_set_cpu(cpu, &sib_leaf->shared_cpu_map);
+				cpumask_set_cpu(i, &this_leaf->shared_cpu_map);
+			}
+		}
+	}
+
+	return 0;
+}
+#else
+static inline int of_cache_shared_cpu_map_setup(unsigned int cpu)
+{
+	return 0;
+}
+#endif
+
+static void cache_shared_cpu_map_remove(unsigned int cpu)
+{
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sib_leaf;
+	unsigned int sibling, index;
+
+	for (index = 0; index < cache_leaves(cpu); index++) {
+		this_leaf = this_cpu_ci->info_list + index;
+		for_each_cpu(sibling, &this_leaf->shared_cpu_map) {
+			struct cpu_cacheinfo *sib_cpu_ci;
+
+			if (sibling == cpu) /* skip itself */
+				continue;
+			sib_cpu_ci = get_cpu_cacheinfo(sibling);
+			sib_leaf = sib_cpu_ci->info_list + index;
+			cpumask_clear_cpu(cpu, &sib_leaf->shared_cpu_map);
+			cpumask_clear_cpu(sibling, &this_leaf->shared_cpu_map);
+		}
+		of_node_put(this_leaf->of_node);
+	}
+}
+
+static void free_cache_attributes(unsigned int cpu)
+{
+	cache_shared_cpu_map_remove(cpu);
+
+	kfree(per_cpu_cacheinfo(cpu));
+	per_cpu_cacheinfo(cpu) = NULL;
+}
+
+int __weak init_cache_level(unsigned int cpu)
+{
+	return -ENOENT;
+}
+
+int __weak populate_cache_leaves(unsigned int cpu)
+{
+	return -ENOENT;
+}
+
+static int detect_cache_attributes(unsigned int cpu)
+{
+	int ret;
+
+	if (init_cache_level(cpu))
+		return -ENOENT;
+
+	per_cpu_cacheinfo(cpu) = kcalloc(cache_leaves(cpu),
+					 sizeof(struct cacheinfo), GFP_KERNEL);
+	if (per_cpu_cacheinfo(cpu) == NULL)
+		return -ENOMEM;
+
+	ret = populate_cache_leaves(cpu);
+	if (ret)
+		goto free_ci;
+	/*
+	 * For systems using DT for cache hierarcy, of_node and shared_cpu_map
+	 * will be set up here. Otherwise populate_cache_leaves needs to set
+	 * shared_cpu_map and next-level-cache should not be specified in DT
+	 */
+	ret = of_cache_shared_cpu_map_setup(cpu);
+	if (ret)
+		goto free_ci;
+	return 0;
+
+free_ci:
+	free_cache_attributes(cpu);
+	return ret;
+}
+
+#ifdef CONFIG_SYSFS
+
+/* pointer to cpuX/cache device */
+static DEFINE_PER_CPU(struct device *, ci_cache_dev);
+#define per_cpu_cache_dev(cpu)	(per_cpu(ci_cache_dev, cpu))
+
+static cpumask_t cache_dev_map;
+
+/* pointer to array of devices for cpuX/cache/indexY */
+static DEFINE_PER_CPU(struct device **, ci_index_dev);
+#define per_cpu_index_dev(cpu)	(per_cpu(ci_index_dev, cpu))
+#define per_cache_index_dev(cpu, idx)	((per_cpu_index_dev(cpu))[idx])
+
+#define show_one(file_name, object)				\
+static ssize_t file_name##_show(struct device *dev,		\
+		struct device_attribute *attr, char *buf)	\
+{								\
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);	\
+	if (!this_leaf->object)					\
+		return -EINVAL;					\
+	return sprintf(buf, "%u\n", this_leaf->object);		\
+}
+
+show_one(level, level);
+show_one(coherency_line_size, coherency_line_size);
+show_one(number_of_sets, number_of_sets);
+show_one(physical_line_partition, physical_line_partition);
+
+static ssize_t ways_of_associativity_show(struct device *dev,
+					  struct device_attribute *attr,
+					  char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	/* will be zero for fully associative cache, but check for size */
+	if (!this_leaf->size)
+		return -EINVAL;
+	return sprintf(buf, "%u\n", this_leaf->ways_of_associativity);
+}
+
+static ssize_t size_show(struct device *dev,
+			 struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	if (!this_leaf->size)
+		return -EINVAL;
+	return sprintf(buf, "%uK\n", this_leaf->size >> 10);
+}
+
+static ssize_t shared_cpumap_show_func(struct device *dev, int type, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf;
+	int n = 0;
+
+	if (len > 1) {
+		const struct cpumask *mask = &this_leaf->shared_cpu_map;
+
+		n = type ? cpulist_scnprintf(buf, len - 2, mask) :
+			   cpumask_scnprintf(buf, len - 2, mask);
+		buf[n++] = '\n';
+		buf[n] = '\0';
+	}
+	return n;
+}
+
+static ssize_t shared_cpu_map_show(struct device *dev,
+				   struct device_attribute *attr, char *buf)
+{
+	return shared_cpumap_show_func(dev, 0, buf);
+}
+
+static ssize_t shared_cpu_list_show(struct device *dev,
+				    struct device_attribute *attr, char *buf)
+{
+	return shared_cpumap_show_func(dev, 1, buf);
+}
+
+static ssize_t type_show(struct device *dev,
+			 struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	switch (this_leaf->type) {
+	case CACHE_TYPE_DATA:
+		return sprintf(buf, "Data\n");
+	case CACHE_TYPE_INST:
+		return sprintf(buf, "Instruction\n");
+	case CACHE_TYPE_UNIFIED:
+		return sprintf(buf, "Unified\n");
+	default:
+		return -EINVAL;
+	}
+}
+
+static ssize_t attributes_show(struct device *dev,
+			       struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	unsigned int ci_attr = this_leaf->attributes;
+	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf - 2;
+	int n = 0;
+
+	if (!ci_attr)
+		return -EINVAL;
+
+	if (ci_attr & CACHE_WRITE_THROUGH)
+		n += snprintf(buf + n, len - n, "WriteThrough\n");
+	if (ci_attr & CACHE_WRITE_BACK)
+		n += snprintf(buf + n, len - n, "WriteBack\n");
+	if (ci_attr & CACHE_READ_ALLOCATE)
+		n += snprintf(buf + n, len - n, "ReadAllocate\n");
+	if (ci_attr & CACHE_WRITE_ALLOCATE)
+		n += snprintf(buf + n, len - n, "WriteAllocate\n");
+	buf[n] = '\0';
+	return n;
+}
+
+static umode_t
+cache_default_attrs_is_visible(struct kobject *kobj,
+			       struct attribute *attr, int unused)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct device_attribute *dev_attr;
+	umode_t mode = attr->mode;
+	char *buf;
+
+	dev_attr = container_of(attr, struct device_attribute, attr);
+	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
+	if (!buf)
+		return 0;
+
+	/* create attributes that provides meaningful value */
+	if (dev_attr->show && dev_attr->show(dev, dev_attr, buf) < 0)
+		mode = 0;
+
+	kfree(buf);
+	return mode;
+}
+
+static DEVICE_ATTR_RO(level);
+static DEVICE_ATTR_RO(type);
+static DEVICE_ATTR_RO(coherency_line_size);
+static DEVICE_ATTR_RO(ways_of_associativity);
+static DEVICE_ATTR_RO(number_of_sets);
+static DEVICE_ATTR_RO(size);
+static DEVICE_ATTR_RO(attributes);
+static DEVICE_ATTR_RO(shared_cpu_map);
+static DEVICE_ATTR_RO(shared_cpu_list);
+static DEVICE_ATTR_RO(physical_line_partition);
+
+static struct attribute *cache_default_attrs[] = {
+	&dev_attr_type.attr,
+	&dev_attr_level.attr,
+	&dev_attr_shared_cpu_map.attr,
+	&dev_attr_shared_cpu_list.attr,
+	&dev_attr_coherency_line_size.attr,
+	&dev_attr_ways_of_associativity.attr,
+	&dev_attr_number_of_sets.attr,
+	&dev_attr_size.attr,
+	&dev_attr_attributes.attr,
+	&dev_attr_physical_line_partition.attr,
+	NULL
+};
+
+static const struct attribute_group cache_default_group = {
+	.attrs = cache_default_attrs,
+	.is_visible = cache_default_attrs_is_visible,
+};
+
+static const struct attribute_group *cache_default_groups[] = {
+	&cache_default_group,
+	NULL,
+};
+
+static struct attribute_group cache_private_group = {
+	.is_visible = cache_default_attrs_is_visible,
+};
+
+static const struct attribute_group *cache_private_groups[] = {
+	&cache_default_group,
+	(const struct attribute_group *)&cache_private_group,
+	NULL,
+};
+
+const struct attribute **
+__weak cache_get_priv_attr(struct cacheinfo *this_leaf)
+{
+	return NULL;
+}
+
+static const struct attribute_group **
+cache_get_attribute_groups(struct cacheinfo *this_leaf)
+{
+	const struct attribute **priv_attr = cache_get_priv_attr(this_leaf);
+
+	if (!priv_attr)
+		return cache_default_groups;
+
+	if (!cache_private_group.attrs)
+		cache_private_group.attrs = (struct attribute **)priv_attr;
+
+	return cache_private_groups;
+}
+
+/* Add/Remove cache interface for CPU device */
+static void cpu_cache_sysfs_exit(unsigned int cpu)
+{
+	int i;
+	struct device *ci_dev;
+
+	if (per_cpu_index_dev(cpu)) {
+		for (i = 0; i < cache_leaves(cpu); i++) {
+			ci_dev = per_cache_index_dev(cpu, i);
+			if (!ci_dev)
+				continue;
+			device_unregister(ci_dev);
+		}
+		kfree(per_cpu_index_dev(cpu));
+		per_cpu_index_dev(cpu) = NULL;
+	}
+	device_unregister(per_cpu_cache_dev(cpu));
+	per_cpu_cache_dev(cpu) = NULL;
+}
+
+static int cpu_cache_sysfs_init(unsigned int cpu)
+{
+	struct device *dev = get_cpu_device(cpu);
+
+	if (per_cpu_cacheinfo(cpu) == NULL)
+		return -ENOENT;
+
+	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
+					       NULL, "cache");
+	if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
+		return PTR_ERR(per_cpu_cache_dev(cpu));
+
+	/* Allocate all required memory */
+	per_cpu_index_dev(cpu) = kcalloc(cache_leaves(cpu),
+					 sizeof(struct device *), GFP_KERNEL);
+	if (unlikely(per_cpu_index_dev(cpu) == NULL))
+		goto err_out;
+
+	return 0;
+
+err_out:
+	cpu_cache_sysfs_exit(cpu);
+	return -ENOMEM;
+}
+
+static int cache_add_dev(unsigned int cpu)
+{
+	unsigned short i;
+	int rc;
+	struct device *ci_dev, *parent;
+	struct cacheinfo *this_leaf;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	const struct attribute_group **cache_groups;
+
+	rc = cpu_cache_sysfs_init(cpu);
+	if (unlikely(rc < 0))
+		return rc;
+
+	parent = per_cpu_cache_dev(cpu);
+	for (i = 0; i < cache_leaves(cpu); i++) {
+		this_leaf = this_cpu_ci->info_list + i;
+		if (this_leaf->disable_sysfs)
+			continue;
+		cache_groups = cache_get_attribute_groups(this_leaf);
+		ci_dev = device_create_with_groups(parent->class, parent, i,
+						   this_leaf, cache_groups,
+						   "index%1u", i);
+		if (IS_ERR_OR_NULL(ci_dev)) {
+			rc = PTR_ERR(ci_dev);
+			goto err;
+		}
+		per_cache_index_dev(cpu, i) = ci_dev;
+	}
+	cpumask_set_cpu(cpu, &cache_dev_map);
+
+	return 0;
+err:
+	cpu_cache_sysfs_exit(cpu);
+	return rc;
+}
+
+static void cache_remove_dev(unsigned int cpu)
+{
+	if (!cpumask_test_cpu(cpu, &cache_dev_map))
+		return;
+	cpumask_clear_cpu(cpu, &cache_dev_map);
+
+	cpu_cache_sysfs_exit(cpu);
+}
+
+static int cacheinfo_cpu_callback(struct notifier_block *nfb,
+				  unsigned long action, void *hcpu)
+{
+	unsigned int cpu = (unsigned long)hcpu;
+	int rc = 0;
+
+	switch (action) {
+	case CPU_ONLINE:
+	case CPU_ONLINE_FROZEN:
+		rc = detect_cache_attributes(cpu);
+		if (!rc)
+			rc = cache_add_dev(cpu);
+		break;
+	case CPU_DEAD:
+	case CPU_DEAD_FROZEN:
+		cache_remove_dev(cpu);
+		if (per_cpu_cacheinfo(cpu))
+			free_cache_attributes(cpu);
+		break;
+	}
+	return notifier_from_errno(rc);
+}
+
+static int __init cacheinfo_sysfs_init(void)
+{
+	int cpu, rc = 0;
+
+	cpu_notifier_register_begin();
+
+	for_each_online_cpu(cpu) {
+		rc = detect_cache_attributes(cpu);
+		if (rc) {
+			pr_err("error detecting cacheinfo..cpu%d\n", cpu);
+			goto out;
+		}
+		rc = cache_add_dev(cpu);
+		if (rc) {
+			free_cache_attributes(cpu);
+			pr_err("error populating cacheinfo..cpu%d\n", cpu);
+			goto out;
+		}
+	}
+	__hotcpu_notifier(cacheinfo_cpu_callback, 0);
+
+out:
+	cpu_notifier_register_done();
+	return rc;
+}
+
+device_initcall(cacheinfo_sysfs_init);
+
+#endif	/* CONFIG_SYSFS */
diff --git a/include/linux/cacheinfo.h b/include/linux/cacheinfo.h
new file mode 100644
index 000000000000..3d67b4910aa8
--- /dev/null
+++ b/include/linux/cacheinfo.h
@@ -0,0 +1,73 @@
+#ifndef _LINUX_CACHEINFO_H
+#define _LINUX_CACHEINFO_H
+
+#include <linux/bitops.h>
+#include <linux/compiler.h>
+#include <linux/cpumask.h>
+#include <linux/device.h>
+#include <linux/of.h>
+#include <linux/sysfs.h>
+
+enum cache_type {
+	CACHE_TYPE_NOCACHE = 0,
+	CACHE_TYPE_INST = BIT(0),
+	CACHE_TYPE_DATA = BIT(1),
+	CACHE_TYPE_SEPARATE = CACHE_TYPE_INST | CACHE_TYPE_DATA,
+	CACHE_TYPE_UNIFIED = BIT(2),
+};
+
+struct cacheinfo {
+	/* core properties */
+	enum cache_type type; /* data, inst or unified */
+	unsigned int level;
+	unsigned int coherency_line_size; /* cache line size  */
+	unsigned int number_of_sets; /* no. of sets per way */
+	unsigned int ways_of_associativity; /* no. of ways */
+	unsigned int physical_line_partition; /* no. of lines per tag */
+	unsigned int size; /* total cache size */
+	cpumask_t shared_cpu_map;
+	unsigned int attributes;
+#define CACHE_WRITE_THROUGH	BIT(0)
+#define CACHE_WRITE_BACK	BIT(1)
+#define CACHE_READ_ALLOCATE	BIT(2)
+#define CACHE_WRITE_ALLOCATE	BIT(3)
+
+	/* book keeping */
+	struct device_node *of_node;	/* cpu if no explicit cache node */
+	bool disable_sysfs; /* don't expose this leaf through sysfs */
+	void *priv;
+};
+
+struct cpu_cacheinfo {
+	struct cacheinfo *info_list;
+	unsigned int num_levels;
+	unsigned int num_leaves;
+};
+
+/*
+ * Helpers to make sure "func" is executed on the cpu whose cache
+ * attributes are being detected
+ */
+#define DEFINE_SMP_CALL_FUNCTION(func)				\
+static void _##func(void *ret)					\
+{								\
+	int cpu = smp_processor_id();				\
+	*(int *)ret = __##func(cpu);				\
+}								\
+								\
+int func(unsigned int cpu)					\
+{								\
+	int ret;						\
+	smp_call_function_single(cpu, _##func, &ret, true);	\
+	return ret;						\
+}
+
+struct cpu_cacheinfo *get_cpu_cacheinfo(unsigned int cpu);
+int init_cache_level(unsigned int cpu);
+int populate_cache_leaves(unsigned int cpu);
+
+#ifdef CONFIG_SYSFS
+const struct attribute **cache_get_priv_attr(struct cacheinfo *this_leaf);
+#endif
+
+#endif /* _LINUX_CACHEINFO_H */
-- 
1.8.3.2

Re: [PATCH v2 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Stephen Boyd]

On 07/25/14 09:44, Sudeep Holla wrote:
> diff --git a/Documentation/ABI/testing/sysfs-devices-system-cpu b/Documentation/ABI/testing/sysfs-devices-system-cpu
> index acb9bfc89b48..832b7f2ed6d2 100644
> --- a/Documentation/ABI/testing/sysfs-devices-system-cpu
> +++ b/Documentation/ABI/testing/sysfs-devices-system-cpu
> @@ -224,3 +224,44 @@ Description:	Parameters for the Intel P-state driver
>  		frequency range.
>  
>  		More details can be found in Documentation/cpu-freq/intel-pstate.txt
> +
> +What:		/sys/devices/system/cpu/cpu*/cache/index*/<set_of_attributes_mentioned_below>
> +Date:		July 2014(documented, existed before August 2008)
> +Contact:	Sudeep Holla <sudeep.holla@arm.com>
> +		Linux kernel mailing list <linux-kernel@vger.kernel.org>
> +Description:	Parameters for the CPU cache attributes
> +
> +		attributes:
> +			- writethrough: data is written to both the cache line
> +					and to the block in the lower-level memory
> +			- writeback: data is written only to the cache line and
> +				     the modified cache line is written to main
> +				     memory only when it is replaced
> +			- writeallocate: allocate a memory location to a cache line
> +					 on a cache miss because of a write
> +			- readallocate: allocate a memory location to a cache line
> +					on a cache miss because of a read
> +
> +		coherency_line_size: the minimum amount of data that gets transferred

Is this in bytes?

> +
> +		level: the cache hierarcy in the multi-level cache configuration
> +
> +		number_of_sets: total number of sets in the cache, a set is a
> +				collection of cache lines with the same cache index
> +
> +		physical_line_partition: number of physical cache line per cache tag
> +
> +		shared_cpu_list: the list of cpus sharing the cache

This is a logical list, not physical right?

> +
> +		shared_cpu_map: logical cpu mask containing the list of cpus sharing
> +				the cache
> +
> +		size: the total cache size in kB
> +
> +		type:
> +			- instruction: cache that only holds instructions
> +			- data: cache that only caches data
> +			- unified: cache that holds both data and instructions
> +
> +		ways_of_associativity: degree of freedom in placing a particular block
> +					of memory in the cache
> diff --git a/drivers/base/cacheinfo.c b/drivers/base/cacheinfo.c
> new file mode 100644
> index 000000000000..983728a919ec
> --- /dev/null
> +++ b/drivers/base/cacheinfo.c
> @@ -0,0 +1,539 @@
[...]
> +
> +static int detect_cache_attributes(unsigned int cpu)

Unused if sysfs is disabled? Actually it looks like everything except
the weak functions are unused in such a case.

> +static ssize_t shared_cpumap_show_func(struct device *dev, int type, char *buf)
> +{
> +	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
> +	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf;
> +	int n = 0;
> +
> +	if (len > 1) {
> +		const struct cpumask *mask = &this_leaf->shared_cpu_map;
> +
> +		n = type ? cpulist_scnprintf(buf, len - 2, mask) :
> +			   cpumask_scnprintf(buf, len - 2, mask);
> +		buf[n++] = '\n';
> +		buf[n] = '\0';
> +	}
> +	return n;
> +}

This looks to be lifted from show_cpumap() (well maybe that function was
lifted from x86, not sure). Perhaps it should be extracted out into an
inline function in cpumask.h? Future work I guess.

> +
> +static ssize_t shared_cpu_map_show(struct device *dev,
> +				   struct device_attribute *attr, char *buf)
> +{
> +	return shared_cpumap_show_func(dev, 0, buf);
> +}
> +
> +static ssize_t shared_cpu_list_show(struct device *dev,
> +				    struct device_attribute *attr, char *buf)
> +{
> +	return shared_cpumap_show_func(dev, 1, buf);
> +}
> +
> +static ssize_t type_show(struct device *dev,
> +			 struct device_attribute *attr, char *buf)
> +{
> +	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
> +
> +	switch (this_leaf->type) {
> +	case CACHE_TYPE_DATA:
> +		return sprintf(buf, "Data\n");
> +	case CACHE_TYPE_INST:
> +		return sprintf(buf, "Instruction\n");
> +	case CACHE_TYPE_UNIFIED:
> +		return sprintf(buf, "Unified\n");
> +	default:
> +		return -EINVAL;
> +	}
> +}
> +
> +static ssize_t attributes_show(struct device *dev,
> +			       struct device_attribute *attr, char *buf)
> +{
> +	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
> +	unsigned int ci_attr = this_leaf->attributes;
> +	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf - 2;
> +	int n = 0;
> +
> +	if (!ci_attr)
> +		return -EINVAL;
> +
> +	if (ci_attr & CACHE_WRITE_THROUGH)
> +		n += snprintf(buf + n, len - n, "WriteThrough\n");
> +	if (ci_attr & CACHE_WRITE_BACK)
> +		n += snprintf(buf + n, len - n, "WriteBack\n");
> +	if (ci_attr & CACHE_READ_ALLOCATE)
> +		n += snprintf(buf + n, len - n, "ReadAllocate\n");
> +	if (ci_attr & CACHE_WRITE_ALLOCATE)
> +		n += snprintf(buf + n, len - n, "WriteAllocate\n");

I see that ia64 has this attributes file, but in that case only two
attributes exist (write through and write back) and only one value is
ever shown. When we have multiple attributes we'll have multiple lines
to parse here. What if we left attributes around for the ia64 case
(possibly even hiding that entirely within that architecture specific
code) and then have files like "allocation_policy" and "storage_method"
that correspond to whether its read/write allocation and write through
or write back? The goal being to make only one value exist in any sysfs
attribute.

> +	buf[n] = '\0';
> +	return n;
> +}
> +
> +static umode_t
> +cache_default_attrs_is_visible(struct kobject *kobj,
> +			       struct attribute *attr, int unused)
> +{
> +	struct device *dev = kobj_to_dev(kobj);
> +	struct device_attribute *dev_attr;
> +	umode_t mode = attr->mode;
> +	char *buf;
> +
> +	dev_attr = container_of(attr, struct device_attribute, attr);
> +	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
> +	if (!buf)
> +		return 0;
> +
> +	/* create attributes that provides meaningful value */
> +	if (dev_attr->show && dev_attr->show(dev, dev_attr, buf) < 0)
> +		mode = 0;
> +
> +	kfree(buf);

This is sort of sad. We have to allocate a whole page and call the show
function to figure out if the attribute is visible? Why don't we
actually look at what the attribute is and check for the structure
members we care about? It looks like there are only a few combinations.

> +	return mode;
> +}
> +
> +static DEVICE_ATTR_RO(level);
> +static DEVICE_ATTR_RO(type);
> +static DEVICE_ATTR_RO(coherency_line_size);
> +static DEVICE_ATTR_RO(ways_of_associativity);
> +static DEVICE_ATTR_RO(number_of_sets);
> +static DEVICE_ATTR_RO(size);
> +static DEVICE_ATTR_RO(attributes);
> +static DEVICE_ATTR_RO(shared_cpu_map);
> +static DEVICE_ATTR_RO(shared_cpu_list);
> +static DEVICE_ATTR_RO(physical_line_partition);
> +
> +static struct attribute *cache_default_attrs[] = {
> +	&dev_attr_type.attr,
> +	&dev_attr_level.attr,
> +	&dev_attr_shared_cpu_map.attr,
> +	&dev_attr_shared_cpu_list.attr,
> +	&dev_attr_coherency_line_size.attr,
> +	&dev_attr_ways_of_associativity.attr,
> +	&dev_attr_number_of_sets.attr,
> +	&dev_attr_size.attr,
> +	&dev_attr_attributes.attr,
> +	&dev_attr_physical_line_partition.attr,
> +	NULL
> +};
> +
> +static const struct attribute_group cache_default_group = {
> +	.attrs = cache_default_attrs,
> +	.is_visible = cache_default_attrs_is_visible,
> +};
> +
> +static const struct attribute_group *cache_default_groups[] = {
> +	&cache_default_group,
> +	NULL,
> +};
> +
> +static struct attribute_group cache_private_group = {
> +	.is_visible = cache_default_attrs_is_visible,
> +};
> +
> +static const struct attribute_group *cache_private_groups[] = {
> +	&cache_default_group,
> +	(const struct attribute_group *)&cache_private_group,
> +	NULL,
> +};
> +
> +const struct attribute **
> +__weak cache_get_priv_attr(struct cacheinfo *this_leaf)
> +{
> +	return NULL;
> +}
> +
> +static const struct attribute_group **
> +cache_get_attribute_groups(struct cacheinfo *this_leaf)
> +{
> +	const struct attribute **priv_attr = cache_get_priv_attr(this_leaf);
> +
> +	if (!priv_attr)
> +		return cache_default_groups;
> +
> +	if (!cache_private_group.attrs)
> +		cache_private_group.attrs = (struct attribute **)priv_attr;
> +
> +	return cache_private_groups;
> +}

There's lots of odd casting going on here. Can we change
cache_get_priv_attr() to cache_get_priv_attr_group() and then rework x86
code to return a group instead of a bunch of attribute pointers? Then
drop const from cache_default_groups, add an extra blank NULL entry for
the possible arch specific groups and then use ARRAY_SIZE() - 1 to set
that entry with whatever we get from cache_get_priv_attr_group(). Also
drop the whole cache_private_groups thing.

> +
> +/* Add/Remove cache interface for CPU device */
> +static void cpu_cache_sysfs_exit(unsigned int cpu)
> +{
> +	int i;
> +	struct device *ci_dev;
> +
> +	if (per_cpu_index_dev(cpu)) {
> +		for (i = 0; i < cache_leaves(cpu); i++) {
> +			ci_dev = per_cache_index_dev(cpu, i);
> +			if (!ci_dev)
> +				continue;
> +			device_unregister(ci_dev);
> +		}
> +		kfree(per_cpu_index_dev(cpu));
> +		per_cpu_index_dev(cpu) = NULL;
> +	}
> +	device_unregister(per_cpu_cache_dev(cpu));
> +	per_cpu_cache_dev(cpu) = NULL;
> +}
> +
> +static int cpu_cache_sysfs_init(unsigned int cpu)
> +{
> +	struct device *dev = get_cpu_device(cpu);
> +
> +	if (per_cpu_cacheinfo(cpu) == NULL)
> +		return -ENOENT;
> +
> +	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
> +					       NULL, "cache");
> +	if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
> +		return PTR_ERR(per_cpu_cache_dev(cpu));
> +
> +	/* Allocate all required memory */
> +	per_cpu_index_dev(cpu) = kcalloc(cache_leaves(cpu),
> +					 sizeof(struct device *), GFP_KERNEL);
> +	if (unlikely(per_cpu_index_dev(cpu) == NULL))
> +		goto err_out;
> +
> +	return 0;
> +
> +err_out:
> +	cpu_cache_sysfs_exit(cpu);
> +	return -ENOMEM;
> +}
> +
> +static int cache_add_dev(unsigned int cpu)
> +{
> +	unsigned short i;

unsigned int?

> +	int rc;
> +	struct device *ci_dev, *parent;
> +	struct cacheinfo *this_leaf;
> +	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
> +	const struct attribute_group **cache_groups;
> +
> +	rc = cpu_cache_sysfs_init(cpu);
> +	if (unlikely(rc < 0))
> +		return rc;
> +
> +	parent = per_cpu_cache_dev(cpu);
> +	for (i = 0; i < cache_leaves(cpu); i++) {
> +		this_leaf = this_cpu_ci->info_list + i;
> +		if (this_leaf->disable_sysfs)
> +			continue;
> +		cache_groups = cache_get_attribute_groups(this_leaf);
> +		ci_dev = device_create_with_groups(parent->class, parent, i,
> +						   this_leaf, cache_groups,
> +						   "index%1u", i);
> +		if (IS_ERR_OR_NULL(ci_dev)) {

IS_ERR?

> +			rc = PTR_ERR(ci_dev);
> +			goto err;
> +		}
> +		per_cache_index_dev(cpu, i) = ci_dev;
> +	}
> +	cpumask_set_cpu(cpu, &cache_dev_map);
> +
> +	return 0;
> +err:
> +	cpu_cache_sysfs_exit(cpu);
> +	return rc;
> +}
> +
> +static void cache_remove_dev(unsigned int cpu)
> +{
> +	if (!cpumask_test_cpu(cpu, &cache_dev_map))
> +		return;
> +	cpumask_clear_cpu(cpu, &cache_dev_map);
> +
> +	cpu_cache_sysfs_exit(cpu);
> +}
> +
> +static int cacheinfo_cpu_callback(struct notifier_block *nfb,
> +				  unsigned long action, void *hcpu)
> +{
> +	unsigned int cpu = (unsigned long)hcpu;
> +	int rc = 0;
> +
> +	switch (action) {

Looks like we can do action & ~CPU_TASKS_FROZEN to save two lines here

> +	case CPU_ONLINE:
> +	case CPU_ONLINE_FROZEN:
> +		rc = detect_cache_attributes(cpu);
> +		if (!rc)
> +			rc = cache_add_dev(cpu);
> +		break;
> +	case CPU_DEAD:
> +	case CPU_DEAD_FROZEN:
> +		cache_remove_dev(cpu);
> +		if (per_cpu_cacheinfo(cpu))
> +			free_cache_attributes(cpu);
> +		break;
> +	}
> +	return notifier_from_errno(rc);
> +}

Hm... adding/detecting/destroying this stuff every time a CPU is
logically hotplugged seems like a waste of time and energy. Why can't we
only do this work when the CPU is actually physically removed? The path
for that is via the subsys_interface and it would make it easier on
programs that want to learn about cache info as long as the CPU is
present in the system even if it isn't online at the time of reading.

> +
> +static int __init cacheinfo_sysfs_init(void)
> +{
> +	int cpu, rc = 0;
> +
> +	cpu_notifier_register_begin();
> +
> +	for_each_online_cpu(cpu) {
> +		rc = detect_cache_attributes(cpu);
> +		if (rc) {
> +			pr_err("error detecting cacheinfo..cpu%d\n", cpu);
> +			goto out;
> +		}
> +		rc = cache_add_dev(cpu);
> +		if (rc) {
> +			free_cache_attributes(cpu);
> +			pr_err("error populating cacheinfo..cpu%d\n", cpu);
> +			goto out;
> +		}
> +	}
> +	__hotcpu_notifier(cacheinfo_cpu_callback, 0);
> +
> +out:
> +	cpu_notifier_register_done();
> +	return rc;
> +}
> +
> +device_initcall(cacheinfo_sysfs_init);
> +
> +#endif	/* CONFIG_SYSFS */
> diff --git a/include/linux/cacheinfo.h b/include/linux/cacheinfo.h
> new file mode 100644
> index 000000000000..3d67b4910aa8
> --- /dev/null
> +++ b/include/linux/cacheinfo.h
> @@ -0,0 +1,73 @@
> +#ifndef _LINUX_CACHEINFO_H
> +#define _LINUX_CACHEINFO_H
> +
> +#include <linux/bitops.h>
> +#include <linux/compiler.h>
> +#include <linux/cpumask.h>
> +#include <linux/device.h>
> +#include <linux/of.h>
> +#include <linux/sysfs.h>

These last three look like they could be removed and we could have
forward declarations for required types. What's compiler.h for?

> +
> +enum cache_type {
> +	CACHE_TYPE_NOCACHE = 0,
> +	CACHE_TYPE_INST = BIT(0),
> +	CACHE_TYPE_DATA = BIT(1),
> +	CACHE_TYPE_SEPARATE = CACHE_TYPE_INST | CACHE_TYPE_DATA,
> +	CACHE_TYPE_UNIFIED = BIT(2),
> +};
> +
> +struct cacheinfo {
> +	/* core properties */
> +	enum cache_type type; /* data, inst or unified */
> +	unsigned int level;
> +	unsigned int coherency_line_size; /* cache line size  */
> +	unsigned int number_of_sets; /* no. of sets per way */
> +	unsigned int ways_of_associativity; /* no. of ways */
> +	unsigned int physical_line_partition; /* no. of lines per tag */
> +	unsigned int size; /* total cache size */
> +	cpumask_t shared_cpu_map;
> +	unsigned int attributes;
> +#define CACHE_WRITE_THROUGH	BIT(0)
> +#define CACHE_WRITE_BACK	BIT(1)
> +#define CACHE_READ_ALLOCATE	BIT(2)
> +#define CACHE_WRITE_ALLOCATE	BIT(3)
> +
> +	/* book keeping */
> +	struct device_node *of_node;	/* cpu if no explicit cache node */
> +	bool disable_sysfs; /* don't expose this leaf through sysfs */
> +	void *priv;
> +};

Can we use real kernel doc notation here please?

> +
> +struct cpu_cacheinfo {
> +	struct cacheinfo *info_list;
> +	unsigned int num_levels;
> +	unsigned int num_leaves;
> +};
> +
> +/*
> + * Helpers to make sure "func" is executed on the cpu whose cache
> + * attributes are being detected
> + */
> +#define DEFINE_SMP_CALL_FUNCTION(func)				\
> +static void _##func(void *ret)					\
> +{								\
> +	int cpu = smp_processor_id();				\
> +	*(int *)ret = __##func(cpu);				\
> +}								\
> +								\
> +int func(unsigned int cpu)					\
> +{								\
> +	int ret;						\
> +	smp_call_function_single(cpu, _##func, &ret, true);	\
> +	return ret;						\
> +}
> +

Bikeshed: Maybe this should be called DEFINE_SMP_CALL_CACHE_FUNCTION()?
Missing include for <linux/smp.h> here?

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation

Re: [PATCH v2 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

Hi Stephen,

Thanks for reviewing this.

On 30/07/14 00:09, Stephen Boyd wrote:
> On 07/25/14 09:44, Sudeep Holla wrote:
>> diff --git a/Documentation/ABI/testing/sysfs-devices-system-cpu b/Documentation/ABI/testing/sysfs-devices-system-cpu
>> index acb9bfc89b48..832b7f2ed6d2 100644
>> --- a/Documentation/ABI/testing/sysfs-devices-system-cpu
>> +++ b/Documentation/ABI/testing/sysfs-devices-system-cpu
>> @@ -224,3 +224,44 @@ Description:     Parameters for the Intel P-state driver
>>                frequency range.
>>
>>                More details can be found in Documentation/cpu-freq/intel-pstate.txt
>> +
>> +What:                /sys/devices/system/cpu/cpu*/cache/index*/<set_of_attributes_mentioned_below>
>> +Date:                July 2014(documented, existed before August 2008)
>> +Contact:     Sudeep Holla <sudeep.holla@arm.com>
>> +             Linux kernel mailing list <linux-kernel@vger.kernel.org>
>> +Description: Parameters for the CPU cache attributes
>> +
>> +             attributes:
>> +                     - writethrough: data is written to both the cache line
>> +                                     and to the block in the lower-level memory
>> +                     - writeback: data is written only to the cache line and
>> +                                  the modified cache line is written to main
>> +                                  memory only when it is replaced
>> +                     - writeallocate: allocate a memory location to a cache line
>> +                                      on a cache miss because of a write
>> +                     - readallocate: allocate a memory location to a cache line
>> +                                     on a cache miss because of a read
>> +
>> +             coherency_line_size: the minimum amount of data that gets transferred
>
> Is this in bytes?
>

Right, fixed now.

>> +
>> +             level: the cache hierarcy in the multi-level cache configuration
>> +
>> +             number_of_sets: total number of sets in the cache, a set is a
>> +                             collection of cache lines with the same cache index
>> +
>> +             physical_line_partition: number of physical cache line per cache tag
>> +
>> +             shared_cpu_list: the list of cpus sharing the cache
>
> This is a logical list, not physical right?
>

Right, done.

>> +
>> +             shared_cpu_map: logical cpu mask containing the list of cpus sharing
>> +                             the cache
>> +
>> +             size: the total cache size in kB
>> +
>> +             type:
>> +                     - instruction: cache that only holds instructions
>> +                     - data: cache that only caches data
>> +                     - unified: cache that holds both data and instructions
>> +
>> +             ways_of_associativity: degree of freedom in placing a particular block
>> +                                     of memory in the cache
>> diff --git a/drivers/base/cacheinfo.c b/drivers/base/cacheinfo.c
>> new file mode 100644
>> index 000000000000..983728a919ec
>> --- /dev/null
>> +++ b/drivers/base/cacheinfo.c
>> @@ -0,0 +1,539 @@
> [...]
>> +
>> +static int detect_cache_attributes(unsigned int cpu)
>
> Unused if sysfs is disabled? Actually it looks like everything except
> the weak functions are unused in such a case.
>



>> +static ssize_t shared_cpumap_show_func(struct device *dev, int type, char *buf)
>> +{
>> +     struct cacheinfo *this_leaf = dev_get_drvdata(dev);
>> +     ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf;
>> +     int n = 0;
>> +
>> +     if (len > 1) {
>> +             const struct cpumask *mask = &this_leaf->shared_cpu_map;
>> +
>> +             n = type ? cpulist_scnprintf(buf, len - 2, mask) :
>> +                        cpumask_scnprintf(buf, len - 2, mask);
>> +             buf[n++] = '\n';
>> +             buf[n] = '\0';
>> +     }
>> +     return n;
>> +}
>
> This looks to be lifted from show_cpumap() (well maybe that function was
> lifted from x86, not sure). Perhaps it should be extracted out into an
> inline function in cpumask.h? Future work I guess.
>

Right looks like this is copied from drivers/base/topology.c, will move to
cpumask.h

>> +
>> +static ssize_t shared_cpu_map_show(struct device *dev,
>> +                                struct device_attribute *attr, char *buf)
>> +{
>> +     return shared_cpumap_show_func(dev, 0, buf);
>> +}
>> +
>> +static ssize_t shared_cpu_list_show(struct device *dev,
>> +                                 struct device_attribute *attr, char *buf)
>> +{
>> +     return shared_cpumap_show_func(dev, 1, buf);
>> +}
>> +
>> +static ssize_t type_show(struct device *dev,
>> +                      struct device_attribute *attr, char *buf)
>> +{
>> +     struct cacheinfo *this_leaf = dev_get_drvdata(dev);
>> +
>> +     switch (this_leaf->type) {
>> +     case CACHE_TYPE_DATA:
>> +             return sprintf(buf, "Data\n");
>> +     case CACHE_TYPE_INST:
>> +             return sprintf(buf, "Instruction\n");
>> +     case CACHE_TYPE_UNIFIED:
>> +             return sprintf(buf, "Unified\n");
>> +     default:
>> +             return -EINVAL;
>> +     }
>> +}
>> +
>> +static ssize_t attributes_show(struct device *dev,
>> +                            struct device_attribute *attr, char *buf)
>> +{
>> +     struct cacheinfo *this_leaf = dev_get_drvdata(dev);
>> +     unsigned int ci_attr = this_leaf->attributes;
>> +     ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf - 2;
>> +     int n = 0;
>> +
>> +     if (!ci_attr)
>> +             return -EINVAL;
>> +
>> +     if (ci_attr & CACHE_WRITE_THROUGH)
>> +             n += snprintf(buf + n, len - n, "WriteThrough\n");
>> +     if (ci_attr & CACHE_WRITE_BACK)
>> +             n += snprintf(buf + n, len - n, "WriteBack\n");
>> +     if (ci_attr & CACHE_READ_ALLOCATE)
>> +             n += snprintf(buf + n, len - n, "ReadAllocate\n");
>> +     if (ci_attr & CACHE_WRITE_ALLOCATE)
>> +             n += snprintf(buf + n, len - n, "WriteAllocate\n");
>
> I see that ia64 has this attributes file, but in that case only two
> attributes exist (write through and write back) and only one value is
> ever shown. When we have multiple attributes we'll have multiple lines
> to parse here. What if we left attributes around for the ia64 case
> (possibly even hiding that entirely within that architecture specific
> code) and then have files like "allocation_policy" and "storage_method"
> that correspond to whether its read/write allocation and write through
> or write back? The goal being to make only one value exist in any sysfs
> attribute.
>

I like your idea, but is it hard rule to have only one value in any
sysfs attribute ? Though one concern I have is if different cache designs
make have different features and like to express that, 'attributes' is a
unified place to do that similar to cpu features in /proc/cpuinfo.

Anyways if we decide to split it, how about write_policy instead of
storage_method ?

>> +     buf[n] = '\0';
>> +     return n;
>> +}
>> +
>> +static umode_t
>> +cache_default_attrs_is_visible(struct kobject *kobj,
>> +                            struct attribute *attr, int unused)
>> +{
>> +     struct device *dev = kobj_to_dev(kobj);
>> +     struct device_attribute *dev_attr;
>> +     umode_t mode = attr->mode;
>> +     char *buf;
>> +
>> +     dev_attr = container_of(attr, struct device_attribute, attr);
>> +     buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
>> +     if (!buf)
>> +             return 0;
>> +
>> +     /* create attributes that provides meaningful value */
>> +     if (dev_attr->show && dev_attr->show(dev, dev_attr, buf) < 0)
>> +             mode = 0;
>> +
>> +     kfree(buf);
>
> This is sort of sad. We have to allocate a whole page and call the show
> function to figure out if the attribute is visible? Why don't we
> actually look at what the attribute is and check for the structure
> members we care about? It looks like there are only a few combinations.
>

Yes I thought about that, as even I didn't like that allocation. But if
we want the private attributes also use the same is_visible callback, we
can't check member directly as we don't know the details of the
individual element.

Even if we have compare elements we need to compare the attribute and
then the value for each element in the structure, requiring changes if
elements are added/removed. I am fine either way, just explaining why
it's done so.

>> +     return mode;
>> +}
>> +
>> +static DEVICE_ATTR_RO(level);
>> +static DEVICE_ATTR_RO(type);
>> +static DEVICE_ATTR_RO(coherency_line_size);
>> +static DEVICE_ATTR_RO(ways_of_associativity);
>> +static DEVICE_ATTR_RO(number_of_sets);
>> +static DEVICE_ATTR_RO(size);
>> +static DEVICE_ATTR_RO(attributes);
>> +static DEVICE_ATTR_RO(shared_cpu_map);
>> +static DEVICE_ATTR_RO(shared_cpu_list);
>> +static DEVICE_ATTR_RO(physical_line_partition);
>> +
>> +static struct attribute *cache_default_attrs[] = {
>> +     &dev_attr_type.attr,
>> +     &dev_attr_level.attr,
>> +     &dev_attr_shared_cpu_map.attr,
>> +     &dev_attr_shared_cpu_list.attr,
>> +     &dev_attr_coherency_line_size.attr,
>> +     &dev_attr_ways_of_associativity.attr,
>> +     &dev_attr_number_of_sets.attr,
>> +     &dev_attr_size.attr,
>> +     &dev_attr_attributes.attr,
>> +     &dev_attr_physical_line_partition.attr,
>> +     NULL
>> +};
>> +
>> +static const struct attribute_group cache_default_group = {
>> +     .attrs = cache_default_attrs,
>> +     .is_visible = cache_default_attrs_is_visible,
>> +};
>> +
>> +static const struct attribute_group *cache_default_groups[] = {
>> +     &cache_default_group,
>> +     NULL,
>> +};
>> +
>> +static struct attribute_group cache_private_group = {
>> +     .is_visible = cache_default_attrs_is_visible,
>> +};
>> +
>> +static const struct attribute_group *cache_private_groups[] = {
>> +     &cache_default_group,
>> +     (const struct attribute_group *)&cache_private_group,
>> +     NULL,
>> +};
>> +
>> +const struct attribute **
>> +__weak cache_get_priv_attr(struct cacheinfo *this_leaf)
>> +{
>> +     return NULL;
>> +}
>> +
>> +static const struct attribute_group **
>> +cache_get_attribute_groups(struct cacheinfo *this_leaf)
>> +{
>> +     const struct attribute **priv_attr = cache_get_priv_attr(this_leaf);
>> +
>> +     if (!priv_attr)
>> +             return cache_default_groups;
>> +
>> +     if (!cache_private_group.attrs)
>> +             cache_private_group.attrs = (struct attribute **)priv_attr;
>> +
>> +     return cache_private_groups;
>> +}
>
> There's lots of odd casting going on here. Can we change
> cache_get_priv_attr() to cache_get_priv_attr_group() and then rework x86
> code to return a group instead of a bunch of attribute pointers? Then
> drop const from cache_default_groups, add an extra blank NULL entry for
> the possible arch specific groups and then use ARRAY_SIZE() - 1 to set
> that entry with whatever we get from cache_get_priv_attr_group(). Also
> drop the whole cache_private_groups thing.
>

I initially had exactly same. I later changed to this mainly to avoid:
1. is_visible duplication
2. run-time updates to cache_groups for each node

No strong opinion again, I am fine either way. Let's see what Greg has
to say.

>> +
>> +/* Add/Remove cache interface for CPU device */
>> +static void cpu_cache_sysfs_exit(unsigned int cpu)
>> +{
>> +     int i;
>> +     struct device *ci_dev;
>> +
>> +     if (per_cpu_index_dev(cpu)) {
>> +             for (i = 0; i < cache_leaves(cpu); i++) {
>> +                     ci_dev = per_cache_index_dev(cpu, i);
>> +                     if (!ci_dev)
>> +                             continue;
>> +                     device_unregister(ci_dev);
>> +             }
>> +             kfree(per_cpu_index_dev(cpu));
>> +             per_cpu_index_dev(cpu) = NULL;
>> +     }
>> +     device_unregister(per_cpu_cache_dev(cpu));
>> +     per_cpu_cache_dev(cpu) = NULL;
>> +}
>> +
>> +static int cpu_cache_sysfs_init(unsigned int cpu)
>> +{
>> +     struct device *dev = get_cpu_device(cpu);
>> +
>> +     if (per_cpu_cacheinfo(cpu) == NULL)
>> +             return -ENOENT;
>> +
>> +     per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
>> +                                            NULL, "cache");
>> +     if (IS_ERR_OR_NULL(per_cpu_cache_dev(cpu)))
>> +             return PTR_ERR(per_cpu_cache_dev(cpu));
>> +
>> +     /* Allocate all required memory */
>> +     per_cpu_index_dev(cpu) = kcalloc(cache_leaves(cpu),
>> +                                      sizeof(struct device *), GFP_KERNEL);
>> +     if (unlikely(per_cpu_index_dev(cpu) == NULL))
>> +             goto err_out;
>> +
>> +     return 0;
>> +
>> +err_out:
>> +     cpu_cache_sysfs_exit(cpu);
>> +     return -ENOMEM;
>> +}
>> +
>> +static int cache_add_dev(unsigned int cpu)
>> +{
>> +     unsigned short i;
>
> unsigned int?
>

Right, left over from early developments when num_leaves were short.

>> +     int rc;
>> +     struct device *ci_dev, *parent;
>> +     struct cacheinfo *this_leaf;
>> +     struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
>> +     const struct attribute_group **cache_groups;
>> +
>> +     rc = cpu_cache_sysfs_init(cpu);
>> +     if (unlikely(rc < 0))
>> +             return rc;
>> +
>> +     parent = per_cpu_cache_dev(cpu);
>> +     for (i = 0; i < cache_leaves(cpu); i++) {
>> +             this_leaf = this_cpu_ci->info_list + i;
>> +             if (this_leaf->disable_sysfs)
>> +                     continue;
>> +             cache_groups = cache_get_attribute_groups(this_leaf);
>> +             ci_dev = device_create_with_groups(parent->class, parent, i,
>> +                                                this_leaf, cache_groups,
>> +                                                "index%1u", i);
>> +             if (IS_ERR_OR_NULL(ci_dev)) {
>
> IS_ERR?
>

Right, done.

>> +                     rc = PTR_ERR(ci_dev);
>> +                     goto err;
>> +             }
>> +             per_cache_index_dev(cpu, i) = ci_dev;
>> +     }
>> +     cpumask_set_cpu(cpu, &cache_dev_map);
>> +
>> +     return 0;
>> +err:
>> +     cpu_cache_sysfs_exit(cpu);
>> +     return rc;
>> +}
>> +
>> +static void cache_remove_dev(unsigned int cpu)
>> +{
>> +     if (!cpumask_test_cpu(cpu, &cache_dev_map))
>> +             return;
>> +     cpumask_clear_cpu(cpu, &cache_dev_map);
>> +
>> +     cpu_cache_sysfs_exit(cpu);
>> +}
>> +
>> +static int cacheinfo_cpu_callback(struct notifier_block *nfb,
>> +                               unsigned long action, void *hcpu)
>> +{
>> +     unsigned int cpu = (unsigned long)hcpu;
>> +     int rc = 0;
>> +
>> +     switch (action) {
>
> Looks like we can do action & ~CPU_TASKS_FROZEN to save two lines here
>

Done.

>> +     case CPU_ONLINE:
>> +     case CPU_ONLINE_FROZEN:
>> +             rc = detect_cache_attributes(cpu);
>> +             if (!rc)
>> +                     rc = cache_add_dev(cpu);
>> +             break;
>> +     case CPU_DEAD:
>> +     case CPU_DEAD_FROZEN:
>> +             cache_remove_dev(cpu);
>> +             if (per_cpu_cacheinfo(cpu))
>> +                     free_cache_attributes(cpu);
>> +             break;
>> +     }
>> +     return notifier_from_errno(rc);
>> +}
>
> Hm... adding/detecting/destroying this stuff every time a CPU is
> logically hotplugged seems like a waste of time and energy. Why can't we
> only do this work when the CPU is actually physically removed? The path
> for that is via the subsys_interface and it would make it easier on
> programs that want to learn about cache info as long as the CPU is
> present in the system even if it isn't online at the time of reading.
>

I agree, but the main reason I retained it as most of the existing
architectures implement this way and I didn't want tho change that
behaviour.

>> +
>> +static int __init cacheinfo_sysfs_init(void)
>> +{
>> +     int cpu, rc = 0;
>> +
>> +     cpu_notifier_register_begin();
>> +
>> +     for_each_online_cpu(cpu) {
>> +             rc = detect_cache_attributes(cpu);
>> +             if (rc) {
>> +                     pr_err("error detecting cacheinfo..cpu%d\n", cpu);
>> +                     goto out;
>> +             }
>> +             rc = cache_add_dev(cpu);
>> +             if (rc) {
>> +                     free_cache_attributes(cpu);
>> +                     pr_err("error populating cacheinfo..cpu%d\n", cpu);
>> +                     goto out;
>> +             }
>> +     }
>> +     __hotcpu_notifier(cacheinfo_cpu_callback, 0);
>> +
>> +out:
>> +     cpu_notifier_register_done();
>> +     return rc;
>> +}
>> +
>> +device_initcall(cacheinfo_sysfs_init);
>> +
>> +#endif       /* CONFIG_SYSFS */
>> diff --git a/include/linux/cacheinfo.h b/include/linux/cacheinfo.h
>> new file mode 100644
>> index 000000000000..3d67b4910aa8
>> --- /dev/null
>> +++ b/include/linux/cacheinfo.h
>> @@ -0,0 +1,73 @@
>> +#ifndef _LINUX_CACHEINFO_H
>> +#define _LINUX_CACHEINFO_H
>> +
>> +#include <linux/bitops.h>
>> +#include <linux/compiler.h>
>> +#include <linux/cpumask.h>
>> +#include <linux/device.h>
>> +#include <linux/of.h>
>> +#include <linux/sysfs.h>
>
> These last three look like they could be removed and we could have
> forward declarations for required types. What's compiler.h for?
>

IIRC I added it for inline attribute, but looks like I dropped inline :(

>> +
>> +enum cache_type {
>> +     CACHE_TYPE_NOCACHE = 0,
>> +     CACHE_TYPE_INST = BIT(0),
>> +     CACHE_TYPE_DATA = BIT(1),
>> +     CACHE_TYPE_SEPARATE = CACHE_TYPE_INST | CACHE_TYPE_DATA,
>> +     CACHE_TYPE_UNIFIED = BIT(2),
>> +};
>> +
>> +struct cacheinfo {
>> +     /* core properties */
>> +     enum cache_type type; /* data, inst or unified */
>> +     unsigned int level;
>> +     unsigned int coherency_line_size; /* cache line size  */
>> +     unsigned int number_of_sets; /* no. of sets per way */
>> +     unsigned int ways_of_associativity; /* no. of ways */
>> +     unsigned int physical_line_partition; /* no. of lines per tag */
>> +     unsigned int size; /* total cache size */
>> +     cpumask_t shared_cpu_map;
>> +     unsigned int attributes;
>> +#define CACHE_WRITE_THROUGH  BIT(0)
>> +#define CACHE_WRITE_BACK     BIT(1)
>> +#define CACHE_READ_ALLOCATE  BIT(2)
>> +#define CACHE_WRITE_ALLOCATE BIT(3)
>> +
>> +     /* book keeping */
>> +     struct device_node *of_node;    /* cpu if no explicit cache node */
>> +     bool disable_sysfs; /* don't expose this leaf through sysfs */
>> +     void *priv;
>> +};
>
> Can we use real kernel doc notation here please?
>

OK

>> +
>> +struct cpu_cacheinfo {
>> +     struct cacheinfo *info_list;
>> +     unsigned int num_levels;
>> +     unsigned int num_leaves;
>> +};
>> +
>> +/*
>> + * Helpers to make sure "func" is executed on the cpu whose cache
>> + * attributes are being detected
>> + */
>> +#define DEFINE_SMP_CALL_FUNCTION(func)                               \
>> +static void _##func(void *ret)                                       \
>> +{                                                            \
>> +     int cpu = smp_processor_id();                           \
>> +     *(int *)ret = __##func(cpu);                            \
>> +}                                                            \
>> +                                                             \
>> +int func(unsigned int cpu)                                   \
>> +{                                                            \
>> +     int ret;                                                \
>> +     smp_call_function_single(cpu, _##func, &ret, true);     \
>> +     return ret;                                             \
>> +}
>> +
>
> Bikeshed: Maybe this should be called DEFINE_SMP_CALL_CACHE_FUNCTION()?
> Missing include for <linux/smp.h> here?

Makes sense, will change.

>
> --
> Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
> hosted by The Linux Foundation
>
>

Re: [PATCH v2 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Stephen Boyd]

On 07/30/14 09:23, Sudeep Holla wrote:
> Hi Stephen,
>
> Thanks for reviewing this.
>
> On 30/07/14 00:09, Stephen Boyd wrote:
>> On 07/25/14 09:44, Sudeep Holla wrote:
>
>>> +
>>> +             shared_cpu_map: logical cpu mask containing the list
>>> of cpus sharing
>>> +                             the cache
>>> +
>>> +             size: the total cache size in kB
>>> +
>>> +             type:
>>> +                     - instruction: cache that only holds instructions
>>> +                     - data: cache that only caches data
>>> +                     - unified: cache that holds both data and
>>> instructions
>>> +
>>> +             ways_of_associativity: degree of freedom in placing a
>>> particular block
>>> +                                     of memory in the cache
>>> diff --git a/drivers/base/cacheinfo.c b/drivers/base/cacheinfo.c
>>> new file mode 100644
>>> index 000000000000..983728a919ec
>>> --- /dev/null
>>> +++ b/drivers/base/cacheinfo.c
>>> @@ -0,0 +1,539 @@
>> [...]
>>> +
>>> +static int detect_cache_attributes(unsigned int cpu)
>>
>> Unused if sysfs is disabled? Actually it looks like everything except
>> the weak functions are unused in such a case.
>>
>
>> I see that ia64 has this attributes file, but in that case only two
>> attributes exist (write through and write back) and only one value is
>> ever shown. When we have multiple attributes we'll have multiple lines
>> to parse here. What if we left attributes around for the ia64 case
>> (possibly even hiding that entirely within that architecture specific
>> code) and then have files like "allocation_policy" and "storage_method"
>> that correspond to whether its read/write allocation and write through
>> or write back? The goal being to make only one value exist in any sysfs
>> attribute.
>>
>
> I like your idea, but is it hard rule to have only one value in any
> sysfs attribute ? Though one concern I have is if different cache designs
> make have different features and like to express that, 'attributes' is a
> unified place to do that similar to cpu features in /proc/cpuinfo.

'attributes' seems too generic. Pretty much anything is an attribute.

>
> Anyways if we decide to split it, how about write_policy instead of
> storage_method ?

Sounds good.

>
>>> +     buf[n] = '\0';
>>> +     return n;
>>> +}
>>> +
>>> +static umode_t
>>> +cache_default_attrs_is_visible(struct kobject *kobj,
>>> +                            struct attribute *attr, int unused)
>>> +{
>>> +     struct device *dev = kobj_to_dev(kobj);
>>> +     struct device_attribute *dev_attr;
>>> +     umode_t mode = attr->mode;
>>> +     char *buf;
>>> +
>>> +     dev_attr = container_of(attr, struct device_attribute, attr);
>>> +     buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
>>> +     if (!buf)
>>> +             return 0;
>>> +
>>> +     /* create attributes that provides meaningful value */
>>> +     if (dev_attr->show && dev_attr->show(dev, dev_attr, buf) < 0)
>>> +             mode = 0;
>>> +
>>> +     kfree(buf);
>>
>> This is sort of sad. We have to allocate a whole page and call the show
>> function to figure out if the attribute is visible? Why don't we
>> actually look at what the attribute is and check for the structure
>> members we care about? It looks like there are only a few combinations.
>>
>
> Yes I thought about that, as even I didn't like that allocation. But if
> we want the private attributes also use the same is_visible callback, we
> can't check member directly as we don't know the details of the
> individual element.
>
> Even if we have compare elements we need to compare the attribute and
> then the value for each element in the structure, requiring changes if
> elements are added/removed. I am fine either way, just explaining why
> it's done so.

Does any other sysfs attribute group do this? If it was desired I would
think someone else would have done this already, or we wouldn't have
even had an is_visible in the first place as this generic code would
replace it.

>
>
>>> +     case CPU_ONLINE:
>>> +     case CPU_ONLINE_FROZEN:
>>> +             rc = detect_cache_attributes(cpu);
>>> +             if (!rc)
>>> +                     rc = cache_add_dev(cpu);
>>> +             break;
>>> +     case CPU_DEAD:
>>> +     case CPU_DEAD_FROZEN:
>>> +             cache_remove_dev(cpu);
>>> +             if (per_cpu_cacheinfo(cpu))
>>> +                     free_cache_attributes(cpu);
>>> +             break;
>>> +     }
>>> +     return notifier_from_errno(rc);
>>> +}
>>
>> Hm... adding/detecting/destroying this stuff every time a CPU is
>> logically hotplugged seems like a waste of time and energy. Why can't we
>> only do this work when the CPU is actually physically removed? The path
>> for that is via the subsys_interface and it would make it easier on
>> programs that want to learn about cache info as long as the CPU is
>> present in the system even if it isn't online at the time of reading.
>>
>
> I agree, but the main reason I retained it as most of the existing
> architectures implement this way and I didn't want tho change that
> behaviour.

Would anything bad happen if we loosened the behavior so that the
directory is always present as long as the CPU is present? I doubt it.
Seems like a low risk change.

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation

Re: [PATCH v2 2/9] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

Hi Stephen,

On 31/07/14 20:46, Stephen Boyd wrote:
> On 07/30/14 09:23, Sudeep Holla wrote:
>> Hi Stephen,
>>
>> Thanks for reviewing this.
>>
>> On 30/07/14 00:09, Stephen Boyd wrote:
>>> On 07/25/14 09:44, Sudeep Holla wrote:
>>
>>>> +
>>>> +             shared_cpu_map: logical cpu mask containing the list
>>>> of cpus sharing
>>>> +                             the cache
>>>> +
>>>> +             size: the total cache size in kB
>>>> +
>>>> +             type:
>>>> +                     - instruction: cache that only holds instructions
>>>> +                     - data: cache that only caches data
>>>> +                     - unified: cache that holds both data and
>>>> instructions
>>>> +
>>>> +             ways_of_associativity: degree of freedom in placing a
>>>> particular block
>>>> +                                     of memory in the cache
>>>> diff --git a/drivers/base/cacheinfo.c b/drivers/base/cacheinfo.c
>>>> new file mode 100644
>>>> index 000000000000..983728a919ec
>>>> --- /dev/null
>>>> +++ b/drivers/base/cacheinfo.c
>>>> @@ -0,0 +1,539 @@
>>> [...]
>>>> +
>>>> +static int detect_cache_attributes(unsigned int cpu)
>>>
>>> Unused if sysfs is disabled? Actually it looks like everything except
>>> the weak functions are unused in such a case.
>>>

I see that sysfs has dummy implementations, probably I can remove #ifdef

>>
>>> I see that ia64 has this attributes file, but in that case only two
>>> attributes exist (write through and write back) and only one value is
>>> ever shown. When we have multiple attributes we'll have multiple lines
>>> to parse here. What if we left attributes around for the ia64 case
>>> (possibly even hiding that entirely within that architecture specific
>>> code) and then have files like "allocation_policy" and "storage_method"
>>> that correspond to whether its read/write allocation and write through
>>> or write back? The goal being to make only one value exist in any sysfs
>>> attribute.
>>>
>>
>> I like your idea, but is it hard rule to have only one value in any
>> sysfs attribute ? Though one concern I have is if different cache designs
>> make have different features and like to express that, 'attributes' is a
>> unified place to do that similar to cpu features in /proc/cpuinfo.
>
> 'attributes' seems too generic. Pretty much anything is an attribute.
>

Yes I agree and hence I compared it to /proc/cpuinfo.
As I said I am fine with new single value sysfs, but my main concern is
the extendability. If we don't for-see any changes in near future, then
we can go with new files as you suggested.

>>
>> Anyways if we decide to split it, how about write_policy instead of
>> storage_method ?
>
> Sounds good.
>

Thanks.

>>
>>>> +     buf[n] = '\0';
>>>> +     return n;
>>>> +}
>>>> +
>>>> +static umode_t
>>>> +cache_default_attrs_is_visible(struct kobject *kobj,
>>>> +                            struct attribute *attr, int unused)
>>>> +{
>>>> +     struct device *dev = kobj_to_dev(kobj);
>>>> +     struct device_attribute *dev_attr;
>>>> +     umode_t mode = attr->mode;
>>>> +     char *buf;
>>>> +
>>>> +     dev_attr = container_of(attr, struct device_attribute, attr);
>>>> +     buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
>>>> +     if (!buf)
>>>> +             return 0;
>>>> +
>>>> +     /* create attributes that provides meaningful value */
>>>> +     if (dev_attr->show && dev_attr->show(dev, dev_attr, buf) < 0)
>>>> +             mode = 0;
>>>> +
>>>> +     kfree(buf);
>>>
>>> This is sort of sad. We have to allocate a whole page and call the show
>>> function to figure out if the attribute is visible? Why don't we
>>> actually look at what the attribute is and check for the structure
>>> members we care about? It looks like there are only a few combinations.
>>>
>>
>> Yes I thought about that, as even I didn't like that allocation. But if
>> we want the private attributes also use the same is_visible callback, we
>> can't check member directly as we don't know the details of the
>> individual element.
>>
>> Even if we have compare elements we need to compare the attribute and
>> then the value for each element in the structure, requiring changes if
>> elements are added/removed. I am fine either way, just explaining why
>> it's done so.
>
> Does any other sysfs attribute group do this? If it was desired I would
> think someone else would have done this already, or we wouldn't have
> even had an is_visible in the first place as this generic code would
> replace it.
>

I saw this first in PPC cacheinfo. Not sure who else have done that.

>>
>>
>>>> +     case CPU_ONLINE:
>>>> +     case CPU_ONLINE_FROZEN:
>>>> +             rc = detect_cache_attributes(cpu);
>>>> +             if (!rc)
>>>> +                     rc = cache_add_dev(cpu);
>>>> +             break;
>>>> +     case CPU_DEAD:
>>>> +     case CPU_DEAD_FROZEN:
>>>> +             cache_remove_dev(cpu);
>>>> +             if (per_cpu_cacheinfo(cpu))
>>>> +                     free_cache_attributes(cpu);
>>>> +             break;
>>>> +     }
>>>> +     return notifier_from_errno(rc);
>>>> +}
>>>
>>> Hm... adding/detecting/destroying this stuff every time a CPU is
>>> logically hotplugged seems like a waste of time and energy. Why can't we
>>> only do this work when the CPU is actually physically removed? The path
>>> for that is via the subsys_interface and it would make it easier on
>>> programs that want to learn about cache info as long as the CPU is
>>> present in the system even if it isn't online at the time of reading.
>>>
>>
>> I agree, but the main reason I retained it as most of the existing
>> architectures implement this way and I didn't want tho change that
>> behaviour.
>
> Would anything bad happen if we loosened the behavior so that the
> directory is always present as long as the CPU is present? I doubt it.
> Seems like a low risk change.
>

Yes, but before I change, I would like to see people are fine with that.
I don't want to move existing implementations into this generic one and
cause breakage.

Regards,
Sudeep

[PATCH v2 3/9] ia64: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: linux-ia64@vger.kernel.org
---
 arch/ia64/kernel/topology.c | 401 +++++++++-----------------------------------
 1 file changed, 79 insertions(+), 322 deletions(-)

diff --git a/arch/ia64/kernel/topology.c b/arch/ia64/kernel/topology.c
index f295f9abba4b..48dc46daf22e 100644
--- a/arch/ia64/kernel/topology.c
+++ b/arch/ia64/kernel/topology.c
@@ -13,6 +13,7 @@
  *	Populate cpu cache entries in sysfs for cpu cache info
  */
 
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
 #include <linux/kernel.h>
 #include <linux/mm.h>
@@ -21,7 +22,6 @@
 #include <linux/init.h>
 #include <linux/bootmem.h>
 #include <linux/nodemask.h>
-#include <linux/notifier.h>
 #include <linux/export.h>
 #include <asm/mmzone.h>
 #include <asm/numa.h>
@@ -103,60 +103,25 @@ subsys_initcall(topology_init);
 /*
  * Export cpu cache information through sysfs
  */
-
-/*
- *  A bunch of string array to get pretty printing
- */
-static const char *cache_types[] = {
-	"",			/* not used */
-	"Instruction",
-	"Data",
-	"Unified"	/* unified */
-};
-
-static const char *cache_mattrib[]={
-	"WriteThrough",
-	"WriteBack",
-	"",		/* reserved */
-	""		/* reserved */
-};
-
-struct cache_info {
-	pal_cache_config_info_t	cci;
-	cpumask_t shared_cpu_map;
-	int level;
-	int type;
-	struct kobject kobj;
-};
-
-struct cpu_cache_info {
-	struct cache_info *cache_leaves;
-	int	num_cache_leaves;
-	struct kobject kobj;
-};
-
-static struct cpu_cache_info	all_cpu_cache_info[NR_CPUS];
-#define LEAF_KOBJECT_PTR(x,y)    (&all_cpu_cache_info[x].cache_leaves[y])
-
 #ifdef CONFIG_SMP
-static void cache_shared_cpu_map_setup(unsigned int cpu,
-		struct cache_info * this_leaf)
+static int __cache_cpumap_setup(unsigned int cpu, struct cacheinfo *this_leaf)
 {
 	pal_cache_shared_info_t	csi;
-	int num_shared, i = 0;
-	unsigned int j;
+	int num_shared, i = 0, j;
+	enum cache_type type = this_leaf->type;
 
 	if (cpu_data(cpu)->threads_per_core <= 1 &&
 		cpu_data(cpu)->cores_per_socket <= 1) {
 		cpu_set(cpu, this_leaf->shared_cpu_map);
-		return;
+		return 0;
 	}
 
-	if (ia64_pal_cache_shared_info(this_leaf->level,
-					this_leaf->type,
-					0,
-					&csi) != PAL_STATUS_SUCCESS)
-		return;
+	if (type == CACHE_TYPE_UNIFIED)
+		type = CACHE_TYPE_DATA;
+
+	if (ia64_pal_cache_shared_info(this_leaf->level, type, 0,
+				       &csi) != PAL_STATUS_SUCCESS)
+		return -EIO;
 
 	num_shared = (int) csi.num_shared;
 	do {
@@ -168,305 +133,97 @@ static void cache_shared_cpu_map_setup(unsigned int cpu,
 
 		i++;
 	} while (i < num_shared &&
-		ia64_pal_cache_shared_info(this_leaf->level,
-				this_leaf->type,
-				i,
-				&csi) == PAL_STATUS_SUCCESS);
-}
-#else
-static void cache_shared_cpu_map_setup(unsigned int cpu,
-		struct cache_info * this_leaf)
-{
-	cpu_set(cpu, this_leaf->shared_cpu_map);
-	return;
-}
-#endif
-
-static ssize_t show_coherency_line_size(struct cache_info *this_leaf,
-					char *buf)
-{
-	return sprintf(buf, "%u\n", 1 << this_leaf->cci.pcci_line_size);
-}
-
-static ssize_t show_ways_of_associativity(struct cache_info *this_leaf,
-					char *buf)
-{
-	return sprintf(buf, "%u\n", this_leaf->cci.pcci_assoc);
-}
-
-static ssize_t show_attributes(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf,
-			"%s\n",
-			cache_mattrib[this_leaf->cci.pcci_cache_attr]);
-}
-
-static ssize_t show_size(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf, "%uK\n", this_leaf->cci.pcci_cache_size / 1024);
-}
-
-static ssize_t show_number_of_sets(struct cache_info *this_leaf, char *buf)
-{
-	unsigned number_of_sets = this_leaf->cci.pcci_cache_size;
-	number_of_sets /= this_leaf->cci.pcci_assoc;
-	number_of_sets /= 1 << this_leaf->cci.pcci_line_size;
-
-	return sprintf(buf, "%u\n", number_of_sets);
-}
-
-static ssize_t show_shared_cpu_map(struct cache_info *this_leaf, char *buf)
-{
-	ssize_t	len;
-	cpumask_t shared_cpu_map;
-
-	cpumask_and(&shared_cpu_map,
-				&this_leaf->shared_cpu_map, cpu_online_mask);
-	len = cpumask_scnprintf(buf, NR_CPUS+1, &shared_cpu_map);
-	len += sprintf(buf+len, "\n");
-	return len;
-}
-
-static ssize_t show_type(struct cache_info *this_leaf, char *buf)
-{
-	int type = this_leaf->type + this_leaf->cci.pcci_unified;
-	return sprintf(buf, "%s\n", cache_types[type]);
-}
-
-static ssize_t show_level(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf, "%u\n", this_leaf->level);
+		 ia64_pal_cache_shared_info(this_leaf->level, type, i,
+					    &csi) == PAL_STATUS_SUCCESS);
+	return 0;
 }
 
-struct cache_attr {
-	struct attribute attr;
-	ssize_t (*show)(struct cache_info *, char *);
-	ssize_t (*store)(struct cache_info *, const char *, size_t count);
-};
-
-#ifdef define_one_ro
-	#undef define_one_ro
-#endif
-#define define_one_ro(_name) \
-	static struct cache_attr _name = \
-__ATTR(_name, 0444, show_##_name, NULL)
-
-define_one_ro(level);
-define_one_ro(type);
-define_one_ro(coherency_line_size);
-define_one_ro(ways_of_associativity);
-define_one_ro(size);
-define_one_ro(number_of_sets);
-define_one_ro(shared_cpu_map);
-define_one_ro(attributes);
-
-static struct attribute * cache_default_attrs[] = {
-	&type.attr,
-	&level.attr,
-	&coherency_line_size.attr,
-	&ways_of_associativity.attr,
-	&attributes.attr,
-	&size.attr,
-	&number_of_sets.attr,
-	&shared_cpu_map.attr,
-	NULL
-};
-
-#define to_object(k) container_of(k, struct cache_info, kobj)
-#define to_attr(a) container_of(a, struct cache_attr, attr)
-
-static ssize_t ia64_cache_show(struct kobject * kobj, struct attribute * attr, char * buf)
+static int cache_shared_cpu_map_setup(unsigned int cpu)
 {
-	struct cache_attr *fattr = to_attr(attr);
-	struct cache_info *this_leaf = to_object(kobj);
-	ssize_t ret;
+	unsigned int idx;
+	int ret = 0;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
 
-	ret = fattr->show ? fattr->show(this_leaf, buf) : 0;
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++, this_leaf++) {
+		ret = __cache_cpumap_setup(cpu, this_leaf);
+		if (ret)
+			break;
+	}
 	return ret;
 }
-
-static const struct sysfs_ops cache_sysfs_ops = {
-	.show   = ia64_cache_show
-};
-
-static struct kobj_type cache_ktype = {
-	.sysfs_ops	= &cache_sysfs_ops,
-	.default_attrs	= cache_default_attrs,
-};
-
-static struct kobj_type cache_ktype_percpu_entry = {
-	.sysfs_ops	= &cache_sysfs_ops,
-};
-
-static void cpu_cache_sysfs_exit(unsigned int cpu)
-{
-	kfree(all_cpu_cache_info[cpu].cache_leaves);
-	all_cpu_cache_info[cpu].cache_leaves = NULL;
-	all_cpu_cache_info[cpu].num_cache_leaves = 0;
-	memset(&all_cpu_cache_info[cpu].kobj, 0, sizeof(struct kobject));
-	return;
-}
-
-static int cpu_cache_sysfs_init(unsigned int cpu)
+#else
+static int cache_shared_cpu_map_setup(unsigned int cpu)
 {
-	unsigned long i, levels, unique_caches;
-	pal_cache_config_info_t cci;
-	int j;
-	long status;
-	struct cache_info *this_cache;
-	int num_cache_leaves = 0;
-
-	if ((status = ia64_pal_cache_summary(&levels, &unique_caches)) != 0) {
-		printk(KERN_ERR "ia64_pal_cache_summary=%ld\n", status);
-		return -1;
-	}
-
-	this_cache=kzalloc(sizeof(struct cache_info)*unique_caches,
-			GFP_KERNEL);
-	if (this_cache == NULL)
-		return -ENOMEM;
-
-	for (i=0; i < levels; i++) {
-		for (j=2; j >0 ; j--) {
-			if ((status=ia64_pal_cache_config_info(i,j, &cci)) !=
-					PAL_STATUS_SUCCESS)
-				continue;
-
-			this_cache[num_cache_leaves].cci = cci;
-			this_cache[num_cache_leaves].level = i + 1;
-			this_cache[num_cache_leaves].type = j;
-
-			cache_shared_cpu_map_setup(cpu,
-					&this_cache[num_cache_leaves]);
-			num_cache_leaves ++;
-		}
-	}
-
-	all_cpu_cache_info[cpu].cache_leaves = this_cache;
-	all_cpu_cache_info[cpu].num_cache_leaves = num_cache_leaves;
-
-	memset(&all_cpu_cache_info[cpu].kobj, 0, sizeof(struct kobject));
+	int idx;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
 
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++, this_leaf++)
+		cpu_set(cpu, this_leaf->shared_cpu_map);
 	return 0;
 }
+#endif
 
-/* Add cache interface for CPU device */
-static int cache_add_dev(struct device *sys_dev)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 pal_cache_config_info_t *cci,
+			 enum cache_type type, unsigned int level)
 {
-	unsigned int cpu = sys_dev->id;
-	unsigned long i, j;
-	struct cache_info *this_object;
-	int retval = 0;
-	cpumask_t oldmask;
-
-	if (all_cpu_cache_info[cpu].kobj.parent)
-		return 0;
-
-	oldmask = current->cpus_allowed;
-	retval = set_cpus_allowed_ptr(current, cpumask_of(cpu));
-	if (unlikely(retval))
-		return retval;
+	unsigned number_of_sets = cci->pcci_cache_size;
 
-	retval = cpu_cache_sysfs_init(cpu);
-	set_cpus_allowed_ptr(current, &oldmask);
-	if (unlikely(retval < 0))
-		return retval;
-
-	retval = kobject_init_and_add(&all_cpu_cache_info[cpu].kobj,
-				      &cache_ktype_percpu_entry, &sys_dev->kobj,
-				      "%s", "cache");
-	if (unlikely(retval < 0)) {
-		cpu_cache_sysfs_exit(cpu);
-		return retval;
-	}
-
-	for (i = 0; i < all_cpu_cache_info[cpu].num_cache_leaves; i++) {
-		this_object = LEAF_KOBJECT_PTR(cpu,i);
-		retval = kobject_init_and_add(&(this_object->kobj),
-					      &cache_ktype,
-					      &all_cpu_cache_info[cpu].kobj,
-					      "index%1lu", i);
-		if (unlikely(retval)) {
-			for (j = 0; j < i; j++) {
-				kobject_put(&(LEAF_KOBJECT_PTR(cpu,j)->kobj));
-			}
-			kobject_put(&all_cpu_cache_info[cpu].kobj);
-			cpu_cache_sysfs_exit(cpu);
-			return retval;
-		}
-		kobject_uevent(&(this_object->kobj), KOBJ_ADD);
-	}
-	kobject_uevent(&all_cpu_cache_info[cpu].kobj, KOBJ_ADD);
-	return retval;
+	number_of_sets /= cci->pcci_assoc;
+	number_of_sets /= 1 << cci->pcci_line_size;
+	this_leaf->level = level;
+	this_leaf->type = cci->pcci_unified ? CACHE_TYPE_UNIFIED : type;
+	this_leaf->coherency_line_size = cci->pcci_line_size;
+	this_leaf->ways_of_associativity = cci->pcci_assoc;
+	this_leaf->size = cci->pcci_cache_size;
+	this_leaf->attributes = cci->pcci_cache_attr;
+	this_leaf->number_of_sets = number_of_sets;
 }
 
-/* Remove cache interface for CPU device */
-static int cache_remove_dev(struct device *sys_dev)
+int init_cache_level(unsigned int cpu)
 {
-	unsigned int cpu = sys_dev->id;
-	unsigned long i;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned long levels, unique_caches;
+	long status;
 
-	for (i = 0; i < all_cpu_cache_info[cpu].num_cache_leaves; i++)
-		kobject_put(&(LEAF_KOBJECT_PTR(cpu,i)->kobj));
+	if (!this_cpu_ci)
+		return -EINVAL;
 
-	if (all_cpu_cache_info[cpu].kobj.parent) {
-		kobject_put(&all_cpu_cache_info[cpu].kobj);
-		memset(&all_cpu_cache_info[cpu].kobj,
-			0,
-			sizeof(struct kobject));
+	status = ia64_pal_cache_summary(&levels, &unique_caches);
+	if (status != PAL_STATUS_SUCCESS) {
+		pr_err("ia64_pal_cache_summary = %ld\n", status);
+		return -EIO;
 	}
-
-	cpu_cache_sysfs_exit(cpu);
+	this_cpu_ci->num_levels = levels;
+	this_cpu_ci->num_leaves = unique_caches;
 
 	return 0;
 }
 
-/*
- * When a cpu is hot-plugged, do a check and initiate
- * cache kobject if necessary
- */
-static int cache_cpu_callback(struct notifier_block *nfb,
-		unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (unsigned long)hcpu;
-	struct device *sys_dev;
-
-	sys_dev = get_cpu_device(cpu);
-	switch (action) {
-	case CPU_ONLINE:
-	case CPU_ONLINE_FROZEN:
-		cache_add_dev(sys_dev);
-		break;
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		cache_remove_dev(sys_dev);
-		break;
-	}
-	return NOTIFY_OK;
-}
-
-static struct notifier_block cache_cpu_notifier =
-{
-	.notifier_call = cache_cpu_callback
-};
-
-static int __init cache_sysfs_init(void)
+int populate_cache_leaves(unsigned int cpu)
 {
-	int i;
-
-	cpu_notifier_register_begin();
-
-	for_each_online_cpu(i) {
-		struct device *sys_dev = get_cpu_device((unsigned int)i);
-		cache_add_dev(sys_dev);
+	unsigned int level, idx;
+	s64 status;
+	pal_cache_config_info_t cci;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		type = CACHE_TYPE_INST;
+		status = ia64_pal_cache_config_info(level - 1, type, &cci);
+		if (status == PAL_STATUS_SUCCESS)
+			ci_leaf_init(this_leaf++, &cci, type, level);
+		type = CACHE_TYPE_DATA;
+		status = ia64_pal_cache_config_info(level - 1, type, &cci);
+		if (status == PAL_STATUS_SUCCESS)
+			ci_leaf_init(this_leaf++, &cci, type, level);
 	}
-
-	__register_hotcpu_notifier(&cache_cpu_notifier);
-
-	cpu_notifier_register_done();
-
-	return 0;
+	return cache_shared_cpu_map_setup(cpu);
 }
-
-device_initcall(cache_sysfs_init);
-
-- 
1.8.3.2

[PATCH v2 3/9] ia64: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: linux-ia64@vger.kernel.org
---
 arch/ia64/kernel/topology.c | 401 +++++++++-----------------------------------
 1 file changed, 79 insertions(+), 322 deletions(-)

diff --git a/arch/ia64/kernel/topology.c b/arch/ia64/kernel/topology.c
index f295f9abba4b..48dc46daf22e 100644
--- a/arch/ia64/kernel/topology.c
+++ b/arch/ia64/kernel/topology.c
@@ -13,6 +13,7 @@
  *	Populate cpu cache entries in sysfs for cpu cache info
  */
 
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
 #include <linux/kernel.h>
 #include <linux/mm.h>
@@ -21,7 +22,6 @@
 #include <linux/init.h>
 #include <linux/bootmem.h>
 #include <linux/nodemask.h>
-#include <linux/notifier.h>
 #include <linux/export.h>
 #include <asm/mmzone.h>
 #include <asm/numa.h>
@@ -103,60 +103,25 @@ subsys_initcall(topology_init);
 /*
  * Export cpu cache information through sysfs
  */
-
-/*
- *  A bunch of string array to get pretty printing
- */
-static const char *cache_types[] = {
-	"",			/* not used */
-	"Instruction",
-	"Data",
-	"Unified"	/* unified */
-};
-
-static const char *cache_mattrib[]={
-	"WriteThrough",
-	"WriteBack",
-	"",		/* reserved */
-	""		/* reserved */
-};
-
-struct cache_info {
-	pal_cache_config_info_t	cci;
-	cpumask_t shared_cpu_map;
-	int level;
-	int type;
-	struct kobject kobj;
-};
-
-struct cpu_cache_info {
-	struct cache_info *cache_leaves;
-	int	num_cache_leaves;
-	struct kobject kobj;
-};
-
-static struct cpu_cache_info	all_cpu_cache_info[NR_CPUS];
-#define LEAF_KOBJECT_PTR(x,y)    (&all_cpu_cache_info[x].cache_leaves[y])
-
 #ifdef CONFIG_SMP
-static void cache_shared_cpu_map_setup(unsigned int cpu,
-		struct cache_info * this_leaf)
+static int __cache_cpumap_setup(unsigned int cpu, struct cacheinfo *this_leaf)
 {
 	pal_cache_shared_info_t	csi;
-	int num_shared, i = 0;
-	unsigned int j;
+	int num_shared, i = 0, j;
+	enum cache_type type = this_leaf->type;
 
 	if (cpu_data(cpu)->threads_per_core <= 1 &&
 		cpu_data(cpu)->cores_per_socket <= 1) {
 		cpu_set(cpu, this_leaf->shared_cpu_map);
-		return;
+		return 0;
 	}
 
-	if (ia64_pal_cache_shared_info(this_leaf->level,
-					this_leaf->type,
-					0,
-					&csi) != PAL_STATUS_SUCCESS)
-		return;
+	if (type = CACHE_TYPE_UNIFIED)
+		type = CACHE_TYPE_DATA;
+
+	if (ia64_pal_cache_shared_info(this_leaf->level, type, 0,
+				       &csi) != PAL_STATUS_SUCCESS)
+		return -EIO;
 
 	num_shared = (int) csi.num_shared;
 	do {
@@ -168,305 +133,97 @@ static void cache_shared_cpu_map_setup(unsigned int cpu,
 
 		i++;
 	} while (i < num_shared &&
-		ia64_pal_cache_shared_info(this_leaf->level,
-				this_leaf->type,
-				i,
-				&csi) = PAL_STATUS_SUCCESS);
-}
-#else
-static void cache_shared_cpu_map_setup(unsigned int cpu,
-		struct cache_info * this_leaf)
-{
-	cpu_set(cpu, this_leaf->shared_cpu_map);
-	return;
-}
-#endif
-
-static ssize_t show_coherency_line_size(struct cache_info *this_leaf,
-					char *buf)
-{
-	return sprintf(buf, "%u\n", 1 << this_leaf->cci.pcci_line_size);
-}
-
-static ssize_t show_ways_of_associativity(struct cache_info *this_leaf,
-					char *buf)
-{
-	return sprintf(buf, "%u\n", this_leaf->cci.pcci_assoc);
-}
-
-static ssize_t show_attributes(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf,
-			"%s\n",
-			cache_mattrib[this_leaf->cci.pcci_cache_attr]);
-}
-
-static ssize_t show_size(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf, "%uK\n", this_leaf->cci.pcci_cache_size / 1024);
-}
-
-static ssize_t show_number_of_sets(struct cache_info *this_leaf, char *buf)
-{
-	unsigned number_of_sets = this_leaf->cci.pcci_cache_size;
-	number_of_sets /= this_leaf->cci.pcci_assoc;
-	number_of_sets /= 1 << this_leaf->cci.pcci_line_size;
-
-	return sprintf(buf, "%u\n", number_of_sets);
-}
-
-static ssize_t show_shared_cpu_map(struct cache_info *this_leaf, char *buf)
-{
-	ssize_t	len;
-	cpumask_t shared_cpu_map;
-
-	cpumask_and(&shared_cpu_map,
-				&this_leaf->shared_cpu_map, cpu_online_mask);
-	len = cpumask_scnprintf(buf, NR_CPUS+1, &shared_cpu_map);
-	len += sprintf(buf+len, "\n");
-	return len;
-}
-
-static ssize_t show_type(struct cache_info *this_leaf, char *buf)
-{
-	int type = this_leaf->type + this_leaf->cci.pcci_unified;
-	return sprintf(buf, "%s\n", cache_types[type]);
-}
-
-static ssize_t show_level(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf, "%u\n", this_leaf->level);
+		 ia64_pal_cache_shared_info(this_leaf->level, type, i,
+					    &csi) = PAL_STATUS_SUCCESS);
+	return 0;
 }
 
-struct cache_attr {
-	struct attribute attr;
-	ssize_t (*show)(struct cache_info *, char *);
-	ssize_t (*store)(struct cache_info *, const char *, size_t count);
-};
-
-#ifdef define_one_ro
-	#undef define_one_ro
-#endif
-#define define_one_ro(_name) \
-	static struct cache_attr _name = \
-__ATTR(_name, 0444, show_##_name, NULL)
-
-define_one_ro(level);
-define_one_ro(type);
-define_one_ro(coherency_line_size);
-define_one_ro(ways_of_associativity);
-define_one_ro(size);
-define_one_ro(number_of_sets);
-define_one_ro(shared_cpu_map);
-define_one_ro(attributes);
-
-static struct attribute * cache_default_attrs[] = {
-	&type.attr,
-	&level.attr,
-	&coherency_line_size.attr,
-	&ways_of_associativity.attr,
-	&attributes.attr,
-	&size.attr,
-	&number_of_sets.attr,
-	&shared_cpu_map.attr,
-	NULL
-};
-
-#define to_object(k) container_of(k, struct cache_info, kobj)
-#define to_attr(a) container_of(a, struct cache_attr, attr)
-
-static ssize_t ia64_cache_show(struct kobject * kobj, struct attribute * attr, char * buf)
+static int cache_shared_cpu_map_setup(unsigned int cpu)
 {
-	struct cache_attr *fattr = to_attr(attr);
-	struct cache_info *this_leaf = to_object(kobj);
-	ssize_t ret;
+	unsigned int idx;
+	int ret = 0;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
 
-	ret = fattr->show ? fattr->show(this_leaf, buf) : 0;
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++, this_leaf++) {
+		ret = __cache_cpumap_setup(cpu, this_leaf);
+		if (ret)
+			break;
+	}
 	return ret;
 }
-
-static const struct sysfs_ops cache_sysfs_ops = {
-	.show   = ia64_cache_show
-};
-
-static struct kobj_type cache_ktype = {
-	.sysfs_ops	= &cache_sysfs_ops,
-	.default_attrs	= cache_default_attrs,
-};
-
-static struct kobj_type cache_ktype_percpu_entry = {
-	.sysfs_ops	= &cache_sysfs_ops,
-};
-
-static void cpu_cache_sysfs_exit(unsigned int cpu)
-{
-	kfree(all_cpu_cache_info[cpu].cache_leaves);
-	all_cpu_cache_info[cpu].cache_leaves = NULL;
-	all_cpu_cache_info[cpu].num_cache_leaves = 0;
-	memset(&all_cpu_cache_info[cpu].kobj, 0, sizeof(struct kobject));
-	return;
-}
-
-static int cpu_cache_sysfs_init(unsigned int cpu)
+#else
+static int cache_shared_cpu_map_setup(unsigned int cpu)
 {
-	unsigned long i, levels, unique_caches;
-	pal_cache_config_info_t cci;
-	int j;
-	long status;
-	struct cache_info *this_cache;
-	int num_cache_leaves = 0;
-
-	if ((status = ia64_pal_cache_summary(&levels, &unique_caches)) != 0) {
-		printk(KERN_ERR "ia64_pal_cache_summary=%ld\n", status);
-		return -1;
-	}
-
-	this_cache=kzalloc(sizeof(struct cache_info)*unique_caches,
-			GFP_KERNEL);
-	if (this_cache = NULL)
-		return -ENOMEM;
-
-	for (i=0; i < levels; i++) {
-		for (j=2; j >0 ; j--) {
-			if ((status=ia64_pal_cache_config_info(i,j, &cci)) !-					PAL_STATUS_SUCCESS)
-				continue;
-
-			this_cache[num_cache_leaves].cci = cci;
-			this_cache[num_cache_leaves].level = i + 1;
-			this_cache[num_cache_leaves].type = j;
-
-			cache_shared_cpu_map_setup(cpu,
-					&this_cache[num_cache_leaves]);
-			num_cache_leaves ++;
-		}
-	}
-
-	all_cpu_cache_info[cpu].cache_leaves = this_cache;
-	all_cpu_cache_info[cpu].num_cache_leaves = num_cache_leaves;
-
-	memset(&all_cpu_cache_info[cpu].kobj, 0, sizeof(struct kobject));
+	int idx;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
 
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++, this_leaf++)
+		cpu_set(cpu, this_leaf->shared_cpu_map);
 	return 0;
 }
+#endif
 
-/* Add cache interface for CPU device */
-static int cache_add_dev(struct device *sys_dev)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 pal_cache_config_info_t *cci,
+			 enum cache_type type, unsigned int level)
 {
-	unsigned int cpu = sys_dev->id;
-	unsigned long i, j;
-	struct cache_info *this_object;
-	int retval = 0;
-	cpumask_t oldmask;
-
-	if (all_cpu_cache_info[cpu].kobj.parent)
-		return 0;
-
-	oldmask = current->cpus_allowed;
-	retval = set_cpus_allowed_ptr(current, cpumask_of(cpu));
-	if (unlikely(retval))
-		return retval;
+	unsigned number_of_sets = cci->pcci_cache_size;
 
-	retval = cpu_cache_sysfs_init(cpu);
-	set_cpus_allowed_ptr(current, &oldmask);
-	if (unlikely(retval < 0))
-		return retval;
-
-	retval = kobject_init_and_add(&all_cpu_cache_info[cpu].kobj,
-				      &cache_ktype_percpu_entry, &sys_dev->kobj,
-				      "%s", "cache");
-	if (unlikely(retval < 0)) {
-		cpu_cache_sysfs_exit(cpu);
-		return retval;
-	}
-
-	for (i = 0; i < all_cpu_cache_info[cpu].num_cache_leaves; i++) {
-		this_object = LEAF_KOBJECT_PTR(cpu,i);
-		retval = kobject_init_and_add(&(this_object->kobj),
-					      &cache_ktype,
-					      &all_cpu_cache_info[cpu].kobj,
-					      "index%1lu", i);
-		if (unlikely(retval)) {
-			for (j = 0; j < i; j++) {
-				kobject_put(&(LEAF_KOBJECT_PTR(cpu,j)->kobj));
-			}
-			kobject_put(&all_cpu_cache_info[cpu].kobj);
-			cpu_cache_sysfs_exit(cpu);
-			return retval;
-		}
-		kobject_uevent(&(this_object->kobj), KOBJ_ADD);
-	}
-	kobject_uevent(&all_cpu_cache_info[cpu].kobj, KOBJ_ADD);
-	return retval;
+	number_of_sets /= cci->pcci_assoc;
+	number_of_sets /= 1 << cci->pcci_line_size;
+	this_leaf->level = level;
+	this_leaf->type = cci->pcci_unified ? CACHE_TYPE_UNIFIED : type;
+	this_leaf->coherency_line_size = cci->pcci_line_size;
+	this_leaf->ways_of_associativity = cci->pcci_assoc;
+	this_leaf->size = cci->pcci_cache_size;
+	this_leaf->attributes = cci->pcci_cache_attr;
+	this_leaf->number_of_sets = number_of_sets;
 }
 
-/* Remove cache interface for CPU device */
-static int cache_remove_dev(struct device *sys_dev)
+int init_cache_level(unsigned int cpu)
 {
-	unsigned int cpu = sys_dev->id;
-	unsigned long i;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned long levels, unique_caches;
+	long status;
 
-	for (i = 0; i < all_cpu_cache_info[cpu].num_cache_leaves; i++)
-		kobject_put(&(LEAF_KOBJECT_PTR(cpu,i)->kobj));
+	if (!this_cpu_ci)
+		return -EINVAL;
 
-	if (all_cpu_cache_info[cpu].kobj.parent) {
-		kobject_put(&all_cpu_cache_info[cpu].kobj);
-		memset(&all_cpu_cache_info[cpu].kobj,
-			0,
-			sizeof(struct kobject));
+	status = ia64_pal_cache_summary(&levels, &unique_caches);
+	if (status != PAL_STATUS_SUCCESS) {
+		pr_err("ia64_pal_cache_summary = %ld\n", status);
+		return -EIO;
 	}
-
-	cpu_cache_sysfs_exit(cpu);
+	this_cpu_ci->num_levels = levels;
+	this_cpu_ci->num_leaves = unique_caches;
 
 	return 0;
 }
 
-/*
- * When a cpu is hot-plugged, do a check and initiate
- * cache kobject if necessary
- */
-static int cache_cpu_callback(struct notifier_block *nfb,
-		unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (unsigned long)hcpu;
-	struct device *sys_dev;
-
-	sys_dev = get_cpu_device(cpu);
-	switch (action) {
-	case CPU_ONLINE:
-	case CPU_ONLINE_FROZEN:
-		cache_add_dev(sys_dev);
-		break;
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		cache_remove_dev(sys_dev);
-		break;
-	}
-	return NOTIFY_OK;
-}
-
-static struct notifier_block cache_cpu_notifier -{
-	.notifier_call = cache_cpu_callback
-};
-
-static int __init cache_sysfs_init(void)
+int populate_cache_leaves(unsigned int cpu)
 {
-	int i;
-
-	cpu_notifier_register_begin();
-
-	for_each_online_cpu(i) {
-		struct device *sys_dev = get_cpu_device((unsigned int)i);
-		cache_add_dev(sys_dev);
+	unsigned int level, idx;
+	s64 status;
+	pal_cache_config_info_t cci;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		type = CACHE_TYPE_INST;
+		status = ia64_pal_cache_config_info(level - 1, type, &cci);
+		if (status = PAL_STATUS_SUCCESS)
+			ci_leaf_init(this_leaf++, &cci, type, level);
+		type = CACHE_TYPE_DATA;
+		status = ia64_pal_cache_config_info(level - 1, type, &cci);
+		if (status = PAL_STATUS_SUCCESS)
+			ci_leaf_init(this_leaf++, &cci, type, level);
 	}
-
-	__register_hotcpu_notifier(&cache_cpu_notifier);
-
-	cpu_notifier_register_done();
-
-	return 0;
+	return cache_shared_cpu_map_setup(cpu);
 }
-
-device_initcall(cache_sysfs_init);
-
-- 
1.8.3.2

[PATCH v2 4/9] s390: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
---
 arch/s390/kernel/cache.c | 388 +++++++++++------------------------------------
 1 file changed, 92 insertions(+), 296 deletions(-)

diff --git a/arch/s390/kernel/cache.c b/arch/s390/kernel/cache.c
index c0b03c28d157..fe21f074cf9f 100644
--- a/arch/s390/kernel/cache.c
+++ b/arch/s390/kernel/cache.c
@@ -5,37 +5,11 @@
  *    Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>
  */
 
-#include <linux/notifier.h>
 #include <linux/seq_file.h>
-#include <linux/init.h>
-#include <linux/list.h>
-#include <linux/slab.h>
 #include <linux/cpu.h>
+#include <linux/cacheinfo.h>
 #include <asm/facility.h>
 
-struct cache {
-	unsigned long size;
-	unsigned int line_size;
-	unsigned int associativity;
-	unsigned int nr_sets;
-	unsigned int level   : 3;
-	unsigned int type    : 2;
-	unsigned int private : 1;
-	struct list_head list;
-};
-
-struct cache_dir {
-	struct kobject *kobj;
-	struct cache_index_dir *index;
-};
-
-struct cache_index_dir {
-	struct kobject kobj;
-	int cpu;
-	struct cache *cache;
-	struct cache_index_dir *next;
-};
-
 enum {
 	CACHE_SCOPE_NOTEXISTS,
 	CACHE_SCOPE_PRIVATE,
@@ -44,10 +18,10 @@ enum {
 };
 
 enum {
-	CACHE_TYPE_SEPARATE,
-	CACHE_TYPE_DATA,
-	CACHE_TYPE_INSTRUCTION,
-	CACHE_TYPE_UNIFIED,
+	CTYPE_SEPARATE,
+	CTYPE_DATA,
+	CTYPE_INSTRUCTION,
+	CTYPE_UNIFIED,
 };
 
 enum {
@@ -70,39 +44,59 @@ struct cache_info {
 };
 
 #define CACHE_MAX_LEVEL 8
-
 union cache_topology {
 	struct cache_info ci[CACHE_MAX_LEVEL];
 	unsigned long long raw;
 };
 
 static const char * const cache_type_string[] = {
-	"Data",
+	"",
 	"Instruction",
+	"Data",
+	"",
 	"Unified",
 };
 
-static struct cache_dir *cache_dir_cpu[NR_CPUS];
-static LIST_HEAD(cache_list);
+static const enum cache_type cache_type_map[] = {
+	[CTYPE_SEPARATE] = CACHE_TYPE_SEPARATE,
+	[CTYPE_DATA] = CACHE_TYPE_DATA,
+	[CTYPE_INSTRUCTION] = CACHE_TYPE_INST,
+	[CTYPE_UNIFIED] = CACHE_TYPE_UNIFIED,
+};
 
 void show_cacheinfo(struct seq_file *m)
 {
-	struct cache *cache;
-	int index = 0;
+	int cpu = smp_processor_id(), idx;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *cache;
 
-	list_for_each_entry(cache, &cache_list, list) {
-		seq_printf(m, "cache%-11d: ", index);
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++) {
+		cache = this_cpu_ci->info_list + idx;
+		seq_printf(m, "cache%-11d: ", idx);
 		seq_printf(m, "level=%d ", cache->level);
 		seq_printf(m, "type=%s ", cache_type_string[cache->type]);
-		seq_printf(m, "scope=%s ", cache->private ? "Private" : "Shared");
-		seq_printf(m, "size=%luK ", cache->size >> 10);
-		seq_printf(m, "line_size=%u ", cache->line_size);
-		seq_printf(m, "associativity=%d", cache->associativity);
+		seq_printf(m, "scope=%s ",
+			   cache->disable_sysfs ? "Shared" : "Private");
+		seq_printf(m, "size=%dK ", cache->size >> 10);
+		seq_printf(m, "line_size=%u ", cache->coherency_line_size);
+		seq_printf(m, "associativity=%d", cache->ways_of_associativity);
 		seq_puts(m, "\n");
-		index++;
 	}
 }
 
+static inline enum cache_type get_cache_type(struct cache_info *ci, int level)
+{
+	if (level >= CACHE_MAX_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+
+	ci += level;
+
+	if (ci->scope != CACHE_SCOPE_SHARED && ci->scope != CACHE_SCOPE_PRIVATE)
+		return CACHE_TYPE_NOCACHE;
+
+	return cache_type_map[ci->type];
+}
+
 static inline unsigned long ecag(int ai, int li, int ti)
 {
 	unsigned long cmd, val;
@@ -113,277 +107,79 @@ static inline unsigned long ecag(int ai, int li, int ti)
 	return val;
 }
 
-static int __init cache_add(int level, int private, int type)
+static void ci_leaf_init(struct cacheinfo *this_leaf, int private,
+			 enum cache_type type, unsigned int level)
 {
-	struct cache *cache;
-	int ti;
+	int ti, num_sets;
+	int cpu = smp_processor_id();
 
-	cache = kzalloc(sizeof(*cache), GFP_KERNEL);
-	if (!cache)
-		return -ENOMEM;
-	if (type == CACHE_TYPE_INSTRUCTION)
+	if (type == CACHE_TYPE_INST)
 		ti = CACHE_TI_INSTRUCTION;
 	else
 		ti = CACHE_TI_UNIFIED;
-	cache->size = ecag(EXTRACT_SIZE, level, ti);
-	cache->line_size = ecag(EXTRACT_LINE_SIZE, level, ti);
-	cache->associativity = ecag(EXTRACT_ASSOCIATIVITY, level, ti);
-	cache->nr_sets = cache->size / cache->associativity;
-	cache->nr_sets /= cache->line_size;
-	cache->private = private;
-	cache->level = level + 1;
-	cache->type = type - 1;
-	list_add_tail(&cache->list, &cache_list);
-	return 0;
-}
-
-static void __init cache_build_info(void)
-{
-	struct cache *cache, *next;
-	union cache_topology ct;
-	int level, private, rc;
-
-	ct.raw = ecag(EXTRACT_TOPOLOGY, 0, 0);
-	for (level = 0; level < CACHE_MAX_LEVEL; level++) {
-		switch (ct.ci[level].scope) {
-		case CACHE_SCOPE_SHARED:
-			private = 0;
-			break;
-		case CACHE_SCOPE_PRIVATE:
-			private = 1;
-			break;
-		default:
-			return;
-		}
-		if (ct.ci[level].type == CACHE_TYPE_SEPARATE) {
-			rc  = cache_add(level, private, CACHE_TYPE_DATA);
-			rc |= cache_add(level, private, CACHE_TYPE_INSTRUCTION);
-		} else {
-			rc = cache_add(level, private, ct.ci[level].type);
-		}
-		if (rc)
-			goto error;
-	}
-	return;
-error:
-	list_for_each_entry_safe(cache, next, &cache_list, list) {
-		list_del(&cache->list);
-		kfree(cache);
-	}
-}
-
-static struct cache_dir *cache_create_cache_dir(int cpu)
-{
-	struct cache_dir *cache_dir;
-	struct kobject *kobj = NULL;
-	struct device *dev;
-
-	dev = get_cpu_device(cpu);
-	if (!dev)
-		goto out;
-	kobj = kobject_create_and_add("cache", &dev->kobj);
-	if (!kobj)
-		goto out;
-	cache_dir = kzalloc(sizeof(*cache_dir), GFP_KERNEL);
-	if (!cache_dir)
-		goto out;
-	cache_dir->kobj = kobj;
-	cache_dir_cpu[cpu] = cache_dir;
-	return cache_dir;
-out:
-	kobject_put(kobj);
-	return NULL;
-}
-
-static struct cache_index_dir *kobj_to_cache_index_dir(struct kobject *kobj)
-{
-	return container_of(kobj, struct cache_index_dir, kobj);
-}
-
-static void cache_index_release(struct kobject *kobj)
-{
-	struct cache_index_dir *index;
-
-	index = kobj_to_cache_index_dir(kobj);
-	kfree(index);
-}
-
-static ssize_t cache_index_show(struct kobject *kobj,
-				struct attribute *attr, char *buf)
-{
-	struct kobj_attribute *kobj_attr;
-
-	kobj_attr = container_of(attr, struct kobj_attribute, attr);
-	return kobj_attr->show(kobj, kobj_attr, buf);
-}
-
-#define DEFINE_CACHE_ATTR(_name, _format, _value)			\
-static ssize_t cache_##_name##_show(struct kobject *kobj,		\
-				    struct kobj_attribute *attr,	\
-				    char *buf)				\
-{									\
-	struct cache_index_dir *index;					\
-									\
-	index = kobj_to_cache_index_dir(kobj);				\
-	return sprintf(buf, _format, _value);				\
-}									\
-static struct kobj_attribute cache_##_name##_attr =			\
-	__ATTR(_name, 0444, cache_##_name##_show, NULL);
 
-DEFINE_CACHE_ATTR(size, "%luK\n", index->cache->size >> 10);
-DEFINE_CACHE_ATTR(coherency_line_size, "%u\n", index->cache->line_size);
-DEFINE_CACHE_ATTR(number_of_sets, "%u\n", index->cache->nr_sets);
-DEFINE_CACHE_ATTR(ways_of_associativity, "%u\n", index->cache->associativity);
-DEFINE_CACHE_ATTR(type, "%s\n", cache_type_string[index->cache->type]);
-DEFINE_CACHE_ATTR(level, "%d\n", index->cache->level);
+	this_leaf->level = level + 1;
+	this_leaf->type = type;
+	this_leaf->coherency_line_size = ecag(EXTRACT_LINE_SIZE, level, ti);
+	this_leaf->ways_of_associativity = ecag(EXTRACT_ASSOCIATIVITY,
+						level, ti);
+	this_leaf->size = ecag(EXTRACT_SIZE, level, ti);
 
-static ssize_t shared_cpu_map_func(struct kobject *kobj, int type, char *buf)
-{
-	struct cache_index_dir *index;
-	int len;
-
-	index = kobj_to_cache_index_dir(kobj);
-	len = type ?
-		cpulist_scnprintf(buf, PAGE_SIZE - 2, cpumask_of(index->cpu)) :
-		cpumask_scnprintf(buf, PAGE_SIZE - 2, cpumask_of(index->cpu));
-	len += sprintf(&buf[len], "\n");
-	return len;
-}
-
-static ssize_t shared_cpu_map_show(struct kobject *kobj,
-				   struct kobj_attribute *attr, char *buf)
-{
-	return shared_cpu_map_func(kobj, 0, buf);
+	num_sets = this_leaf->size / this_leaf->coherency_line_size;
+	num_sets /= this_leaf->ways_of_associativity;
+	this_leaf->number_of_sets = num_sets;
+	cpumask_set_cpu(cpu, &this_leaf->shared_cpu_map);
+	if (!private)
+		this_leaf->disable_sysfs = true;
 }
-static struct kobj_attribute cache_shared_cpu_map_attr =
-	__ATTR(shared_cpu_map, 0444, shared_cpu_map_show, NULL);
 
-static ssize_t shared_cpu_list_show(struct kobject *kobj,
-				    struct kobj_attribute *attr, char *buf)
+int init_cache_level(unsigned int cpu)
 {
-	return shared_cpu_map_func(kobj, 1, buf);
-}
-static struct kobj_attribute cache_shared_cpu_list_attr =
-	__ATTR(shared_cpu_list, 0444, shared_cpu_list_show, NULL);
-
-static struct attribute *cache_index_default_attrs[] = {
-	&cache_type_attr.attr,
-	&cache_size_attr.attr,
-	&cache_number_of_sets_attr.attr,
-	&cache_ways_of_associativity_attr.attr,
-	&cache_level_attr.attr,
-	&cache_coherency_line_size_attr.attr,
-	&cache_shared_cpu_map_attr.attr,
-	&cache_shared_cpu_list_attr.attr,
-	NULL,
-};
-
-static const struct sysfs_ops cache_index_ops = {
-	.show = cache_index_show,
-};
-
-static struct kobj_type cache_index_type = {
-	.sysfs_ops = &cache_index_ops,
-	.release = cache_index_release,
-	.default_attrs = cache_index_default_attrs,
-};
-
-static int cache_create_index_dir(struct cache_dir *cache_dir,
-				  struct cache *cache, int index, int cpu)
-{
-	struct cache_index_dir *index_dir;
-	int rc;
-
-	index_dir = kzalloc(sizeof(*index_dir), GFP_KERNEL);
-	if (!index_dir)
-		return -ENOMEM;
-	index_dir->cache = cache;
-	index_dir->cpu = cpu;
-	rc = kobject_init_and_add(&index_dir->kobj, &cache_index_type,
-				  cache_dir->kobj, "index%d", index);
-	if (rc)
-		goto out;
-	index_dir->next = cache_dir->index;
-	cache_dir->index = index_dir;
-	return 0;
-out:
-	kfree(index_dir);
-	return rc;
-}
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int level = 0, leaves = 0;
+	union cache_topology ct;
+	enum cache_type ctype;
 
-static int cache_add_cpu(int cpu)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-	int rc, index = 0;
+	if (!this_cpu_ci)
+		return -EINVAL;
 
-	if (list_empty(&cache_list))
-		return 0;
-	cache_dir = cache_create_cache_dir(cpu);
-	if (!cache_dir)
-		return -ENOMEM;
-	list_for_each_entry(cache, &cache_list, list) {
-		if (!cache->private)
+	ct.raw = ecag(EXTRACT_TOPOLOGY, 0, 0);
+	do {
+		ctype = get_cache_type(&ct.ci[0], level);
+		if (ctype == CACHE_TYPE_NOCACHE)
 			break;
-		rc = cache_create_index_dir(cache_dir, cache, index, cpu);
-		if (rc)
-			return rc;
-		index++;
-	}
-	return 0;
-}
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	} while (++level < CACHE_MAX_LEVEL);
 
-static void cache_remove_cpu(int cpu)
-{
-	struct cache_index_dir *index, *next;
-	struct cache_dir *cache_dir;
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
 
-	cache_dir = cache_dir_cpu[cpu];
-	if (!cache_dir)
-		return;
-	index = cache_dir->index;
-	while (index) {
-		next = index->next;
-		kobject_put(&index->kobj);
-		index = next;
-	}
-	kobject_put(cache_dir->kobj);
-	kfree(cache_dir);
-	cache_dir_cpu[cpu] = NULL;
+	return 0;
 }
 
-static int cache_hotplug(struct notifier_block *nfb, unsigned long action,
-			 void *hcpu)
+int populate_cache_leaves(unsigned int cpu)
 {
-	int cpu = (long)hcpu;
-	int rc = 0;
+	unsigned int level, idx, pvt;
+	union cache_topology ct;
+	enum cache_type ctype;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_ONLINE:
-		rc = cache_add_cpu(cpu);
-		if (rc)
-			cache_remove_cpu(cpu);
-		break;
-	case CPU_DEAD:
-		cache_remove_cpu(cpu);
-		break;
+	ct.raw = ecag(EXTRACT_TOPOLOGY, 0, 0);
+	for (idx = 0, level = 0; level < this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		pvt = (ct.ci[level].scope == CACHE_SCOPE_PRIVATE) ? 1 : 0;
+		ctype = get_cache_type(&ct.ci[0], level);
+		if (ctype == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, pvt, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, pvt, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, pvt, ctype, level);
+		}
 	}
-	return rc ? NOTIFY_BAD : NOTIFY_OK;
-}
-
-static int __init cache_init(void)
-{
-	int cpu;
-
-	if (!test_facility(34))
-		return 0;
-	cache_build_info();
-
-	cpu_notifier_register_begin();
-	for_each_online_cpu(cpu)
-		cache_add_cpu(cpu);
-	__hotcpu_notifier(cache_hotplug, 0);
-	cpu_notifier_register_done();
 	return 0;
 }
-device_initcall(cache_init);
-- 
1.8.3.2

[PATCH v2 5/9] x86: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

The private pointer provided by the cacheinfo is used to implement
the AMD L3 cache specific attributes.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: x86@kernel.org
---
 arch/x86/kernel/cpu/intel_cacheinfo.c | 680 +++++++++-------------------------
 1 file changed, 166 insertions(+), 514 deletions(-)

diff --git a/arch/x86/kernel/cpu/intel_cacheinfo.c b/arch/x86/kernel/cpu/intel_cacheinfo.c
index a952e9c85b6f..7aa9c2a7e320 100644
--- a/arch/x86/kernel/cpu/intel_cacheinfo.c
+++ b/arch/x86/kernel/cpu/intel_cacheinfo.c
@@ -7,16 +7,14 @@
  *	Andi Kleen / Andreas Herrmann	: CPUID4 emulation on AMD.
  */
 
-#include <linux/init.h>
 #include <linux/slab.h>
-#include <linux/device.h>
-#include <linux/compiler.h>
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
 #include <linux/sched.h>
+#include <linux/sysfs.h>
 #include <linux/pci.h>
 
 #include <asm/processor.h>
-#include <linux/smp.h>
 #include <asm/amd_nb.h>
 #include <asm/smp.h>
 
@@ -116,10 +114,10 @@ static const struct _cache_table cache_table[] =
 
 
 enum _cache_type {
-	CACHE_TYPE_NULL	= 0,
-	CACHE_TYPE_DATA = 1,
-	CACHE_TYPE_INST = 2,
-	CACHE_TYPE_UNIFIED = 3
+	CTYPE_NULL = 0,
+	CTYPE_DATA = 1,
+	CTYPE_INST = 2,
+	CTYPE_UNIFIED = 3
 };
 
 union _cpuid4_leaf_eax {
@@ -159,11 +157,6 @@ struct _cpuid4_info_regs {
 	struct amd_northbridge *nb;
 };
 
-struct _cpuid4_info {
-	struct _cpuid4_info_regs base;
-	DECLARE_BITMAP(shared_cpu_map, NR_CPUS);
-};
-
 unsigned short			num_cache_leaves;
 
 /* AMD doesn't have CPUID4. Emulate it here to report the same
@@ -220,6 +213,13 @@ static const unsigned short assocs[] = {
 static const unsigned char levels[] = { 1, 1, 2, 3 };
 static const unsigned char types[] = { 1, 2, 3, 3 };
 
+static const enum cache_type cache_type_map[] = {
+	[CTYPE_NULL] = CACHE_TYPE_NOCACHE,
+	[CTYPE_DATA] = CACHE_TYPE_DATA,
+	[CTYPE_INST] = CACHE_TYPE_INST,
+	[CTYPE_UNIFIED] = CACHE_TYPE_UNIFIED,
+};
+
 static void
 amd_cpuid4(int leaf, union _cpuid4_leaf_eax *eax,
 		     union _cpuid4_leaf_ebx *ebx,
@@ -291,14 +291,9 @@ amd_cpuid4(int leaf, union _cpuid4_leaf_eax *eax,
 		(ebx->split.ways_of_associativity + 1) - 1;
 }
 
-struct _cache_attr {
-	struct attribute attr;
-	ssize_t (*show)(struct _cpuid4_info *, char *, unsigned int);
-	ssize_t (*store)(struct _cpuid4_info *, const char *, size_t count,
-			 unsigned int);
-};
-
 #if defined(CONFIG_AMD_NB) && defined(CONFIG_SYSFS)
+
+static struct attribute **amd_l3_attrs;
 /*
  * L3 cache descriptors
  */
@@ -325,20 +320,6 @@ static void amd_calc_l3_indices(struct amd_northbridge *nb)
 	l3->indices = (max(max3(sc0, sc1, sc2), sc3) << 10) - 1;
 }
 
-static void amd_init_l3_cache(struct _cpuid4_info_regs *this_leaf, int index)
-{
-	int node;
-
-	/* only for L3, and not in virtualized environments */
-	if (index < 3)
-		return;
-
-	node = amd_get_nb_id(smp_processor_id());
-	this_leaf->nb = node_to_amd_nb(node);
-	if (this_leaf->nb && !this_leaf->nb->l3_cache.indices)
-		amd_calc_l3_indices(this_leaf->nb);
-}
-
 /*
  * check whether a slot used for disabling an L3 index is occupied.
  * @l3: L3 cache descriptor
@@ -359,15 +340,16 @@ int amd_get_l3_disable_slot(struct amd_northbridge *nb, unsigned slot)
 	return -1;
 }
 
-static ssize_t show_cache_disable(struct _cpuid4_info *this_leaf, char *buf,
+static ssize_t show_cache_disable(struct cacheinfo *this_leaf, char *buf,
 				  unsigned int slot)
 {
 	int index;
+	struct amd_northbridge *nb = this_leaf->priv;
 
-	if (!this_leaf->base.nb || !amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
+	if (!nb || !amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
 		return -EINVAL;
 
-	index = amd_get_l3_disable_slot(this_leaf->base.nb, slot);
+	index = amd_get_l3_disable_slot(nb, slot);
 	if (index >= 0)
 		return sprintf(buf, "%d\n", index);
 
@@ -376,9 +358,10 @@ static ssize_t show_cache_disable(struct _cpuid4_info *this_leaf, char *buf,
 
 #define SHOW_CACHE_DISABLE(slot)					\
 static ssize_t								\
-show_cache_disable_##slot(struct _cpuid4_info *this_leaf, char *buf,	\
-			  unsigned int cpu)				\
+cache_disable_##slot##_show(struct device *dev,				\
+			    struct device_attribute *attr, char *buf)	\
 {									\
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);		\
 	return show_cache_disable(this_leaf, buf, slot);		\
 }
 SHOW_CACHE_DISABLE(0)
@@ -446,25 +429,26 @@ int amd_set_l3_disable_slot(struct amd_northbridge *nb, int cpu, unsigned slot,
 	return 0;
 }
 
-static ssize_t store_cache_disable(struct _cpuid4_info *this_leaf,
-				  const char *buf, size_t count,
-				  unsigned int slot)
+static ssize_t store_cache_disable(struct cacheinfo *this_leaf,
+				   const char *buf, size_t count,
+				   unsigned int slot)
 {
 	unsigned long val = 0;
 	int cpu, err = 0;
+	struct amd_northbridge *nb = this_leaf->priv;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
-	if (!this_leaf->base.nb || !amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
+	if (!nb || !amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
 		return -EINVAL;
 
-	cpu = cpumask_first(to_cpumask(this_leaf->shared_cpu_map));
+	cpu = cpumask_first(&this_leaf->shared_cpu_map);
 
 	if (strict_strtoul(buf, 10, &val) < 0)
 		return -EINVAL;
 
-	err = amd_set_l3_disable_slot(this_leaf->base.nb, cpu, slot, val);
+	err = amd_set_l3_disable_slot(nb, cpu, slot, val);
 	if (err) {
 		if (err == -EEXIST)
 			pr_warning("L3 slot %d in use/index already disabled!\n",
@@ -476,39 +460,39 @@ static ssize_t store_cache_disable(struct _cpuid4_info *this_leaf,
 
 #define STORE_CACHE_DISABLE(slot)					\
 static ssize_t								\
-store_cache_disable_##slot(struct _cpuid4_info *this_leaf,		\
-			   const char *buf, size_t count,		\
-			   unsigned int cpu)				\
+cache_disable_##slot##_store(struct device *dev,			\
+			     struct device_attribute *attr,		\
+			     const char *buf, size_t count)		\
 {									\
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);		\
 	return store_cache_disable(this_leaf, buf, count, slot);	\
 }
 STORE_CACHE_DISABLE(0)
 STORE_CACHE_DISABLE(1)
 
-static struct _cache_attr cache_disable_0 = __ATTR(cache_disable_0, 0644,
-		show_cache_disable_0, store_cache_disable_0);
-static struct _cache_attr cache_disable_1 = __ATTR(cache_disable_1, 0644,
-		show_cache_disable_1, store_cache_disable_1);
-
-static ssize_t
-show_subcaches(struct _cpuid4_info *this_leaf, char *buf, unsigned int cpu)
+static ssize_t subcaches_show(struct device *dev,
+			      struct device_attribute *attr, char *buf)
 {
-	if (!this_leaf->base.nb || !amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
-		return -EINVAL;
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	int cpu = cpumask_first(&this_leaf->shared_cpu_map);
 
+	if (!this_leaf->priv || !amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
+		return -EINVAL;
 	return sprintf(buf, "%x\n", amd_get_subcaches(cpu));
 }
 
-static ssize_t
-store_subcaches(struct _cpuid4_info *this_leaf, const char *buf, size_t count,
-		unsigned int cpu)
+static ssize_t subcaches_store(struct device *dev,
+			       struct device_attribute *attr,
+			       const char *buf, size_t count)
 {
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	int cpu = cpumask_first(&this_leaf->shared_cpu_map);
 	unsigned long val;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
-	if (!this_leaf->base.nb || !amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
+	if (!this_leaf->priv || !amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
 		return -EINVAL;
 
 	if (strict_strtoul(buf, 16, &val) < 0)
@@ -520,9 +504,59 @@ store_subcaches(struct _cpuid4_info *this_leaf, const char *buf, size_t count,
 	return count;
 }
 
-static struct _cache_attr subcaches =
-	__ATTR(subcaches, 0644, show_subcaches, store_subcaches);
+static DEVICE_ATTR_RW(cache_disable_0);
+static DEVICE_ATTR_RW(cache_disable_1);
+static DEVICE_ATTR_RW(subcaches);
+
+static void init_amd_l3_attrs(void)
+{
+	int n = 1;
+
+	if (amd_l3_attrs) /* already initialized */
+		return;
+
+	if (amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
+		n += 2;
+	if (amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
+		n += 1;
+
+	amd_l3_attrs = kcalloc(n, sizeof(*amd_l3_attrs), GFP_KERNEL);
+	if (!amd_l3_attrs) {
+		pr_err("%s: failed to alloc memory!\n", __func__);
+		return;
+	}
+
+	n = 0;
+	if (amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE)) {
+		amd_l3_attrs[n++] = &dev_attr_cache_disable_0.attr;
+		amd_l3_attrs[n++] = &dev_attr_cache_disable_1.attr;
+	}
+	if (amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
+		amd_l3_attrs[n++] = &dev_attr_subcaches.attr;
+}
+
+const struct attribute **cache_get_priv_attr(struct cacheinfo *this_leaf)
+{
+	if (!this_leaf->priv || !amd_l3_attrs)
+		return NULL;
+	return (const struct attribute **)amd_l3_attrs;
+}
+
+static void amd_init_l3_cache(struct _cpuid4_info_regs *this_leaf, int index)
+{
+	int node;
+
+	/* only for L3, and not in virtualized environments */
+	if (index < 3)
+		return;
 
+	node = amd_get_nb_id(smp_processor_id());
+	this_leaf->nb = node_to_amd_nb(node);
+	if (this_leaf->nb && !this_leaf->nb->l3_cache.indices) {
+		amd_calc_l3_indices(this_leaf->nb);
+		init_amd_l3_attrs();
+	}
+}
 #else
 #define amd_init_l3_cache(x, y)
 #endif  /* CONFIG_AMD_NB && CONFIG_SYSFS */
@@ -546,7 +580,7 @@ cpuid4_cache_lookup_regs(int index, struct _cpuid4_info_regs *this_leaf)
 		cpuid_count(4, index, &eax.full, &ebx.full, &ecx.full, &edx);
 	}
 
-	if (eax.split.type == CACHE_TYPE_NULL)
+	if (eax.split.type == CTYPE_NULL)
 		return -EIO; /* better error ? */
 
 	this_leaf->eax = eax;
@@ -575,7 +609,7 @@ static int find_num_cache_leaves(struct cpuinfo_x86 *c)
 		/* Do cpuid(op) loop to find out num_cache_leaves */
 		cpuid_count(op, i, &eax, &ebx, &ecx, &edx);
 		cache_eax.full = eax;
-	} while (cache_eax.split.type != CACHE_TYPE_NULL);
+	} while (cache_eax.split.type != CTYPE_NULL);
 	return i;
 }
 
@@ -626,9 +660,9 @@ unsigned int init_intel_cacheinfo(struct cpuinfo_x86 *c)
 
 			switch (this_leaf.eax.split.level) {
 			case 1:
-				if (this_leaf.eax.split.type == CACHE_TYPE_DATA)
+				if (this_leaf.eax.split.type == CTYPE_DATA)
 					new_l1d = this_leaf.size/1024;
-				else if (this_leaf.eax.split.type == CACHE_TYPE_INST)
+				else if (this_leaf.eax.split.type == CTYPE_INST)
 					new_l1i = this_leaf.size/1024;
 				break;
 			case 2:
@@ -735,55 +769,46 @@ unsigned int init_intel_cacheinfo(struct cpuinfo_x86 *c)
 	return l2;
 }
 
-#ifdef CONFIG_SYSFS
-
-/* pointer to _cpuid4_info array (for each cache leaf) */
-static DEFINE_PER_CPU(struct _cpuid4_info *, ici_cpuid4_info);
-#define CPUID4_INFO_IDX(x, y)	(&((per_cpu(ici_cpuid4_info, x))[y]))
-
-#ifdef CONFIG_SMP
-
-static int cache_shared_amd_cpu_map_setup(unsigned int cpu, int index)
+static int __cache_amd_cpumap_setup(unsigned int cpu, int index,
+				    struct _cpuid4_info_regs *base)
 {
-	struct _cpuid4_info *this_leaf;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf;
 	int i, sibling;
 
 	if (cpu_has_topoext) {
 		unsigned int apicid, nshared, first, last;
 
-		if (!per_cpu(ici_cpuid4_info, cpu))
-			return 0;
-
-		this_leaf = CPUID4_INFO_IDX(cpu, index);
-		nshared = this_leaf->base.eax.split.num_threads_sharing + 1;
+		this_leaf = this_cpu_ci->info_list + index;
+		nshared = base->eax.split.num_threads_sharing + 1;
 		apicid = cpu_data(cpu).apicid;
 		first = apicid - (apicid % nshared);
 		last = first + nshared - 1;
 
 		for_each_online_cpu(i) {
+			this_cpu_ci = get_cpu_cacheinfo(i);
 			apicid = cpu_data(i).apicid;
 			if ((apicid < first) || (apicid > last))
 				continue;
-			if (!per_cpu(ici_cpuid4_info, i))
-				continue;
-			this_leaf = CPUID4_INFO_IDX(i, index);
+			this_leaf = this_cpu_ci->info_list + index;
 
 			for_each_online_cpu(sibling) {
 				apicid = cpu_data(sibling).apicid;
 				if ((apicid < first) || (apicid > last))
 					continue;
-				set_bit(sibling, this_leaf->shared_cpu_map);
+				cpumask_set_cpu(sibling,
+						&this_leaf->shared_cpu_map);
 			}
 		}
 	} else if (index == 3) {
 		for_each_cpu(i, cpu_llc_shared_mask(cpu)) {
-			if (!per_cpu(ici_cpuid4_info, i))
-				continue;
-			this_leaf = CPUID4_INFO_IDX(i, index);
+			this_cpu_ci = get_cpu_cacheinfo(i);
+			this_leaf = this_cpu_ci->info_list + index;
 			for_each_cpu(sibling, cpu_llc_shared_mask(cpu)) {
 				if (!cpu_online(sibling))
 					continue;
-				set_bit(sibling, this_leaf->shared_cpu_map);
+				cpumask_set_cpu(sibling,
+						&this_leaf->shared_cpu_map);
 			}
 		}
 	} else
@@ -792,459 +817,86 @@ static int cache_shared_amd_cpu_map_setup(unsigned int cpu, int index)
 	return 1;
 }
 
-static void cache_shared_cpu_map_setup(unsigned int cpu, int index)
+static void __cache_cpumap_setup(unsigned int cpu, int index,
+				 struct _cpuid4_info_regs *base)
 {
-	struct _cpuid4_info *this_leaf, *sibling_leaf;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sibling_leaf;
 	unsigned long num_threads_sharing;
 	int index_msb, i;
 	struct cpuinfo_x86 *c = &cpu_data(cpu);
 
 	if (c->x86_vendor == X86_VENDOR_AMD) {
-		if (cache_shared_amd_cpu_map_setup(cpu, index))
+		if (__cache_amd_cpumap_setup(cpu, index, base))
 			return;
 	}
 
-	this_leaf = CPUID4_INFO_IDX(cpu, index);
-	num_threads_sharing = 1 + this_leaf->base.eax.split.num_threads_sharing;
+	this_leaf = this_cpu_ci->info_list + index;
+	num_threads_sharing = 1 + base->eax.split.num_threads_sharing;
 
+	cpumask_set_cpu(cpu, &this_leaf->shared_cpu_map);
 	if (num_threads_sharing == 1)
-		cpumask_set_cpu(cpu, to_cpumask(this_leaf->shared_cpu_map));
-	else {
-		index_msb = get_count_order(num_threads_sharing);
-
-		for_each_online_cpu(i) {
-			if (cpu_data(i).apicid >> index_msb ==
-			    c->apicid >> index_msb) {
-				cpumask_set_cpu(i,
-					to_cpumask(this_leaf->shared_cpu_map));
-				if (i != cpu && per_cpu(ici_cpuid4_info, i))  {
-					sibling_leaf =
-						CPUID4_INFO_IDX(i, index);
-					cpumask_set_cpu(cpu, to_cpumask(
-						sibling_leaf->shared_cpu_map));
-				}
-			}
-		}
-	}
-}
-static void cache_remove_shared_cpu_map(unsigned int cpu, int index)
-{
-	struct _cpuid4_info	*this_leaf, *sibling_leaf;
-	int sibling;
-
-	this_leaf = CPUID4_INFO_IDX(cpu, index);
-	for_each_cpu(sibling, to_cpumask(this_leaf->shared_cpu_map)) {
-		sibling_leaf = CPUID4_INFO_IDX(sibling, index);
-		cpumask_clear_cpu(cpu,
-				  to_cpumask(sibling_leaf->shared_cpu_map));
-	}
-}
-#else
-static void cache_shared_cpu_map_setup(unsigned int cpu, int index)
-{
-}
-
-static void cache_remove_shared_cpu_map(unsigned int cpu, int index)
-{
-}
-#endif
-
-static void free_cache_attributes(unsigned int cpu)
-{
-	int i;
-
-	for (i = 0; i < num_cache_leaves; i++)
-		cache_remove_shared_cpu_map(cpu, i);
-
-	kfree(per_cpu(ici_cpuid4_info, cpu));
-	per_cpu(ici_cpuid4_info, cpu) = NULL;
-}
-
-static void get_cpu_leaves(void *_retval)
-{
-	int j, *retval = _retval, cpu = smp_processor_id();
+		return;
 
-	/* Do cpuid and store the results */
-	for (j = 0; j < num_cache_leaves; j++) {
-		struct _cpuid4_info *this_leaf = CPUID4_INFO_IDX(cpu, j);
+	index_msb = get_count_order(num_threads_sharing);
 
-		*retval = cpuid4_cache_lookup_regs(j, &this_leaf->base);
-		if (unlikely(*retval < 0)) {
-			int i;
+	for_each_online_cpu(i)
+		if (cpu_data(i).apicid >> index_msb == c->apicid >> index_msb) {
+			struct cpu_cacheinfo *sib_cpu_ci = get_cpu_cacheinfo(i);
 
-			for (i = 0; i < j; i++)
-				cache_remove_shared_cpu_map(cpu, i);
-			break;
+			if (i == cpu || !sib_cpu_ci->info_list)
+				continue;/* skip if itself or no cacheinfo */
+			sibling_leaf = sib_cpu_ci->info_list + index;
+			cpumask_set_cpu(i, &this_leaf->shared_cpu_map);
+			cpumask_set_cpu(cpu, &sibling_leaf->shared_cpu_map);
 		}
-		cache_shared_cpu_map_setup(cpu, j);
-	}
 }
 
-static int detect_cache_attributes(unsigned int cpu)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 struct _cpuid4_info_regs *base)
 {
-	int			retval;
-
-	if (num_cache_leaves == 0)
-		return -ENOENT;
-
-	per_cpu(ici_cpuid4_info, cpu) = kzalloc(
-	    sizeof(struct _cpuid4_info) * num_cache_leaves, GFP_KERNEL);
-	if (per_cpu(ici_cpuid4_info, cpu) == NULL)
-		return -ENOMEM;
-
-	smp_call_function_single(cpu, get_cpu_leaves, &retval, true);
-	if (retval) {
-		kfree(per_cpu(ici_cpuid4_info, cpu));
-		per_cpu(ici_cpuid4_info, cpu) = NULL;
-	}
-
-	return retval;
+	this_leaf->level = base->eax.split.level;
+	this_leaf->type = cache_type_map[base->eax.split.type];
+	this_leaf->coherency_line_size =
+				base->ebx.split.coherency_line_size + 1;
+	this_leaf->ways_of_associativity =
+				base->ebx.split.ways_of_associativity + 1;
+	this_leaf->size = base->size;
+	this_leaf->number_of_sets = base->ecx.split.number_of_sets + 1;
+	this_leaf->physical_line_partition =
+				base->ebx.split.physical_line_partition + 1;
+	this_leaf->priv = base->nb;
 }
 
-#include <linux/kobject.h>
-#include <linux/sysfs.h>
-#include <linux/cpu.h>
-
-/* pointer to kobject for cpuX/cache */
-static DEFINE_PER_CPU(struct kobject *, ici_cache_kobject);
-
-struct _index_kobject {
-	struct kobject kobj;
-	unsigned int cpu;
-	unsigned short index;
-};
-
-/* pointer to array of kobjects for cpuX/cache/indexY */
-static DEFINE_PER_CPU(struct _index_kobject *, ici_index_kobject);
-#define INDEX_KOBJECT_PTR(x, y)		(&((per_cpu(ici_index_kobject, x))[y]))
-
-#define show_one_plus(file_name, object, val)				\
-static ssize_t show_##file_name(struct _cpuid4_info *this_leaf, char *buf, \
-				unsigned int cpu)			\
-{									\
-	return sprintf(buf, "%lu\n", (unsigned long)this_leaf->object + val); \
-}
-
-show_one_plus(level, base.eax.split.level, 0);
-show_one_plus(coherency_line_size, base.ebx.split.coherency_line_size, 1);
-show_one_plus(physical_line_partition, base.ebx.split.physical_line_partition, 1);
-show_one_plus(ways_of_associativity, base.ebx.split.ways_of_associativity, 1);
-show_one_plus(number_of_sets, base.ecx.split.number_of_sets, 1);
-
-static ssize_t show_size(struct _cpuid4_info *this_leaf, char *buf,
-			 unsigned int cpu)
-{
-	return sprintf(buf, "%luK\n", this_leaf->base.size / 1024);
-}
-
-static ssize_t show_shared_cpu_map_func(struct _cpuid4_info *this_leaf,
-					int type, char *buf)
-{
-	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf;
-	int n = 0;
-
-	if (len > 1) {
-		const struct cpumask *mask;
-
-		mask = to_cpumask(this_leaf->shared_cpu_map);
-		n = type ?
-			cpulist_scnprintf(buf, len-2, mask) :
-			cpumask_scnprintf(buf, len-2, mask);
-		buf[n++] = '\n';
-		buf[n] = '\0';
-	}
-	return n;
-}
-
-static inline ssize_t show_shared_cpu_map(struct _cpuid4_info *leaf, char *buf,
-					  unsigned int cpu)
-{
-	return show_shared_cpu_map_func(leaf, 0, buf);
-}
-
-static inline ssize_t show_shared_cpu_list(struct _cpuid4_info *leaf, char *buf,
-					   unsigned int cpu)
-{
-	return show_shared_cpu_map_func(leaf, 1, buf);
-}
-
-static ssize_t show_type(struct _cpuid4_info *this_leaf, char *buf,
-			 unsigned int cpu)
-{
-	switch (this_leaf->base.eax.split.type) {
-	case CACHE_TYPE_DATA:
-		return sprintf(buf, "Data\n");
-	case CACHE_TYPE_INST:
-		return sprintf(buf, "Instruction\n");
-	case CACHE_TYPE_UNIFIED:
-		return sprintf(buf, "Unified\n");
-	default:
-		return sprintf(buf, "Unknown\n");
-	}
-}
-
-#define to_object(k)	container_of(k, struct _index_kobject, kobj)
-#define to_attr(a)	container_of(a, struct _cache_attr, attr)
-
-#define define_one_ro(_name) \
-static struct _cache_attr _name = \
-	__ATTR(_name, 0444, show_##_name, NULL)
-
-define_one_ro(level);
-define_one_ro(type);
-define_one_ro(coherency_line_size);
-define_one_ro(physical_line_partition);
-define_one_ro(ways_of_associativity);
-define_one_ro(number_of_sets);
-define_one_ro(size);
-define_one_ro(shared_cpu_map);
-define_one_ro(shared_cpu_list);
-
-static struct attribute *default_attrs[] = {
-	&type.attr,
-	&level.attr,
-	&coherency_line_size.attr,
-	&physical_line_partition.attr,
-	&ways_of_associativity.attr,
-	&number_of_sets.attr,
-	&size.attr,
-	&shared_cpu_map.attr,
-	&shared_cpu_list.attr,
-	NULL
-};
-
-#ifdef CONFIG_AMD_NB
-static struct attribute **amd_l3_attrs(void)
+static int __init_cache_level(unsigned int cpu)
 {
-	static struct attribute **attrs;
-	int n;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
 
-	if (attrs)
-		return attrs;
-
-	n = ARRAY_SIZE(default_attrs);
-
-	if (amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
-		n += 2;
-
-	if (amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
-		n += 1;
-
-	attrs = kzalloc(n * sizeof (struct attribute *), GFP_KERNEL);
-	if (attrs == NULL)
-		return attrs = default_attrs;
-
-	for (n = 0; default_attrs[n]; n++)
-		attrs[n] = default_attrs[n];
-
-	if (amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE)) {
-		attrs[n++] = &cache_disable_0.attr;
-		attrs[n++] = &cache_disable_1.attr;
-	}
-
-	if (amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
-		attrs[n++] = &subcaches.attr;
-
-	return attrs;
-}
-#endif
-
-static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
-{
-	struct _cache_attr *fattr = to_attr(attr);
-	struct _index_kobject *this_leaf = to_object(kobj);
-	ssize_t ret;
-
-	ret = fattr->show ?
-		fattr->show(CPUID4_INFO_IDX(this_leaf->cpu, this_leaf->index),
-			buf, this_leaf->cpu) :
-		0;
-	return ret;
-}
-
-static ssize_t store(struct kobject *kobj, struct attribute *attr,
-		     const char *buf, size_t count)
-{
-	struct _cache_attr *fattr = to_attr(attr);
-	struct _index_kobject *this_leaf = to_object(kobj);
-	ssize_t ret;
-
-	ret = fattr->store ?
-		fattr->store(CPUID4_INFO_IDX(this_leaf->cpu, this_leaf->index),
-			buf, count, this_leaf->cpu) :
-		0;
-	return ret;
-}
-
-static const struct sysfs_ops sysfs_ops = {
-	.show   = show,
-	.store  = store,
-};
-
-static struct kobj_type ktype_cache = {
-	.sysfs_ops	= &sysfs_ops,
-	.default_attrs	= default_attrs,
-};
-
-static struct kobj_type ktype_percpu_entry = {
-	.sysfs_ops	= &sysfs_ops,
-};
-
-static void cpuid4_cache_sysfs_exit(unsigned int cpu)
-{
-	kfree(per_cpu(ici_cache_kobject, cpu));
-	kfree(per_cpu(ici_index_kobject, cpu));
-	per_cpu(ici_cache_kobject, cpu) = NULL;
-	per_cpu(ici_index_kobject, cpu) = NULL;
-	free_cache_attributes(cpu);
-}
-
-static int cpuid4_cache_sysfs_init(unsigned int cpu)
-{
-	int err;
-
-	if (num_cache_leaves == 0)
+	if (!num_cache_leaves)
 		return -ENOENT;
-
-	err = detect_cache_attributes(cpu);
-	if (err)
-		return err;
-
-	/* Allocate all required memory */
-	per_cpu(ici_cache_kobject, cpu) =
-		kzalloc(sizeof(struct kobject), GFP_KERNEL);
-	if (unlikely(per_cpu(ici_cache_kobject, cpu) == NULL))
-		goto err_out;
-
-	per_cpu(ici_index_kobject, cpu) = kzalloc(
-	    sizeof(struct _index_kobject) * num_cache_leaves, GFP_KERNEL);
-	if (unlikely(per_cpu(ici_index_kobject, cpu) == NULL))
-		goto err_out;
-
+	if (!this_cpu_ci)
+		return -EINVAL;
+	this_cpu_ci->num_levels = 3;
+	this_cpu_ci->num_leaves = num_cache_leaves;
 	return 0;
-
-err_out:
-	cpuid4_cache_sysfs_exit(cpu);
-	return -ENOMEM;
 }
 
-static DECLARE_BITMAP(cache_dev_map, NR_CPUS);
-
-/* Add/Remove cache interface for CPU device */
-static int cache_add_dev(struct device *dev)
+static int __populate_cache_leaves(unsigned int cpu)
 {
-	unsigned int cpu = dev->id;
-	unsigned long i, j;
-	struct _index_kobject *this_object;
-	struct _cpuid4_info   *this_leaf;
-	int retval;
-
-	retval = cpuid4_cache_sysfs_init(cpu);
-	if (unlikely(retval < 0))
-		return retval;
-
-	retval = kobject_init_and_add(per_cpu(ici_cache_kobject, cpu),
-				      &ktype_percpu_entry,
-				      &dev->kobj, "%s", "cache");
-	if (retval < 0) {
-		cpuid4_cache_sysfs_exit(cpu);
-		return retval;
-	}
+	unsigned int idx, ret;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+	struct _cpuid4_info_regs id4_regs = {};
 
-	for (i = 0; i < num_cache_leaves; i++) {
-		this_object = INDEX_KOBJECT_PTR(cpu, i);
-		this_object->cpu = cpu;
-		this_object->index = i;
-
-		this_leaf = CPUID4_INFO_IDX(cpu, i);
-
-		ktype_cache.default_attrs = default_attrs;
-#ifdef CONFIG_AMD_NB
-		if (this_leaf->base.nb)
-			ktype_cache.default_attrs = amd_l3_attrs();
-#endif
-		retval = kobject_init_and_add(&(this_object->kobj),
-					      &ktype_cache,
-					      per_cpu(ici_cache_kobject, cpu),
-					      "index%1lu", i);
-		if (unlikely(retval)) {
-			for (j = 0; j < i; j++)
-				kobject_put(&(INDEX_KOBJECT_PTR(cpu, j)->kobj));
-			kobject_put(per_cpu(ici_cache_kobject, cpu));
-			cpuid4_cache_sysfs_exit(cpu);
-			return retval;
-		}
-		kobject_uevent(&(this_object->kobj), KOBJ_ADD);
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++) {
+		ret = cpuid4_cache_lookup_regs(idx, &id4_regs);
+		if (ret)
+			return ret;
+		ci_leaf_init(this_leaf++, &id4_regs);
+		__cache_cpumap_setup(cpu, idx, &id4_regs);
 	}
-	cpumask_set_cpu(cpu, to_cpumask(cache_dev_map));
-
-	kobject_uevent(per_cpu(ici_cache_kobject, cpu), KOBJ_ADD);
 	return 0;
 }
 
-static void cache_remove_dev(struct device *dev)
-{
-	unsigned int cpu = dev->id;
-	unsigned long i;
-
-	if (per_cpu(ici_cpuid4_info, cpu) == NULL)
-		return;
-	if (!cpumask_test_cpu(cpu, to_cpumask(cache_dev_map)))
-		return;
-	cpumask_clear_cpu(cpu, to_cpumask(cache_dev_map));
-
-	for (i = 0; i < num_cache_leaves; i++)
-		kobject_put(&(INDEX_KOBJECT_PTR(cpu, i)->kobj));
-	kobject_put(per_cpu(ici_cache_kobject, cpu));
-	cpuid4_cache_sysfs_exit(cpu);
-}
-
-static int cacheinfo_cpu_callback(struct notifier_block *nfb,
-				  unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (unsigned long)hcpu;
-	struct device *dev;
-
-	dev = get_cpu_device(cpu);
-	switch (action) {
-	case CPU_ONLINE:
-	case CPU_ONLINE_FROZEN:
-		cache_add_dev(dev);
-		break;
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		cache_remove_dev(dev);
-		break;
-	}
-	return NOTIFY_OK;
-}
-
-static struct notifier_block cacheinfo_cpu_notifier = {
-	.notifier_call = cacheinfo_cpu_callback,
-};
-
-static int __init cache_sysfs_init(void)
-{
-	int i, err = 0;
-
-	if (num_cache_leaves == 0)
-		return 0;
-
-	cpu_notifier_register_begin();
-	for_each_online_cpu(i) {
-		struct device *dev = get_cpu_device(i);
-
-		err = cache_add_dev(dev);
-		if (err)
-			goto out;
-	}
-	__register_hotcpu_notifier(&cacheinfo_cpu_notifier);
-
-out:
-	cpu_notifier_register_done();
-	return err;
-}
-
-device_initcall(cache_sysfs_init);
-
-#endif
+DEFINE_SMP_CALL_FUNCTION(init_cache_level)
+DEFINE_SMP_CALL_FUNCTION(populate_cache_leaves)
-- 
1.8.3.2

[PATCH v2 6/9] powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: linuxppc-dev@lists.ozlabs.org
---
 arch/powerpc/kernel/cacheinfo.c | 813 +++++-----------------------------------
 arch/powerpc/kernel/cacheinfo.h |   8 -
 arch/powerpc/kernel/sysfs.c     |  12 +-
 3 files changed, 91 insertions(+), 742 deletions(-)
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h

diff --git a/arch/powerpc/kernel/cacheinfo.c b/arch/powerpc/kernel/cacheinfo.c
index 40198d50b4c2..b871c246d945 100644
--- a/arch/powerpc/kernel/cacheinfo.c
+++ b/arch/powerpc/kernel/cacheinfo.c
@@ -10,38 +10,10 @@
  * 2 as published by the Free Software Foundation.
  */
 
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
-#include <linux/cpumask.h>
 #include <linux/kernel.h>
-#include <linux/kobject.h>
-#include <linux/list.h>
-#include <linux/notifier.h>
 #include <linux/of.h>
-#include <linux/percpu.h>
-#include <linux/slab.h>
-#include <asm/prom.h>
-
-#include "cacheinfo.h"
-
-/* per-cpu object for tracking:
- * - a "cache" kobject for the top-level directory
- * - a list of "index" objects representing the cpu's local cache hierarchy
- */
-struct cache_dir {
-	struct kobject *kobj; /* bare (not embedded) kobject for cache
-			       * directory */
-	struct cache_index_dir *index; /* list of index objects */
-};
-
-/* "index" object: each cpu's cache directory has an index
- * subdirectory corresponding to a cache object associated with the
- * cpu.  This object's lifetime is managed via the embedded kobject.
- */
-struct cache_index_dir {
-	struct kobject kobj;
-	struct cache_index_dir *next; /* next index in parent directory */
-	struct cache *cache;
-};
 
 /* Template for determining which OF properties to query for a given
  * cache type */
@@ -60,11 +32,6 @@ struct cache_type_info {
 	const char *nr_sets_prop;
 };
 
-/* These are used to index the cache_type_info array. */
-#define CACHE_TYPE_UNIFIED     0
-#define CACHE_TYPE_INSTRUCTION 1
-#define CACHE_TYPE_DATA        2
-
 static const struct cache_type_info cache_type_info[] = {
 	{
 		/* PowerPC Processor binding says the [di]-cache-*
@@ -92,231 +59,83 @@ static const struct cache_type_info cache_type_info[] = {
 	},
 };
 
-/* Cache object: each instance of this corresponds to a distinct cache
- * in the system.  There are separate objects for Harvard caches: one
- * each for instruction and data, and each refers to the same OF node.
- * The refcount of the OF node is elevated for the lifetime of the
- * cache object.  A cache object is released when its shared_cpu_map
- * is cleared (see cache_cpu_clear).
- *
- * A cache object is on two lists: an unsorted global list
- * (cache_list) of cache objects; and a singly-linked list
- * representing the local cache hierarchy, which is ordered by level
- * (e.g. L1d -> L1i -> L2 -> L3).
- */
-struct cache {
-	struct device_node *ofnode;    /* OF node for this cache, may be cpu */
-	struct cpumask shared_cpu_map; /* online CPUs using this cache */
-	int type;                      /* split cache disambiguation */
-	int level;                     /* level not explicit in device tree */
-	struct list_head list;         /* global list of cache objects */
-	struct cache *next_local;      /* next cache of >= level */
-};
-
-static DEFINE_PER_CPU(struct cache_dir *, cache_dir_pcpu);
-
-/* traversal/modification of this list occurs only at cpu hotplug time;
- * access is serialized by cpu hotplug locking
- */
-static LIST_HEAD(cache_list);
-
-static struct cache_index_dir *kobj_to_cache_index_dir(struct kobject *k)
-{
-	return container_of(k, struct cache_index_dir, kobj);
-}
-
-static const char *cache_type_string(const struct cache *cache)
+static inline int get_cacheinfo_idx(enum cache_type type)
 {
-	return cache_type_info[cache->type].name;
-}
-
-static void cache_init(struct cache *cache, int type, int level,
-		       struct device_node *ofnode)
-{
-	cache->type = type;
-	cache->level = level;
-	cache->ofnode = of_node_get(ofnode);
-	INIT_LIST_HEAD(&cache->list);
-	list_add(&cache->list, &cache_list);
-}
-
-static struct cache *new_cache(int type, int level, struct device_node *ofnode)
-{
-	struct cache *cache;
-
-	cache = kzalloc(sizeof(*cache), GFP_KERNEL);
-	if (cache)
-		cache_init(cache, type, level, ofnode);
-
-	return cache;
-}
-
-static void release_cache_debugcheck(struct cache *cache)
-{
-	struct cache *iter;
-
-	list_for_each_entry(iter, &cache_list, list)
-		WARN_ONCE(iter->next_local == cache,
-			  "cache for %s(%s) refers to cache for %s(%s)\n",
-			  iter->ofnode->full_name,
-			  cache_type_string(iter),
-			  cache->ofnode->full_name,
-			  cache_type_string(cache));
-}
-
-static void release_cache(struct cache *cache)
-{
-	if (!cache)
-		return;
-
-	pr_debug("freeing L%d %s cache for %s\n", cache->level,
-		 cache_type_string(cache), cache->ofnode->full_name);
-
-	release_cache_debugcheck(cache);
-	list_del(&cache->list);
-	of_node_put(cache->ofnode);
-	kfree(cache);
-}
-
-static void cache_cpu_set(struct cache *cache, int cpu)
-{
-	struct cache *next = cache;
-
-	while (next) {
-		WARN_ONCE(cpumask_test_cpu(cpu, &next->shared_cpu_map),
-			  "CPU %i already accounted in %s(%s)\n",
-			  cpu, next->ofnode->full_name,
-			  cache_type_string(next));
-		cpumask_set_cpu(cpu, &next->shared_cpu_map);
-		next = next->next_local;
-	}
+	if (type == CACHE_TYPE_UNIFIED)
+		return 0;
+	else
+		return type;
 }
 
-static int cache_size(const struct cache *cache, unsigned int *ret)
+static void cache_size(struct cacheinfo *this_leaf)
 {
 	const char *propname;
 	const __be32 *cache_size;
+	int ct_idx;
 
-	propname = cache_type_info[cache->type].size_prop;
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	propname = cache_type_info[ct_idx].size_prop;
 
-	cache_size = of_get_property(cache->ofnode, propname, NULL);
+	cache_size = of_get_property(this_leaf->of_node, propname, NULL);
 	if (!cache_size)
-		return -ENODEV;
-
-	*ret = of_read_number(cache_size, 1);
-	return 0;
-}
-
-static int cache_size_kb(const struct cache *cache, unsigned int *ret)
-{
-	unsigned int size;
-
-	if (cache_size(cache, &size))
-		return -ENODEV;
-
-	*ret = size / 1024;
-	return 0;
+		this_leaf->size = 0;
+	else
+		this_leaf->size = of_read_number(cache_size, 1);
 }
 
 /* not cache_line_size() because that's a macro in include/linux/cache.h */
-static int cache_get_line_size(const struct cache *cache, unsigned int *ret)
+static void cache_get_line_size(struct cacheinfo *this_leaf)
 {
 	const __be32 *line_size;
-	int i, lim;
+	int i, lim, ct_idx;
 
-	lim = ARRAY_SIZE(cache_type_info[cache->type].line_size_props);
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	lim = ARRAY_SIZE(cache_type_info[ct_idx].line_size_props);
 
 	for (i = 0; i < lim; i++) {
 		const char *propname;
 
-		propname = cache_type_info[cache->type].line_size_props[i];
-		line_size = of_get_property(cache->ofnode, propname, NULL);
+		propname = cache_type_info[ct_idx].line_size_props[i];
+		line_size = of_get_property(this_leaf->of_node, propname, NULL);
 		if (line_size)
 			break;
 	}
 
 	if (!line_size)
-		return -ENODEV;
-
-	*ret = of_read_number(line_size, 1);
-	return 0;
+		this_leaf->coherency_line_size = 0;
+	else
+		this_leaf->coherency_line_size = of_read_number(line_size, 1);
 }
 
-static int cache_nr_sets(const struct cache *cache, unsigned int *ret)
+static void cache_nr_sets(struct cacheinfo *this_leaf)
 {
 	const char *propname;
 	const __be32 *nr_sets;
+	int ct_idx;
 
-	propname = cache_type_info[cache->type].nr_sets_prop;
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	propname = cache_type_info[ct_idx].nr_sets_prop;
 
-	nr_sets = of_get_property(cache->ofnode, propname, NULL);
+	nr_sets = of_get_property(this_leaf->of_node, propname, NULL);
 	if (!nr_sets)
-		return -ENODEV;
-
-	*ret = of_read_number(nr_sets, 1);
-	return 0;
+		this_leaf->number_of_sets = 0;
+	else
+		this_leaf->number_of_sets = of_read_number(nr_sets, 1);
 }
 
-static int cache_associativity(const struct cache *cache, unsigned int *ret)
+static void cache_associativity(struct cacheinfo *this_leaf)
 {
-	unsigned int line_size;
-	unsigned int nr_sets;
-	unsigned int size;
-
-	if (cache_nr_sets(cache, &nr_sets))
-		goto err;
+	unsigned int line_size = this_leaf->coherency_line_size;
+	unsigned int nr_sets = this_leaf->number_of_sets;
+	unsigned int size = this_leaf->size;
 
 	/* If the cache is fully associative, there is no need to
 	 * check the other properties.
 	 */
-	if (nr_sets == 1) {
-		*ret = 0;
-		return 0;
-	}
-
-	if (cache_get_line_size(cache, &line_size))
-		goto err;
-	if (cache_size(cache, &size))
-		goto err;
-
-	if (!(nr_sets > 0 && size > 0 && line_size > 0))
-		goto err;
-
-	*ret = (size / nr_sets) / line_size;
-	return 0;
-err:
-	return -ENODEV;
-}
-
-/* helper for dealing with split caches */
-static struct cache *cache_find_first_sibling(struct cache *cache)
-{
-	struct cache *iter;
-
-	if (cache->type == CACHE_TYPE_UNIFIED)
-		return cache;
-
-	list_for_each_entry(iter, &cache_list, list)
-		if (iter->ofnode == cache->ofnode && iter->next_local == cache)
-			return iter;
-
-	return cache;
-}
-
-/* return the first cache on a local list matching node */
-static struct cache *cache_lookup_by_node(const struct device_node *node)
-{
-	struct cache *cache = NULL;
-	struct cache *iter;
-
-	list_for_each_entry(iter, &cache_list, list) {
-		if (iter->ofnode != node)
-			continue;
-		cache = cache_find_first_sibling(iter);
-		break;
-	}
-
-	return cache;
+	if ((nr_sets == 1) || !(nr_sets > 0 && size > 0 && line_size > 0))
+		this_leaf->ways_of_associativity = 0;
+	else
+		this_leaf->ways_of_associativity = (size / nr_sets) / line_size;
 }
 
 static bool cache_node_is_unified(const struct device_node *np)
@@ -324,526 +143,74 @@ static bool cache_node_is_unified(const struct device_node *np)
 	return of_get_property(np, "cache-unified", NULL);
 }
 
-static struct cache *cache_do_one_devnode_unified(struct device_node *node,
-						  int level)
-{
-	struct cache *cache;
-
-	pr_debug("creating L%d ucache for %s\n", level, node->full_name);
-
-	cache = new_cache(CACHE_TYPE_UNIFIED, level, node);
-
-	return cache;
-}
-
-static struct cache *cache_do_one_devnode_split(struct device_node *node,
-						int level)
-{
-	struct cache *dcache, *icache;
-
-	pr_debug("creating L%d dcache and icache for %s\n", level,
-		 node->full_name);
-
-	dcache = new_cache(CACHE_TYPE_DATA, level, node);
-	icache = new_cache(CACHE_TYPE_INSTRUCTION, level, node);
-
-	if (!dcache || !icache)
-		goto err;
-
-	dcache->next_local = icache;
-
-	return dcache;
-err:
-	release_cache(dcache);
-	release_cache(icache);
-	return NULL;
-}
-
-static struct cache *cache_do_one_devnode(struct device_node *node, int level)
-{
-	struct cache *cache;
-
-	if (cache_node_is_unified(node))
-		cache = cache_do_one_devnode_unified(node, level);
-	else
-		cache = cache_do_one_devnode_split(node, level);
-
-	return cache;
-}
-
-static struct cache *cache_lookup_or_instantiate(struct device_node *node,
-						 int level)
-{
-	struct cache *cache;
-
-	cache = cache_lookup_by_node(node);
-
-	WARN_ONCE(cache && cache->level != level,
-		  "cache level mismatch on lookup (got %d, expected %d)\n",
-		  cache->level, level);
-
-	if (!cache)
-		cache = cache_do_one_devnode(node, level);
-
-	return cache;
-}
-
-static void link_cache_lists(struct cache *smaller, struct cache *bigger)
-{
-	while (smaller->next_local) {
-		if (smaller->next_local == bigger)
-			return; /* already linked */
-		smaller = smaller->next_local;
-	}
-
-	smaller->next_local = bigger;
-}
-
-static void do_subsidiary_caches_debugcheck(struct cache *cache)
-{
-	WARN_ON_ONCE(cache->level != 1);
-	WARN_ON_ONCE(strcmp(cache->ofnode->type, "cpu"));
-}
-
-static void do_subsidiary_caches(struct cache *cache)
-{
-	struct device_node *subcache_node;
-	int level = cache->level;
-
-	do_subsidiary_caches_debugcheck(cache);
-
-	while ((subcache_node = of_find_next_cache_node(cache->ofnode))) {
-		struct cache *subcache;
-
-		level++;
-		subcache = cache_lookup_or_instantiate(subcache_node, level);
-		of_node_put(subcache_node);
-		if (!subcache)
-			break;
-
-		link_cache_lists(cache, subcache);
-		cache = subcache;
-	}
-}
-
-static struct cache *cache_chain_instantiate(unsigned int cpu_id)
-{
-	struct device_node *cpu_node;
-	struct cache *cpu_cache = NULL;
-
-	pr_debug("creating cache object(s) for CPU %i\n", cpu_id);
-
-	cpu_node = of_get_cpu_node(cpu_id, NULL);
-	WARN_ONCE(!cpu_node, "no OF node found for CPU %i\n", cpu_id);
-	if (!cpu_node)
-		goto out;
-
-	cpu_cache = cache_lookup_or_instantiate(cpu_node, 1);
-	if (!cpu_cache)
-		goto out;
-
-	do_subsidiary_caches(cpu_cache);
-
-	cache_cpu_set(cpu_cache, cpu_id);
-out:
-	of_node_put(cpu_node);
-
-	return cpu_cache;
-}
-
-static struct cache_dir *cacheinfo_create_cache_dir(unsigned int cpu_id)
-{
-	struct cache_dir *cache_dir;
-	struct device *dev;
-	struct kobject *kobj = NULL;
-
-	dev = get_cpu_device(cpu_id);
-	WARN_ONCE(!dev, "no dev for CPU %i\n", cpu_id);
-	if (!dev)
-		goto err;
-
-	kobj = kobject_create_and_add("cache", &dev->kobj);
-	if (!kobj)
-		goto err;
-
-	cache_dir = kzalloc(sizeof(*cache_dir), GFP_KERNEL);
-	if (!cache_dir)
-		goto err;
-
-	cache_dir->kobj = kobj;
-
-	WARN_ON_ONCE(per_cpu(cache_dir_pcpu, cpu_id) != NULL);
-
-	per_cpu(cache_dir_pcpu, cpu_id) = cache_dir;
-
-	return cache_dir;
-err:
-	kobject_put(kobj);
-	return NULL;
-}
-
-static void cache_index_release(struct kobject *kobj)
-{
-	struct cache_index_dir *index;
-
-	index = kobj_to_cache_index_dir(kobj);
-
-	pr_debug("freeing index directory for L%d %s cache\n",
-		 index->cache->level, cache_type_string(index->cache));
-
-	kfree(index);
-}
-
-static ssize_t cache_index_show(struct kobject *k, struct attribute *attr, char *buf)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
 {
-	struct kobj_attribute *kobj_attr;
-
-	kobj_attr = container_of(attr, struct kobj_attribute, attr);
-
-	return kobj_attr->show(k, kobj_attr, buf);
+	this_leaf->level = level;
+	this_leaf->type = type;
+	cache_size(this_leaf);
+	cache_get_line_size(this_leaf);
+	cache_nr_sets(this_leaf);
+	cache_associativity(this_leaf);
 }
 
-static struct cache *index_kobj_to_cache(struct kobject *k)
+int init_cache_level(unsigned int cpu)
 {
-	struct cache_index_dir *index;
+	struct device_node *np;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int level = 0, leaves = 0;
 
-	index = kobj_to_cache_index_dir(k);
-
-	return index->cache;
-}
-
-static ssize_t size_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int size_kb;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_size_kb(cache, &size_kb))
+	if (!cpu_dev) {
+		pr_err("No cpu device for CPU %d\n", cpu);
 		return -ENODEV;
-
-	return sprintf(buf, "%uK\n", size_kb);
-}
-
-static struct kobj_attribute cache_size_attr =
-	__ATTR(size, 0444, size_show, NULL);
-
-
-static ssize_t line_size_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int line_size;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_get_line_size(cache, &line_size))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", line_size);
-}
-
-static struct kobj_attribute cache_line_size_attr =
-	__ATTR(coherency_line_size, 0444, line_size_show, NULL);
-
-static ssize_t nr_sets_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int nr_sets;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_nr_sets(cache, &nr_sets))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", nr_sets);
-}
-
-static struct kobj_attribute cache_nr_sets_attr =
-	__ATTR(number_of_sets, 0444, nr_sets_show, NULL);
-
-static ssize_t associativity_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int associativity;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_associativity(cache, &associativity))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", associativity);
-}
-
-static struct kobj_attribute cache_assoc_attr =
-	__ATTR(ways_of_associativity, 0444, associativity_show, NULL);
-
-static ssize_t type_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	return sprintf(buf, "%s\n", cache_type_string(cache));
-}
-
-static struct kobj_attribute cache_type_attr =
-	__ATTR(type, 0444, type_show, NULL);
-
-static ssize_t level_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache_index_dir *index;
-	struct cache *cache;
-
-	index = kobj_to_cache_index_dir(k);
-	cache = index->cache;
-
-	return sprintf(buf, "%d\n", cache->level);
-}
-
-static struct kobj_attribute cache_level_attr =
-	__ATTR(level, 0444, level_show, NULL);
-
-static ssize_t shared_cpu_map_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache_index_dir *index;
-	struct cache *cache;
-	int len;
-	int n = 0;
-
-	index = kobj_to_cache_index_dir(k);
-	cache = index->cache;
-	len = PAGE_SIZE - 2;
-
-	if (len > 1) {
-		n = cpumask_scnprintf(buf, len, &cache->shared_cpu_map);
-		buf[n++] = '\n';
-		buf[n] = '\0';
 	}
-	return n;
-}
-
-static struct kobj_attribute cache_shared_cpu_map_attr =
-	__ATTR(shared_cpu_map, 0444, shared_cpu_map_show, NULL);
-
-/* Attributes which should always be created -- the kobject/sysfs core
- * does this automatically via kobj_type->default_attrs.  This is the
- * minimum data required to uniquely identify a cache.
- */
-static struct attribute *cache_index_default_attrs[] = {
-	&cache_type_attr.attr,
-	&cache_level_attr.attr,
-	&cache_shared_cpu_map_attr.attr,
-	NULL,
-};
-
-/* Attributes which should be created if the cache device node has the
- * right properties -- see cacheinfo_create_index_opt_attrs
- */
-static struct kobj_attribute *cache_index_opt_attrs[] = {
-	&cache_size_attr,
-	&cache_line_size_attr,
-	&cache_nr_sets_attr,
-	&cache_assoc_attr,
-};
-
-static const struct sysfs_ops cache_index_ops = {
-	.show = cache_index_show,
-};
-
-static struct kobj_type cache_index_type = {
-	.release = cache_index_release,
-	.sysfs_ops = &cache_index_ops,
-	.default_attrs = cache_index_default_attrs,
-};
-
-static void cacheinfo_create_index_opt_attrs(struct cache_index_dir *dir)
-{
-	const char *cache_name;
-	const char *cache_type;
-	struct cache *cache;
-	char *buf;
-	int i;
-
-	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
-	if (!buf)
-		return;
-
-	cache = dir->cache;
-	cache_name = cache->ofnode->full_name;
-	cache_type = cache_type_string(cache);
-
-	/* We don't want to create an attribute that can't provide a
-	 * meaningful value.  Check the return value of each optional
-	 * attribute's ->show method before registering the
-	 * attribute.
-	 */
-	for (i = 0; i < ARRAY_SIZE(cache_index_opt_attrs); i++) {
-		struct kobj_attribute *attr;
-		ssize_t rc;
-
-		attr = cache_index_opt_attrs[i];
-
-		rc = attr->show(&dir->kobj, attr, buf);
-		if (rc <= 0) {
-			pr_debug("not creating %s attribute for "
-				 "%s(%s) (rc = %zd)\n",
-				 attr->attr.name, cache_name,
-				 cache_type, rc);
-			continue;
-		}
-		if (sysfs_create_file(&dir->kobj, &attr->attr))
-			pr_debug("could not create %s attribute for %s(%s)\n",
-				 attr->attr.name, cache_name, cache_type);
+	np = cpu_dev->of_node;
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
 	}
 
-	kfree(buf);
-}
-
-static void cacheinfo_create_index_dir(struct cache *cache, int index,
-				       struct cache_dir *cache_dir)
-{
-	struct cache_index_dir *index_dir;
-	int rc;
-
-	index_dir = kzalloc(sizeof(*index_dir), GFP_KERNEL);
-	if (!index_dir)
-		goto err;
-
-	index_dir->cache = cache;
-
-	rc = kobject_init_and_add(&index_dir->kobj, &cache_index_type,
-				  cache_dir->kobj, "index%d", index);
-	if (rc)
-		goto err;
-
-	index_dir->next = cache_dir->index;
-	cache_dir->index = index_dir;
-
-	cacheinfo_create_index_opt_attrs(index_dir);
-
-	return;
-err:
-	kfree(index_dir);
-}
-
-static void cacheinfo_sysfs_populate(unsigned int cpu_id,
-				     struct cache *cache_list)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-	int index = 0;
-
-	cache_dir = cacheinfo_create_cache_dir(cpu_id);
-	if (!cache_dir)
-		return;
-
-	cache = cache_list;
-	while (cache) {
-		cacheinfo_create_index_dir(cache, index, cache_dir);
-		index++;
-		cache = cache->next_local;
+	while (np) {
+		leaves += cache_node_is_unified(np) ? 1 : 2;
+		level++;
+		of_node_put(np);
+		np = of_find_next_cache_node(np);
 	}
-}
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
 
-void cacheinfo_cpu_online(unsigned int cpu_id)
-{
-	struct cache *cache;
-
-	cache = cache_chain_instantiate(cpu_id);
-	if (!cache)
-		return;
-
-	cacheinfo_sysfs_populate(cpu_id, cache);
-}
-
-/* functions needed to remove cache entry for cpu offline or suspend/resume */
-
-#if (defined(CONFIG_PPC_PSERIES) && defined(CONFIG_SUSPEND)) || \
-    defined(CONFIG_HOTPLUG_CPU)
-
-static struct cache *cache_lookup_by_cpu(unsigned int cpu_id)
-{
-	struct device_node *cpu_node;
-	struct cache *cache;
-
-	cpu_node = of_get_cpu_node(cpu_id, NULL);
-	WARN_ONCE(!cpu_node, "no OF node found for CPU %i\n", cpu_id);
-	if (!cpu_node)
-		return NULL;
-
-	cache = cache_lookup_by_node(cpu_node);
-	of_node_put(cpu_node);
-
-	return cache;
+	return 0;
 }
 
-static void remove_index_dirs(struct cache_dir *cache_dir)
+int populate_cache_leaves(unsigned int cpu)
 {
-	struct cache_index_dir *index;
-
-	index = cache_dir->index;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct device_node *np;
+	unsigned int level, idx;
 
-	while (index) {
-		struct cache_index_dir *next;
-
-		next = index->next;
-		kobject_put(&index->kobj);
-		index = next;
+	np = of_node_get(cpu_dev->of_node);
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
 	}
-}
-
-static void remove_cache_dir(struct cache_dir *cache_dir)
-{
-	remove_index_dirs(cache_dir);
 
-	/* Remove cache dir from sysfs */
-	kobject_del(cache_dir->kobj);
-
-	kobject_put(cache_dir->kobj);
-
-	kfree(cache_dir);
-}
-
-static void cache_cpu_clear(struct cache *cache, int cpu)
-{
-	while (cache) {
-		struct cache *next = cache->next_local;
-
-		WARN_ONCE(!cpumask_test_cpu(cpu, &cache->shared_cpu_map),
-			  "CPU %i not accounted in %s(%s)\n",
-			  cpu, cache->ofnode->full_name,
-			  cache_type_string(cache));
-
-		cpumask_clear_cpu(cpu, &cache->shared_cpu_map);
-
-		/* Release the cache object if all the cpus using it
-		 * are offline */
-		if (cpumask_empty(&cache->shared_cpu_map))
-			release_cache(cache);
-
-		cache = next;
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		this_leaf->of_node = np;
+		if (cache_node_is_unified(np)) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_UNIFIED, level);
+		} else {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		}
+		np = of_find_next_cache_node(np);
 	}
+	return 0;
 }
 
-void cacheinfo_cpu_offline(unsigned int cpu_id)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-
-	/* Prevent userspace from seeing inconsistent state - remove
-	 * the sysfs hierarchy first */
-	cache_dir = per_cpu(cache_dir_pcpu, cpu_id);
-
-	/* careful, sysfs population may have failed */
-	if (cache_dir)
-		remove_cache_dir(cache_dir);
-
-	per_cpu(cache_dir_pcpu, cpu_id) = NULL;
-
-	/* clear the CPU's bit in its cache chain, possibly freeing
-	 * cache objects */
-	cache = cache_lookup_by_cpu(cpu_id);
-	if (cache)
-		cache_cpu_clear(cache, cpu_id);
-}
-#endif /* (CONFIG_PPC_PSERIES && CONFIG_SUSPEND) || CONFIG_HOTPLUG_CPU */
diff --git a/arch/powerpc/kernel/cacheinfo.h b/arch/powerpc/kernel/cacheinfo.h
deleted file mode 100644
index a7b74d36acd7..000000000000
--- a/arch/powerpc/kernel/cacheinfo.h
+++ /dev/null
@@ -1,8 +0,0 @@
-#ifndef _PPC_CACHEINFO_H
-#define _PPC_CACHEINFO_H
-
-/* These are just hooks for sysfs.c to use. */
-extern void cacheinfo_cpu_online(unsigned int cpu_id);
-extern void cacheinfo_cpu_offline(unsigned int cpu_id);
-
-#endif /* _PPC_CACHEINFO_H */
diff --git a/arch/powerpc/kernel/sysfs.c b/arch/powerpc/kernel/sysfs.c
index 67fd2fd2620a..6e9c5a8141bb 100644
--- a/arch/powerpc/kernel/sysfs.c
+++ b/arch/powerpc/kernel/sysfs.c
@@ -19,8 +19,6 @@
 #include <asm/pmc.h>
 #include <asm/firmware.h>
 
-#include "cacheinfo.h"
-
 #ifdef CONFIG_PPC64
 #include <asm/paca.h>
 #include <asm/lppaca.h>
@@ -743,7 +741,6 @@ static void register_cpu_online(unsigned int cpu)
 		device_create_file(s, &dev_attr_altivec_idle_wait_time);
 	}
 #endif
-	cacheinfo_cpu_online(cpu);
 }
 
 #ifdef CONFIG_HOTPLUG_CPU
@@ -824,7 +821,6 @@ static void unregister_cpu_online(unsigned int cpu)
 		device_remove_file(s, &dev_attr_altivec_idle_wait_time);
 	}
 #endif
-	cacheinfo_cpu_offline(cpu);
 }
 
 #ifdef CONFIG_ARCH_CPU_PROBE_RELEASE
@@ -988,8 +984,7 @@ static int __init topology_init(void)
 	int cpu;
 
 	register_nodes();
-
-	cpu_notifier_register_begin();
+	register_cpu_notifier(&sysfs_cpu_nb);
 
 	for_each_possible_cpu(cpu) {
 		struct cpu *c = &per_cpu(cpu_devices, cpu);
@@ -1013,11 +1008,6 @@ static int __init topology_init(void)
 		if (cpu_online(cpu))
 			register_cpu_online(cpu);
 	}
-
-	__register_cpu_notifier(&sysfs_cpu_nb);
-
-	cpu_notifier_register_done();
-
 #ifdef CONFIG_PPC64
 	sysfs_create_dscr_default();
 #endif /* CONFIG_PPC64 */
-- 
1.8.3.2

[PATCH v2 6/9] powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: linuxppc-dev@lists.ozlabs.org
---
 arch/powerpc/kernel/cacheinfo.c | 813 +++++-----------------------------------
 arch/powerpc/kernel/cacheinfo.h |   8 -
 arch/powerpc/kernel/sysfs.c     |  12 +-
 3 files changed, 91 insertions(+), 742 deletions(-)
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h

diff --git a/arch/powerpc/kernel/cacheinfo.c b/arch/powerpc/kernel/cacheinfo.c
index 40198d50b4c2..b871c246d945 100644
--- a/arch/powerpc/kernel/cacheinfo.c
+++ b/arch/powerpc/kernel/cacheinfo.c
@@ -10,38 +10,10 @@
  * 2 as published by the Free Software Foundation.
  */
 
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
-#include <linux/cpumask.h>
 #include <linux/kernel.h>
-#include <linux/kobject.h>
-#include <linux/list.h>
-#include <linux/notifier.h>
 #include <linux/of.h>
-#include <linux/percpu.h>
-#include <linux/slab.h>
-#include <asm/prom.h>
-
-#include "cacheinfo.h"
-
-/* per-cpu object for tracking:
- * - a "cache" kobject for the top-level directory
- * - a list of "index" objects representing the cpu's local cache hierarchy
- */
-struct cache_dir {
-	struct kobject *kobj; /* bare (not embedded) kobject for cache
-			       * directory */
-	struct cache_index_dir *index; /* list of index objects */
-};
-
-/* "index" object: each cpu's cache directory has an index
- * subdirectory corresponding to a cache object associated with the
- * cpu.  This object's lifetime is managed via the embedded kobject.
- */
-struct cache_index_dir {
-	struct kobject kobj;
-	struct cache_index_dir *next; /* next index in parent directory */
-	struct cache *cache;
-};
 
 /* Template for determining which OF properties to query for a given
  * cache type */
@@ -60,11 +32,6 @@ struct cache_type_info {
 	const char *nr_sets_prop;
 };
 
-/* These are used to index the cache_type_info array. */
-#define CACHE_TYPE_UNIFIED     0
-#define CACHE_TYPE_INSTRUCTION 1
-#define CACHE_TYPE_DATA        2
-
 static const struct cache_type_info cache_type_info[] = {
 	{
 		/* PowerPC Processor binding says the [di]-cache-*
@@ -92,231 +59,83 @@ static const struct cache_type_info cache_type_info[] = {
 	},
 };
 
-/* Cache object: each instance of this corresponds to a distinct cache
- * in the system.  There are separate objects for Harvard caches: one
- * each for instruction and data, and each refers to the same OF node.
- * The refcount of the OF node is elevated for the lifetime of the
- * cache object.  A cache object is released when its shared_cpu_map
- * is cleared (see cache_cpu_clear).
- *
- * A cache object is on two lists: an unsorted global list
- * (cache_list) of cache objects; and a singly-linked list
- * representing the local cache hierarchy, which is ordered by level
- * (e.g. L1d -> L1i -> L2 -> L3).
- */
-struct cache {
-	struct device_node *ofnode;    /* OF node for this cache, may be cpu */
-	struct cpumask shared_cpu_map; /* online CPUs using this cache */
-	int type;                      /* split cache disambiguation */
-	int level;                     /* level not explicit in device tree */
-	struct list_head list;         /* global list of cache objects */
-	struct cache *next_local;      /* next cache of >= level */
-};
-
-static DEFINE_PER_CPU(struct cache_dir *, cache_dir_pcpu);
-
-/* traversal/modification of this list occurs only at cpu hotplug time;
- * access is serialized by cpu hotplug locking
- */
-static LIST_HEAD(cache_list);
-
-static struct cache_index_dir *kobj_to_cache_index_dir(struct kobject *k)
-{
-	return container_of(k, struct cache_index_dir, kobj);
-}
-
-static const char *cache_type_string(const struct cache *cache)
+static inline int get_cacheinfo_idx(enum cache_type type)
 {
-	return cache_type_info[cache->type].name;
-}
-
-static void cache_init(struct cache *cache, int type, int level,
-		       struct device_node *ofnode)
-{
-	cache->type = type;
-	cache->level = level;
-	cache->ofnode = of_node_get(ofnode);
-	INIT_LIST_HEAD(&cache->list);
-	list_add(&cache->list, &cache_list);
-}
-
-static struct cache *new_cache(int type, int level, struct device_node *ofnode)
-{
-	struct cache *cache;
-
-	cache = kzalloc(sizeof(*cache), GFP_KERNEL);
-	if (cache)
-		cache_init(cache, type, level, ofnode);
-
-	return cache;
-}
-
-static void release_cache_debugcheck(struct cache *cache)
-{
-	struct cache *iter;
-
-	list_for_each_entry(iter, &cache_list, list)
-		WARN_ONCE(iter->next_local == cache,
-			  "cache for %s(%s) refers to cache for %s(%s)\n",
-			  iter->ofnode->full_name,
-			  cache_type_string(iter),
-			  cache->ofnode->full_name,
-			  cache_type_string(cache));
-}
-
-static void release_cache(struct cache *cache)
-{
-	if (!cache)
-		return;
-
-	pr_debug("freeing L%d %s cache for %s\n", cache->level,
-		 cache_type_string(cache), cache->ofnode->full_name);
-
-	release_cache_debugcheck(cache);
-	list_del(&cache->list);
-	of_node_put(cache->ofnode);
-	kfree(cache);
-}
-
-static void cache_cpu_set(struct cache *cache, int cpu)
-{
-	struct cache *next = cache;
-
-	while (next) {
-		WARN_ONCE(cpumask_test_cpu(cpu, &next->shared_cpu_map),
-			  "CPU %i already accounted in %s(%s)\n",
-			  cpu, next->ofnode->full_name,
-			  cache_type_string(next));
-		cpumask_set_cpu(cpu, &next->shared_cpu_map);
-		next = next->next_local;
-	}
+	if (type == CACHE_TYPE_UNIFIED)
+		return 0;
+	else
+		return type;
 }
 
-static int cache_size(const struct cache *cache, unsigned int *ret)
+static void cache_size(struct cacheinfo *this_leaf)
 {
 	const char *propname;
 	const __be32 *cache_size;
+	int ct_idx;
 
-	propname = cache_type_info[cache->type].size_prop;
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	propname = cache_type_info[ct_idx].size_prop;
 
-	cache_size = of_get_property(cache->ofnode, propname, NULL);
+	cache_size = of_get_property(this_leaf->of_node, propname, NULL);
 	if (!cache_size)
-		return -ENODEV;
-
-	*ret = of_read_number(cache_size, 1);
-	return 0;
-}
-
-static int cache_size_kb(const struct cache *cache, unsigned int *ret)
-{
-	unsigned int size;
-
-	if (cache_size(cache, &size))
-		return -ENODEV;
-
-	*ret = size / 1024;
-	return 0;
+		this_leaf->size = 0;
+	else
+		this_leaf->size = of_read_number(cache_size, 1);
 }
 
 /* not cache_line_size() because that's a macro in include/linux/cache.h */
-static int cache_get_line_size(const struct cache *cache, unsigned int *ret)
+static void cache_get_line_size(struct cacheinfo *this_leaf)
 {
 	const __be32 *line_size;
-	int i, lim;
+	int i, lim, ct_idx;
 
-	lim = ARRAY_SIZE(cache_type_info[cache->type].line_size_props);
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	lim = ARRAY_SIZE(cache_type_info[ct_idx].line_size_props);
 
 	for (i = 0; i < lim; i++) {
 		const char *propname;
 
-		propname = cache_type_info[cache->type].line_size_props[i];
-		line_size = of_get_property(cache->ofnode, propname, NULL);
+		propname = cache_type_info[ct_idx].line_size_props[i];
+		line_size = of_get_property(this_leaf->of_node, propname, NULL);
 		if (line_size)
 			break;
 	}
 
 	if (!line_size)
-		return -ENODEV;
-
-	*ret = of_read_number(line_size, 1);
-	return 0;
+		this_leaf->coherency_line_size = 0;
+	else
+		this_leaf->coherency_line_size = of_read_number(line_size, 1);
 }
 
-static int cache_nr_sets(const struct cache *cache, unsigned int *ret)
+static void cache_nr_sets(struct cacheinfo *this_leaf)
 {
 	const char *propname;
 	const __be32 *nr_sets;
+	int ct_idx;
 
-	propname = cache_type_info[cache->type].nr_sets_prop;
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	propname = cache_type_info[ct_idx].nr_sets_prop;
 
-	nr_sets = of_get_property(cache->ofnode, propname, NULL);
+	nr_sets = of_get_property(this_leaf->of_node, propname, NULL);
 	if (!nr_sets)
-		return -ENODEV;
-
-	*ret = of_read_number(nr_sets, 1);
-	return 0;
+		this_leaf->number_of_sets = 0;
+	else
+		this_leaf->number_of_sets = of_read_number(nr_sets, 1);
 }
 
-static int cache_associativity(const struct cache *cache, unsigned int *ret)
+static void cache_associativity(struct cacheinfo *this_leaf)
 {
-	unsigned int line_size;
-	unsigned int nr_sets;
-	unsigned int size;
-
-	if (cache_nr_sets(cache, &nr_sets))
-		goto err;
+	unsigned int line_size = this_leaf->coherency_line_size;
+	unsigned int nr_sets = this_leaf->number_of_sets;
+	unsigned int size = this_leaf->size;
 
 	/* If the cache is fully associative, there is no need to
 	 * check the other properties.
 	 */
-	if (nr_sets == 1) {
-		*ret = 0;
-		return 0;
-	}
-
-	if (cache_get_line_size(cache, &line_size))
-		goto err;
-	if (cache_size(cache, &size))
-		goto err;
-
-	if (!(nr_sets > 0 && size > 0 && line_size > 0))
-		goto err;
-
-	*ret = (size / nr_sets) / line_size;
-	return 0;
-err:
-	return -ENODEV;
-}
-
-/* helper for dealing with split caches */
-static struct cache *cache_find_first_sibling(struct cache *cache)
-{
-	struct cache *iter;
-
-	if (cache->type == CACHE_TYPE_UNIFIED)
-		return cache;
-
-	list_for_each_entry(iter, &cache_list, list)
-		if (iter->ofnode == cache->ofnode && iter->next_local == cache)
-			return iter;
-
-	return cache;
-}
-
-/* return the first cache on a local list matching node */
-static struct cache *cache_lookup_by_node(const struct device_node *node)
-{
-	struct cache *cache = NULL;
-	struct cache *iter;
-
-	list_for_each_entry(iter, &cache_list, list) {
-		if (iter->ofnode != node)
-			continue;
-		cache = cache_find_first_sibling(iter);
-		break;
-	}
-
-	return cache;
+	if ((nr_sets == 1) || !(nr_sets > 0 && size > 0 && line_size > 0))
+		this_leaf->ways_of_associativity = 0;
+	else
+		this_leaf->ways_of_associativity = (size / nr_sets) / line_size;
 }
 
 static bool cache_node_is_unified(const struct device_node *np)
@@ -324,526 +143,74 @@ static bool cache_node_is_unified(const struct device_node *np)
 	return of_get_property(np, "cache-unified", NULL);
 }
 
-static struct cache *cache_do_one_devnode_unified(struct device_node *node,
-						  int level)
-{
-	struct cache *cache;
-
-	pr_debug("creating L%d ucache for %s\n", level, node->full_name);
-
-	cache = new_cache(CACHE_TYPE_UNIFIED, level, node);
-
-	return cache;
-}
-
-static struct cache *cache_do_one_devnode_split(struct device_node *node,
-						int level)
-{
-	struct cache *dcache, *icache;
-
-	pr_debug("creating L%d dcache and icache for %s\n", level,
-		 node->full_name);
-
-	dcache = new_cache(CACHE_TYPE_DATA, level, node);
-	icache = new_cache(CACHE_TYPE_INSTRUCTION, level, node);
-
-	if (!dcache || !icache)
-		goto err;
-
-	dcache->next_local = icache;
-
-	return dcache;
-err:
-	release_cache(dcache);
-	release_cache(icache);
-	return NULL;
-}
-
-static struct cache *cache_do_one_devnode(struct device_node *node, int level)
-{
-	struct cache *cache;
-
-	if (cache_node_is_unified(node))
-		cache = cache_do_one_devnode_unified(node, level);
-	else
-		cache = cache_do_one_devnode_split(node, level);
-
-	return cache;
-}
-
-static struct cache *cache_lookup_or_instantiate(struct device_node *node,
-						 int level)
-{
-	struct cache *cache;
-
-	cache = cache_lookup_by_node(node);
-
-	WARN_ONCE(cache && cache->level != level,
-		  "cache level mismatch on lookup (got %d, expected %d)\n",
-		  cache->level, level);
-
-	if (!cache)
-		cache = cache_do_one_devnode(node, level);
-
-	return cache;
-}
-
-static void link_cache_lists(struct cache *smaller, struct cache *bigger)
-{
-	while (smaller->next_local) {
-		if (smaller->next_local == bigger)
-			return; /* already linked */
-		smaller = smaller->next_local;
-	}
-
-	smaller->next_local = bigger;
-}
-
-static void do_subsidiary_caches_debugcheck(struct cache *cache)
-{
-	WARN_ON_ONCE(cache->level != 1);
-	WARN_ON_ONCE(strcmp(cache->ofnode->type, "cpu"));
-}
-
-static void do_subsidiary_caches(struct cache *cache)
-{
-	struct device_node *subcache_node;
-	int level = cache->level;
-
-	do_subsidiary_caches_debugcheck(cache);
-
-	while ((subcache_node = of_find_next_cache_node(cache->ofnode))) {
-		struct cache *subcache;
-
-		level++;
-		subcache = cache_lookup_or_instantiate(subcache_node, level);
-		of_node_put(subcache_node);
-		if (!subcache)
-			break;
-
-		link_cache_lists(cache, subcache);
-		cache = subcache;
-	}
-}
-
-static struct cache *cache_chain_instantiate(unsigned int cpu_id)
-{
-	struct device_node *cpu_node;
-	struct cache *cpu_cache = NULL;
-
-	pr_debug("creating cache object(s) for CPU %i\n", cpu_id);
-
-	cpu_node = of_get_cpu_node(cpu_id, NULL);
-	WARN_ONCE(!cpu_node, "no OF node found for CPU %i\n", cpu_id);
-	if (!cpu_node)
-		goto out;
-
-	cpu_cache = cache_lookup_or_instantiate(cpu_node, 1);
-	if (!cpu_cache)
-		goto out;
-
-	do_subsidiary_caches(cpu_cache);
-
-	cache_cpu_set(cpu_cache, cpu_id);
-out:
-	of_node_put(cpu_node);
-
-	return cpu_cache;
-}
-
-static struct cache_dir *cacheinfo_create_cache_dir(unsigned int cpu_id)
-{
-	struct cache_dir *cache_dir;
-	struct device *dev;
-	struct kobject *kobj = NULL;
-
-	dev = get_cpu_device(cpu_id);
-	WARN_ONCE(!dev, "no dev for CPU %i\n", cpu_id);
-	if (!dev)
-		goto err;
-
-	kobj = kobject_create_and_add("cache", &dev->kobj);
-	if (!kobj)
-		goto err;
-
-	cache_dir = kzalloc(sizeof(*cache_dir), GFP_KERNEL);
-	if (!cache_dir)
-		goto err;
-
-	cache_dir->kobj = kobj;
-
-	WARN_ON_ONCE(per_cpu(cache_dir_pcpu, cpu_id) != NULL);
-
-	per_cpu(cache_dir_pcpu, cpu_id) = cache_dir;
-
-	return cache_dir;
-err:
-	kobject_put(kobj);
-	return NULL;
-}
-
-static void cache_index_release(struct kobject *kobj)
-{
-	struct cache_index_dir *index;
-
-	index = kobj_to_cache_index_dir(kobj);
-
-	pr_debug("freeing index directory for L%d %s cache\n",
-		 index->cache->level, cache_type_string(index->cache));
-
-	kfree(index);
-}
-
-static ssize_t cache_index_show(struct kobject *k, struct attribute *attr, char *buf)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
 {
-	struct kobj_attribute *kobj_attr;
-
-	kobj_attr = container_of(attr, struct kobj_attribute, attr);
-
-	return kobj_attr->show(k, kobj_attr, buf);
+	this_leaf->level = level;
+	this_leaf->type = type;
+	cache_size(this_leaf);
+	cache_get_line_size(this_leaf);
+	cache_nr_sets(this_leaf);
+	cache_associativity(this_leaf);
 }
 
-static struct cache *index_kobj_to_cache(struct kobject *k)
+int init_cache_level(unsigned int cpu)
 {
-	struct cache_index_dir *index;
+	struct device_node *np;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int level = 0, leaves = 0;
 
-	index = kobj_to_cache_index_dir(k);
-
-	return index->cache;
-}
-
-static ssize_t size_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int size_kb;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_size_kb(cache, &size_kb))
+	if (!cpu_dev) {
+		pr_err("No cpu device for CPU %d\n", cpu);
 		return -ENODEV;
-
-	return sprintf(buf, "%uK\n", size_kb);
-}
-
-static struct kobj_attribute cache_size_attr =
-	__ATTR(size, 0444, size_show, NULL);
-
-
-static ssize_t line_size_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int line_size;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_get_line_size(cache, &line_size))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", line_size);
-}
-
-static struct kobj_attribute cache_line_size_attr =
-	__ATTR(coherency_line_size, 0444, line_size_show, NULL);
-
-static ssize_t nr_sets_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int nr_sets;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_nr_sets(cache, &nr_sets))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", nr_sets);
-}
-
-static struct kobj_attribute cache_nr_sets_attr =
-	__ATTR(number_of_sets, 0444, nr_sets_show, NULL);
-
-static ssize_t associativity_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int associativity;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_associativity(cache, &associativity))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", associativity);
-}
-
-static struct kobj_attribute cache_assoc_attr =
-	__ATTR(ways_of_associativity, 0444, associativity_show, NULL);
-
-static ssize_t type_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	return sprintf(buf, "%s\n", cache_type_string(cache));
-}
-
-static struct kobj_attribute cache_type_attr =
-	__ATTR(type, 0444, type_show, NULL);
-
-static ssize_t level_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache_index_dir *index;
-	struct cache *cache;
-
-	index = kobj_to_cache_index_dir(k);
-	cache = index->cache;
-
-	return sprintf(buf, "%d\n", cache->level);
-}
-
-static struct kobj_attribute cache_level_attr =
-	__ATTR(level, 0444, level_show, NULL);
-
-static ssize_t shared_cpu_map_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache_index_dir *index;
-	struct cache *cache;
-	int len;
-	int n = 0;
-
-	index = kobj_to_cache_index_dir(k);
-	cache = index->cache;
-	len = PAGE_SIZE - 2;
-
-	if (len > 1) {
-		n = cpumask_scnprintf(buf, len, &cache->shared_cpu_map);
-		buf[n++] = '\n';
-		buf[n] = '\0';
 	}
-	return n;
-}
-
-static struct kobj_attribute cache_shared_cpu_map_attr =
-	__ATTR(shared_cpu_map, 0444, shared_cpu_map_show, NULL);
-
-/* Attributes which should always be created -- the kobject/sysfs core
- * does this automatically via kobj_type->default_attrs.  This is the
- * minimum data required to uniquely identify a cache.
- */
-static struct attribute *cache_index_default_attrs[] = {
-	&cache_type_attr.attr,
-	&cache_level_attr.attr,
-	&cache_shared_cpu_map_attr.attr,
-	NULL,
-};
-
-/* Attributes which should be created if the cache device node has the
- * right properties -- see cacheinfo_create_index_opt_attrs
- */
-static struct kobj_attribute *cache_index_opt_attrs[] = {
-	&cache_size_attr,
-	&cache_line_size_attr,
-	&cache_nr_sets_attr,
-	&cache_assoc_attr,
-};
-
-static const struct sysfs_ops cache_index_ops = {
-	.show = cache_index_show,
-};
-
-static struct kobj_type cache_index_type = {
-	.release = cache_index_release,
-	.sysfs_ops = &cache_index_ops,
-	.default_attrs = cache_index_default_attrs,
-};
-
-static void cacheinfo_create_index_opt_attrs(struct cache_index_dir *dir)
-{
-	const char *cache_name;
-	const char *cache_type;
-	struct cache *cache;
-	char *buf;
-	int i;
-
-	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
-	if (!buf)
-		return;
-
-	cache = dir->cache;
-	cache_name = cache->ofnode->full_name;
-	cache_type = cache_type_string(cache);
-
-	/* We don't want to create an attribute that can't provide a
-	 * meaningful value.  Check the return value of each optional
-	 * attribute's ->show method before registering the
-	 * attribute.
-	 */
-	for (i = 0; i < ARRAY_SIZE(cache_index_opt_attrs); i++) {
-		struct kobj_attribute *attr;
-		ssize_t rc;
-
-		attr = cache_index_opt_attrs[i];
-
-		rc = attr->show(&dir->kobj, attr, buf);
-		if (rc <= 0) {
-			pr_debug("not creating %s attribute for "
-				 "%s(%s) (rc = %zd)\n",
-				 attr->attr.name, cache_name,
-				 cache_type, rc);
-			continue;
-		}
-		if (sysfs_create_file(&dir->kobj, &attr->attr))
-			pr_debug("could not create %s attribute for %s(%s)\n",
-				 attr->attr.name, cache_name, cache_type);
+	np = cpu_dev->of_node;
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
 	}
 
-	kfree(buf);
-}
-
-static void cacheinfo_create_index_dir(struct cache *cache, int index,
-				       struct cache_dir *cache_dir)
-{
-	struct cache_index_dir *index_dir;
-	int rc;
-
-	index_dir = kzalloc(sizeof(*index_dir), GFP_KERNEL);
-	if (!index_dir)
-		goto err;
-
-	index_dir->cache = cache;
-
-	rc = kobject_init_and_add(&index_dir->kobj, &cache_index_type,
-				  cache_dir->kobj, "index%d", index);
-	if (rc)
-		goto err;
-
-	index_dir->next = cache_dir->index;
-	cache_dir->index = index_dir;
-
-	cacheinfo_create_index_opt_attrs(index_dir);
-
-	return;
-err:
-	kfree(index_dir);
-}
-
-static void cacheinfo_sysfs_populate(unsigned int cpu_id,
-				     struct cache *cache_list)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-	int index = 0;
-
-	cache_dir = cacheinfo_create_cache_dir(cpu_id);
-	if (!cache_dir)
-		return;
-
-	cache = cache_list;
-	while (cache) {
-		cacheinfo_create_index_dir(cache, index, cache_dir);
-		index++;
-		cache = cache->next_local;
+	while (np) {
+		leaves += cache_node_is_unified(np) ? 1 : 2;
+		level++;
+		of_node_put(np);
+		np = of_find_next_cache_node(np);
 	}
-}
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
 
-void cacheinfo_cpu_online(unsigned int cpu_id)
-{
-	struct cache *cache;
-
-	cache = cache_chain_instantiate(cpu_id);
-	if (!cache)
-		return;
-
-	cacheinfo_sysfs_populate(cpu_id, cache);
-}
-
-/* functions needed to remove cache entry for cpu offline or suspend/resume */
-
-#if (defined(CONFIG_PPC_PSERIES) && defined(CONFIG_SUSPEND)) || \
-    defined(CONFIG_HOTPLUG_CPU)
-
-static struct cache *cache_lookup_by_cpu(unsigned int cpu_id)
-{
-	struct device_node *cpu_node;
-	struct cache *cache;
-
-	cpu_node = of_get_cpu_node(cpu_id, NULL);
-	WARN_ONCE(!cpu_node, "no OF node found for CPU %i\n", cpu_id);
-	if (!cpu_node)
-		return NULL;
-
-	cache = cache_lookup_by_node(cpu_node);
-	of_node_put(cpu_node);
-
-	return cache;
+	return 0;
 }
 
-static void remove_index_dirs(struct cache_dir *cache_dir)
+int populate_cache_leaves(unsigned int cpu)
 {
-	struct cache_index_dir *index;
-
-	index = cache_dir->index;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct device_node *np;
+	unsigned int level, idx;
 
-	while (index) {
-		struct cache_index_dir *next;
-
-		next = index->next;
-		kobject_put(&index->kobj);
-		index = next;
+	np = of_node_get(cpu_dev->of_node);
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
 	}
-}
-
-static void remove_cache_dir(struct cache_dir *cache_dir)
-{
-	remove_index_dirs(cache_dir);
 
-	/* Remove cache dir from sysfs */
-	kobject_del(cache_dir->kobj);
-
-	kobject_put(cache_dir->kobj);
-
-	kfree(cache_dir);
-}
-
-static void cache_cpu_clear(struct cache *cache, int cpu)
-{
-	while (cache) {
-		struct cache *next = cache->next_local;
-
-		WARN_ONCE(!cpumask_test_cpu(cpu, &cache->shared_cpu_map),
-			  "CPU %i not accounted in %s(%s)\n",
-			  cpu, cache->ofnode->full_name,
-			  cache_type_string(cache));
-
-		cpumask_clear_cpu(cpu, &cache->shared_cpu_map);
-
-		/* Release the cache object if all the cpus using it
-		 * are offline */
-		if (cpumask_empty(&cache->shared_cpu_map))
-			release_cache(cache);
-
-		cache = next;
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		this_leaf->of_node = np;
+		if (cache_node_is_unified(np)) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_UNIFIED, level);
+		} else {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		}
+		np = of_find_next_cache_node(np);
 	}
+	return 0;
 }
 
-void cacheinfo_cpu_offline(unsigned int cpu_id)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-
-	/* Prevent userspace from seeing inconsistent state - remove
-	 * the sysfs hierarchy first */
-	cache_dir = per_cpu(cache_dir_pcpu, cpu_id);
-
-	/* careful, sysfs population may have failed */
-	if (cache_dir)
-		remove_cache_dir(cache_dir);
-
-	per_cpu(cache_dir_pcpu, cpu_id) = NULL;
-
-	/* clear the CPU's bit in its cache chain, possibly freeing
-	 * cache objects */
-	cache = cache_lookup_by_cpu(cpu_id);
-	if (cache)
-		cache_cpu_clear(cache, cpu_id);
-}
-#endif /* (CONFIG_PPC_PSERIES && CONFIG_SUSPEND) || CONFIG_HOTPLUG_CPU */
diff --git a/arch/powerpc/kernel/cacheinfo.h b/arch/powerpc/kernel/cacheinfo.h
deleted file mode 100644
index a7b74d36acd7..000000000000
--- a/arch/powerpc/kernel/cacheinfo.h
+++ /dev/null
@@ -1,8 +0,0 @@
-#ifndef _PPC_CACHEINFO_H
-#define _PPC_CACHEINFO_H
-
-/* These are just hooks for sysfs.c to use. */
-extern void cacheinfo_cpu_online(unsigned int cpu_id);
-extern void cacheinfo_cpu_offline(unsigned int cpu_id);
-
-#endif /* _PPC_CACHEINFO_H */
diff --git a/arch/powerpc/kernel/sysfs.c b/arch/powerpc/kernel/sysfs.c
index 67fd2fd2620a..6e9c5a8141bb 100644
--- a/arch/powerpc/kernel/sysfs.c
+++ b/arch/powerpc/kernel/sysfs.c
@@ -19,8 +19,6 @@
 #include <asm/pmc.h>
 #include <asm/firmware.h>
 
-#include "cacheinfo.h"
-
 #ifdef CONFIG_PPC64
 #include <asm/paca.h>
 #include <asm/lppaca.h>
@@ -743,7 +741,6 @@ static void register_cpu_online(unsigned int cpu)
 		device_create_file(s, &dev_attr_altivec_idle_wait_time);
 	}
 #endif
-	cacheinfo_cpu_online(cpu);
 }
 
 #ifdef CONFIG_HOTPLUG_CPU
@@ -824,7 +821,6 @@ static void unregister_cpu_online(unsigned int cpu)
 		device_remove_file(s, &dev_attr_altivec_idle_wait_time);
 	}
 #endif
-	cacheinfo_cpu_offline(cpu);
 }
 
 #ifdef CONFIG_ARCH_CPU_PROBE_RELEASE
@@ -988,8 +984,7 @@ static int __init topology_init(void)
 	int cpu;
 
 	register_nodes();
-
-	cpu_notifier_register_begin();
+	register_cpu_notifier(&sysfs_cpu_nb);
 
 	for_each_possible_cpu(cpu) {
 		struct cpu *c = &per_cpu(cpu_devices, cpu);
@@ -1013,11 +1008,6 @@ static int __init topology_init(void)
 		if (cpu_online(cpu))
 			register_cpu_online(cpu);
 	}
-
-	__register_cpu_notifier(&sysfs_cpu_nb);
-
-	cpu_notifier_register_done();
-
 #ifdef CONFIG_PPC64
 	sysfs_create_dscr_default();
 #endif /* CONFIG_PPC64 */
-- 
1.8.3.2

[PATCH v2 7/9] ARM64: kernel: add support for cpu cache information

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds support for cacheinfo on ARM64.

On ARMv8, the cache hierarchy can be identified through Cache Level ID
(CLIDR) register while the cache geometry is provided by Cache Size ID
(CCSIDR) register.

Since the architecture doesn't provide any way of detecting the cpus
sharing particular cache, device tree is used for the same purpose.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
---
 arch/arm64/kernel/Makefile    |   3 +-
 arch/arm64/kernel/cacheinfo.c | 142 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 144 insertions(+), 1 deletion(-)
 create mode 100644 arch/arm64/kernel/cacheinfo.c

diff --git a/arch/arm64/kernel/Makefile b/arch/arm64/kernel/Makefile
index cdaedad3afe5..754a3d07b6b8 100644
--- a/arch/arm64/kernel/Makefile
+++ b/arch/arm64/kernel/Makefile
@@ -15,7 +15,8 @@ CFLAGS_REMOVE_return_address.o = -pg
 arm64-obj-y		:= cputable.o debug-monitors.o entry.o irq.o fpsimd.o	\
 			   entry-fpsimd.o process.o ptrace.o setup.o signal.o	\
 			   sys.o stacktrace.o time.o traps.o io.o vdso.o	\
-			   hyp-stub.o psci.o cpu_ops.o insn.o return_address.o
+			   hyp-stub.o psci.o cpu_ops.o insn.o return_address.o	\
+			   cacheinfo.o
 
 arm64-obj-$(CONFIG_COMPAT)		+= sys32.o kuser32.o signal32.o 	\
 					   sys_compat.o
diff --git a/arch/arm64/kernel/cacheinfo.c b/arch/arm64/kernel/cacheinfo.c
new file mode 100644
index 000000000000..6c559659b460
--- /dev/null
+++ b/arch/arm64/kernel/cacheinfo.c
@@ -0,0 +1,142 @@
+/*
+ *  ARM64 cacheinfo support
+ *
+ *  Copyright (C) 2014 ARM Ltd.
+ *  All Rights Reserved
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/cpu.h>
+#include <linux/compiler.h>
+#include <linux/of.h>
+
+#include <asm/processor.h>
+
+#define MAX_CACHE_LEVEL			7	/* Max 7 level supported */
+/* Ctypen, bits[3(n - 1) + 2 : 3(n - 1)], for n = 1 to 7 */
+#define CLIDR_CTYPE_SHIFT(level)	(3 * (level - 1))
+#define CLIDR_CTYPE_MASK(level)		(7 << CLIDR_CTYPE_SHIFT(level))
+#define CLIDR_CTYPE(clidr, level)	\
+	(((clidr) & CLIDR_CTYPE_MASK(level)) >> CLIDR_CTYPE_SHIFT(level))
+
+static inline enum cache_type get_cache_type(int level)
+{
+	u64 clidr;
+
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	asm volatile ("mrs     %x0, clidr_el1" : "=r" (clidr));
+	return CLIDR_CTYPE(clidr, level);
+}
+
+/*
+ * NumSets, bits[27:13] - (Number of sets in cache) - 1
+ * Associativity, bits[12:3] - (Associativity of cache) - 1
+ * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
+ */
+#define CCSIDR_WRITE_THROUGH		BIT(31)
+#define CCSIDR_WRITE_BACK		BIT(30)
+#define CCSIDR_READ_ALLOCATE		BIT(29)
+#define CCSIDR_WRITE_ALLOCATE		BIT(28)
+#define CCSIDR_LINESIZE_MASK		0x7
+#define CCSIDR_ASSOCIATIVITY_SHIFT	3
+#define CCSIDR_ASSOCIATIVITY_MASK	0x3FF
+#define CCSIDR_NUMSETS_SHIFT		13
+#define CCSIDR_NUMSETS_MASK		0x7FF
+
+/*
+ * Which cache CCSIDR represents depends on CSSELR value
+ * Make sure no one else changes CSSELR during this
+ * smp_call_function_single prevents preemption for us
+ */
+static inline u32 get_ccsidr(u64 csselr)
+{
+	u64 ccsidr;
+
+	/* Put value into CSSELR */
+	asm volatile("msr csselr_el1, %x0" : : "r" (csselr));
+	isb();
+	/* Read result out of CCSIDR */
+	asm volatile("mrs %x0, ccsidr_el1" : "=r" (ccsidr));
+
+	return (u32)ccsidr;
+}
+
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
+{
+	bool is_instr_cache = type & CACHE_TYPE_INST;
+	u32 tmp = get_ccsidr((level - 1) << 1 | is_instr_cache);
+
+	this_leaf->level = level;
+	this_leaf->type = type;
+	this_leaf->coherency_line_size =
+	    (1 << ((tmp & CCSIDR_LINESIZE_MASK) + 2)) * 4;
+	this_leaf->number_of_sets =
+	    ((tmp >> CCSIDR_NUMSETS_SHIFT) & CCSIDR_NUMSETS_MASK) + 1;
+	this_leaf->ways_of_associativity = ((tmp >> CCSIDR_ASSOCIATIVITY_SHIFT)
+					    & CCSIDR_ASSOCIATIVITY_MASK) + 1;
+	this_leaf->size = this_leaf->number_of_sets *
+	    this_leaf->coherency_line_size * this_leaf->ways_of_associativity;
+	this_leaf->attributes =
+		((tmp & CCSIDR_WRITE_THROUGH) ? CACHE_WRITE_THROUGH : 0) |
+		((tmp & CCSIDR_WRITE_BACK) ? CACHE_WRITE_BACK : 0) |
+		((tmp & CCSIDR_READ_ALLOCATE) ? CACHE_READ_ALLOCATE : 0) |
+		((tmp & CCSIDR_WRITE_ALLOCATE) ? CACHE_WRITE_ALLOCATE : 0);
+}
+
+static int __init_cache_level(unsigned int cpu)
+{
+	unsigned int ctype, level, leaves;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	for (level = 1, leaves = 0; level <= MAX_CACHE_LEVEL; level++) {
+		ctype = get_cache_type(level);
+		if (ctype == CACHE_TYPE_NOCACHE) {
+			level--;
+			break;
+		}
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	}
+
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
+	return 0;
+}
+
+static int __populate_cache_leaves(unsigned int cpu)
+{
+	unsigned int level, idx;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		type = get_cache_type(level);
+		if (type == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, type, level);
+		}
+	}
+	return 0;
+}
+
+DEFINE_SMP_CALL_FUNCTION(init_cache_level)
+DEFINE_SMP_CALL_FUNCTION(populate_cache_leaves)
-- 
1.8.3.2

[PATCH v2 7/9] ARM64: kernel: add support for cpu cache information

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds support for cacheinfo on ARM64.

On ARMv8, the cache hierarchy can be identified through Cache Level ID
(CLIDR) register while the cache geometry is provided by Cache Size ID
(CCSIDR) register.

Since the architecture doesn't provide any way of detecting the cpus
sharing particular cache, device tree is used for the same purpose.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel at lists.infradead.org
---
 arch/arm64/kernel/Makefile    |   3 +-
 arch/arm64/kernel/cacheinfo.c | 142 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 144 insertions(+), 1 deletion(-)
 create mode 100644 arch/arm64/kernel/cacheinfo.c

diff --git a/arch/arm64/kernel/Makefile b/arch/arm64/kernel/Makefile
index cdaedad3afe5..754a3d07b6b8 100644
--- a/arch/arm64/kernel/Makefile
+++ b/arch/arm64/kernel/Makefile
@@ -15,7 +15,8 @@ CFLAGS_REMOVE_return_address.o = -pg
 arm64-obj-y		:= cputable.o debug-monitors.o entry.o irq.o fpsimd.o	\
 			   entry-fpsimd.o process.o ptrace.o setup.o signal.o	\
 			   sys.o stacktrace.o time.o traps.o io.o vdso.o	\
-			   hyp-stub.o psci.o cpu_ops.o insn.o return_address.o
+			   hyp-stub.o psci.o cpu_ops.o insn.o return_address.o	\
+			   cacheinfo.o
 
 arm64-obj-$(CONFIG_COMPAT)		+= sys32.o kuser32.o signal32.o 	\
 					   sys_compat.o
diff --git a/arch/arm64/kernel/cacheinfo.c b/arch/arm64/kernel/cacheinfo.c
new file mode 100644
index 000000000000..6c559659b460
--- /dev/null
+++ b/arch/arm64/kernel/cacheinfo.c
@@ -0,0 +1,142 @@
+/*
+ *  ARM64 cacheinfo support
+ *
+ *  Copyright (C) 2014 ARM Ltd.
+ *  All Rights Reserved
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/cpu.h>
+#include <linux/compiler.h>
+#include <linux/of.h>
+
+#include <asm/processor.h>
+
+#define MAX_CACHE_LEVEL			7	/* Max 7 level supported */
+/* Ctypen, bits[3(n - 1) + 2 : 3(n - 1)], for n = 1 to 7 */
+#define CLIDR_CTYPE_SHIFT(level)	(3 * (level - 1))
+#define CLIDR_CTYPE_MASK(level)		(7 << CLIDR_CTYPE_SHIFT(level))
+#define CLIDR_CTYPE(clidr, level)	\
+	(((clidr) & CLIDR_CTYPE_MASK(level)) >> CLIDR_CTYPE_SHIFT(level))
+
+static inline enum cache_type get_cache_type(int level)
+{
+	u64 clidr;
+
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	asm volatile ("mrs     %x0, clidr_el1" : "=r" (clidr));
+	return CLIDR_CTYPE(clidr, level);
+}
+
+/*
+ * NumSets, bits[27:13] - (Number of sets in cache) - 1
+ * Associativity, bits[12:3] - (Associativity of cache) - 1
+ * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
+ */
+#define CCSIDR_WRITE_THROUGH		BIT(31)
+#define CCSIDR_WRITE_BACK		BIT(30)
+#define CCSIDR_READ_ALLOCATE		BIT(29)
+#define CCSIDR_WRITE_ALLOCATE		BIT(28)
+#define CCSIDR_LINESIZE_MASK		0x7
+#define CCSIDR_ASSOCIATIVITY_SHIFT	3
+#define CCSIDR_ASSOCIATIVITY_MASK	0x3FF
+#define CCSIDR_NUMSETS_SHIFT		13
+#define CCSIDR_NUMSETS_MASK		0x7FF
+
+/*
+ * Which cache CCSIDR represents depends on CSSELR value
+ * Make sure no one else changes CSSELR during this
+ * smp_call_function_single prevents preemption for us
+ */
+static inline u32 get_ccsidr(u64 csselr)
+{
+	u64 ccsidr;
+
+	/* Put value into CSSELR */
+	asm volatile("msr csselr_el1, %x0" : : "r" (csselr));
+	isb();
+	/* Read result out of CCSIDR */
+	asm volatile("mrs %x0, ccsidr_el1" : "=r" (ccsidr));
+
+	return (u32)ccsidr;
+}
+
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
+{
+	bool is_instr_cache = type & CACHE_TYPE_INST;
+	u32 tmp = get_ccsidr((level - 1) << 1 | is_instr_cache);
+
+	this_leaf->level = level;
+	this_leaf->type = type;
+	this_leaf->coherency_line_size =
+	    (1 << ((tmp & CCSIDR_LINESIZE_MASK) + 2)) * 4;
+	this_leaf->number_of_sets =
+	    ((tmp >> CCSIDR_NUMSETS_SHIFT) & CCSIDR_NUMSETS_MASK) + 1;
+	this_leaf->ways_of_associativity = ((tmp >> CCSIDR_ASSOCIATIVITY_SHIFT)
+					    & CCSIDR_ASSOCIATIVITY_MASK) + 1;
+	this_leaf->size = this_leaf->number_of_sets *
+	    this_leaf->coherency_line_size * this_leaf->ways_of_associativity;
+	this_leaf->attributes =
+		((tmp & CCSIDR_WRITE_THROUGH) ? CACHE_WRITE_THROUGH : 0) |
+		((tmp & CCSIDR_WRITE_BACK) ? CACHE_WRITE_BACK : 0) |
+		((tmp & CCSIDR_READ_ALLOCATE) ? CACHE_READ_ALLOCATE : 0) |
+		((tmp & CCSIDR_WRITE_ALLOCATE) ? CACHE_WRITE_ALLOCATE : 0);
+}
+
+static int __init_cache_level(unsigned int cpu)
+{
+	unsigned int ctype, level, leaves;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	for (level = 1, leaves = 0; level <= MAX_CACHE_LEVEL; level++) {
+		ctype = get_cache_type(level);
+		if (ctype == CACHE_TYPE_NOCACHE) {
+			level--;
+			break;
+		}
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	}
+
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
+	return 0;
+}
+
+static int __populate_cache_leaves(unsigned int cpu)
+{
+	unsigned int level, idx;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		type = get_cache_type(level);
+		if (type == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, type, level);
+		}
+	}
+	return 0;
+}
+
+DEFINE_SMP_CALL_FUNCTION(init_cache_level)
+DEFINE_SMP_CALL_FUNCTION(populate_cache_leaves)
-- 
1.8.3.2

[PATCH v2 8/9] ARM: kernel: add support for cpu cache information

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds support for cacheinfo on ARM platforms.

On ARMv7, the cache hierarchy can be identified through Cache Level ID
register(CLIDR) while the cache geometry is provided by Cache Size ID
register(CCSIDR).

On architecture versions before ARMv7, CLIDR and CCSIDR is not
implemented. The cache type register(CTR) provides both cache hierarchy
and geometry if implemented. For implementations that doesn't support
CTR, we need to list the probable value of CTR if it was implemented
along with the cpuid for the sake of simplicity to handle them.

Since the architecture doesn't provide any way of detecting the cpus
sharing particular cache, device tree is used fo the same purpose.
On non-DT platforms, first level caches are per-cpu while higher level
caches are assumed system-wide.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Nicolas Pitre <nicolas.pitre@linaro.org>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
---
 arch/arm/kernel/Makefile    |   1 +
 arch/arm/kernel/cacheinfo.c | 272 ++++++++++++++++++++++++++++++++++++++++++++
 arch/arm/mm/Kconfig         |  13 +++
 3 files changed, 286 insertions(+)
 create mode 100644 arch/arm/kernel/cacheinfo.c

--->8

Hi Russell,

Since for few CPUs like ARM11MP core which implements VMSAv6 + Advanced OS
Features, Linux returns cpu_architecture as armv7, I have added a list
for that as they are armv6 and don't implement CLIDR and group.
Let me know if there's any alternative to handle that.

This also depends on your patch
"ARM: make it easier to check the CPU part number correctly"

Regards,
Sudeep

diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile
index 38ddd9f83d0e..2c5ff0efb670 100644
--- a/arch/arm/kernel/Makefile
+++ b/arch/arm/kernel/Makefile
@@ -29,6 +29,7 @@ obj-y		+= entry-v7m.o v7m.o
 else
 obj-y		+= entry-armv.o
 endif
+obj-$(CONFIG_CPU_HAS_CACHE) += cacheinfo.o
 
 obj-$(CONFIG_OC_ETM)		+= etm.o
 obj-$(CONFIG_CPU_IDLE)		+= cpuidle.o
diff --git a/arch/arm/kernel/cacheinfo.c b/arch/arm/kernel/cacheinfo.c
new file mode 100644
index 000000000000..427ba641b33a
--- /dev/null
+++ b/arch/arm/kernel/cacheinfo.c
@@ -0,0 +1,272 @@
+/*
+ *  ARM cacheinfo support
+ *
+ *  Copyright (C) 2014 ARM Ltd.
+ *  All Rights Reserved
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/cpu.h>
+#include <linux/compiler.h>
+#include <linux/of.h>
+
+#include <asm/cputype.h>
+#include <asm/processor.h>
+#include <asm/system_info.h>
+
+#define cache_is_armv7() \
+	(cpu_architecture() >= CPU_ARCH_ARMv7 && !armv6_extended())
+#define MAX_CACHE_LEVEL		(cache_is_armv7() ? 7 : 1)
+
+#define CTR_CTYPE_SHIFT		24
+#define CTR_CTYPE_MASK		(1 << CTR_CTYPE_SHIFT)
+
+struct ctr_info {
+	unsigned int cpuid_part;
+	unsigned int ctr;
+};
+/*
+ *  Cache Type Register
+ *  +---------------------------------+
+ *  | 31 29 | 28 25 |24| 23 12 | 11 0 |
+ *  +---------------------------------+
+ *  | 0 0 0 | Ctype | S| Dsize | Isize|
+ *  +---------------------------------+
+ * The table below encodes only Dsize and Isize
+ */
+static struct ctr_info cache_ctr_list[] = {
+	{0x4400a100, 0x0016A16A }, /* 32kB D$, 32kB I$ */
+	{0x4400a110, 0x0012A16A }, /* 16kB D$, 32kB I$ */
+	{0x6900b110, 0x0012A16A }, /* 16kB D$, 32kB I$ */
+};
+
+/*
+ * List of CPUs reported as ARMv7 but don't implement CLIDR,
+ * CSSELR and CCSIDR. Cache information is still available from CTR
+ */
+static int armv6_ext_cpuid_part[] = { 0x4100b020, };
+
+static bool armv6_extended(void)
+{
+	int i, cpuid_part = read_cpuid_part();
+
+	for (i = 0; i < ARRAY_SIZE(armv6_ext_cpuid_part); i++)
+		if (armv6_ext_cpuid_part[i] == cpuid_part)
+			return true;
+	return false;
+}
+
+static int get_unimplemented_ctr(unsigned int *ctr)
+{
+	int i, cpuid_part = read_cpuid_part();
+
+	for (i = 0; i < ARRAY_SIZE(cache_ctr_list); i++)
+		if (cache_ctr_list[i].cpuid_part == cpuid_part) {
+			*ctr = cache_ctr_list[i].ctr;
+			return 0;
+		}
+	return -ENOENT;
+}
+
+static unsigned int get_ctr(void)
+{
+	unsigned int ctr;
+
+	if (get_unimplemented_ctr(&ctr))
+		ctr = read_cpuid_cachetype();
+	return ctr;
+}
+
+static enum cache_type __get_cache_type(int level)
+{
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	return get_ctr() & CTR_CTYPE_MASK ?
+		CACHE_TYPE_SEPARATE : CACHE_TYPE_UNIFIED;
+}
+
+/*
+ *  +---------------------------------+
+ *  | 9  8  7  6 | 5  4  3 | 2 | 1  0 |
+ *  +---------------------------------+
+ *  |    size    |  assoc  | m |  len |
+ *  +---------------------------------+
+ * linelen        = 1 << (len + 3)
+ * multiplier     = 2 + m
+ * nsets          = 1 << (size + 6 - assoc - len)
+ * associativity  = multiplier << (assoc - 1)
+ * cache_size     = multiplier << (size + 8)
+ */
+#define CTR_LINESIZE_MASK	0x3
+#define CTR_MULTIPLIER_SHIFT	2
+#define CTR_MULTIPLIER_MASK	0x1
+#define CTR_ASSOCIAT_SHIFT	3
+#define CTR_ASSOCIAT_MASK	0x7
+#define CTR_SIZE_SHIFT		6
+#define CTR_SIZE_MASK		0xF
+#define CTR_DCACHE_SHIFT	12
+
+static void __ci_leaf_init(enum cache_type type, struct cacheinfo *this_leaf)
+{
+	unsigned int size, multiplier, assoc, len, tmp = get_ctr();
+
+	if (type == CACHE_TYPE_DATA)
+		tmp >>= CTR_DCACHE_SHIFT;
+
+	len = tmp & CTR_LINESIZE_MASK;
+	size = (tmp >> CTR_SIZE_SHIFT) & CTR_SIZE_MASK;
+	assoc = (tmp >> CTR_ASSOCIAT_SHIFT) & CTR_ASSOCIAT_MASK;
+	multiplier = ((tmp >> CTR_MULTIPLIER_SHIFT) & CTR_MULTIPLIER_MASK) + 2;
+
+	this_leaf->type = type;
+	this_leaf->coherency_line_size = 1 << (len + 3);
+	this_leaf->number_of_sets = 1 << (size + 6 - assoc - len);
+	this_leaf->ways_of_associativity = multiplier << (assoc - 1);
+	this_leaf->size = multiplier << (size + 8);
+}
+
+/* Ctypen, bits[3(n - 1) + 2 : 3(n - 1)], for n = 1 to 7 */
+#define CLIDR_CTYPE_SHIFT(level)	(3 * (level - 1))
+#define CLIDR_CTYPE_MASK(level)		(7 << CLIDR_CTYPE_SHIFT(level))
+#define CLIDR_CTYPE(clidr, level)	\
+	(((clidr) & CLIDR_CTYPE_MASK(level)) >> CLIDR_CTYPE_SHIFT(level))
+
+static inline enum cache_type __armv7_get_cache_type(int level)
+{
+	unsigned int clidr;
+
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	asm volatile ("mrc p15, 1, %0, c0, c0, 1" : "=r" (clidr));
+	return CLIDR_CTYPE(clidr, level);
+}
+
+/*
+ * NumSets, bits[27:13] - (Number of sets in cache) - 1
+ * Associativity, bits[12:3] - (Associativity of cache) - 1
+ * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
+ */
+#define CCSIDR_WRITE_THROUGH		BIT(31)
+#define CCSIDR_WRITE_BACK		BIT(30)
+#define CCSIDR_READ_ALLOCATE		BIT(29)
+#define CCSIDR_WRITE_ALLOCATE		BIT(28)
+#define CCSIDR_LINESIZE_MASK		0x7
+#define CCSIDR_ASSOCIATIVITY_SHIFT	3
+#define CCSIDR_ASSOCIATIVITY_MASK	0x3FF
+#define CCSIDR_NUMSETS_SHIFT		13
+#define CCSIDR_NUMSETS_MASK		0x7FF
+
+/*
+ * Which cache CCSIDR represents depends on CSSELR value
+ * Make sure no one else changes CSSELR during this
+ * smp_call_function_single prevents preemption for us
+ */
+static inline u32 get_ccsidr(u32 csselr)
+{
+	u32 ccsidr;
+
+	/* Put value into CSSELR */
+	asm volatile ("mcr p15, 2, %0, c0, c0, 0" : : "r" (csselr));
+	isb();
+	/* Read result out of CCSIDR */
+	asm volatile ("mrc p15, 1, %0, c0, c0, 0" : "=r" (ccsidr));
+
+	return ccsidr;
+}
+
+static void __armv7_ci_leaf_init(enum cache_type type,
+				 struct cacheinfo *this_leaf)
+{
+	bool is_instr_cache = type & CACHE_TYPE_INST;
+	u32 tmp = get_ccsidr((this_leaf->level - 1) << 1 | is_instr_cache);
+
+	this_leaf->type = type;
+	this_leaf->coherency_line_size =
+	    (1 << ((tmp & CCSIDR_LINESIZE_MASK) + 2)) * 4;
+	this_leaf->number_of_sets =
+	    ((tmp >> CCSIDR_NUMSETS_SHIFT) & CCSIDR_NUMSETS_MASK) + 1;
+	this_leaf->ways_of_associativity = ((tmp >> CCSIDR_ASSOCIATIVITY_SHIFT)
+					    & CCSIDR_ASSOCIATIVITY_MASK) + 1;
+	this_leaf->size = this_leaf->number_of_sets *
+	    this_leaf->coherency_line_size * this_leaf->ways_of_associativity;
+	this_leaf->attributes =
+		((tmp & CCSIDR_WRITE_THROUGH) ? CACHE_WRITE_THROUGH : 0) |
+		((tmp & CCSIDR_WRITE_BACK) ? CACHE_WRITE_BACK : 0) |
+		((tmp & CCSIDR_READ_ALLOCATE) ? CACHE_READ_ALLOCATE : 0) |
+		((tmp & CCSIDR_WRITE_ALLOCATE) ? CACHE_WRITE_ALLOCATE : 0);
+}
+
+static inline enum cache_type get_cache_type(int level)
+{
+	if (cache_is_armv7())
+		return __armv7_get_cache_type(level);
+	else
+		return __get_cache_type(level);
+}
+
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
+{
+	this_leaf->level = level;
+	if (cache_is_armv7())
+		__armv7_ci_leaf_init(type, this_leaf);
+	else
+		__ci_leaf_init(type, this_leaf);
+}
+
+static int __init_cache_level(unsigned int cpu)
+{
+	unsigned int ctype, level, leaves;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	for (level = 1, leaves = 0; level <= MAX_CACHE_LEVEL; level++) {
+		ctype = get_cache_type(level);
+		if (ctype == CACHE_TYPE_NOCACHE) {
+			level--;
+			break;
+		}
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	}
+
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
+
+	return 0;
+}
+
+static int __populate_cache_leaves(unsigned int cpu)
+{
+	unsigned int level, idx;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		type = get_cache_type(level);
+		if (type == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, type, level);
+		}
+	}
+	return 0;
+}
+
+DEFINE_SMP_CALL_FUNCTION(init_cache_level)
+DEFINE_SMP_CALL_FUNCTION(populate_cache_leaves)
diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig
index c348eaee7ee2..153abc3bac4e 100644
--- a/arch/arm/mm/Kconfig
+++ b/arch/arm/mm/Kconfig
@@ -494,30 +494,42 @@ config CPU_PABRT_V7
 # The cache model
 config CPU_CACHE_V4
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V4WT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V4WB
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V6
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V7
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_NOP
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_VIVT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_VIPT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_FA
 	bool
+	select CPU_HAS_CACHE
+
+config CPU_HAS_CACHE
+	bool
 
 if MMU
 # The copy-page model
@@ -845,6 +857,7 @@ config DMA_CACHE_RWFO
 
 config OUTER_CACHE
 	bool
+	select CPU_HAS_CACHE
 
 config OUTER_CACHE_SYNC
 	bool
-- 
1.8.3.2

[PATCH v2 8/9] ARM: kernel: add support for cpu cache information

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds support for cacheinfo on ARM platforms.

On ARMv7, the cache hierarchy can be identified through Cache Level ID
register(CLIDR) while the cache geometry is provided by Cache Size ID
register(CCSIDR).

On architecture versions before ARMv7, CLIDR and CCSIDR is not
implemented. The cache type register(CTR) provides both cache hierarchy
and geometry if implemented. For implementations that doesn't support
CTR, we need to list the probable value of CTR if it was implemented
along with the cpuid for the sake of simplicity to handle them.

Since the architecture doesn't provide any way of detecting the cpus
sharing particular cache, device tree is used fo the same purpose.
On non-DT platforms, first level caches are per-cpu while higher level
caches are assumed system-wide.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Nicolas Pitre <nicolas.pitre@linaro.org>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel at lists.infradead.org
---
 arch/arm/kernel/Makefile    |   1 +
 arch/arm/kernel/cacheinfo.c | 272 ++++++++++++++++++++++++++++++++++++++++++++
 arch/arm/mm/Kconfig         |  13 +++
 3 files changed, 286 insertions(+)
 create mode 100644 arch/arm/kernel/cacheinfo.c

--->8

Hi Russell,

Since for few CPUs like ARM11MP core which implements VMSAv6 + Advanced OS
Features, Linux returns cpu_architecture as armv7, I have added a list
for that as they are armv6 and don't implement CLIDR and group.
Let me know if there's any alternative to handle that.

This also depends on your patch
"ARM: make it easier to check the CPU part number correctly"

Regards,
Sudeep

diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile
index 38ddd9f83d0e..2c5ff0efb670 100644
--- a/arch/arm/kernel/Makefile
+++ b/arch/arm/kernel/Makefile
@@ -29,6 +29,7 @@ obj-y		+= entry-v7m.o v7m.o
 else
 obj-y		+= entry-armv.o
 endif
+obj-$(CONFIG_CPU_HAS_CACHE) += cacheinfo.o
 
 obj-$(CONFIG_OC_ETM)		+= etm.o
 obj-$(CONFIG_CPU_IDLE)		+= cpuidle.o
diff --git a/arch/arm/kernel/cacheinfo.c b/arch/arm/kernel/cacheinfo.c
new file mode 100644
index 000000000000..427ba641b33a
--- /dev/null
+++ b/arch/arm/kernel/cacheinfo.c
@@ -0,0 +1,272 @@
+/*
+ *  ARM cacheinfo support
+ *
+ *  Copyright (C) 2014 ARM Ltd.
+ *  All Rights Reserved
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/cpu.h>
+#include <linux/compiler.h>
+#include <linux/of.h>
+
+#include <asm/cputype.h>
+#include <asm/processor.h>
+#include <asm/system_info.h>
+
+#define cache_is_armv7() \
+	(cpu_architecture() >= CPU_ARCH_ARMv7 && !armv6_extended())
+#define MAX_CACHE_LEVEL		(cache_is_armv7() ? 7 : 1)
+
+#define CTR_CTYPE_SHIFT		24
+#define CTR_CTYPE_MASK		(1 << CTR_CTYPE_SHIFT)
+
+struct ctr_info {
+	unsigned int cpuid_part;
+	unsigned int ctr;
+};
+/*
+ *  Cache Type Register
+ *  +---------------------------------+
+ *  | 31 29 | 28 25 |24| 23 12 | 11 0 |
+ *  +---------------------------------+
+ *  | 0 0 0 | Ctype | S| Dsize | Isize|
+ *  +---------------------------------+
+ * The table below encodes only Dsize and Isize
+ */
+static struct ctr_info cache_ctr_list[] = {
+	{0x4400a100, 0x0016A16A }, /* 32kB D$, 32kB I$ */
+	{0x4400a110, 0x0012A16A }, /* 16kB D$, 32kB I$ */
+	{0x6900b110, 0x0012A16A }, /* 16kB D$, 32kB I$ */
+};
+
+/*
+ * List of CPUs reported as ARMv7 but don't implement CLIDR,
+ * CSSELR and CCSIDR. Cache information is still available from CTR
+ */
+static int armv6_ext_cpuid_part[] = { 0x4100b020, };
+
+static bool armv6_extended(void)
+{
+	int i, cpuid_part = read_cpuid_part();
+
+	for (i = 0; i < ARRAY_SIZE(armv6_ext_cpuid_part); i++)
+		if (armv6_ext_cpuid_part[i] == cpuid_part)
+			return true;
+	return false;
+}
+
+static int get_unimplemented_ctr(unsigned int *ctr)
+{
+	int i, cpuid_part = read_cpuid_part();
+
+	for (i = 0; i < ARRAY_SIZE(cache_ctr_list); i++)
+		if (cache_ctr_list[i].cpuid_part == cpuid_part) {
+			*ctr = cache_ctr_list[i].ctr;
+			return 0;
+		}
+	return -ENOENT;
+}
+
+static unsigned int get_ctr(void)
+{
+	unsigned int ctr;
+
+	if (get_unimplemented_ctr(&ctr))
+		ctr = read_cpuid_cachetype();
+	return ctr;
+}
+
+static enum cache_type __get_cache_type(int level)
+{
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	return get_ctr() & CTR_CTYPE_MASK ?
+		CACHE_TYPE_SEPARATE : CACHE_TYPE_UNIFIED;
+}
+
+/*
+ *  +---------------------------------+
+ *  | 9  8  7  6 | 5  4  3 | 2 | 1  0 |
+ *  +---------------------------------+
+ *  |    size    |  assoc  | m |  len |
+ *  +---------------------------------+
+ * linelen        = 1 << (len + 3)
+ * multiplier     = 2 + m
+ * nsets          = 1 << (size + 6 - assoc - len)
+ * associativity  = multiplier << (assoc - 1)
+ * cache_size     = multiplier << (size + 8)
+ */
+#define CTR_LINESIZE_MASK	0x3
+#define CTR_MULTIPLIER_SHIFT	2
+#define CTR_MULTIPLIER_MASK	0x1
+#define CTR_ASSOCIAT_SHIFT	3
+#define CTR_ASSOCIAT_MASK	0x7
+#define CTR_SIZE_SHIFT		6
+#define CTR_SIZE_MASK		0xF
+#define CTR_DCACHE_SHIFT	12
+
+static void __ci_leaf_init(enum cache_type type, struct cacheinfo *this_leaf)
+{
+	unsigned int size, multiplier, assoc, len, tmp = get_ctr();
+
+	if (type == CACHE_TYPE_DATA)
+		tmp >>= CTR_DCACHE_SHIFT;
+
+	len = tmp & CTR_LINESIZE_MASK;
+	size = (tmp >> CTR_SIZE_SHIFT) & CTR_SIZE_MASK;
+	assoc = (tmp >> CTR_ASSOCIAT_SHIFT) & CTR_ASSOCIAT_MASK;
+	multiplier = ((tmp >> CTR_MULTIPLIER_SHIFT) & CTR_MULTIPLIER_MASK) + 2;
+
+	this_leaf->type = type;
+	this_leaf->coherency_line_size = 1 << (len + 3);
+	this_leaf->number_of_sets = 1 << (size + 6 - assoc - len);
+	this_leaf->ways_of_associativity = multiplier << (assoc - 1);
+	this_leaf->size = multiplier << (size + 8);
+}
+
+/* Ctypen, bits[3(n - 1) + 2 : 3(n - 1)], for n = 1 to 7 */
+#define CLIDR_CTYPE_SHIFT(level)	(3 * (level - 1))
+#define CLIDR_CTYPE_MASK(level)		(7 << CLIDR_CTYPE_SHIFT(level))
+#define CLIDR_CTYPE(clidr, level)	\
+	(((clidr) & CLIDR_CTYPE_MASK(level)) >> CLIDR_CTYPE_SHIFT(level))
+
+static inline enum cache_type __armv7_get_cache_type(int level)
+{
+	unsigned int clidr;
+
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	asm volatile ("mrc p15, 1, %0, c0, c0, 1" : "=r" (clidr));
+	return CLIDR_CTYPE(clidr, level);
+}
+
+/*
+ * NumSets, bits[27:13] - (Number of sets in cache) - 1
+ * Associativity, bits[12:3] - (Associativity of cache) - 1
+ * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
+ */
+#define CCSIDR_WRITE_THROUGH		BIT(31)
+#define CCSIDR_WRITE_BACK		BIT(30)
+#define CCSIDR_READ_ALLOCATE		BIT(29)
+#define CCSIDR_WRITE_ALLOCATE		BIT(28)
+#define CCSIDR_LINESIZE_MASK		0x7
+#define CCSIDR_ASSOCIATIVITY_SHIFT	3
+#define CCSIDR_ASSOCIATIVITY_MASK	0x3FF
+#define CCSIDR_NUMSETS_SHIFT		13
+#define CCSIDR_NUMSETS_MASK		0x7FF
+
+/*
+ * Which cache CCSIDR represents depends on CSSELR value
+ * Make sure no one else changes CSSELR during this
+ * smp_call_function_single prevents preemption for us
+ */
+static inline u32 get_ccsidr(u32 csselr)
+{
+	u32 ccsidr;
+
+	/* Put value into CSSELR */
+	asm volatile ("mcr p15, 2, %0, c0, c0, 0" : : "r" (csselr));
+	isb();
+	/* Read result out of CCSIDR */
+	asm volatile ("mrc p15, 1, %0, c0, c0, 0" : "=r" (ccsidr));
+
+	return ccsidr;
+}
+
+static void __armv7_ci_leaf_init(enum cache_type type,
+				 struct cacheinfo *this_leaf)
+{
+	bool is_instr_cache = type & CACHE_TYPE_INST;
+	u32 tmp = get_ccsidr((this_leaf->level - 1) << 1 | is_instr_cache);
+
+	this_leaf->type = type;
+	this_leaf->coherency_line_size =
+	    (1 << ((tmp & CCSIDR_LINESIZE_MASK) + 2)) * 4;
+	this_leaf->number_of_sets =
+	    ((tmp >> CCSIDR_NUMSETS_SHIFT) & CCSIDR_NUMSETS_MASK) + 1;
+	this_leaf->ways_of_associativity = ((tmp >> CCSIDR_ASSOCIATIVITY_SHIFT)
+					    & CCSIDR_ASSOCIATIVITY_MASK) + 1;
+	this_leaf->size = this_leaf->number_of_sets *
+	    this_leaf->coherency_line_size * this_leaf->ways_of_associativity;
+	this_leaf->attributes =
+		((tmp & CCSIDR_WRITE_THROUGH) ? CACHE_WRITE_THROUGH : 0) |
+		((tmp & CCSIDR_WRITE_BACK) ? CACHE_WRITE_BACK : 0) |
+		((tmp & CCSIDR_READ_ALLOCATE) ? CACHE_READ_ALLOCATE : 0) |
+		((tmp & CCSIDR_WRITE_ALLOCATE) ? CACHE_WRITE_ALLOCATE : 0);
+}
+
+static inline enum cache_type get_cache_type(int level)
+{
+	if (cache_is_armv7())
+		return __armv7_get_cache_type(level);
+	else
+		return __get_cache_type(level);
+}
+
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
+{
+	this_leaf->level = level;
+	if (cache_is_armv7())
+		__armv7_ci_leaf_init(type, this_leaf);
+	else
+		__ci_leaf_init(type, this_leaf);
+}
+
+static int __init_cache_level(unsigned int cpu)
+{
+	unsigned int ctype, level, leaves;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	for (level = 1, leaves = 0; level <= MAX_CACHE_LEVEL; level++) {
+		ctype = get_cache_type(level);
+		if (ctype == CACHE_TYPE_NOCACHE) {
+			level--;
+			break;
+		}
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	}
+
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
+
+	return 0;
+}
+
+static int __populate_cache_leaves(unsigned int cpu)
+{
+	unsigned int level, idx;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		type = get_cache_type(level);
+		if (type == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, type, level);
+		}
+	}
+	return 0;
+}
+
+DEFINE_SMP_CALL_FUNCTION(init_cache_level)
+DEFINE_SMP_CALL_FUNCTION(populate_cache_leaves)
diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig
index c348eaee7ee2..153abc3bac4e 100644
--- a/arch/arm/mm/Kconfig
+++ b/arch/arm/mm/Kconfig
@@ -494,30 +494,42 @@ config CPU_PABRT_V7
 # The cache model
 config CPU_CACHE_V4
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V4WT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V4WB
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V6
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V7
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_NOP
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_VIVT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_VIPT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_FA
 	bool
+	select CPU_HAS_CACHE
+
+config CPU_HAS_CACHE
+	bool
 
 if MMU
 # The copy-page model
@@ -845,6 +857,7 @@ config DMA_CACHE_RWFO
 
 config OUTER_CACHE
 	bool
+	select CPU_HAS_CACHE
 
 config OUTER_CACHE_SYNC
 	bool
-- 
1.8.3.2

[PATCH v2 9/9] ARM: kernel: add outer cache support for cacheinfo implementation

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

In order to support outer cache in the cacheinfo infrastructure, a new
function 'get_info' is added to outer_cache_fns. This function is used
to get the outer cache information namely: line size, number of ways of
associativity and number of sets.

This patch adds 'get_info' supports to all L2 cache implementations on
ARM except Marvell's Feroceon L2 cache.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
---
 arch/arm/include/asm/outercache.h |  9 +++++++++
 arch/arm/kernel/cacheinfo.c       | 14 +++++++++++++-
 arch/arm/mm/cache-l2x0.c          | 35 ++++++++++++++++++++++++++++++++++-
 arch/arm/mm/cache-tauros2.c       | 35 +++++++++++++++++++++++++++++++++++
 arch/arm/mm/cache-xsc3l2.c        | 16 ++++++++++++++++
 5 files changed, 107 insertions(+), 2 deletions(-)

diff --git a/arch/arm/include/asm/outercache.h b/arch/arm/include/asm/outercache.h
index 891a56b35bcf..2765c8c61c8c 100644
--- a/arch/arm/include/asm/outercache.h
+++ b/arch/arm/include/asm/outercache.h
@@ -23,7 +23,10 @@
 
 #include <linux/types.h>
 
+struct cacheinfo;
+
 struct outer_cache_fns {
+	void (*get_info)(struct cacheinfo *info);
 	void (*inv_range)(unsigned long, unsigned long);
 	void (*clean_range)(unsigned long, unsigned long);
 	void (*flush_range)(unsigned long, unsigned long);
@@ -112,6 +115,11 @@ static inline void outer_resume(void)
 		outer_cache.resume();
 }
 
+static inline void outer_get_info(struct cacheinfo *info)
+{
+	if (outer_cache.get_info)
+		outer_cache.get_info(info);
+}
 #else
 
 static inline void outer_inv_range(phys_addr_t start, phys_addr_t end)
@@ -123,6 +131,7 @@ static inline void outer_flush_range(phys_addr_t start, phys_addr_t end)
 static inline void outer_flush_all(void) { }
 static inline void outer_disable(void) { }
 static inline void outer_resume(void) { }
+static inline void outer_get_info(struct outer_cache_info *info) { }
 
 #endif
 
diff --git a/arch/arm/kernel/cacheinfo.c b/arch/arm/kernel/cacheinfo.c
index 427ba641b33a..9a516e43bbf5 100644
--- a/arch/arm/kernel/cacheinfo.c
+++ b/arch/arm/kernel/cacheinfo.c
@@ -24,6 +24,7 @@
 #include <linux/of.h>
 
 #include <asm/cputype.h>
+#include <asm/outercache.h>
 #include <asm/processor.h>
 #include <asm/system_info.h>
 
@@ -217,11 +218,19 @@ static inline enum cache_type get_cache_type(int level)
 		return __get_cache_type(level);
 }
 
+static inline void __outer_ci_leaf_init(struct cacheinfo *this_leaf)
+{
+	outer_get_info(this_leaf);
+	BUG_ON(this_leaf->type == CACHE_TYPE_SEPARATE);
+}
+
 static void ci_leaf_init(struct cacheinfo *this_leaf,
 			 enum cache_type type, unsigned int level)
 {
 	this_leaf->level = level;
-	if (cache_is_armv7())
+	if (type == CACHE_TYPE_NOCACHE)	/* must be outer cache */
+		__outer_ci_leaf_init(this_leaf);
+	else if (cache_is_armv7())
 		__armv7_ci_leaf_init(type, this_leaf);
 	else
 		__ci_leaf_init(type, this_leaf);
@@ -245,6 +254,9 @@ static int __init_cache_level(unsigned int cpu)
 	this_cpu_ci->num_levels = level;
 	this_cpu_ci->num_leaves = leaves;
 
+	if (IS_ENABLED(CONFIG_OUTER_CACHE) && outer_cache.get_info)
+		this_cpu_ci->num_leaves++, this_cpu_ci->num_levels++;
+
 	return 0;
 }
 
diff --git a/arch/arm/mm/cache-l2x0.c b/arch/arm/mm/cache-l2x0.c
index 7c3fb41a462e..503bafdeb25b 100644
--- a/arch/arm/mm/cache-l2x0.c
+++ b/arch/arm/mm/cache-l2x0.c
@@ -17,6 +17,7 @@
  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  */
 #include <linux/cpu.h>
+#include <linux/cacheinfo.h>
 #include <linux/err.h>
 #include <linux/init.h>
 #include <linux/smp.h>
@@ -105,6 +106,22 @@ static inline void l2c_unlock(void __iomem *base, unsigned num)
 	}
 }
 
+static void __l2x0_getinfo(struct cacheinfo *this_leaf)
+{
+	unsigned int assoc = get_count_order(l2x0_way_mask);
+
+	this_leaf->size = l2x0_size;
+	this_leaf->coherency_line_size = CACHE_LINE_SIZE;
+	this_leaf->ways_of_associativity = assoc;
+	this_leaf->number_of_sets = l2x0_size / (assoc * CACHE_LINE_SIZE);
+}
+
+static void l2x0_getinfo(struct cacheinfo *this_leaf)
+{
+	this_leaf->type = CACHE_TYPE_UNIFIED;
+	__l2x0_getinfo(this_leaf);
+}
+
 /*
  * Enable the L2 cache controller.  This function must only be
  * called when the cache controller is known to be disabled.
@@ -309,6 +326,7 @@ static const struct l2c_init_data l2c210_data __initconst = {
 		.disable = l2c_disable,
 		.sync = l2c210_sync,
 		.resume = l2c210_resume,
+		.get_info = l2x0_getinfo,
 	},
 };
 
@@ -466,6 +484,7 @@ static const struct l2c_init_data l2c220_data = {
 		.disable = l2c_disable,
 		.sync = l2c220_sync,
 		.resume = l2c210_resume,
+		.get_info = l2x0_getinfo,
 	},
 };
 
@@ -814,6 +833,7 @@ static const struct l2c_init_data l2c310_init_fns __initconst = {
 		.disable = l2c310_disable,
 		.sync = l2c210_sync,
 		.resume = l2c310_resume,
+		.get_info = l2x0_getinfo,
 	},
 };
 
@@ -894,7 +914,6 @@ static void __init __l2c_init(const struct l2c_init_data *data,
 		data->enable(l2x0_base, aux, data->num_lock);
 
 	outer_cache = fns;
-
 	/*
 	 * It is strange to save the register state before initialisation,
 	 * but hey, this is what the DT implementations decided to do.
@@ -994,6 +1013,7 @@ static const struct l2c_init_data of_l2c210_data __initconst = {
 		.disable     = l2c_disable,
 		.sync        = l2c210_sync,
 		.resume      = l2c210_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1012,6 +1032,7 @@ static const struct l2c_init_data of_l2c220_data __initconst = {
 		.disable     = l2c_disable,
 		.sync        = l2c220_sync,
 		.resume      = l2c210_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1065,6 +1086,7 @@ static const struct l2c_init_data of_l2c310_data __initconst = {
 		.disable     = l2c310_disable,
 		.sync        = l2c210_sync,
 		.resume      = l2c310_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1092,6 +1114,7 @@ static const struct l2c_init_data of_l2c310_coherent_data __initconst = {
 		.flush_all   = l2c210_flush_all,
 		.disable     = l2c310_disable,
 		.resume      = l2c310_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1255,6 +1278,12 @@ static void __init aurora_of_parse(const struct device_node *np,
 	*aux_mask &= ~mask;
 }
 
+static void aurora_no_outer_data_getinfo(struct cacheinfo *this_leaf)
+{
+	this_leaf->type = CACHE_TYPE_INST;
+	__l2x0_getinfo(this_leaf);
+}
+
 static const struct l2c_init_data of_aurora_with_outer_data __initconst = {
 	.type = "Aurora",
 	.way_size_0 = SZ_4K,
@@ -1271,6 +1300,7 @@ static const struct l2c_init_data of_aurora_with_outer_data __initconst = {
 		.disable     = l2x0_disable,
 		.sync        = l2x0_cache_sync,
 		.resume      = aurora_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1284,6 +1314,7 @@ static const struct l2c_init_data of_aurora_no_outer_data __initconst = {
 	.save  = aurora_save,
 	.outer_cache = {
 		.resume      = aurora_resume,
+		.get_info    = aurora_no_outer_data_getinfo,
 	},
 };
 
@@ -1439,6 +1470,7 @@ static const struct l2c_init_data of_bcm_l2x0_data __initconst = {
 		.disable     = l2c310_disable,
 		.sync        = l2c210_sync,
 		.resume      = l2c310_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1475,6 +1507,7 @@ static const struct l2c_init_data of_tauros3_data __initconst = {
 	/* Tauros3 broadcasts L1 cache operations to L2 */
 	.outer_cache = {
 		.resume      = tauros3_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
diff --git a/arch/arm/mm/cache-tauros2.c b/arch/arm/mm/cache-tauros2.c
index b273739e6359..98efe3d12d94 100644
--- a/arch/arm/mm/cache-tauros2.c
+++ b/arch/arm/mm/cache-tauros2.c
@@ -60,6 +60,7 @@ static inline void tauros2_inv_pa(unsigned long addr)
  * noninclusive.
  */
 #define CACHE_LINE_SIZE		32
+#define CACHE_LINE_SHIFT	5
 
 static void tauros2_inv_range(unsigned long start, unsigned long end)
 {
@@ -131,6 +132,39 @@ static void tauros2_resume(void)
 	"mcr	p15, 0, %0, c1, c0, 0 @Enable L2 Cache\n\t"
 	: : "r" (0x0));
 }
+
+/*
+ *  +----------------------------------------+
+ *  | 11 10 9  8 | 7  6  5  4  3 | 2 |  1  0 |
+ *  +----------------------------------------+
+ *  |  way size  | associativity | - |line_sz|
+ *  +----------------------------------------+
+ */
+#define L2CTR_ASSOCIAT_SHIFT	3
+#define L2CTR_ASSOCIAT_MASK	0x1F
+#define L2CTR_WAYSIZE_SHIFT	8
+#define L2CTR_WAYSIZE_MASK	0xF
+#define CACHE_WAY_PER_SET(l2ctr)	\
+	(((l2_ctr) >> L2CTR_ASSOCIAT_SHIFT) & L2CTR_ASSOCIAT_MASK)
+#define CACHE_WAY_SIZE(l2ctr)		\
+	(8192 << (((l2ctr) >> L2CTR_WAYSIZE_SHIFT) & L2CTR_WAYSIZE_MASK))
+#define CACHE_SET_SIZE(l2ctr)	(CACHE_WAY_SIZE(l2ctr) >> CACHE_LINE_SHIFT)
+
+static void tauros2_getinfo(struct cacheinfo *info)
+{
+	unsigned int l2_ctr;
+
+	__asm__("mrc p15, 1, %0, c0, c0, 1" : "=r" (l2_ctr));
+
+	this_leaf->type = CACHE_TYPE_UNIFIED;
+	this_leaf->coherency_line_size = CACHE_LINE_SIZE;
+	this_leaf->ways_of_associativity = CACHE_WAY_PER_SET(l2_ctr);
+	this_leaf->number_of_sets = CACHE_SET_SIZE(l2_ctr);
+	this_leaf->size = this_leaf->coherency_line_size *
+			  this_leaf->number_of_sets *
+			  this_leaf->ways_of_associativity;
+}
+
 #endif
 
 static inline u32 __init read_extra_features(void)
@@ -226,6 +260,7 @@ static void __init tauros2_internal_init(unsigned int features)
 		outer_cache.flush_range = tauros2_flush_range;
 		outer_cache.disable = tauros2_disable;
 		outer_cache.resume = tauros2_resume;
+		outer_cache.get_info = tauros2_getinfo;
 	}
 #endif
 
diff --git a/arch/arm/mm/cache-xsc3l2.c b/arch/arm/mm/cache-xsc3l2.c
index 6c3edeb66e74..0f8ad0431a1b 100644
--- a/arch/arm/mm/cache-xsc3l2.c
+++ b/arch/arm/mm/cache-xsc3l2.c
@@ -201,6 +201,21 @@ static void xsc3_l2_flush_range(unsigned long start, unsigned long end)
 	dsb();
 }
 
+static void xsc3_l2_getinfo(struct outer_cache_info *info)
+{
+	unsigned long l2ctype;
+
+	__asm__("mrc p15, 1, %0, c0, c0, 1" : "=r" (l2ctype));
+
+	this_leaf->type = CACHE_TYPE_UNIFIED;
+	this_leaf->coherency_line_size = CACHE_LINE_SIZE;
+	this_leaf->ways_of_associativity = CACHE_WAY_PER_SET;
+	this_leaf->number_of_sets = CACHE_SET_SIZE(l2ctype);
+	this_leaf->size = this_leaf->coherency_line_size *
+			  this_leaf->number_of_sets *
+			  this_leaf->ways_of_associativity;
+}
+
 static int __init xsc3_l2_init(void)
 {
 	if (!cpu_is_xsc3() || !xsc3_l2_present())
@@ -213,6 +228,7 @@ static int __init xsc3_l2_init(void)
 		outer_cache.inv_range = xsc3_l2_inv_range;
 		outer_cache.clean_range = xsc3_l2_clean_range;
 		outer_cache.flush_range = xsc3_l2_flush_range;
+		outer_cache.get_info    = xsc3_l2_getinfo;
 	}
 
 	return 0;
-- 
1.8.3.2

[PATCH v2 9/9] ARM: kernel: add outer cache support for cacheinfo implementation

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

In order to support outer cache in the cacheinfo infrastructure, a new
function 'get_info' is added to outer_cache_fns. This function is used
to get the outer cache information namely: line size, number of ways of
associativity and number of sets.

This patch adds 'get_info' supports to all L2 cache implementations on
ARM except Marvell's Feroceon L2 cache.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel at lists.infradead.org
---
 arch/arm/include/asm/outercache.h |  9 +++++++++
 arch/arm/kernel/cacheinfo.c       | 14 +++++++++++++-
 arch/arm/mm/cache-l2x0.c          | 35 ++++++++++++++++++++++++++++++++++-
 arch/arm/mm/cache-tauros2.c       | 35 +++++++++++++++++++++++++++++++++++
 arch/arm/mm/cache-xsc3l2.c        | 16 ++++++++++++++++
 5 files changed, 107 insertions(+), 2 deletions(-)

diff --git a/arch/arm/include/asm/outercache.h b/arch/arm/include/asm/outercache.h
index 891a56b35bcf..2765c8c61c8c 100644
--- a/arch/arm/include/asm/outercache.h
+++ b/arch/arm/include/asm/outercache.h
@@ -23,7 +23,10 @@
 
 #include <linux/types.h>
 
+struct cacheinfo;
+
 struct outer_cache_fns {
+	void (*get_info)(struct cacheinfo *info);
 	void (*inv_range)(unsigned long, unsigned long);
 	void (*clean_range)(unsigned long, unsigned long);
 	void (*flush_range)(unsigned long, unsigned long);
@@ -112,6 +115,11 @@ static inline void outer_resume(void)
 		outer_cache.resume();
 }
 
+static inline void outer_get_info(struct cacheinfo *info)
+{
+	if (outer_cache.get_info)
+		outer_cache.get_info(info);
+}
 #else
 
 static inline void outer_inv_range(phys_addr_t start, phys_addr_t end)
@@ -123,6 +131,7 @@ static inline void outer_flush_range(phys_addr_t start, phys_addr_t end)
 static inline void outer_flush_all(void) { }
 static inline void outer_disable(void) { }
 static inline void outer_resume(void) { }
+static inline void outer_get_info(struct outer_cache_info *info) { }
 
 #endif
 
diff --git a/arch/arm/kernel/cacheinfo.c b/arch/arm/kernel/cacheinfo.c
index 427ba641b33a..9a516e43bbf5 100644
--- a/arch/arm/kernel/cacheinfo.c
+++ b/arch/arm/kernel/cacheinfo.c
@@ -24,6 +24,7 @@
 #include <linux/of.h>
 
 #include <asm/cputype.h>
+#include <asm/outercache.h>
 #include <asm/processor.h>
 #include <asm/system_info.h>
 
@@ -217,11 +218,19 @@ static inline enum cache_type get_cache_type(int level)
 		return __get_cache_type(level);
 }
 
+static inline void __outer_ci_leaf_init(struct cacheinfo *this_leaf)
+{
+	outer_get_info(this_leaf);
+	BUG_ON(this_leaf->type == CACHE_TYPE_SEPARATE);
+}
+
 static void ci_leaf_init(struct cacheinfo *this_leaf,
 			 enum cache_type type, unsigned int level)
 {
 	this_leaf->level = level;
-	if (cache_is_armv7())
+	if (type == CACHE_TYPE_NOCACHE)	/* must be outer cache */
+		__outer_ci_leaf_init(this_leaf);
+	else if (cache_is_armv7())
 		__armv7_ci_leaf_init(type, this_leaf);
 	else
 		__ci_leaf_init(type, this_leaf);
@@ -245,6 +254,9 @@ static int __init_cache_level(unsigned int cpu)
 	this_cpu_ci->num_levels = level;
 	this_cpu_ci->num_leaves = leaves;
 
+	if (IS_ENABLED(CONFIG_OUTER_CACHE) && outer_cache.get_info)
+		this_cpu_ci->num_leaves++, this_cpu_ci->num_levels++;
+
 	return 0;
 }
 
diff --git a/arch/arm/mm/cache-l2x0.c b/arch/arm/mm/cache-l2x0.c
index 7c3fb41a462e..503bafdeb25b 100644
--- a/arch/arm/mm/cache-l2x0.c
+++ b/arch/arm/mm/cache-l2x0.c
@@ -17,6 +17,7 @@
  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  */
 #include <linux/cpu.h>
+#include <linux/cacheinfo.h>
 #include <linux/err.h>
 #include <linux/init.h>
 #include <linux/smp.h>
@@ -105,6 +106,22 @@ static inline void l2c_unlock(void __iomem *base, unsigned num)
 	}
 }
 
+static void __l2x0_getinfo(struct cacheinfo *this_leaf)
+{
+	unsigned int assoc = get_count_order(l2x0_way_mask);
+
+	this_leaf->size = l2x0_size;
+	this_leaf->coherency_line_size = CACHE_LINE_SIZE;
+	this_leaf->ways_of_associativity = assoc;
+	this_leaf->number_of_sets = l2x0_size / (assoc * CACHE_LINE_SIZE);
+}
+
+static void l2x0_getinfo(struct cacheinfo *this_leaf)
+{
+	this_leaf->type = CACHE_TYPE_UNIFIED;
+	__l2x0_getinfo(this_leaf);
+}
+
 /*
  * Enable the L2 cache controller.  This function must only be
  * called when the cache controller is known to be disabled.
@@ -309,6 +326,7 @@ static const struct l2c_init_data l2c210_data __initconst = {
 		.disable = l2c_disable,
 		.sync = l2c210_sync,
 		.resume = l2c210_resume,
+		.get_info = l2x0_getinfo,
 	},
 };
 
@@ -466,6 +484,7 @@ static const struct l2c_init_data l2c220_data = {
 		.disable = l2c_disable,
 		.sync = l2c220_sync,
 		.resume = l2c210_resume,
+		.get_info = l2x0_getinfo,
 	},
 };
 
@@ -814,6 +833,7 @@ static const struct l2c_init_data l2c310_init_fns __initconst = {
 		.disable = l2c310_disable,
 		.sync = l2c210_sync,
 		.resume = l2c310_resume,
+		.get_info = l2x0_getinfo,
 	},
 };
 
@@ -894,7 +914,6 @@ static void __init __l2c_init(const struct l2c_init_data *data,
 		data->enable(l2x0_base, aux, data->num_lock);
 
 	outer_cache = fns;
-
 	/*
 	 * It is strange to save the register state before initialisation,
 	 * but hey, this is what the DT implementations decided to do.
@@ -994,6 +1013,7 @@ static const struct l2c_init_data of_l2c210_data __initconst = {
 		.disable     = l2c_disable,
 		.sync        = l2c210_sync,
 		.resume      = l2c210_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1012,6 +1032,7 @@ static const struct l2c_init_data of_l2c220_data __initconst = {
 		.disable     = l2c_disable,
 		.sync        = l2c220_sync,
 		.resume      = l2c210_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1065,6 +1086,7 @@ static const struct l2c_init_data of_l2c310_data __initconst = {
 		.disable     = l2c310_disable,
 		.sync        = l2c210_sync,
 		.resume      = l2c310_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1092,6 +1114,7 @@ static const struct l2c_init_data of_l2c310_coherent_data __initconst = {
 		.flush_all   = l2c210_flush_all,
 		.disable     = l2c310_disable,
 		.resume      = l2c310_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1255,6 +1278,12 @@ static void __init aurora_of_parse(const struct device_node *np,
 	*aux_mask &= ~mask;
 }
 
+static void aurora_no_outer_data_getinfo(struct cacheinfo *this_leaf)
+{
+	this_leaf->type = CACHE_TYPE_INST;
+	__l2x0_getinfo(this_leaf);
+}
+
 static const struct l2c_init_data of_aurora_with_outer_data __initconst = {
 	.type = "Aurora",
 	.way_size_0 = SZ_4K,
@@ -1271,6 +1300,7 @@ static const struct l2c_init_data of_aurora_with_outer_data __initconst = {
 		.disable     = l2x0_disable,
 		.sync        = l2x0_cache_sync,
 		.resume      = aurora_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1284,6 +1314,7 @@ static const struct l2c_init_data of_aurora_no_outer_data __initconst = {
 	.save  = aurora_save,
 	.outer_cache = {
 		.resume      = aurora_resume,
+		.get_info    = aurora_no_outer_data_getinfo,
 	},
 };
 
@@ -1439,6 +1470,7 @@ static const struct l2c_init_data of_bcm_l2x0_data __initconst = {
 		.disable     = l2c310_disable,
 		.sync        = l2c210_sync,
 		.resume      = l2c310_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1475,6 +1507,7 @@ static const struct l2c_init_data of_tauros3_data __initconst = {
 	/* Tauros3 broadcasts L1 cache operations to L2 */
 	.outer_cache = {
 		.resume      = tauros3_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
diff --git a/arch/arm/mm/cache-tauros2.c b/arch/arm/mm/cache-tauros2.c
index b273739e6359..98efe3d12d94 100644
--- a/arch/arm/mm/cache-tauros2.c
+++ b/arch/arm/mm/cache-tauros2.c
@@ -60,6 +60,7 @@ static inline void tauros2_inv_pa(unsigned long addr)
  * noninclusive.
  */
 #define CACHE_LINE_SIZE		32
+#define CACHE_LINE_SHIFT	5
 
 static void tauros2_inv_range(unsigned long start, unsigned long end)
 {
@@ -131,6 +132,39 @@ static void tauros2_resume(void)
 	"mcr	p15, 0, %0, c1, c0, 0 @Enable L2 Cache\n\t"
 	: : "r" (0x0));
 }
+
+/*
+ *  +----------------------------------------+
+ *  | 11 10 9  8 | 7  6  5  4  3 | 2 |  1  0 |
+ *  +----------------------------------------+
+ *  |  way size  | associativity | - |line_sz|
+ *  +----------------------------------------+
+ */
+#define L2CTR_ASSOCIAT_SHIFT	3
+#define L2CTR_ASSOCIAT_MASK	0x1F
+#define L2CTR_WAYSIZE_SHIFT	8
+#define L2CTR_WAYSIZE_MASK	0xF
+#define CACHE_WAY_PER_SET(l2ctr)	\
+	(((l2_ctr) >> L2CTR_ASSOCIAT_SHIFT) & L2CTR_ASSOCIAT_MASK)
+#define CACHE_WAY_SIZE(l2ctr)		\
+	(8192 << (((l2ctr) >> L2CTR_WAYSIZE_SHIFT) & L2CTR_WAYSIZE_MASK))
+#define CACHE_SET_SIZE(l2ctr)	(CACHE_WAY_SIZE(l2ctr) >> CACHE_LINE_SHIFT)
+
+static void tauros2_getinfo(struct cacheinfo *info)
+{
+	unsigned int l2_ctr;
+
+	__asm__("mrc p15, 1, %0, c0, c0, 1" : "=r" (l2_ctr));
+
+	this_leaf->type = CACHE_TYPE_UNIFIED;
+	this_leaf->coherency_line_size = CACHE_LINE_SIZE;
+	this_leaf->ways_of_associativity = CACHE_WAY_PER_SET(l2_ctr);
+	this_leaf->number_of_sets = CACHE_SET_SIZE(l2_ctr);
+	this_leaf->size = this_leaf->coherency_line_size *
+			  this_leaf->number_of_sets *
+			  this_leaf->ways_of_associativity;
+}
+
 #endif
 
 static inline u32 __init read_extra_features(void)
@@ -226,6 +260,7 @@ static void __init tauros2_internal_init(unsigned int features)
 		outer_cache.flush_range = tauros2_flush_range;
 		outer_cache.disable = tauros2_disable;
 		outer_cache.resume = tauros2_resume;
+		outer_cache.get_info = tauros2_getinfo;
 	}
 #endif
 
diff --git a/arch/arm/mm/cache-xsc3l2.c b/arch/arm/mm/cache-xsc3l2.c
index 6c3edeb66e74..0f8ad0431a1b 100644
--- a/arch/arm/mm/cache-xsc3l2.c
+++ b/arch/arm/mm/cache-xsc3l2.c
@@ -201,6 +201,21 @@ static void xsc3_l2_flush_range(unsigned long start, unsigned long end)
 	dsb();
 }
 
+static void xsc3_l2_getinfo(struct outer_cache_info *info)
+{
+	unsigned long l2ctype;
+
+	__asm__("mrc p15, 1, %0, c0, c0, 1" : "=r" (l2ctype));
+
+	this_leaf->type = CACHE_TYPE_UNIFIED;
+	this_leaf->coherency_line_size = CACHE_LINE_SIZE;
+	this_leaf->ways_of_associativity = CACHE_WAY_PER_SET;
+	this_leaf->number_of_sets = CACHE_SET_SIZE(l2ctype);
+	this_leaf->size = this_leaf->coherency_line_size *
+			  this_leaf->number_of_sets *
+			  this_leaf->ways_of_associativity;
+}
+
 static int __init xsc3_l2_init(void)
 {
 	if (!cpu_is_xsc3() || !xsc3_l2_present())
@@ -213,6 +228,7 @@ static int __init xsc3_l2_init(void)
 		outer_cache.inv_range = xsc3_l2_inv_range;
 		outer_cache.clean_range = xsc3_l2_clean_range;
 		outer_cache.flush_range = xsc3_l2_flush_range;
+		outer_cache.get_info    = xsc3_l2_getinfo;
 	}
 
 	return 0;
-- 
1.8.3.2

[PATCH v3 00/11] drivers: cacheinfo support

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This series adds a generic cacheinfo support similar to topology. The
implementation is based on x86 cacheinfo support. Currently x86, powerpc,
ia64 and s390 have their own implementations. While adding similar support
to ARM and ARM64, here is the attempt to make it generic quite similar to
topology info support. It also adds the missing ABI documentation for
the cacheinfo sysfs which is already being used.

It moves all the existing different implementations on x86, ia64, powerpc
and s390 to use the generic cacheinfo infrastructure introduced here.
These changes on non-ARM platforms are only compile tested and tested on x86.

This series also adds support for ARM and ARM64 architectures based on
the generic support.

The code can be fetched from:
 git://linux-arm.org/linux-skn cacheinfo

Changes v2->v3:
	- Added {allocation,write}_policy instead of single attributes sysfs
	  (attributes retained on ia64 privately as it was used only on that)
	- factored out show_cpumap into separate helper in cpumask.h
	- populate cpu_{map,list} for non-DT system if they are not populated
	  by arch specific callbacks
	- removed use of sysfs *_show callback in cache_attrs_is_visible
	- all the review comments from Stephen Boyd implemented

Changes v1->v2:
	- removed custom device_{add,remove}_attrs, using is_visible callback
	  instead(suggested by GregKH)
	- arm64: changes as per MarkR review comments
	- Moved smp_call_function_single to architectures using it(arm, arm64,
	  x86) (suggested by Stephen Boyd)
	- arm (mostly changes as per RMK's review comments)
		- fixed to allow v7 + v6 build
		- l2 cache changes to remove extra structure
		- populated CTR for few StrongARM CPU's not implementing CTR

[v1] https://lkml.org/lkml/2014/6/25/603
[v2] https://lkml.org/lkml/2014/7/25/467

Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-ia64@vger.kernel.org
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-arm-kernel@lists.infradead.org

Sudeep Holla (11):
  cpumask: factor out show_cpumap into separate helper function
  topology: replace custom attribute macros with standard DEVICE_ATTR*
  drivers: base: add new class "cpu" to group cpu devices
  drivers: base: support cpu cache information interface to userspace
    via sysfs
  ia64: move cacheinfo sysfs to generic cacheinfo infrastructure
  s390: move cacheinfo sysfs to generic cacheinfo infrastructure
  x86: move cacheinfo sysfs to generic cacheinfo infrastructure
  powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure
  ARM64: kernel: add support for cpu cache information
  ARM: kernel: add support for cpu cache information
  ARM: kernel: add outer cache support for cacheinfo implementation

 Documentation/ABI/testing/sysfs-devices-system-cpu |  47 ++
 arch/arm/include/asm/outercache.h                  |   9 +
 arch/arm/kernel/Makefile                           |   1 +
 arch/arm/kernel/cacheinfo.c                        | 287 ++++++++
 arch/arm/mm/Kconfig                                |  13 +
 arch/arm/mm/cache-l2x0.c                           |  35 +-
 arch/arm/mm/cache-tauros2.c                        |  36 +
 arch/arm/mm/cache-xsc3l2.c                         |  17 +
 arch/arm64/kernel/Makefile                         |   2 +-
 arch/arm64/kernel/cacheinfo.c                      | 142 ++++
 arch/ia64/kernel/topology.c                        | 421 +++--------
 arch/powerpc/kernel/cacheinfo.c                    | 812 +++------------------
 arch/powerpc/kernel/cacheinfo.h                    |   8 -
 arch/powerpc/kernel/sysfs.c                        |  12 +-
 arch/s390/kernel/cache.c                           | 388 +++-------
 arch/x86/kernel/cpu/intel_cacheinfo.c              | 709 +++++-------------
 arch/x86/kernel/cpu/perf_event_amd_iommu.c         |   5 +-
 arch/x86/kernel/cpu/perf_event_amd_uncore.c        |   6 +-
 arch/x86/kernel/cpu/perf_event_intel_rapl.c        |   6 +-
 arch/x86/kernel/cpu/perf_event_intel_uncore.c      |   6 +-
 drivers/acpi/acpi_pad.c                            |   6 +-
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 543 ++++++++++++++
 drivers/base/core.c                                |  39 +-
 drivers/base/cpu.c                                 |  12 +-
 drivers/base/node.c                                |  14 +-
 drivers/base/topology.c                            |  71 +-
 drivers/pci/pci-sysfs.c                            |  39 +-
 include/linux/cacheinfo.h                          | 100 +++
 include/linux/cpu.h                                |   2 +
 include/linux/cpumask.h                            |  27 +
 31 files changed, 1826 insertions(+), 1991 deletions(-)
 create mode 100644 arch/arm/kernel/cacheinfo.c
 create mode 100644 arch/arm64/kernel/cacheinfo.c
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

-- 
1.8.3.2

[PATCH v3 00/11] drivers: cacheinfo support

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This series adds a generic cacheinfo support similar to topology. The
implementation is based on x86 cacheinfo support. Currently x86, powerpc,
ia64 and s390 have their own implementations. While adding similar support
to ARM and ARM64, here is the attempt to make it generic quite similar to
topology info support. It also adds the missing ABI documentation for
the cacheinfo sysfs which is already being used.

It moves all the existing different implementations on x86, ia64, powerpc
and s390 to use the generic cacheinfo infrastructure introduced here.
These changes on non-ARM platforms are only compile tested and tested on x86.

This series also adds support for ARM and ARM64 architectures based on
the generic support.

The code can be fetched from:
 git://linux-arm.org/linux-skn cacheinfo

Changes v2->v3:
	- Added {allocation,write}_policy instead of single attributes sysfs
	  (attributes retained on ia64 privately as it was used only on that)
	- factored out show_cpumap into separate helper in cpumask.h
	- populate cpu_{map,list} for non-DT system if they are not populated
	  by arch specific callbacks
	- removed use of sysfs *_show callback in cache_attrs_is_visible
	- all the review comments from Stephen Boyd implemented

Changes v1->v2:
	- removed custom device_{add,remove}_attrs, using is_visible callback
	  instead(suggested by GregKH)
	- arm64: changes as per MarkR review comments
	- Moved smp_call_function_single to architectures using it(arm, arm64,
	  x86) (suggested by Stephen Boyd)
	- arm (mostly changes as per RMK's review comments)
		- fixed to allow v7 + v6 build
		- l2 cache changes to remove extra structure
		- populated CTR for few StrongARM CPU's not implementing CTR

[v1] https://lkml.org/lkml/2014/6/25/603
[v2] https://lkml.org/lkml/2014/7/25/467

Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-ia64@vger.kernel.org
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-arm-kernel@lists.infradead.org

Sudeep Holla (11):
  cpumask: factor out show_cpumap into separate helper function
  topology: replace custom attribute macros with standard DEVICE_ATTR*
  drivers: base: add new class "cpu" to group cpu devices
  drivers: base: support cpu cache information interface to userspace
    via sysfs
  ia64: move cacheinfo sysfs to generic cacheinfo infrastructure
  s390: move cacheinfo sysfs to generic cacheinfo infrastructure
  x86: move cacheinfo sysfs to generic cacheinfo infrastructure
  powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure
  ARM64: kernel: add support for cpu cache information
  ARM: kernel: add support for cpu cache information
  ARM: kernel: add outer cache support for cacheinfo implementation

 Documentation/ABI/testing/sysfs-devices-system-cpu |  47 ++
 arch/arm/include/asm/outercache.h                  |   9 +
 arch/arm/kernel/Makefile                           |   1 +
 arch/arm/kernel/cacheinfo.c                        | 287 ++++++++
 arch/arm/mm/Kconfig                                |  13 +
 arch/arm/mm/cache-l2x0.c                           |  35 +-
 arch/arm/mm/cache-tauros2.c                        |  36 +
 arch/arm/mm/cache-xsc3l2.c                         |  17 +
 arch/arm64/kernel/Makefile                         |   2 +-
 arch/arm64/kernel/cacheinfo.c                      | 142 ++++
 arch/ia64/kernel/topology.c                        | 421 +++--------
 arch/powerpc/kernel/cacheinfo.c                    | 812 +++------------------
 arch/powerpc/kernel/cacheinfo.h                    |   8 -
 arch/powerpc/kernel/sysfs.c                        |  12 +-
 arch/s390/kernel/cache.c                           | 388 +++-------
 arch/x86/kernel/cpu/intel_cacheinfo.c              | 709 +++++-------------
 arch/x86/kernel/cpu/perf_event_amd_iommu.c         |   5 +-
 arch/x86/kernel/cpu/perf_event_amd_uncore.c        |   6 +-
 arch/x86/kernel/cpu/perf_event_intel_rapl.c        |   6 +-
 arch/x86/kernel/cpu/perf_event_intel_uncore.c      |   6 +-
 drivers/acpi/acpi_pad.c                            |   6 +-
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 543 ++++++++++++++
 drivers/base/core.c                                |  39 +-
 drivers/base/cpu.c                                 |  12 +-
 drivers/base/node.c                                |  14 +-
 drivers/base/topology.c                            |  71 +-
 drivers/pci/pci-sysfs.c                            |  39 +-
 include/linux/cacheinfo.h                          | 100 +++
 include/linux/cpu.h                                |   2 +
 include/linux/cpumask.h                            |  27 +
 31 files changed, 1826 insertions(+), 1991 deletions(-)
 create mode 100644 arch/arm/kernel/cacheinfo.c
 create mode 100644 arch/arm64/kernel/cacheinfo.c
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

-- 
1.8.3.2

[PATCH v3 00/11] drivers: cacheinfo support

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This series adds a generic cacheinfo support similar to topology. The
implementation is based on x86 cacheinfo support. Currently x86, powerpc,
ia64 and s390 have their own implementations. While adding similar support
to ARM and ARM64, here is the attempt to make it generic quite similar to
topology info support. It also adds the missing ABI documentation for
the cacheinfo sysfs which is already being used.

It moves all the existing different implementations on x86, ia64, powerpc
and s390 to use the generic cacheinfo infrastructure introduced here.
These changes on non-ARM platforms are only compile tested and tested on x86.

This series also adds support for ARM and ARM64 architectures based on
the generic support.

The code can be fetched from:
 git://linux-arm.org/linux-skn cacheinfo

Changes v2->v3:
	- Added {allocation,write}_policy instead of single attributes sysfs
	  (attributes retained on ia64 privately as it was used only on that)
	- factored out show_cpumap into separate helper in cpumask.h
	- populate cpu_{map,list} for non-DT system if they are not populated
	  by arch specific callbacks
	- removed use of sysfs *_show callback in cache_attrs_is_visible
	- all the review comments from Stephen Boyd implemented

Changes v1->v2:
	- removed custom device_{add,remove}_attrs, using is_visible callback
	  instead(suggested by GregKH)
	- arm64: changes as per MarkR review comments
	- Moved smp_call_function_single to architectures using it(arm, arm64,
	  x86) (suggested by Stephen Boyd)
	- arm (mostly changes as per RMK's review comments)
		- fixed to allow v7 + v6 build
		- l2 cache changes to remove extra structure
		- populated CTR for few StrongARM CPU's not implementing CTR

[v1] https://lkml.org/lkml/2014/6/25/603
[v2] https://lkml.org/lkml/2014/7/25/467

Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-ia64 at vger.kernel.org
Cc: linux390 at de.ibm.com
Cc: linux-s390 at vger.kernel.org
Cc: x86 at kernel.org
Cc: linuxppc-dev at lists.ozlabs.org
Cc: linux-arm-kernel at lists.infradead.org

Sudeep Holla (11):
  cpumask: factor out show_cpumap into separate helper function
  topology: replace custom attribute macros with standard DEVICE_ATTR*
  drivers: base: add new class "cpu" to group cpu devices
  drivers: base: support cpu cache information interface to userspace
    via sysfs
  ia64: move cacheinfo sysfs to generic cacheinfo infrastructure
  s390: move cacheinfo sysfs to generic cacheinfo infrastructure
  x86: move cacheinfo sysfs to generic cacheinfo infrastructure
  powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure
  ARM64: kernel: add support for cpu cache information
  ARM: kernel: add support for cpu cache information
  ARM: kernel: add outer cache support for cacheinfo implementation

 Documentation/ABI/testing/sysfs-devices-system-cpu |  47 ++
 arch/arm/include/asm/outercache.h                  |   9 +
 arch/arm/kernel/Makefile                           |   1 +
 arch/arm/kernel/cacheinfo.c                        | 287 ++++++++
 arch/arm/mm/Kconfig                                |  13 +
 arch/arm/mm/cache-l2x0.c                           |  35 +-
 arch/arm/mm/cache-tauros2.c                        |  36 +
 arch/arm/mm/cache-xsc3l2.c                         |  17 +
 arch/arm64/kernel/Makefile                         |   2 +-
 arch/arm64/kernel/cacheinfo.c                      | 142 ++++
 arch/ia64/kernel/topology.c                        | 421 +++--------
 arch/powerpc/kernel/cacheinfo.c                    | 812 +++------------------
 arch/powerpc/kernel/cacheinfo.h                    |   8 -
 arch/powerpc/kernel/sysfs.c                        |  12 +-
 arch/s390/kernel/cache.c                           | 388 +++-------
 arch/x86/kernel/cpu/intel_cacheinfo.c              | 709 +++++-------------
 arch/x86/kernel/cpu/perf_event_amd_iommu.c         |   5 +-
 arch/x86/kernel/cpu/perf_event_amd_uncore.c        |   6 +-
 arch/x86/kernel/cpu/perf_event_intel_rapl.c        |   6 +-
 arch/x86/kernel/cpu/perf_event_intel_uncore.c      |   6 +-
 drivers/acpi/acpi_pad.c                            |   6 +-
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 543 ++++++++++++++
 drivers/base/core.c                                |  39 +-
 drivers/base/cpu.c                                 |  12 +-
 drivers/base/node.c                                |  14 +-
 drivers/base/topology.c                            |  71 +-
 drivers/pci/pci-sysfs.c                            |  39 +-
 include/linux/cacheinfo.h                          | 100 +++
 include/linux/cpu.h                                |   2 +
 include/linux/cpumask.h                            |  27 +
 31 files changed, 1826 insertions(+), 1991 deletions(-)
 create mode 100644 arch/arm/kernel/cacheinfo.c
 create mode 100644 arch/arm64/kernel/cacheinfo.c
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

-- 
1.8.3.2

[PATCH v3 00/11] drivers: cacheinfo support

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This series adds a generic cacheinfo support similar to topology. The
implementation is based on x86 cacheinfo support. Currently x86, powerpc,
ia64 and s390 have their own implementations. While adding similar support
to ARM and ARM64, here is the attempt to make it generic quite similar to
topology info support. It also adds the missing ABI documentation for
the cacheinfo sysfs which is already being used.

It moves all the existing different implementations on x86, ia64, powerpc
and s390 to use the generic cacheinfo infrastructure introduced here.
These changes on non-ARM platforms are only compile tested and tested on x86.

This series also adds support for ARM and ARM64 architectures based on
the generic support.

The code can be fetched from:
 git://linux-arm.org/linux-skn cacheinfo

Changes v2->v3:
	- Added {allocation,write}_policy instead of single attributes sysfs
	  (attributes retained on ia64 privately as it was used only on that)
	- factored out show_cpumap into separate helper in cpumask.h
	- populate cpu_{map,list} for non-DT system if they are not populated
	  by arch specific callbacks
	- removed use of sysfs *_show callback in cache_attrs_is_visible
	- all the review comments from Stephen Boyd implemented

Changes v1->v2:
	- removed custom device_{add,remove}_attrs, using is_visible callback
	  instead(suggested by GregKH)
	- arm64: changes as per MarkR review comments
	- Moved smp_call_function_single to architectures using it(arm, arm64,
	  x86) (suggested by Stephen Boyd)
	- arm (mostly changes as per RMK's review comments)
		- fixed to allow v7 + v6 build
		- l2 cache changes to remove extra structure
		- populated CTR for few StrongARM CPU's not implementing CTR

[v1] https://lkml.org/lkml/2014/6/25/603
[v2] https://lkml.org/lkml/2014/7/25/467

Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-ia64@vger.kernel.org
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-arm-kernel@lists.infradead.org

Sudeep Holla (11):
  cpumask: factor out show_cpumap into separate helper function
  topology: replace custom attribute macros with standard DEVICE_ATTR*
  drivers: base: add new class "cpu" to group cpu devices
  drivers: base: support cpu cache information interface to userspace
    via sysfs
  ia64: move cacheinfo sysfs to generic cacheinfo infrastructure
  s390: move cacheinfo sysfs to generic cacheinfo infrastructure
  x86: move cacheinfo sysfs to generic cacheinfo infrastructure
  powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure
  ARM64: kernel: add support for cpu cache information
  ARM: kernel: add support for cpu cache information
  ARM: kernel: add outer cache support for cacheinfo implementation

 Documentation/ABI/testing/sysfs-devices-system-cpu |  47 ++
 arch/arm/include/asm/outercache.h                  |   9 +
 arch/arm/kernel/Makefile                           |   1 +
 arch/arm/kernel/cacheinfo.c                        | 287 ++++++++
 arch/arm/mm/Kconfig                                |  13 +
 arch/arm/mm/cache-l2x0.c                           |  35 +-
 arch/arm/mm/cache-tauros2.c                        |  36 +
 arch/arm/mm/cache-xsc3l2.c                         |  17 +
 arch/arm64/kernel/Makefile                         |   2 +-
 arch/arm64/kernel/cacheinfo.c                      | 142 ++++
 arch/ia64/kernel/topology.c                        | 421 +++--------
 arch/powerpc/kernel/cacheinfo.c                    | 812 +++------------------
 arch/powerpc/kernel/cacheinfo.h                    |   8 -
 arch/powerpc/kernel/sysfs.c                        |  12 +-
 arch/s390/kernel/cache.c                           | 388 +++-------
 arch/x86/kernel/cpu/intel_cacheinfo.c              | 709 +++++-------------
 arch/x86/kernel/cpu/perf_event_amd_iommu.c         |   5 +-
 arch/x86/kernel/cpu/perf_event_amd_uncore.c        |   6 +-
 arch/x86/kernel/cpu/perf_event_intel_rapl.c        |   6 +-
 arch/x86/kernel/cpu/perf_event_intel_uncore.c      |   6 +-
 drivers/acpi/acpi_pad.c                            |   6 +-
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 543 ++++++++++++++
 drivers/base/core.c                                |  39 +-
 drivers/base/cpu.c                                 |  12 +-
 drivers/base/node.c                                |  14 +-
 drivers/base/topology.c                            |  71 +-
 drivers/pci/pci-sysfs.c                            |  39 +-
 include/linux/cacheinfo.h                          | 100 +++
 include/linux/cpu.h                                |   2 +
 include/linux/cpumask.h                            |  27 +
 31 files changed, 1826 insertions(+), 1991 deletions(-)
 create mode 100644 arch/arm/kernel/cacheinfo.c
 create mode 100644 arch/arm64/kernel/cacheinfo.c
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

-- 
1.8.3.2

[PATCH v3 01/11] cpumask: factor out show_cpumap into separate helper function

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

Many sysfs *_show use cpu{list,mask}_scnprintf to copy cpumap to the
buffer aligned to PAGE_SIZE, append '\n' and '\0' to return null
terminated buffer with newline.

This patch creates a new helper function cpumap_copy_to_buf in
cpumask.h and consolidates most of them using this new helper.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 arch/x86/kernel/cpu/perf_event_amd_iommu.c    |  5 +---
 arch/x86/kernel/cpu/perf_event_amd_uncore.c   |  6 +----
 arch/x86/kernel/cpu/perf_event_intel_rapl.c   |  6 +----
 arch/x86/kernel/cpu/perf_event_intel_uncore.c |  6 +----
 drivers/acpi/acpi_pad.c                       |  6 +----
 drivers/base/cpu.c                            |  5 +---
 drivers/base/node.c                           | 14 +++-------
 drivers/base/topology.c                       | 22 ++-------------
 drivers/pci/pci-sysfs.c                       | 39 +++++++--------------------
 include/linux/cpumask.h                       | 27 +++++++++++++++++++
 10 files changed, 49 insertions(+), 87 deletions(-)

diff --git a/arch/x86/kernel/cpu/perf_event_amd_iommu.c b/arch/x86/kernel/cpu/perf_event_amd_iommu.c
index 639d1289b1ba..61a670de0ef2 100644
--- a/arch/x86/kernel/cpu/perf_event_amd_iommu.c
+++ b/arch/x86/kernel/cpu/perf_event_amd_iommu.c
@@ -130,10 +130,7 @@ static ssize_t _iommu_cpumask_show(struct device *dev,
 				   struct device_attribute *attr,
 				   char *buf)
 {
-	int n = cpulist_scnprintf(buf, PAGE_SIZE - 2, &iommu_cpumask);
-	buf[n++] = '\n';
-	buf[n] = '\0';
-	return n;
+	return cpumap_copy_to_buf(true, &iommu_cpumask, buf);
 }
 static DEVICE_ATTR(cpumask, S_IRUGO, _iommu_cpumask_show, NULL);
 
diff --git a/arch/x86/kernel/cpu/perf_event_amd_uncore.c b/arch/x86/kernel/cpu/perf_event_amd_uncore.c
index 30790d798e6b..d96a37ed8570 100644
--- a/arch/x86/kernel/cpu/perf_event_amd_uncore.c
+++ b/arch/x86/kernel/cpu/perf_event_amd_uncore.c
@@ -219,7 +219,6 @@ static ssize_t amd_uncore_attr_show_cpumask(struct device *dev,
 					    struct device_attribute *attr,
 					    char *buf)
 {
-	int n;
 	cpumask_t *active_mask;
 	struct pmu *pmu = dev_get_drvdata(dev);
 
@@ -230,10 +229,7 @@ static ssize_t amd_uncore_attr_show_cpumask(struct device *dev,
 	else
 		return 0;
 
-	n = cpulist_scnprintf(buf, PAGE_SIZE - 2, active_mask);
-	buf[n++] = '\n';
-	buf[n] = '\0';
-	return n;
+	return cpumap_copy_to_buf(true, active_mask, buf);
 }
 static DEVICE_ATTR(cpumask, S_IRUGO, amd_uncore_attr_show_cpumask, NULL);
 
diff --git a/arch/x86/kernel/cpu/perf_event_intel_rapl.c b/arch/x86/kernel/cpu/perf_event_intel_rapl.c
index 619f7699487a..8a673e138e2e 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_rapl.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_rapl.c
@@ -365,11 +365,7 @@ static void rapl_pmu_event_read(struct perf_event *event)
 static ssize_t rapl_get_attr_cpumask(struct device *dev,
 				struct device_attribute *attr, char *buf)
 {
-	int n = cpulist_scnprintf(buf, PAGE_SIZE - 2, &rapl_cpu_mask);
-
-	buf[n++] = '\n';
-	buf[n] = '\0';
-	return n;
+	return cpumap_copy_to_buf(true, &rapl_cpu_mask, buf);
 }
 
 static DEVICE_ATTR(cpumask, S_IRUGO, rapl_get_attr_cpumask, NULL);
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.c b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
index 0939f86f543d..a0b810951743 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
@@ -3617,11 +3617,7 @@ static int uncore_pmu_event_init(struct perf_event *event)
 static ssize_t uncore_get_attr_cpumask(struct device *dev,
 				struct device_attribute *attr, char *buf)
 {
-	int n = cpulist_scnprintf(buf, PAGE_SIZE - 2, &uncore_cpu_mask);
-
-	buf[n++] = '\n';
-	buf[n] = '\0';
-	return n;
+	return cpumap_copy_to_buf(true, &uncore_cpu_mask, buf);
 }
 
 static DEVICE_ATTR(cpumask, S_IRUGO, uncore_get_attr_cpumask, NULL);
diff --git a/drivers/acpi/acpi_pad.c b/drivers/acpi/acpi_pad.c
index f148a0580e04..0dbe52ed6a0d 100644
--- a/drivers/acpi/acpi_pad.c
+++ b/drivers/acpi/acpi_pad.c
@@ -350,11 +350,7 @@ static ssize_t acpi_pad_idlecpus_store(struct device *dev,
 static ssize_t acpi_pad_idlecpus_show(struct device *dev,
 	struct device_attribute *attr, char *buf)
 {
-	int n = 0;
-	n = cpumask_scnprintf(buf, PAGE_SIZE-2, to_cpumask(pad_busy_cpus_bits));
-	buf[n++] = '\n';
-	buf[n] = '\0';
-	return n;
+	return cpumap_copy_to_buf(false, to_cpumask(pad_busy_cpus_bits), buf);
 }
 static DEVICE_ATTR(idlecpus, S_IRUGO|S_IWUSR,
 	acpi_pad_idlecpus_show,
diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c
index 006b1bc5297d..277a9cfa9040 100644
--- a/drivers/base/cpu.c
+++ b/drivers/base/cpu.c
@@ -207,11 +207,8 @@ static ssize_t show_cpus_attr(struct device *dev,
 			      char *buf)
 {
 	struct cpu_attr *ca = container_of(attr, struct cpu_attr, attr);
-	int n = cpulist_scnprintf(buf, PAGE_SIZE-2, *(ca->map));
 
-	buf[n++] = '\n';
-	buf[n] = '\0';
-	return n;
+	return cpumap_copy_to_buf(true, *ca->map, buf);
 }
 
 #define _CPU_ATTR(name, map) \
diff --git a/drivers/base/node.c b/drivers/base/node.c
index c6d3ae05f1ca..6e58ad566c13 100644
--- a/drivers/base/node.c
+++ b/drivers/base/node.c
@@ -25,32 +25,26 @@ static struct bus_type node_subsys = {
 };
 
 
-static ssize_t node_read_cpumap(struct device *dev, int type, char *buf)
+static ssize_t node_read_cpumap(struct device *dev, bool list, char *buf)
 {
 	struct node *node_dev = to_node(dev);
 	const struct cpumask *mask = cpumask_of_node(node_dev->dev.id);
-	int len;
 
 	/* 2008/04/07: buf currently PAGE_SIZE, need 9 chars per 32 bits. */
 	BUILD_BUG_ON((NR_CPUS/32 * 9) > (PAGE_SIZE-1));
 
-	len = type?
-		cpulist_scnprintf(buf, PAGE_SIZE-2, mask) :
-		cpumask_scnprintf(buf, PAGE_SIZE-2, mask);
- 	buf[len++] = '\n';
- 	buf[len] = '\0';
-	return len;
+	return cpumap_copy_to_buf(list, mask, buf);
 }
 
 static inline ssize_t node_read_cpumask(struct device *dev,
 				struct device_attribute *attr, char *buf)
 {
-	return node_read_cpumap(dev, 0, buf);
+	return node_read_cpumap(dev, false, buf);
 }
 static inline ssize_t node_read_cpulist(struct device *dev,
 				struct device_attribute *attr, char *buf)
 {
-	return node_read_cpumap(dev, 1, buf);
+	return node_read_cpumap(dev, true, buf);
 }
 
 static DEVICE_ATTR(cpumap,  S_IRUGO, node_read_cpumask, NULL);
diff --git a/drivers/base/topology.c b/drivers/base/topology.c
index be7c1fb7c0c9..39b294bd769b 100644
--- a/drivers/base/topology.c
+++ b/drivers/base/topology.c
@@ -42,29 +42,11 @@ static ssize_t show_##name(struct device *dev,			\
 	return sprintf(buf, "%d\n", topology_##name(dev->id));	\
 }
 
-#if defined(topology_thread_cpumask) || defined(topology_core_cpumask) || \
-    defined(topology_book_cpumask)
-static ssize_t show_cpumap(int type, const struct cpumask *mask, char *buf)
-{
-	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf;
-	int n = 0;
-
-	if (len > 1) {
-		n = type?
-			cpulist_scnprintf(buf, len-2, mask) :
-			cpumask_scnprintf(buf, len-2, mask);
-		buf[n++] = '\n';
-		buf[n] = '\0';
-	}
-	return n;
-}
-#endif
-
 #define define_siblings_show_map(name)					\
 static ssize_t show_##name(struct device *dev,				\
 			   struct device_attribute *attr, char *buf)	\
 {									\
-	return show_cpumap(0, topology_##name(dev->id), buf);		\
+	return cpumap_copy_to_buf(false, topology_##name(dev->id), buf);\
 }
 
 #define define_siblings_show_list(name)					\
@@ -72,7 +54,7 @@ static ssize_t show_##name##_list(struct device *dev,			\
 				  struct device_attribute *attr,	\
 				  char *buf)				\
 {									\
-	return show_cpumap(1, topology_##name(dev->id), buf);		\
+	return cpumap_copy_to_buf(true, topology_##name(dev->id), buf);	\
 }
 
 #define define_siblings_show_func(name)		\
diff --git a/drivers/pci/pci-sysfs.c b/drivers/pci/pci-sysfs.c
index 9ff0a901ecf7..aa0f1359ef30 100644
--- a/drivers/pci/pci-sysfs.c
+++ b/drivers/pci/pci-sysfs.c
@@ -77,11 +77,10 @@ static ssize_t broken_parity_status_store(struct device *dev,
 }
 static DEVICE_ATTR_RW(broken_parity_status);
 
-static ssize_t pci_dev_show_local_cpu(struct device *dev, int type,
+static ssize_t pci_dev_show_local_cpu(struct device *dev, bool list,
 				      struct device_attribute *attr, char *buf)
 {
 	const struct cpumask *mask;
-	int len;
 
 #ifdef CONFIG_NUMA
 	mask = (dev_to_node(dev) == -1) ? cpu_online_mask :
@@ -89,59 +88,41 @@ static ssize_t pci_dev_show_local_cpu(struct device *dev, int type,
 #else
 	mask = cpumask_of_pcibus(to_pci_dev(dev)->bus);
 #endif
-	len = type ?
-		cpumask_scnprintf(buf, PAGE_SIZE-2, mask) :
-		cpulist_scnprintf(buf, PAGE_SIZE-2, mask);
-
-	buf[len++] = '\n';
-	buf[len] = '\0';
-	return len;
+	return cpumap_copy_to_buf(list, mask, buf);
 }
 
 static ssize_t local_cpus_show(struct device *dev,
 			       struct device_attribute *attr, char *buf)
 {
-	return pci_dev_show_local_cpu(dev, 1, attr, buf);
+	return pci_dev_show_local_cpu(dev, false, attr, buf);
 }
 static DEVICE_ATTR_RO(local_cpus);
 
 static ssize_t local_cpulist_show(struct device *dev,
 				  struct device_attribute *attr, char *buf)
 {
-	return pci_dev_show_local_cpu(dev, 0, attr, buf);
+	return pci_dev_show_local_cpu(dev, true, attr, buf);
 }
 static DEVICE_ATTR_RO(local_cpulist);
 
 /*
  * PCI Bus Class Devices
  */
-static ssize_t pci_bus_show_cpuaffinity(struct device *dev, int type,
-					struct device_attribute *attr,
-					char *buf)
-{
-	int ret;
-	const struct cpumask *cpumask;
-
-	cpumask = cpumask_of_pcibus(to_pci_bus(dev));
-	ret = type ?
-		cpulist_scnprintf(buf, PAGE_SIZE-2, cpumask) :
-		cpumask_scnprintf(buf, PAGE_SIZE-2, cpumask);
-	buf[ret++] = '\n';
-	buf[ret] = '\0';
-	return ret;
-}
-
 static ssize_t cpuaffinity_show(struct device *dev,
 				struct device_attribute *attr, char *buf)
 {
-	return pci_bus_show_cpuaffinity(dev, 0, attr, buf);
+	const struct cpumask *cpumask = cpumask_of_pcibus(to_pci_bus(dev));
+
+	return cpumap_copy_to_buf(false, cpumask, buf);
 }
 static DEVICE_ATTR_RO(cpuaffinity);
 
 static ssize_t cpulistaffinity_show(struct device *dev,
 				    struct device_attribute *attr, char *buf)
 {
-	return pci_bus_show_cpuaffinity(dev, 1, attr, buf);
+	const struct cpumask *cpumask = cpumask_of_pcibus(to_pci_bus(dev));
+
+	return cpumap_copy_to_buf(true, cpumask, buf);
 }
 static DEVICE_ATTR_RO(cpulistaffinity);
 
diff --git a/include/linux/cpumask.h b/include/linux/cpumask.h
index 2997af6d2ccd..26d8348292dd 100644
--- a/include/linux/cpumask.h
+++ b/include/linux/cpumask.h
@@ -11,6 +11,8 @@
 #include <linux/bitmap.h>
 #include <linux/bug.h>
 
+#include <asm/page.h>
+
 typedef struct cpumask { DECLARE_BITMAP(bits, NR_CPUS); } cpumask_t;
 
 /**
@@ -792,6 +794,31 @@ static inline const struct cpumask *get_cpu_mask(unsigned int cpu)
 }
 #endif /* NR_CPUS > BITS_PER_LONG */
 
+/**
+ * cpumap_copy_to_buf  - copies the cpumask into the buffer either
+ *	as comma-separated list of cpus or hex values of cpumask
+ * @list: indicates whether the cpumap must be list
+ * @mask: the cpumask to copy
+ * @buf: the buffer to copy into
+ *
+ * Returns the length of the (null-terminated) @buf string, zero if
+ * nothing is copied.
+ */
+static inline ssize_t
+cpumap_copy_to_buf(bool list, const struct cpumask *mask, char *buf)
+{
+	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf - 2;
+	int n = 0;
+
+	if (len > 1) {
+		n = list ? cpulist_scnprintf(buf, len, mask) :
+			   cpumask_scnprintf(buf, len, mask);
+		buf[n++] = '\n';
+		buf[n] = '\0';
+	}
+	return n;
+}
+
 /*
  *
  * From here down, all obsolete.  Use cpumask_ variants!
-- 
1.8.3.2

[PATCH v3 02/11] topology: replace custom attribute macros with standard DEVICE_ATTR*

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

Currently couple of custom macros are defined to declare the
device attributes. However there are already standard macros
defined in device.h that suffice the need and these custom
macros can be removed.

This patch replaces custom attribute macros with standard
DEVICE_ATTR_RO attribute

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 drivers/base/topology.c | 53 ++++++++++++++++++++++---------------------------
 1 file changed, 24 insertions(+), 29 deletions(-)

diff --git a/drivers/base/topology.c b/drivers/base/topology.c
index 39b294bd769b..71251ef96e0a 100644
--- a/drivers/base/topology.c
+++ b/drivers/base/topology.c
@@ -29,57 +29,52 @@
 #include <linux/hardirq.h>
 #include <linux/topology.h>
 
-#define define_one_ro_named(_name, _func)				\
-	static DEVICE_ATTR(_name, 0444, _func, NULL)
-
-#define define_one_ro(_name)				\
-	static DEVICE_ATTR(_name, 0444, show_##_name, NULL)
-
 #define define_id_show_func(name)				\
-static ssize_t show_##name(struct device *dev,			\
+static ssize_t name##_show(struct device *dev,			\
 		struct device_attribute *attr, char *buf)	\
 {								\
 	return sprintf(buf, "%d\n", topology_##name(dev->id));	\
 }
 
-#define define_siblings_show_map(name)					\
-static ssize_t show_##name(struct device *dev,				\
+#define define_siblings_show_map(name, mask)				\
+static ssize_t name##_show(struct device *dev,				\
 			   struct device_attribute *attr, char *buf)	\
 {									\
-	return cpumap_copy_to_buf(false, topology_##name(dev->id), buf);\
+	return cpumap_copy_to_buf(false, topology_##mask(dev->id), buf);\
 }
 
-#define define_siblings_show_list(name)					\
-static ssize_t show_##name##_list(struct device *dev,			\
-				  struct device_attribute *attr,	\
-				  char *buf)				\
+#define define_siblings_show_list(name, mask)				\
+static ssize_t name##_list_show(struct device *dev,			\
+				struct device_attribute *attr,		\
+				char *buf)				\
 {									\
-	return cpumap_copy_to_buf(true, topology_##name(dev->id), buf);	\
+	return cpumap_copy_to_buf(true, topology_##mask(dev->id), buf);	\
 }
 
-#define define_siblings_show_func(name)		\
-	define_siblings_show_map(name); define_siblings_show_list(name)
+#define define_siblings_show_func(name, mask)	\
+	define_siblings_show_map(name, mask);	\
+	define_siblings_show_list(name, mask)
 
 define_id_show_func(physical_package_id);
-define_one_ro(physical_package_id);
+static DEVICE_ATTR_RO(physical_package_id);
 
 define_id_show_func(core_id);
-define_one_ro(core_id);
+static DEVICE_ATTR_RO(core_id);
 
-define_siblings_show_func(thread_cpumask);
-define_one_ro_named(thread_siblings, show_thread_cpumask);
-define_one_ro_named(thread_siblings_list, show_thread_cpumask_list);
+define_siblings_show_func(thread_siblings, thread_cpumask);
+static DEVICE_ATTR_RO(thread_siblings);
+static DEVICE_ATTR_RO(thread_siblings_list);
 
-define_siblings_show_func(core_cpumask);
-define_one_ro_named(core_siblings, show_core_cpumask);
-define_one_ro_named(core_siblings_list, show_core_cpumask_list);
+define_siblings_show_func(core_siblings, core_cpumask);
+static DEVICE_ATTR_RO(core_siblings);
+static DEVICE_ATTR_RO(core_siblings_list);
 
 #ifdef CONFIG_SCHED_BOOK
 define_id_show_func(book_id);
-define_one_ro(book_id);
-define_siblings_show_func(book_cpumask);
-define_one_ro_named(book_siblings, show_book_cpumask);
-define_one_ro_named(book_siblings_list, show_book_cpumask_list);
+static DEVICE_ATTR_RO(book_id);
+define_siblings_show_func(book_siblings, book_cpumask);
+static DEVICE_ATTR_RO(book_siblings);
+static DEVICE_ATTR_RO(book_siblings_list);
 #endif
 
 static struct attribute *default_attrs[] = {
-- 
1.8.3.2

[PATCH v3 03/11] drivers: base: add new class "cpu" to group cpu devices

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch creates a new class called "cpu" and assigns it to all the
cpu devices. This helps in grouping all the cpu devices and associated
child devices under the same class.

This patch also:
1. modifies the get_parent_device to return the legacy path
   (/sys/devices/system/cpu/..) for the cpu class devices to support
   existing sysfs ABI
2. avoids creating link in the class directory pointing to the device as
   there would be per-cpu instance of these devices with the same name
3. makes sure subsystem symlink continues pointing to cpu bus instead of
   cpu class for cpu devices

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 drivers/base/core.c | 39 +++++++++++++++++++++++++++++++++------
 drivers/base/cpu.c  |  7 +++++++
 include/linux/cpu.h |  2 ++
 3 files changed, 42 insertions(+), 6 deletions(-)

diff --git a/drivers/base/core.c b/drivers/base/core.c
index 20da3ad1696b..fe622e2a48d0 100644
--- a/drivers/base/core.c
+++ b/drivers/base/core.c
@@ -10,6 +10,7 @@
  *
  */
 
+#include <linux/cpu.h>
 #include <linux/device.h>
 #include <linux/err.h>
 #include <linux/init.h>
@@ -742,6 +743,12 @@ static struct kobject *get_device_parent(struct device *dev,
 			return &block_class.p->subsys.kobj;
 		}
 #endif
+		/*
+		 * if the device is in cpu class, then use the default/legacy
+		 * /sys/devices/system/cpu/.. path
+		 */
+		if (dev->class == cpu_class)
+			return &parent->kobj;
 
 		/*
 		 * If we have no parent, we live in "virtual".
@@ -808,11 +815,17 @@ static int device_add_class_symlinks(struct device *dev)
 	if (!dev->class)
 		return 0;
 
-	error = sysfs_create_link(&dev->kobj,
-				  &dev->class->p->subsys.kobj,
-				  "subsystem");
-	if (error)
-		goto out;
+	/*
+	 * the subsystem symlink in each cpu device needs to continue
+	 * pointing to cpu bus
+	 */
+	if (dev->bus != &cpu_subsys) {
+		error = sysfs_create_link(&dev->kobj,
+					  &dev->class->p->subsys.kobj,
+					  "subsystem");
+		if (error)
+			goto out;
+	}
 
 	if (dev->parent && device_is_not_partition(dev)) {
 		error = sysfs_create_link(&dev->kobj, &dev->parent->kobj,
@@ -826,6 +839,13 @@ static int device_add_class_symlinks(struct device *dev)
 	if (sysfs_deprecated && dev->class == &block_class)
 		return 0;
 #endif
+	/*
+	 * don't create a link in the cpu class directory pointing to the
+	 * device as there would be per-cpu instance of these devices with
+	 * the same name
+	 */
+	if (dev->class == cpu_class)
+		return 0;
 
 	/* link in the class directory pointing to the device */
 	error = sysfs_create_link(&dev->class->p->subsys.kobj,
@@ -851,11 +871,18 @@ static void device_remove_class_symlinks(struct device *dev)
 
 	if (dev->parent && device_is_not_partition(dev))
 		sysfs_remove_link(&dev->kobj, "device");
-	sysfs_remove_link(&dev->kobj, "subsystem");
+
+	/* if subsystem points to cpu bus, bus_remove_device will remove it */
+	if (dev->bus != &cpu_subsys)
+		sysfs_remove_link(&dev->kobj, "subsystem");
 #ifdef CONFIG_BLOCK
 	if (sysfs_deprecated && dev->class == &block_class)
 		return;
 #endif
+	/* symlinks are not created for cpu class devices, nothing to remove */
+	if (dev->class == cpu_class)
+		return;
+
 	sysfs_delete_link(&dev->class->p->subsys.kobj, &dev->kobj, dev_name(dev));
 }
 
diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c
index 277a9cfa9040..8e380214625d 100644
--- a/drivers/base/cpu.c
+++ b/drivers/base/cpu.c
@@ -319,6 +319,7 @@ static int cpu_uevent(struct device *dev, struct kobj_uevent_env *env)
 }
 #endif
 
+struct class *cpu_class;
 /*
  * register_cpu - Setup a sysfs device for a CPU.
  * @cpu - cpu->hotpluggable field set to 1 will generate a control file in
@@ -335,6 +336,8 @@ int register_cpu(struct cpu *cpu, int num)
 	memset(&cpu->dev, 0x00, sizeof(struct device));
 	cpu->dev.id = num;
 	cpu->dev.bus = &cpu_subsys;
+	cpu->dev.parent = cpu_subsys.dev_root;
+	cpu->dev.class = cpu_class;
 	cpu->dev.release = cpu_device_release;
 	cpu->dev.offline_disabled = !cpu->hotpluggable;
 	cpu->dev.offline = !cpu_online(num);
@@ -420,5 +423,9 @@ void __init cpu_dev_init(void)
 	if (subsys_system_register(&cpu_subsys, cpu_root_attr_groups))
 		panic("Failed to register CPU subsystem");
 
+	cpu_class = class_create(THIS_MODULE, "cpu");
+	if (IS_ERR(cpu_class))
+		panic("Failed to register CPU class");
+
 	cpu_dev_register_generic();
 }
diff --git a/include/linux/cpu.h b/include/linux/cpu.h
index 95978ad7fcdd..8c0fc9b0acad 100644
--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -39,6 +39,8 @@ extern void cpu_remove_dev_attr(struct device_attribute *attr);
 extern int cpu_add_dev_attr_group(struct attribute_group *attrs);
 extern void cpu_remove_dev_attr_group(struct attribute_group *attrs);
 
+extern struct class *cpu_class;
+
 #ifdef CONFIG_HOTPLUG_CPU
 extern void unregister_cpu(struct cpu *cpu);
 extern ssize_t arch_cpu_probe(const char *, size_t);
-- 
1.8.3.2

Re: [PATCH v3 03/11] drivers: base: add new class "cpu" to group cpu devices

[David Herrmann]

Hi

On Thu, Aug 21, 2014 at 12:59 PM, Sudeep Holla <sudeep.holla@arm.com> wrote:
> From: Sudeep Holla <sudeep.holla@arm.com>
>
> This patch creates a new class called "cpu" and assigns it to all the
> cpu devices. This helps in grouping all the cpu devices and associated
> child devices under the same class.
>
> This patch also:
> 1. modifies the get_parent_device to return the legacy path
>    (/sys/devices/system/cpu/..) for the cpu class devices to support
>    existing sysfs ABI
> 2. avoids creating link in the class directory pointing to the device as
>    there would be per-cpu instance of these devices with the same name
> 3. makes sure subsystem symlink continues pointing to cpu bus instead of
>    cpu class for cpu devices

This patch lacks any explanation _why_ you add a class for CPUs. With
this patch applied, these directories are effectively the same:
  /sys/bus/cpu/devices/
  /sys/class/cpu/

Why do we need a cpu-class if the same set of information is already
available on the cpu-bus? Furthermore, classes are deprecated anyway.
Everything you can do with a class can be solved with a bus. And we
already have a bus for cpus.

Thanks
David

> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> ---
>  drivers/base/core.c | 39 +++++++++++++++++++++++++++++++++------
>  drivers/base/cpu.c  |  7 +++++++
>  include/linux/cpu.h |  2 ++
>  3 files changed, 42 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/base/core.c b/drivers/base/core.c
> index 20da3ad1696b..fe622e2a48d0 100644
> --- a/drivers/base/core.c
> +++ b/drivers/base/core.c
> @@ -10,6 +10,7 @@
>   *
>   */
>
> +#include <linux/cpu.h>
>  #include <linux/device.h>
>  #include <linux/err.h>
>  #include <linux/init.h>
> @@ -742,6 +743,12 @@ static struct kobject *get_device_parent(struct device *dev,
>                         return &block_class.p->subsys.kobj;
>                 }
>  #endif
> +               /*
> +                * if the device is in cpu class, then use the default/legacy
> +                * /sys/devices/system/cpu/.. path
> +                */
> +               if (dev->class == cpu_class)
> +                       return &parent->kobj;
>
>                 /*
>                  * If we have no parent, we live in "virtual".
> @@ -808,11 +815,17 @@ static int device_add_class_symlinks(struct device *dev)
>         if (!dev->class)
>                 return 0;
>
> -       error = sysfs_create_link(&dev->kobj,
> -                                 &dev->class->p->subsys.kobj,
> -                                 "subsystem");
> -       if (error)
> -               goto out;
> +       /*
> +        * the subsystem symlink in each cpu device needs to continue
> +        * pointing to cpu bus
> +        */
> +       if (dev->bus != &cpu_subsys) {
> +               error = sysfs_create_link(&dev->kobj,
> +                                         &dev->class->p->subsys.kobj,
> +                                         "subsystem");
> +               if (error)
> +                       goto out;
> +       }
>
>         if (dev->parent && device_is_not_partition(dev)) {
>                 error = sysfs_create_link(&dev->kobj, &dev->parent->kobj,
> @@ -826,6 +839,13 @@ static int device_add_class_symlinks(struct device *dev)
>         if (sysfs_deprecated && dev->class == &block_class)
>                 return 0;
>  #endif
> +       /*
> +        * don't create a link in the cpu class directory pointing to the
> +        * device as there would be per-cpu instance of these devices with
> +        * the same name
> +        */
> +       if (dev->class == cpu_class)
> +               return 0;
>
>         /* link in the class directory pointing to the device */
>         error = sysfs_create_link(&dev->class->p->subsys.kobj,
> @@ -851,11 +871,18 @@ static void device_remove_class_symlinks(struct device *dev)
>
>         if (dev->parent && device_is_not_partition(dev))
>                 sysfs_remove_link(&dev->kobj, "device");
> -       sysfs_remove_link(&dev->kobj, "subsystem");
> +
> +       /* if subsystem points to cpu bus, bus_remove_device will remove it */
> +       if (dev->bus != &cpu_subsys)
> +               sysfs_remove_link(&dev->kobj, "subsystem");
>  #ifdef CONFIG_BLOCK
>         if (sysfs_deprecated && dev->class == &block_class)
>                 return;
>  #endif
> +       /* symlinks are not created for cpu class devices, nothing to remove */
> +       if (dev->class == cpu_class)
> +               return;
> +
>         sysfs_delete_link(&dev->class->p->subsys.kobj, &dev->kobj, dev_name(dev));
>  }
>
> diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c
> index 277a9cfa9040..8e380214625d 100644
> --- a/drivers/base/cpu.c
> +++ b/drivers/base/cpu.c
> @@ -319,6 +319,7 @@ static int cpu_uevent(struct device *dev, struct kobj_uevent_env *env)
>  }
>  #endif
>
> +struct class *cpu_class;
>  /*
>   * register_cpu - Setup a sysfs device for a CPU.
>   * @cpu - cpu->hotpluggable field set to 1 will generate a control file in
> @@ -335,6 +336,8 @@ int register_cpu(struct cpu *cpu, int num)
>         memset(&cpu->dev, 0x00, sizeof(struct device));
>         cpu->dev.id = num;
>         cpu->dev.bus = &cpu_subsys;
> +       cpu->dev.parent = cpu_subsys.dev_root;
> +       cpu->dev.class = cpu_class;
>         cpu->dev.release = cpu_device_release;
>         cpu->dev.offline_disabled = !cpu->hotpluggable;
>         cpu->dev.offline = !cpu_online(num);
> @@ -420,5 +423,9 @@ void __init cpu_dev_init(void)
>         if (subsys_system_register(&cpu_subsys, cpu_root_attr_groups))
>                 panic("Failed to register CPU subsystem");
>
> +       cpu_class = class_create(THIS_MODULE, "cpu");
> +       if (IS_ERR(cpu_class))
> +               panic("Failed to register CPU class");
> +
>         cpu_dev_register_generic();
>  }
> diff --git a/include/linux/cpu.h b/include/linux/cpu.h
> index 95978ad7fcdd..8c0fc9b0acad 100644
> --- a/include/linux/cpu.h
> +++ b/include/linux/cpu.h
> @@ -39,6 +39,8 @@ extern void cpu_remove_dev_attr(struct device_attribute *attr);
>  extern int cpu_add_dev_attr_group(struct attribute_group *attrs);
>  extern void cpu_remove_dev_attr_group(struct attribute_group *attrs);
>
> +extern struct class *cpu_class;
> +
>  #ifdef CONFIG_HOTPLUG_CPU
>  extern void unregister_cpu(struct cpu *cpu);
>  extern ssize_t arch_cpu_probe(const char *, size_t);
> --
> 1.8.3.2
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: [PATCH v3 03/11] drivers: base: add new class "cpu" to group cpu devices

[Sudeep Holla]

On 21/08/14 12:20, David Herrmann wrote:
> Hi
>
> On Thu, Aug 21, 2014 at 12:59 PM, Sudeep Holla <sudeep.holla@arm.com> wrote:
>> From: Sudeep Holla <sudeep.holla@arm.com>
>>
>> This patch creates a new class called "cpu" and assigns it to all the
>> cpu devices. This helps in grouping all the cpu devices and associated
>> child devices under the same class.
>>
>> This patch also:
>> 1. modifies the get_parent_device to return the legacy path
>>     (/sys/devices/system/cpu/..) for the cpu class devices to support
>>     existing sysfs ABI
>> 2. avoids creating link in the class directory pointing to the device as
>>     there would be per-cpu instance of these devices with the same name
>> 3. makes sure subsystem symlink continues pointing to cpu bus instead of
>>     cpu class for cpu devices
>
> This patch lacks any explanation _why_ you add a class for CPUs. With
> this patch applied, these directories are effectively the same:
>    /sys/bus/cpu/devices/
>    /sys/class/cpu/
>

Yes that's the intention, so that we don't break any existing sysfs/ABI

> Why do we need a cpu-class if the same set of information is already
> available on the cpu-bus? Furthermore, classes are deprecated anyway.
> Everything you can do with a class can be solved with a bus. And we
> already have a bus for cpus.
>

This was suggested[1] by GregKH. The main reason it was added is to
reuse the device attributes rather than creating the raw kobjects.

It helps to move few other cpu related subsystems using raw kobjects to
the device attribute groups.

Regards,
Sudeep

[1] https://lkml.org/lkml/2014/2/10/766

Re: [PATCH v3 03/11] drivers: base: add new class "cpu" to group cpu devices

[David Herrmann]

Hi

On Thu, Aug 21, 2014 at 2:30 PM, Sudeep Holla <sudeep.holla@arm.com> wrote:
> On 21/08/14 12:20, David Herrmann wrote:
>> Why do we need a cpu-class if the same set of information is already
>> available on the cpu-bus? Furthermore, classes are deprecated anyway.
>> Everything you can do with a class can be solved with a bus. And we
>> already have a bus for cpus.
>>
>
> This was suggested[1] by GregKH. The main reason it was added is to
> reuse the device attributes rather than creating the raw kobjects.
>
> It helps to move few other cpu related subsystems using raw kobjects to
> the device attribute groups.

So the only reason to add a class is to get attributes registered
properly with the device? Just set dev->groups before calling
device_add()? This works just fine on buses, too.

Thanks
David

Re: [PATCH v3 03/11] drivers: base: add new class "cpu" to group cpu devices

[Sudeep Holla]

On 21/08/14 13:37, David Herrmann wrote:
> Hi
>
> On Thu, Aug 21, 2014 at 2:30 PM, Sudeep Holla <sudeep.holla@arm.com> wrote:
>> On 21/08/14 12:20, David Herrmann wrote:
>>> Why do we need a cpu-class if the same set of information is already
>>> available on the cpu-bus? Furthermore, classes are deprecated anyway.
>>> Everything you can do with a class can be solved with a bus. And we
>>> already have a bus for cpus.
>>>

Sorry I wasn't aware that classes are deprecated. IIUC its class_device
struct and related APIs that are deprecated not struct class. I may be
wrong but Greg would not have suggested to use class if it was deprecated.

>>
>> This was suggested[1] by GregKH. The main reason it was added is to
>> reuse the device attributes rather than creating the raw kobjects.
>>
>> It helps to move few other cpu related subsystems using raw kobjects to
>> the device attribute groups.
>
> So the only reason to add a class is to get attributes registered
> properly with the device? Just set dev->groups before calling
> device_add()? This works just fine on buses, too.
>

Do you mean to say create bus for each cpu and add cpu devices like
cache under that ? We can't reuse the cpu bus i.e cpu_subsys as we can't
add per-cpu devices with same name under /sys/bus/cpu/devices/

Also adding bus for each cpu device might be more complex than creating
cpu class.

That's the reason why I added checks to avoid creating link in the class
directory pointing to the device. It would be difficult to do
that with bus as it can't be unconditional.

Regards,
Sudeep

Re: [PATCH v3 03/11] drivers: base: add new class "cpu" to group cpu devices

[Kay Sievers]

On Thu, Aug 21, 2014 at 2:30 PM, Sudeep Holla <sudeep.holla@arm.com> wrote:
> On 21/08/14 12:20, David Herrmann wrote:
>> On Thu, Aug 21, 2014 at 12:59 PM, Sudeep Holla <sudeep.holla@arm.com>
>> wrote:
>>>
>>> From: Sudeep Holla <sudeep.holla@arm.com>
>>>
>>> This patch creates a new class called "cpu" and assigns it to all the
>>> cpu devices. This helps in grouping all the cpu devices and associated
>>> child devices under the same class.
>>>
>>> This patch also:
>>> 1. modifies the get_parent_device to return the legacy path
>>>     (/sys/devices/system/cpu/..) for the cpu class devices to support
>>>     existing sysfs ABI
>>> 2. avoids creating link in the class directory pointing to the device as
>>>     there would be per-cpu instance of these devices with the same name
>>> 3. makes sure subsystem symlink continues pointing to cpu bus instead of
>>>     cpu class for cpu devices
>>
>>
>> This patch lacks any explanation _why_ you add a class for CPUs. With
>> this patch applied, these directories are effectively the same:
>>    /sys/bus/cpu/devices/
>>    /sys/class/cpu/
>>
>
> Yes that's the intention, so that we don't break any existing sysfs/ABI
>
>
>> Why do we need a cpu-class if the same set of information is already
>> available on the cpu-bus? Furthermore, classes are deprecated anyway.
>> Everything you can do with a class can be solved with a bus. And we
>> already have a bus for cpus.
>>
>
> This was suggested[1] by GregKH. The main reason it was added is to
> reuse the device attributes rather than creating the raw kobjects.
>
> It helps to move few other cpu related subsystems using raw kobjects to
> the device attribute groups.

No, nothing should ever create a bus and a class with the same name.
This is not supported by userspace tools.

Your problem needs to be addressed by adding things to the existing
"cpu" bus, not by adding a new class.

Kay

[PATCH] drivers: base: add cpu_device_create to support per-cpu devices

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds a new function to create per-cpu devices.
This helps in:
1. reusing the device infrastructure to create any cpu related
   attributes and corresponding sysfs instead of creating and
   dealing with raw kobjects directly
2. retaining the legacy path(/sys/devices/system/cpu/..) to support
   existing sysfs ABI
3. avoiding to create links in the bus directory pointing to the
   device as there would be per-cpu instance of these devices with
   the same name since dev->bus is not populated to cpu_sysbus on
   purpose

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 drivers/base/cpu.c  | 54 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 include/linux/cpu.h |  4 ++++
 2 files changed, 58 insertions(+)

Hi Greg,

Here is the alternate solution I could come up with instead of
creating cpu class. cpu_device_create is very similar to
device_create_groups_vargs w/o class support, but I could not
reuse anything else to avoid creating similar function.

Let me know your thoughts/suggestions on this.

Regards,
Sudeep


diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c
index 277a9cfa9040..53f0c4141d05 100644
--- a/drivers/base/cpu.c
+++ b/drivers/base/cpu.c
@@ -363,6 +363,60 @@ struct device *get_cpu_device(unsigned cpu)
 }
 EXPORT_SYMBOL_GPL(get_cpu_device);
 
+static void device_create_release(struct device *dev)
+{
+	kfree(dev);
+}
+
+static struct device *
+__cpu_device_create(struct device *parent, void *drvdata,
+		    const struct attribute_group **groups,
+		    const char *fmt, va_list args)
+{
+	struct device *dev = NULL;
+	int retval = -ENODEV;
+
+	dev = kzalloc(sizeof(*dev), GFP_KERNEL);
+	if (!dev) {
+		retval = -ENOMEM;
+		goto error;
+	}
+
+	device_initialize(dev);
+	dev->parent = parent;
+	dev->groups = groups;
+	dev->release = device_create_release;
+	dev_set_drvdata(dev, drvdata);
+
+	retval = kobject_set_name_vargs(&dev->kobj, fmt, args);
+	if (retval)
+		goto error;
+
+	retval = device_add(dev);
+	if (retval)
+		goto error;
+
+	return dev;
+
+error:
+	put_device(dev);
+	return ERR_PTR(retval);
+}
+
+struct device *cpu_device_create(struct device *parent, void *drvdata,
+				 const struct attribute_group **groups,
+				 const char *fmt, ...)
+{
+	va_list vargs;
+	struct device *dev;
+
+	va_start(vargs, fmt);
+	dev = __cpu_device_create(parent, drvdata, groups, fmt, vargs);
+	va_end(vargs);
+	return dev;
+}
+EXPORT_SYMBOL_GPL(cpu_device_create);
+
 #ifdef CONFIG_GENERIC_CPU_AUTOPROBE
 static DEVICE_ATTR(modalias, 0444, print_cpu_modalias, NULL);
 #endif
diff --git a/include/linux/cpu.h b/include/linux/cpu.h
index 95978ad7fcdd..bb790a5621c0 100644
--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -19,6 +19,7 @@
 
 struct device;
 struct device_node;
+struct attribute_group;
 
 struct cpu {
 	int node_id;		/* The node which contains the CPU */
@@ -39,6 +40,9 @@ extern void cpu_remove_dev_attr(struct device_attribute *attr);
 extern int cpu_add_dev_attr_group(struct attribute_group *attrs);
 extern void cpu_remove_dev_attr_group(struct attribute_group *attrs);
 
+extern struct device *cpu_device_create(struct device *parent, void *drvdata,
+					const struct attribute_group **groups,
+					const char *fmt, ...);
 #ifdef CONFIG_HOTPLUG_CPU
 extern void unregister_cpu(struct cpu *cpu);
 extern ssize_t arch_cpu_probe(const char *, size_t);
-- 
1.8.3.2

Re: [PATCH] drivers: base: add cpu_device_create to support per-cpu devices

[David Herrmann]

Hi

On Fri, Aug 22, 2014 at 1:29 PM, Sudeep Holla <sudeep.holla@arm.com> wrote:
> From: Sudeep Holla <sudeep.holla@arm.com>
>
> This patch adds a new function to create per-cpu devices.
> This helps in:
> 1. reusing the device infrastructure to create any cpu related
>    attributes and corresponding sysfs instead of creating and
>    dealing with raw kobjects directly
> 2. retaining the legacy path(/sys/devices/system/cpu/..) to support
>    existing sysfs ABI
> 3. avoiding to create links in the bus directory pointing to the
>    device as there would be per-cpu instance of these devices with
>    the same name since dev->bus is not populated to cpu_sysbus on
>    purpose
>
> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> ---
>  drivers/base/cpu.c  | 54 +++++++++++++++++++++++++++++++++++++++++++++++++++++
>  include/linux/cpu.h |  4 ++++
>  2 files changed, 58 insertions(+)
>
> Hi Greg,
>
> Here is the alternate solution I could come up with instead of
> creating cpu class. cpu_device_create is very similar to
> device_create_groups_vargs w/o class support, but I could not
> reuse anything else to avoid creating similar function.
>
> Let me know your thoughts/suggestions on this.
>
> Regards,
> Sudeep
>
>
> diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c
> index 277a9cfa9040..53f0c4141d05 100644
> --- a/drivers/base/cpu.c
> +++ b/drivers/base/cpu.c
> @@ -363,6 +363,60 @@ struct device *get_cpu_device(unsigned cpu)
>  }
>  EXPORT_SYMBOL_GPL(get_cpu_device);
>
> +static void device_create_release(struct device *dev)
> +{
> +       kfree(dev);
> +}
> +
> +static struct device *
> +__cpu_device_create(struct device *parent, void *drvdata,
> +                   const struct attribute_group **groups,
> +                   const char *fmt, va_list args)
> +{
> +       struct device *dev = NULL;
> +       int retval = -ENODEV;
> +
> +       dev = kzalloc(sizeof(*dev), GFP_KERNEL);
> +       if (!dev) {
> +               retval = -ENOMEM;
> +               goto error;
> +       }
> +
> +       device_initialize(dev);
> +       dev->parent = parent;
> +       dev->groups = groups;
> +       dev->release = device_create_release;
> +       dev_set_drvdata(dev, drvdata);
> +
> +       retval = kobject_set_name_vargs(&dev->kobj, fmt, args);
> +       if (retval)
> +               goto error;
> +
> +       retval = device_add(dev);
> +       if (retval)
> +               goto error;

Exactly! As I said, simply setting dev->groups before calling device_add().

However, I really don't understand why we need this as global API.
Skimming over the other patches, you use cpu_device_create() only in
one place. Why not hard-code this all there? It is totally OK to do
device initialization in drivers. All the helpers (like
device_create(), device_create_with_groups(), and so on) are just
convenience functions. The driver-core API explicitly allows drivers
to initialize devices manually.

Nevertheless, this patch looks fine.

Thanks
David

> +
> +       return dev;
> +
> +error:
> +       put_device(dev);
> +       return ERR_PTR(retval);
> +}
> +
> +struct device *cpu_device_create(struct device *parent, void *drvdata,
> +                                const struct attribute_group **groups,
> +                                const char *fmt, ...)
> +{
> +       va_list vargs;
> +       struct device *dev;
> +
> +       va_start(vargs, fmt);
> +       dev = __cpu_device_create(parent, drvdata, groups, fmt, vargs);
> +       va_end(vargs);
> +       return dev;
> +}
> +EXPORT_SYMBOL_GPL(cpu_device_create);
> +
>  #ifdef CONFIG_GENERIC_CPU_AUTOPROBE
>  static DEVICE_ATTR(modalias, 0444, print_cpu_modalias, NULL);
>  #endif
> diff --git a/include/linux/cpu.h b/include/linux/cpu.h
> index 95978ad7fcdd..bb790a5621c0 100644
> --- a/include/linux/cpu.h
> +++ b/include/linux/cpu.h
> @@ -19,6 +19,7 @@
>
>  struct device;
>  struct device_node;
> +struct attribute_group;
>
>  struct cpu {
>         int node_id;            /* The node which contains the CPU */
> @@ -39,6 +40,9 @@ extern void cpu_remove_dev_attr(struct device_attribute *attr);
>  extern int cpu_add_dev_attr_group(struct attribute_group *attrs);
>  extern void cpu_remove_dev_attr_group(struct attribute_group *attrs);
>
> +extern struct device *cpu_device_create(struct device *parent, void *drvdata,
> +                                       const struct attribute_group **groups,
> +                                       const char *fmt, ...);
>  #ifdef CONFIG_HOTPLUG_CPU
>  extern void unregister_cpu(struct cpu *cpu);
>  extern ssize_t arch_cpu_probe(const char *, size_t);
> --
> 1.8.3.2
>

Re: [PATCH] drivers: base: add cpu_device_create to support per-cpu devices

[David Herrmann]

Hi

On Fri, Aug 22, 2014 at 1:37 PM, David Herrmann <dh.herrmann@gmail.com> wrote:
> Hi
>
> On Fri, Aug 22, 2014 at 1:29 PM, Sudeep Holla <sudeep.holla@arm.com> wrote:
>> From: Sudeep Holla <sudeep.holla@arm.com>
>>
>> This patch adds a new function to create per-cpu devices.
>> This helps in:
>> 1. reusing the device infrastructure to create any cpu related
>>    attributes and corresponding sysfs instead of creating and
>>    dealing with raw kobjects directly
>> 2. retaining the legacy path(/sys/devices/system/cpu/..) to support
>>    existing sysfs ABI
>> 3. avoiding to create links in the bus directory pointing to the
>>    device as there would be per-cpu instance of these devices with
>>    the same name since dev->bus is not populated to cpu_sysbus on
>>    purpose
>>
>> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
>> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
>> ---
>>  drivers/base/cpu.c  | 54 +++++++++++++++++++++++++++++++++++++++++++++++++++++
>>  include/linux/cpu.h |  4 ++++
>>  2 files changed, 58 insertions(+)
>>
>> Hi Greg,
>>
>> Here is the alternate solution I could come up with instead of
>> creating cpu class. cpu_device_create is very similar to
>> device_create_groups_vargs w/o class support, but I could not
>> reuse anything else to avoid creating similar function.
>>
>> Let me know your thoughts/suggestions on this.
>>
>> Regards,
>> Sudeep
>>
>>
>> diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c
>> index 277a9cfa9040..53f0c4141d05 100644
>> --- a/drivers/base/cpu.c
>> +++ b/drivers/base/cpu.c
>> @@ -363,6 +363,60 @@ struct device *get_cpu_device(unsigned cpu)
>>  }
>>  EXPORT_SYMBOL_GPL(get_cpu_device);
>>
>> +static void device_create_release(struct device *dev)
>> +{
>> +       kfree(dev);
>> +}
>> +
>> +static struct device *
>> +__cpu_device_create(struct device *parent, void *drvdata,
>> +                   const struct attribute_group **groups,
>> +                   const char *fmt, va_list args)
>> +{
>> +       struct device *dev = NULL;
>> +       int retval = -ENODEV;
>> +
>> +       dev = kzalloc(sizeof(*dev), GFP_KERNEL);
>> +       if (!dev) {
>> +               retval = -ENOMEM;
>> +               goto error;
>> +       }
>> +
>> +       device_initialize(dev);
>> +       dev->parent = parent;
>> +       dev->groups = groups;
>> +       dev->release = device_create_release;
>> +       dev_set_drvdata(dev, drvdata);
>> +
>> +       retval = kobject_set_name_vargs(&dev->kobj, fmt, args);
>> +       if (retval)
>> +               goto error;
>> +
>> +       retval = device_add(dev);
>> +       if (retval)
>> +               goto error;
>
> Exactly! As I said, simply setting dev->groups before calling device_add().
>
> However, I really don't understand why we need this as global API.
> Skimming over the other patches, you use cpu_device_create() only in
> one place. Why not hard-code this all there? It is totally OK to do
> device initialization in drivers. All the helpers (like
> device_create(), device_create_with_groups(), and so on) are just
> convenience functions. The driver-core API explicitly allows drivers
> to initialize devices manually.
>
> Nevertheless, this patch looks fine.

Wait, no. Why don't you set dev->bus to cpu_subsys? Is this thing
supposed to create child-devices of CPUs? Can you describe what your
topology is supposed to look like?

Thanks
David

Re: [PATCH] drivers: base: add cpu_device_create to support per-cpu devices

[Sudeep Holla]

On 22/08/14 12:41, David Herrmann wrote:
> Hi
>
> On Fri, Aug 22, 2014 at 1:37 PM, David Herrmann <dh.herrmann@gmail.com> wrote:
>> Hi
>>
>> On Fri, Aug 22, 2014 at 1:29 PM, Sudeep Holla <sudeep.holla@arm.com> wrote:
>>> From: Sudeep Holla <sudeep.holla@arm.com>
>>>
>>> This patch adds a new function to create per-cpu devices.
>>> This helps in:
>>> 1. reusing the device infrastructure to create any cpu related
>>>     attributes and corresponding sysfs instead of creating and
>>>     dealing with raw kobjects directly
>>> 2. retaining the legacy path(/sys/devices/system/cpu/..) to support
>>>     existing sysfs ABI
>>> 3. avoiding to create links in the bus directory pointing to the
>>>     device as there would be per-cpu instance of these devices with
>>>     the same name since dev->bus is not populated to cpu_sysbus on
>>>     purpose
>>>
>>> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
>>> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
>>> ---
>>>   drivers/base/cpu.c  | 54 +++++++++++++++++++++++++++++++++++++++++++++++++++++
>>>   include/linux/cpu.h |  4 ++++
>>>   2 files changed, 58 insertions(+)
>>>
>>> Hi Greg,
>>>
>>> Here is the alternate solution I could come up with instead of
>>> creating cpu class. cpu_device_create is very similar to
>>> device_create_groups_vargs w/o class support, but I could not
>>> reuse anything else to avoid creating similar function.
>>>
>>> Let me know your thoughts/suggestions on this.
>>>
>>> Regards,
>>> Sudeep
>>>
>>>
>>> diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c
>>> index 277a9cfa9040..53f0c4141d05 100644
>>> --- a/drivers/base/cpu.c
>>> +++ b/drivers/base/cpu.c
>>> @@ -363,6 +363,60 @@ struct device *get_cpu_device(unsigned cpu)
>>>   }
>>>   EXPORT_SYMBOL_GPL(get_cpu_device);
>>>
>>> +static void device_create_release(struct device *dev)
>>> +{
>>> +       kfree(dev);
>>> +}
>>> +
>>> +static struct device *
>>> +__cpu_device_create(struct device *parent, void *drvdata,
>>> +                   const struct attribute_group **groups,
>>> +                   const char *fmt, va_list args)
>>> +{
>>> +       struct device *dev = NULL;
>>> +       int retval = -ENODEV;
>>> +
>>> +       dev = kzalloc(sizeof(*dev), GFP_KERNEL);
>>> +       if (!dev) {
>>> +               retval = -ENOMEM;
>>> +               goto error;
>>> +       }
>>> +
>>> +       device_initialize(dev);
>>> +       dev->parent = parent;
>>> +       dev->groups = groups;
>>> +       dev->release = device_create_release;
>>> +       dev_set_drvdata(dev, drvdata);
>>> +
>>> +       retval = kobject_set_name_vargs(&dev->kobj, fmt, args);
>>> +       if (retval)
>>> +               goto error;
>>> +
>>> +       retval = device_add(dev);
>>> +       if (retval)
>>> +               goto error;
>>
>> Exactly! As I said, simply setting dev->groups before calling device_add().
>>
>> However, I really don't understand why we need this as global API.
>> Skimming over the other patches, you use cpu_device_create() only in
>> one place. Why not hard-code this all there? It is totally OK to do
>> device initialization in drivers. All the helpers (like
>> device_create(), device_create_with_groups(), and so on) are just
>> convenience functions. The driver-core API explicitly allows drivers
>> to initialize devices manually.
>>
>> Nevertheless, this patch looks fine.
>
> Wait, no. Why don't you set dev->bus to cpu_subsys? Is this thing
> supposed to create child-devices of CPUs? Can you describe what your
> topology is supposed to look like?
>

Yes, it's not done on purpose as mentioned in the commit log.
E.g.: cacheinfo topology will be as below

/sys/devices/system/cpu/cpuX/cache/index0/<attributes>
/sys/devices/system/cpu/cpuX/cache/index1/<attributes>
..
/sys/devices/system/cpu/cpuX/cache/index<Y/<attributes>

In this case 'cache' is cpuX's child and index<0..Y> are children of
cache in cpuX. The main problem with per-cpu device is that they have
same name for each cpu's instance and when the bus is set to this
devices, the driver model tries to create symlink to each of these
devices in /sys/bus/cpu/... which fails.

Regards,
Sudeep

Re: [PATCH] drivers: base: add cpu_device_create to support per-cpu devices

[Sudeep Holla]

Hi David,

On 22/08/14 13:33, Sudeep Holla wrote:
>
>
> On 22/08/14 12:41, David Herrmann wrote:
>> Hi
>>
>> On Fri, Aug 22, 2014 at 1:37 PM, David Herrmann <dh.herrmann@gmail.com> wrote:
>>> Hi
>>>
>>> On Fri, Aug 22, 2014 at 1:29 PM, Sudeep Holla <sudeep.holla@arm.com> wrote:
>>>> From: Sudeep Holla <sudeep.holla@arm.com>
>>>>
>>>> This patch adds a new function to create per-cpu devices.
>>>> This helps in:
>>>> 1. reusing the device infrastructure to create any cpu related
>>>>      attributes and corresponding sysfs instead of creating and
>>>>      dealing with raw kobjects directly
>>>> 2. retaining the legacy path(/sys/devices/system/cpu/..) to support
>>>>      existing sysfs ABI
>>>> 3. avoiding to create links in the bus directory pointing to the
>>>>      device as there would be per-cpu instance of these devices with
>>>>      the same name since dev->bus is not populated to cpu_sysbus on
>>>>      purpose
>>>>
>>>> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
>>>> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
>>>> ---
>>>>    drivers/base/cpu.c  | 54 +++++++++++++++++++++++++++++++++++++++++++++++++++++
>>>>    include/linux/cpu.h |  4 ++++
>>>>    2 files changed, 58 insertions(+)
>>>>
>>>> Hi Greg,
>>>>
>>>> Here is the alternate solution I could come up with instead of
>>>> creating cpu class. cpu_device_create is very similar to
>>>> device_create_groups_vargs w/o class support, but I could not
>>>> reuse anything else to avoid creating similar function.
>>>>
>>>> Let me know your thoughts/suggestions on this.
>>>>
>>>> Regards,
>>>> Sudeep
>>>>
>>>>
>>>> diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c
>>>> index 277a9cfa9040..53f0c4141d05 100644
>>>> --- a/drivers/base/cpu.c
>>>> +++ b/drivers/base/cpu.c
>>>> @@ -363,6 +363,60 @@ struct device *get_cpu_device(unsigned cpu)
>>>>    }
>>>>    EXPORT_SYMBOL_GPL(get_cpu_device);
>>>>
>>>> +static void device_create_release(struct device *dev)
>>>> +{
>>>> +       kfree(dev);
>>>> +}
>>>> +
>>>> +static struct device *
>>>> +__cpu_device_create(struct device *parent, void *drvdata,
>>>> +                   const struct attribute_group **groups,
>>>> +                   const char *fmt, va_list args)
>>>> +{
>>>> +       struct device *dev = NULL;
>>>> +       int retval = -ENODEV;
>>>> +
>>>> +       dev = kzalloc(sizeof(*dev), GFP_KERNEL);
>>>> +       if (!dev) {
>>>> +               retval = -ENOMEM;
>>>> +               goto error;
>>>> +       }
>>>> +
>>>> +       device_initialize(dev);
>>>> +       dev->parent = parent;
>>>> +       dev->groups = groups;
>>>> +       dev->release = device_create_release;
>>>> +       dev_set_drvdata(dev, drvdata);
>>>> +
>>>> +       retval = kobject_set_name_vargs(&dev->kobj, fmt, args);
>>>> +       if (retval)
>>>> +               goto error;
>>>> +
>>>> +       retval = device_add(dev);
>>>> +       if (retval)
>>>> +               goto error;
>>>
>>> Exactly! As I said, simply setting dev->groups before calling device_add().
>>>
>>> However, I really don't understand why we need this as global API.
>>> Skimming over the other patches, you use cpu_device_create() only in
>>> one place. Why not hard-code this all there? It is totally OK to do
>>> device initialization in drivers. All the helpers (like
>>> device_create(), device_create_with_groups(), and so on) are just
>>> convenience functions. The driver-core API explicitly allows drivers
>>> to initialize devices manually.
>>>
>>> Nevertheless, this patch looks fine.
>>
>> Wait, no. Why don't you set dev->bus to cpu_subsys? Is this thing
>> supposed to create child-devices of CPUs? Can you describe what your
>> topology is supposed to look like?
>>
>
> Yes, it's not done on purpose as mentioned in the commit log.
> E.g.: cacheinfo topology will be as below
>
> /sys/devices/system/cpu/cpuX/cache/index0/<attributes>
> /sys/devices/system/cpu/cpuX/cache/index1/<attributes>
> ..
> /sys/devices/system/cpu/cpuX/cache/index<Y/<attributes>
>

Does the above topology looks fine to you ? Since the parent is set
properly, not setting bus will not cause any issue to the topology.

> In this case 'cache' is cpuX's child and index<0..Y> are children of
> cache in cpuX. The main problem with per-cpu device is that they have
> same name for each cpu's instance and when the bus is set to this
> devices, the driver model tries to create symlink to each of these
> devices in /sys/bus/cpu/... which fails.
>

Here is the exact issue, probably kernel dump is easier to explain.
It fails for second CPU as the sysfs path already exists.

WARNING: CPU: 1 PID: 1 at fs/sysfs/dir.c:31 sysfs_warn_dup+0x51/0x5c()
sysfs: cannot create duplicate filename '/bus/cpu/devices/cache'
Modules linked in:
CPU: 1 PID: 1 Comm: swapper/0 Not tainted 
3.17.0-rc2-00013-g7956a439b183-dirty #89
[<c0013c3d>] (unwind_backtrace) from [<c0010581>] (show_stack+0x11/0x14)
[<c0010581>] (show_stack) from [<c04e9419>] (dump_stack+0x69/0x74)
[<c04e9419>] (dump_stack) from [<c00204cb>] (warn_slowpath_common+0x5f/0x78)
[<c00204cb>] (warn_slowpath_common) from [<c0020507>] 
(warn_slowpath_fmt+0x23/0x2c)
[<c0020507>] (warn_slowpath_fmt) from [<c013b921>] 
(sysfs_warn_dup+0x51/0x5c)
[<c013b921>] (sysfs_warn_dup) from [<c013bb6d>] 
(sysfs_do_create_link_sd.isra.2+0x91/0x94)
[<c013bb6d>] (sysfs_do_create_link_sd.isra.2) from [<c031d4f3>] 
(bus_add_device+0xab/0x134)
[<c031d4f3>] (bus_add_device) from [<c031c16d>] (device_add+0x2a1/0x3dc)
[<c031c16d>] (device_add) from [<c031f905>] (cpu_device_create+0x85/0x94)

Hi Greg,

Any suggestions to proceed on this ?

Regards,
Sudeep

Re: [PATCH] drivers: base: add cpu_device_create to support per-cpu devices

[David Herrmann]

Hi

On Tue, Aug 26, 2014 at 6:54 PM, Sudeep Holla <sudeep.holla@arm.com> wrote:
>> Yes, it's not done on purpose as mentioned in the commit log.
>> E.g.: cacheinfo topology will be as below
>>
>> /sys/devices/system/cpu/cpuX/cache/index0/<attributes>
>> /sys/devices/system/cpu/cpuX/cache/index1/<attributes>
>> ..
>> /sys/devices/system/cpu/cpuX/cache/index<Y/<attributes>
>>
>
> Does the above topology looks fine to you ? Since the parent is set
> properly, not setting bus will not cause any issue to the topology.

Sure, looks fine.

Thanks
David

Re: [PATCH] drivers: base: add cpu_device_create to support per-cpu devices

[Sudeep Holla]

On 22/08/14 12:37, David Herrmann wrote:
> Hi
>
> On Fri, Aug 22, 2014 at 1:29 PM, Sudeep Holla <sudeep.holla@arm.com> wrote:
>> From: Sudeep Holla <sudeep.holla@arm.com>
>>
>> This patch adds a new function to create per-cpu devices.
>> This helps in:
>> 1. reusing the device infrastructure to create any cpu related
>>     attributes and corresponding sysfs instead of creating and
>>     dealing with raw kobjects directly
>> 2. retaining the legacy path(/sys/devices/system/cpu/..) to support
>>     existing sysfs ABI
>> 3. avoiding to create links in the bus directory pointing to the
>>     device as there would be per-cpu instance of these devices with
>>     the same name since dev->bus is not populated to cpu_sysbus on
>>     purpose
>>
>> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
>> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
>> ---
>>   drivers/base/cpu.c  | 54 +++++++++++++++++++++++++++++++++++++++++++++++++++++
>>   include/linux/cpu.h |  4 ++++
>>   2 files changed, 58 insertions(+)
>>
>> Hi Greg,
>>
>> Here is the alternate solution I could come up with instead of
>> creating cpu class. cpu_device_create is very similar to
>> device_create_groups_vargs w/o class support, but I could not
>> reuse anything else to avoid creating similar function.
>>
>> Let me know your thoughts/suggestions on this.
>>
>> Regards,
>> Sudeep
>>
>>
>> diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c
>> index 277a9cfa9040..53f0c4141d05 100644
>> --- a/drivers/base/cpu.c
>> +++ b/drivers/base/cpu.c
>> @@ -363,6 +363,60 @@ struct device *get_cpu_device(unsigned cpu)
>>   }
>>   EXPORT_SYMBOL_GPL(get_cpu_device);
>>
>> +static void device_create_release(struct device *dev)
>> +{
>> +       kfree(dev);
>> +}
>> +
>> +static struct device *
>> +__cpu_device_create(struct device *parent, void *drvdata,
>> +                   const struct attribute_group **groups,
>> +                   const char *fmt, va_list args)
>> +{
>> +       struct device *dev = NULL;
>> +       int retval = -ENODEV;
>> +
>> +       dev = kzalloc(sizeof(*dev), GFP_KERNEL);
>> +       if (!dev) {
>> +               retval = -ENOMEM;
>> +               goto error;
>> +       }
>> +
>> +       device_initialize(dev);
>> +       dev->parent = parent;
>> +       dev->groups = groups;
>> +       dev->release = device_create_release;
>> +       dev_set_drvdata(dev, drvdata);
>> +
>> +       retval = kobject_set_name_vargs(&dev->kobj, fmt, args);
>> +       if (retval)
>> +               goto error;
>> +
>> +       retval = device_add(dev);
>> +       if (retval)
>> +               goto error;
>
> Exactly! As I said, simply setting dev->groups before calling device_add().
>
> However, I really don't understand why we need this as global API.

Yes right now it's used only in cacheinfo. The main reason why I am 
exporting it is that we can reuse it in some places like cpuidle/freq
which deals with raw kobjects directly and can be moved to device 
attributes.

> Skimming over the other patches, you use cpu_device_create() only in
> one place. Why not hard-code this all there? It is totally OK to do
> device initialization in drivers. All the helpers (like
> device_create(), device_create_with_groups(), and so on) are just
> convenience functions. The driver-core API explicitly allows drivers
> to initialize devices manually.
>
> Nevertheless, this patch looks fine.
>

Thanks for having a look at this.

Regards,
Sudeep

Re: [PATCH] drivers: base: add cpu_device_create to support per-cpu devices

[Sudeep Holla]

Hi Greg,

On 22/08/14 12:29, Sudeep Holla wrote:
> From: Sudeep Holla <sudeep.holla@arm.com>
>
> This patch adds a new function to create per-cpu devices.
> This helps in:
> 1. reusing the device infrastructure to create any cpu related
>     attributes and corresponding sysfs instead of creating and
>     dealing with raw kobjects directly
> 2. retaining the legacy path(/sys/devices/system/cpu/..) to support
>     existing sysfs ABI
> 3. avoiding to create links in the bus directory pointing to the
>     device as there would be per-cpu instance of these devices with
>     the same name since dev->bus is not populated to cpu_sysbus on
>     purpose
>
> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> ---
>   drivers/base/cpu.c  | 54 +++++++++++++++++++++++++++++++++++++++++++++++++++++
>   include/linux/cpu.h |  4 ++++
>   2 files changed, 58 insertions(+)
>
> Hi Greg,
>
> Here is the alternate solution I could come up with instead of
> creating cpu class. cpu_device_create is very similar to
> device_create_groups_vargs w/o class support, but I could not
> reuse anything else to avoid creating similar function.
>
> Let me know your thoughts/suggestions on this.
>

Any feedback on this ? If ok, I will respin cacheinfo series removing
cpu class creation with this patch.

Regards,
Sudeep

Re: [PATCH] drivers: base: add cpu_device_create to support per-cpu devices

[Greg Kroah-Hartman]

On Tue, Sep 02, 2014 at 06:22:07PM +0100, Sudeep Holla wrote:
> Hi Greg,
> 
> On 22/08/14 12:29, Sudeep Holla wrote:
> >From: Sudeep Holla <sudeep.holla@arm.com>
> >
> >This patch adds a new function to create per-cpu devices.
> >This helps in:
> >1. reusing the device infrastructure to create any cpu related
> >    attributes and corresponding sysfs instead of creating and
> >    dealing with raw kobjects directly
> >2. retaining the legacy path(/sys/devices/system/cpu/..) to support
> >    existing sysfs ABI
> >3. avoiding to create links in the bus directory pointing to the
> >    device as there would be per-cpu instance of these devices with
> >    the same name since dev->bus is not populated to cpu_sysbus on
> >    purpose
> >
> >Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
> >Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> >---
> >  drivers/base/cpu.c  | 54 +++++++++++++++++++++++++++++++++++++++++++++++++++++
> >  include/linux/cpu.h |  4 ++++
> >  2 files changed, 58 insertions(+)
> >
> >Hi Greg,
> >
> >Here is the alternate solution I could come up with instead of
> >creating cpu class. cpu_device_create is very similar to
> >device_create_groups_vargs w/o class support, but I could not
> >reuse anything else to avoid creating similar function.
> >
> >Let me know your thoughts/suggestions on this.
> >
> 
> Any feedback on this ? If ok, I will respin cacheinfo series removing
> cpu class creation with this patch.

Please address the issues others have already raised.

Re: [PATCH] drivers: base: add cpu_device_create to support per-cpu devices

[Sudeep Holla]

Hi Greg,

On 02/09/14 18:26, Greg Kroah-Hartman wrote:
> On Tue, Sep 02, 2014 at 06:22:07PM +0100, Sudeep Holla wrote:
>> Hi Greg,
>>
>> On 22/08/14 12:29, Sudeep Holla wrote:
>>> From: Sudeep Holla <sudeep.holla@arm.com>
>>>
>>> This patch adds a new function to create per-cpu devices.
>>> This helps in:
>>> 1. reusing the device infrastructure to create any cpu related
>>>     attributes and corresponding sysfs instead of creating and
>>>     dealing with raw kobjects directly
>>> 2. retaining the legacy path(/sys/devices/system/cpu/..) to support
>>>     existing sysfs ABI
>>> 3. avoiding to create links in the bus directory pointing to the
>>>     device as there would be per-cpu instance of these devices with
>>>     the same name since dev->bus is not populated to cpu_sysbus on
>>>     purpose
>>>
>>> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
>>> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
>>> ---
>>>   drivers/base/cpu.c  | 54 +++++++++++++++++++++++++++++++++++++++++++++++++++++
>>>   include/linux/cpu.h |  4 ++++
>>>   2 files changed, 58 insertions(+)
>>>
>>> Hi Greg,
>>>
>>> Here is the alternate solution I could come up with instead of
>>> creating cpu class. cpu_device_create is very similar to
>>> device_create_groups_vargs w/o class support, but I could not
>>> reuse anything else to avoid creating similar function.
>>>
>>> Let me know your thoughts/suggestions on this.
>>>
>>
>> Any feedback on this ? If ok, I will respin cacheinfo series removing
>> cpu class creation with this patch.
>
> Please address the issues others have already raised.
>

If you meant issues raised on cacheinfo series, I have addressed them
already in v3(raised by you, Stephen Boyd, Mark and Russell). I have
retained hotplug notifiers which Stephen wanted to remove as I don't
know if that's acceptable on all the architectures.

I posted this single patch as a replacement to the patch creating cpu
class as Kay and David suggested that we can't have class and bus with
same name.

Regards,
Sudeep

Re: [PATCH] drivers: base: add cpu_device_create to support per-cpu devices

[Greg Kroah-Hartman]

On Tue, Sep 02, 2014 at 06:40:21PM +0100, Sudeep Holla wrote:
> Hi Greg,
> 
> On 02/09/14 18:26, Greg Kroah-Hartman wrote:
> >On Tue, Sep 02, 2014 at 06:22:07PM +0100, Sudeep Holla wrote:
> >>Hi Greg,
> >>
> >>On 22/08/14 12:29, Sudeep Holla wrote:
> >>>From: Sudeep Holla <sudeep.holla@arm.com>
> >>>
> >>>This patch adds a new function to create per-cpu devices.
> >>>This helps in:
> >>>1. reusing the device infrastructure to create any cpu related
> >>>    attributes and corresponding sysfs instead of creating and
> >>>    dealing with raw kobjects directly
> >>>2. retaining the legacy path(/sys/devices/system/cpu/..) to support
> >>>    existing sysfs ABI
> >>>3. avoiding to create links in the bus directory pointing to the
> >>>    device as there would be per-cpu instance of these devices with
> >>>    the same name since dev->bus is not populated to cpu_sysbus on
> >>>    purpose
> >>>
> >>>Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
> >>>Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> >>>---
> >>>  drivers/base/cpu.c  | 54 +++++++++++++++++++++++++++++++++++++++++++++++++++++
> >>>  include/linux/cpu.h |  4 ++++
> >>>  2 files changed, 58 insertions(+)
> >>>
> >>>Hi Greg,
> >>>
> >>>Here is the alternate solution I could come up with instead of
> >>>creating cpu class. cpu_device_create is very similar to
> >>>device_create_groups_vargs w/o class support, but I could not
> >>>reuse anything else to avoid creating similar function.
> >>>
> >>>Let me know your thoughts/suggestions on this.
> >>>
> >>
> >>Any feedback on this ? If ok, I will respin cacheinfo series removing
> >>cpu class creation with this patch.
> >
> >Please address the issues others have already raised.
> >
> 
> If you meant issues raised on cacheinfo series, I have addressed them
> already in v3(raised by you, Stephen Boyd, Mark and Russell). I have
> retained hotplug notifiers which Stephen wanted to remove as I don't
> know if that's acceptable on all the architectures.
> 
> I posted this single patch as a replacement to the patch creating cpu
> class as Kay and David suggested that we can't have class and bus with
> same name.

Then I have no idea what is going on, sorry, please respin as you see
fit and resend...

thanks,

greg k-h

[PATCH v3 04/11] drivers: base: support cpu cache information interface to userspace via sysfs

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds initial support for providing processor cache information
to userspace through sysfs interface. This is based on already existing
implementations(x86, ia64, s390 and powerpc) and hence the interface is
intended to be fully compatible.

The main purpose of this generic support is to avoid further code
duplication to support new architectures and also to unify all the existing
different implementations.

This implementation maintains the hierarchy of cache objects which reflects
the system's cache topology. Cache devices are instantiated as needed as
CPUs come online. The cache information is replicated per-cpu even if they are
shared. A per-cpu array of cache information maintained is used mainly for
sysfs-related book keeping.

It also implements the shared_cpu_map attribute, which is essential for
enabling both kernel and user-space to discover the system's overall cache
topology.

This patch also add the missing ABI documentation for the cacheinfo sysfs
interface already, which is well defined and widely used.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-doc@vger.kernel.org
---
 Documentation/ABI/testing/sysfs-devices-system-cpu |  47 ++
 drivers/base/Makefile                              |   2 +-
 drivers/base/cacheinfo.c                           | 543 +++++++++++++++++++++
 include/linux/cacheinfo.h                          | 100 ++++
 4 files changed, 691 insertions(+), 1 deletion(-)
 create mode 100644 drivers/base/cacheinfo.c
 create mode 100644 include/linux/cacheinfo.h

diff --git a/Documentation/ABI/testing/sysfs-devices-system-cpu b/Documentation/ABI/testing/sysfs-devices-system-cpu
index acb9bfc89b48..99983e67c13c 100644
--- a/Documentation/ABI/testing/sysfs-devices-system-cpu
+++ b/Documentation/ABI/testing/sysfs-devices-system-cpu
@@ -224,3 +224,50 @@ Description:	Parameters for the Intel P-state driver
 		frequency range.
 
 		More details can be found in Documentation/cpu-freq/intel-pstate.txt
+
+What:		/sys/devices/system/cpu/cpu*/cache/index*/<set_of_attributes_mentioned_below>
+Date:		July 2014(documented, existed before August 2008)
+Contact:	Sudeep Holla <sudeep.holla@arm.com>
+		Linux kernel mailing list <linux-kernel@vger.kernel.org>
+Description:	Parameters for the CPU cache attributes
+
+		allocation_policy:
+			- WriteAllocate: allocate a memory location to a cache line
+					 on a cache miss because of a write
+			- ReadAllocate: allocate a memory location to a cache line
+					on a cache miss because of a read
+			- ReadWriteAllocate: both writeallocate and readallocate
+
+		attributes: LEGACY used only on IA64 and is same as write_policy
+
+		coherency_line_size: the minimum amount of data in bytes that gets
+				     transferred from memory to cache
+
+		level: the cache hierarcy in the multi-level cache configuration
+
+		number_of_sets: total number of sets in the cache, a set is a
+				collection of cache lines with the same cache index
+
+		physical_line_partition: number of physical cache line per cache tag
+
+		shared_cpu_list: the list of logical cpus sharing the cache
+
+		shared_cpu_map: logical cpu mask containing the list of cpus sharing
+				the cache
+
+		size: the total cache size in kB
+
+		type:
+			- Instruction: cache that only holds instructions
+			- Data: cache that only caches data
+			- Unified: cache that holds both data and instructions
+
+		ways_of_associativity: degree of freedom in placing a particular block
+					of memory in the cache
+
+		write_policy:
+			- WriteThrough: data is written to both the cache line
+					and to the block in the lower-level memory
+			- WriteBack: data is written only to the cache line and
+				     the modified cache line is written to main
+				     memory only when it is replaced
diff --git a/drivers/base/Makefile b/drivers/base/Makefile
index 4aab26ec0292..f901bc1cffc8 100644
--- a/drivers/base/Makefile
+++ b/drivers/base/Makefile
@@ -4,7 +4,7 @@ obj-y			:= component.o core.o bus.o dd.o syscore.o \
 			   driver.o class.o platform.o \
 			   cpu.o firmware.o init.o map.o devres.o \
 			   attribute_container.o transport_class.o \
-			   topology.o container.o
+			   topology.o container.o cacheinfo.o
 obj-$(CONFIG_DEVTMPFS)	+= devtmpfs.o
 obj-$(CONFIG_DMA_CMA) += dma-contiguous.o
 obj-y			+= power/
diff --git a/drivers/base/cacheinfo.c b/drivers/base/cacheinfo.c
new file mode 100644
index 000000000000..97728ce19505
--- /dev/null
+++ b/drivers/base/cacheinfo.c
@@ -0,0 +1,543 @@
+/*
+ * cacheinfo support - processor cache information via sysfs
+ *
+ * Based on arch/x86/kernel/cpu/intel_cacheinfo.c
+ * Author: Sudeep Holla <sudeep.holla@arm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/compiler.h>
+#include <linux/cpu.h>
+#include <linux/device.h>
+#include <linux/init.h>
+#include <linux/of.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/smp.h>
+#include <linux/sysfs.h>
+
+/* pointer to per cpu cacheinfo */
+static DEFINE_PER_CPU(struct cpu_cacheinfo, ci_cpu_cacheinfo);
+#define ci_cacheinfo(cpu)	(&per_cpu(ci_cpu_cacheinfo, cpu))
+#define cache_leaves(cpu)	(ci_cacheinfo(cpu)->num_leaves)
+#define per_cpu_cacheinfo(cpu)	(ci_cacheinfo(cpu)->info_list)
+
+struct cpu_cacheinfo *get_cpu_cacheinfo(unsigned int cpu)
+{
+	return ci_cacheinfo(cpu);
+}
+
+#ifdef CONFIG_OF
+static int cache_setup_of_node(unsigned int cpu)
+{
+	struct device_node *np;
+	struct cacheinfo *this_leaf;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int index = 0;
+
+	/* skip if of_node is already populated */
+	if (this_cpu_ci->info_list->of_node)
+		return 0;
+
+	if (!cpu_dev) {
+		pr_err("No cpu device for CPU %d\n", cpu);
+		return -ENODEV;
+	}
+	np = cpu_dev->of_node;
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
+	}
+
+	while (np && index < cache_leaves(cpu)) {
+		this_leaf = this_cpu_ci->info_list + index;
+		if (this_leaf->level != 1)
+			np = of_find_next_cache_node(np);
+		else
+			np = of_node_get(np);/* cpu node itself */
+		this_leaf->of_node = np;
+		index++;
+	}
+	return 0;
+}
+
+static inline bool cache_leaves_are_shared(struct cacheinfo *this_leaf,
+					   struct cacheinfo *sib_leaf)
+{
+	return sib_leaf->of_node == this_leaf->of_node;
+}
+#else
+static inline int cache_setup_of_node(unsigned int cpu) { return 0; }
+static inline bool cache_leaves_are_shared(struct cacheinfo *this_leaf,
+					   struct cacheinfo *sib_leaf)
+{
+	/*
+	 * For non-DT systems, assume unique level 1 cache, system-wide
+	 * shared caches for all other levels. This will be used only if
+	 * arch specific code has not populated shared_cpu_map
+	 */
+	return !(this_leaf->level == 1);
+}
+#endif
+
+static int cache_shared_cpu_map_setup(unsigned int cpu)
+{
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sib_leaf;
+	unsigned int index;
+	int ret;
+
+	ret = cache_setup_of_node(cpu);
+	if (ret)
+		return ret;
+
+	for (index = 0; index < cache_leaves(cpu); index++) {
+		unsigned int i;
+
+		this_leaf = this_cpu_ci->info_list + index;
+		/* skip if shared_cpu_map is already populated */
+		if (!cpumask_empty(&this_leaf->shared_cpu_map))
+			continue;
+
+		cpumask_set_cpu(cpu, &this_leaf->shared_cpu_map);
+		for_each_online_cpu(i) {
+			struct cpu_cacheinfo *sib_cpu_ci = get_cpu_cacheinfo(i);
+
+			if (i == cpu || !sib_cpu_ci->info_list)
+				continue;/* skip if itself or no cacheinfo */
+			sib_leaf = sib_cpu_ci->info_list + index;
+			if (cache_leaves_are_shared(this_leaf, sib_leaf)) {
+				cpumask_set_cpu(cpu, &sib_leaf->shared_cpu_map);
+				cpumask_set_cpu(i, &this_leaf->shared_cpu_map);
+			}
+		}
+	}
+
+	return 0;
+}
+
+static void cache_shared_cpu_map_remove(unsigned int cpu)
+{
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sib_leaf;
+	unsigned int sibling, index;
+
+	for (index = 0; index < cache_leaves(cpu); index++) {
+		this_leaf = this_cpu_ci->info_list + index;
+		for_each_cpu(sibling, &this_leaf->shared_cpu_map) {
+			struct cpu_cacheinfo *sib_cpu_ci;
+
+			if (sibling == cpu) /* skip itself */
+				continue;
+			sib_cpu_ci = get_cpu_cacheinfo(sibling);
+			sib_leaf = sib_cpu_ci->info_list + index;
+			cpumask_clear_cpu(cpu, &sib_leaf->shared_cpu_map);
+			cpumask_clear_cpu(sibling, &this_leaf->shared_cpu_map);
+		}
+		of_node_put(this_leaf->of_node);
+	}
+}
+
+static void free_cache_attributes(unsigned int cpu)
+{
+	cache_shared_cpu_map_remove(cpu);
+
+	kfree(per_cpu_cacheinfo(cpu));
+	per_cpu_cacheinfo(cpu) = NULL;
+}
+
+int __weak init_cache_level(unsigned int cpu)
+{
+	return -ENOENT;
+}
+
+int __weak populate_cache_leaves(unsigned int cpu)
+{
+	return -ENOENT;
+}
+
+static int detect_cache_attributes(unsigned int cpu)
+{
+	int ret;
+
+	if (init_cache_level(cpu))
+		return -ENOENT;
+
+	per_cpu_cacheinfo(cpu) = kcalloc(cache_leaves(cpu),
+					 sizeof(struct cacheinfo), GFP_KERNEL);
+	if (per_cpu_cacheinfo(cpu) == NULL)
+		return -ENOMEM;
+
+	ret = populate_cache_leaves(cpu);
+	if (ret)
+		goto free_ci;
+	/*
+	 * For systems using DT for cache hierarcy, of_node and shared_cpu_map
+	 * will be set up here only if they are not populated already
+	 */
+	ret = cache_shared_cpu_map_setup(cpu);
+	if (ret)
+		goto free_ci;
+	return 0;
+
+free_ci:
+	free_cache_attributes(cpu);
+	return ret;
+}
+
+/* pointer to cpuX/cache device */
+static DEFINE_PER_CPU(struct device *, ci_cache_dev);
+#define per_cpu_cache_dev(cpu)	(per_cpu(ci_cache_dev, cpu))
+
+static cpumask_t cache_dev_map;
+
+/* pointer to array of devices for cpuX/cache/indexY */
+static DEFINE_PER_CPU(struct device **, ci_index_dev);
+#define per_cpu_index_dev(cpu)	(per_cpu(ci_index_dev, cpu))
+#define per_cache_index_dev(cpu, idx)	((per_cpu_index_dev(cpu))[idx])
+
+#define show_one(file_name, object)				\
+static ssize_t file_name##_show(struct device *dev,		\
+		struct device_attribute *attr, char *buf)	\
+{								\
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);	\
+	return sprintf(buf, "%u\n", this_leaf->object);		\
+}
+
+show_one(level, level);
+show_one(coherency_line_size, coherency_line_size);
+show_one(number_of_sets, number_of_sets);
+show_one(physical_line_partition, physical_line_partition);
+show_one(ways_of_associativity, ways_of_associativity);
+
+static ssize_t size_show(struct device *dev,
+			 struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	return sprintf(buf, "%uK\n", this_leaf->size >> 10);
+}
+
+static ssize_t shared_cpumap_show_func(struct device *dev, bool list, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	const struct cpumask *mask = &this_leaf->shared_cpu_map;
+
+	return cpumap_copy_to_buf(list, mask, buf);
+}
+
+static ssize_t shared_cpu_map_show(struct device *dev,
+				   struct device_attribute *attr, char *buf)
+{
+	return shared_cpumap_show_func(dev, false, buf);
+}
+
+static ssize_t shared_cpu_list_show(struct device *dev,
+				    struct device_attribute *attr, char *buf)
+{
+	return shared_cpumap_show_func(dev, true, buf);
+}
+
+static ssize_t type_show(struct device *dev,
+			 struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+
+	switch (this_leaf->type) {
+	case CACHE_TYPE_DATA:
+		return sprintf(buf, "Data\n");
+	case CACHE_TYPE_INST:
+		return sprintf(buf, "Instruction\n");
+	case CACHE_TYPE_UNIFIED:
+		return sprintf(buf, "Unified\n");
+	default:
+		return -EINVAL;
+	}
+}
+
+static ssize_t allocation_policy_show(struct device *dev,
+				      struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	unsigned int ci_attr = this_leaf->attributes;
+	int n = 0;
+
+	if ((ci_attr & CACHE_READ_ALLOCATE) && (ci_attr & CACHE_WRITE_ALLOCATE))
+		n = sprintf(buf, "ReadWriteAllocate\n");
+	else if (ci_attr & CACHE_READ_ALLOCATE)
+		n = sprintf(buf, "ReadAllocate\n");
+	else if (ci_attr & CACHE_WRITE_ALLOCATE)
+		n = sprintf(buf, "WriteAllocate\n");
+	return n;
+}
+
+static ssize_t write_policy_show(struct device *dev,
+				 struct device_attribute *attr, char *buf)
+{
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	unsigned int ci_attr = this_leaf->attributes;
+	int n = 0;
+
+	if (ci_attr & CACHE_WRITE_THROUGH)
+		n = sprintf(buf, "WriteThrough\n");
+	else if (ci_attr & CACHE_WRITE_BACK)
+		n = sprintf(buf, "WriteBack\n");
+	return n;
+}
+
+static DEVICE_ATTR_RO(level);
+static DEVICE_ATTR_RO(type);
+static DEVICE_ATTR_RO(coherency_line_size);
+static DEVICE_ATTR_RO(ways_of_associativity);
+static DEVICE_ATTR_RO(number_of_sets);
+static DEVICE_ATTR_RO(size);
+static DEVICE_ATTR_RO(allocation_policy);
+static DEVICE_ATTR_RO(write_policy);
+static DEVICE_ATTR_RO(shared_cpu_map);
+static DEVICE_ATTR_RO(shared_cpu_list);
+static DEVICE_ATTR_RO(physical_line_partition);
+
+static struct attribute *cache_default_attrs[] = {
+	&dev_attr_type.attr,
+	&dev_attr_level.attr,
+	&dev_attr_shared_cpu_map.attr,
+	&dev_attr_shared_cpu_list.attr,
+	&dev_attr_coherency_line_size.attr,
+	&dev_attr_ways_of_associativity.attr,
+	&dev_attr_number_of_sets.attr,
+	&dev_attr_size.attr,
+	&dev_attr_allocation_policy.attr,
+	&dev_attr_write_policy.attr,
+	&dev_attr_physical_line_partition.attr,
+	NULL
+};
+
+static umode_t
+cache_default_attrs_is_visible(struct kobject *kobj,
+			       struct attribute *attr, int unused)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	const struct cpumask *mask = &this_leaf->shared_cpu_map;
+	umode_t mode = attr->mode;
+
+	if ((attr == &dev_attr_type.attr) && this_leaf->type)
+		return mode;
+	if ((attr == &dev_attr_level.attr) && this_leaf->level)
+		return mode;
+	if ((attr == &dev_attr_shared_cpu_map.attr) && !cpumask_empty(mask))
+		return mode;
+	if ((attr == &dev_attr_shared_cpu_list.attr) && !cpumask_empty(mask))
+		return mode;
+	if ((attr == &dev_attr_coherency_line_size.attr) &&
+	    this_leaf->coherency_line_size)
+		return mode;
+	if ((attr == &dev_attr_ways_of_associativity.attr) &&
+	    this_leaf->size) /* allow 0 = full associativity */
+		return mode;
+	if ((attr == &dev_attr_number_of_sets.attr) &&
+	    this_leaf->number_of_sets)
+		return mode;
+	if ((attr == &dev_attr_size.attr) && this_leaf->size)
+		return mode;
+	if ((attr == &dev_attr_write_policy.attr) &&
+	    (this_leaf->attributes & CACHE_WRITE_POLICY_MASK))
+		return mode;
+	if ((attr == &dev_attr_allocation_policy.attr) &&
+	    (this_leaf->attributes & CACHE_ALLOCATE_POLICY_MASK))
+		return mode;
+	if ((attr == &dev_attr_physical_line_partition.attr) &&
+	    this_leaf->physical_line_partition)
+		return mode;
+
+	return 0;
+}
+
+static const struct attribute_group cache_default_group = {
+	.attrs = cache_default_attrs,
+	.is_visible = cache_default_attrs_is_visible,
+};
+
+static const struct attribute_group *cache_default_groups[] = {
+	&cache_default_group,
+	NULL,
+};
+
+static const struct attribute_group *cache_private_groups[] = {
+	&cache_default_group,
+	NULL, /* Place holder for private group */
+	NULL,
+};
+
+const struct attribute_group *
+__weak cache_get_priv_group(struct cacheinfo *this_leaf)
+{
+	return NULL;
+}
+
+static const struct attribute_group **
+cache_get_attribute_groups(struct cacheinfo *this_leaf)
+{
+	const struct attribute_group *priv_group =
+			cache_get_priv_group(this_leaf);
+
+	if (!priv_group)
+		return cache_default_groups;
+
+	if (!cache_private_groups[1])
+		cache_private_groups[1] = priv_group;
+
+	return cache_private_groups;
+}
+
+/* Add/Remove cache interface for CPU device */
+static void cpu_cache_sysfs_exit(unsigned int cpu)
+{
+	int i;
+	struct device *ci_dev;
+
+	if (per_cpu_index_dev(cpu)) {
+		for (i = 0; i < cache_leaves(cpu); i++) {
+			ci_dev = per_cache_index_dev(cpu, i);
+			if (!ci_dev)
+				continue;
+			device_unregister(ci_dev);
+		}
+		kfree(per_cpu_index_dev(cpu));
+		per_cpu_index_dev(cpu) = NULL;
+	}
+	device_unregister(per_cpu_cache_dev(cpu));
+	per_cpu_cache_dev(cpu) = NULL;
+}
+
+static int cpu_cache_sysfs_init(unsigned int cpu)
+{
+	struct device *dev = get_cpu_device(cpu);
+
+	if (per_cpu_cacheinfo(cpu) == NULL)
+		return -ENOENT;
+
+	per_cpu_cache_dev(cpu) = device_create(dev->class, dev, cpu,
+					       NULL, "cache");
+	if (IS_ERR(per_cpu_cache_dev(cpu)))
+		return PTR_ERR(per_cpu_cache_dev(cpu));
+
+	/* Allocate all required memory */
+	per_cpu_index_dev(cpu) = kcalloc(cache_leaves(cpu),
+					 sizeof(struct device *), GFP_KERNEL);
+	if (unlikely(per_cpu_index_dev(cpu) == NULL))
+		goto err_out;
+
+	return 0;
+
+err_out:
+	cpu_cache_sysfs_exit(cpu);
+	return -ENOMEM;
+}
+
+static int cache_add_dev(unsigned int cpu)
+{
+	unsigned int i;
+	int rc;
+	struct device *ci_dev, *parent;
+	struct cacheinfo *this_leaf;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	const struct attribute_group **cache_groups;
+
+	rc = cpu_cache_sysfs_init(cpu);
+	if (unlikely(rc < 0))
+		return rc;
+
+	parent = per_cpu_cache_dev(cpu);
+	for (i = 0; i < cache_leaves(cpu); i++) {
+		this_leaf = this_cpu_ci->info_list + i;
+		if (this_leaf->disable_sysfs)
+			continue;
+		cache_groups = cache_get_attribute_groups(this_leaf);
+		ci_dev = device_create_with_groups(parent->class, parent, i,
+						   this_leaf, cache_groups,
+						   "index%1u", i);
+		if (IS_ERR(ci_dev)) {
+			rc = PTR_ERR(ci_dev);
+			goto err;
+		}
+		per_cache_index_dev(cpu, i) = ci_dev;
+	}
+	cpumask_set_cpu(cpu, &cache_dev_map);
+
+	return 0;
+err:
+	cpu_cache_sysfs_exit(cpu);
+	return rc;
+}
+
+static void cache_remove_dev(unsigned int cpu)
+{
+	if (!cpumask_test_cpu(cpu, &cache_dev_map))
+		return;
+	cpumask_clear_cpu(cpu, &cache_dev_map);
+
+	cpu_cache_sysfs_exit(cpu);
+}
+
+static int cacheinfo_cpu_callback(struct notifier_block *nfb,
+				  unsigned long action, void *hcpu)
+{
+	unsigned int cpu = (unsigned long)hcpu;
+	int rc = 0;
+
+	switch (action & ~CPU_TASKS_FROZEN) {
+	case CPU_ONLINE:
+		rc = detect_cache_attributes(cpu);
+		if (!rc)
+			rc = cache_add_dev(cpu);
+		break;
+	case CPU_DEAD:
+		cache_remove_dev(cpu);
+		if (per_cpu_cacheinfo(cpu))
+			free_cache_attributes(cpu);
+		break;
+	}
+	return notifier_from_errno(rc);
+}
+
+static int __init cacheinfo_sysfs_init(void)
+{
+	int cpu, rc = 0;
+
+	cpu_notifier_register_begin();
+
+	for_each_online_cpu(cpu) {
+		rc = detect_cache_attributes(cpu);
+		if (rc) {
+			pr_err("error detecting cacheinfo..cpu%d\n", cpu);
+			goto out;
+		}
+		rc = cache_add_dev(cpu);
+		if (rc) {
+			free_cache_attributes(cpu);
+			pr_err("error populating cacheinfo..cpu%d\n", cpu);
+			goto out;
+		}
+	}
+	__hotcpu_notifier(cacheinfo_cpu_callback, 0);
+
+out:
+	cpu_notifier_register_done();
+	return rc;
+}
+
+device_initcall(cacheinfo_sysfs_init);
diff --git a/include/linux/cacheinfo.h b/include/linux/cacheinfo.h
new file mode 100644
index 000000000000..3daf5ed392c9
--- /dev/null
+++ b/include/linux/cacheinfo.h
@@ -0,0 +1,100 @@
+#ifndef _LINUX_CACHEINFO_H
+#define _LINUX_CACHEINFO_H
+
+#include <linux/bitops.h>
+#include <linux/cpumask.h>
+#include <linux/smp.h>
+
+struct device_node;
+struct attribute;
+
+enum cache_type {
+	CACHE_TYPE_NOCACHE = 0,
+	CACHE_TYPE_INST = BIT(0),
+	CACHE_TYPE_DATA = BIT(1),
+	CACHE_TYPE_SEPARATE = CACHE_TYPE_INST | CACHE_TYPE_DATA,
+	CACHE_TYPE_UNIFIED = BIT(2),
+};
+
+/**
+ * struct cacheinfo - represent a cache leaf node
+ * @type: type of the cache - data, inst or unified
+ * @level: represents the hierarcy in the multi-level cache
+ * @coherency_line_size: size of each cache line usually representing
+ *	the minimum amount of data that gets transferred from memory
+ * @number_of_sets: total number of sets, a set is a collection of cache
+ *	lines sharing the same index
+ * @ways_of_associativity: number of ways in which a particular memory
+ *	block can be placed in the cache
+ * @physical_line_partition: number of physical cache lines sharing the
+ *	same cachetag
+ * @size: Total size of the cache
+ * @shared_cpu_map: logical cpumask representing all the cpus sharing
+ *	this cache node
+ * @attributes: bitfield representing various cache attributes
+ * @of_node: if devicetree is used, this represents either the cpu node in
+ *	case there's no explicit cache node or the cache node itself in the
+ *	device tree
+ * @disable_sysfs: indicates whether this node is visible to the user via
+ *	sysfs or not
+ * @priv: pointer to any private data structure specific to particular
+ *	cache design
+ *
+ * While @of_node, @disable_sysfs and @priv are used for internal book
+ * keeping, the remaining members form the core properties of the cache
+ */
+struct cacheinfo {
+	enum cache_type type;
+	unsigned int level;
+	unsigned int coherency_line_size;
+	unsigned int number_of_sets;
+	unsigned int ways_of_associativity;
+	unsigned int physical_line_partition;
+	unsigned int size;
+	cpumask_t shared_cpu_map;
+	unsigned int attributes;
+#define CACHE_WRITE_THROUGH	BIT(0)
+#define CACHE_WRITE_BACK	BIT(1)
+#define CACHE_WRITE_POLICY_MASK		\
+	(CACHE_WRITE_THROUGH | CACHE_WRITE_BACK)
+#define CACHE_READ_ALLOCATE	BIT(2)
+#define CACHE_WRITE_ALLOCATE	BIT(3)
+#define CACHE_ALLOCATE_POLICY_MASK	\
+	(CACHE_READ_ALLOCATE | CACHE_WRITE_ALLOCATE)
+
+	struct device_node *of_node;
+	bool disable_sysfs;
+	void *priv;
+};
+
+struct cpu_cacheinfo {
+	struct cacheinfo *info_list;
+	unsigned int num_levels;
+	unsigned int num_leaves;
+};
+
+/*
+ * Helpers to make sure "func" is executed on the cpu whose cache
+ * attributes are being detected
+ */
+#define DEFINE_SMP_CALL_CACHE_FUNCTION(func)			\
+static inline void _##func(void *ret)				\
+{								\
+	int cpu = smp_processor_id();				\
+	*(int *)ret = __##func(cpu);				\
+}								\
+								\
+int func(unsigned int cpu)					\
+{								\
+	int ret;						\
+	smp_call_function_single(cpu, _##func, &ret, true);	\
+	return ret;						\
+}
+
+struct cpu_cacheinfo *get_cpu_cacheinfo(unsigned int cpu);
+int init_cache_level(unsigned int cpu);
+int populate_cache_leaves(unsigned int cpu);
+
+const struct attribute_group *cache_get_priv_group(struct cacheinfo *this_leaf);
+
+#endif /* _LINUX_CACHEINFO_H */
-- 
1.8.3.2

[PATCH v3 05/11] ia64: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: linux-ia64@vger.kernel.org
---
 arch/ia64/kernel/topology.c | 421 ++++++++++++--------------------------------
 1 file changed, 111 insertions(+), 310 deletions(-)

diff --git a/arch/ia64/kernel/topology.c b/arch/ia64/kernel/topology.c
index f295f9abba4b..3d6749cb53ec 100644
--- a/arch/ia64/kernel/topology.c
+++ b/arch/ia64/kernel/topology.c
@@ -13,6 +13,7 @@
  *	Populate cpu cache entries in sysfs for cpu cache info
  */
 
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
 #include <linux/kernel.h>
 #include <linux/mm.h>
@@ -21,7 +22,6 @@
 #include <linux/init.h>
 #include <linux/bootmem.h>
 #include <linux/nodemask.h>
-#include <linux/notifier.h>
 #include <linux/export.h>
 #include <asm/mmzone.h>
 #include <asm/numa.h>
@@ -103,60 +103,25 @@ subsys_initcall(topology_init);
 /*
  * Export cpu cache information through sysfs
  */
-
-/*
- *  A bunch of string array to get pretty printing
- */
-static const char *cache_types[] = {
-	"",			/* not used */
-	"Instruction",
-	"Data",
-	"Unified"	/* unified */
-};
-
-static const char *cache_mattrib[]={
-	"WriteThrough",
-	"WriteBack",
-	"",		/* reserved */
-	""		/* reserved */
-};
-
-struct cache_info {
-	pal_cache_config_info_t	cci;
-	cpumask_t shared_cpu_map;
-	int level;
-	int type;
-	struct kobject kobj;
-};
-
-struct cpu_cache_info {
-	struct cache_info *cache_leaves;
-	int	num_cache_leaves;
-	struct kobject kobj;
-};
-
-static struct cpu_cache_info	all_cpu_cache_info[NR_CPUS];
-#define LEAF_KOBJECT_PTR(x,y)    (&all_cpu_cache_info[x].cache_leaves[y])
-
 #ifdef CONFIG_SMP
-static void cache_shared_cpu_map_setup(unsigned int cpu,
-		struct cache_info * this_leaf)
+static int __cache_cpumap_setup(unsigned int cpu, struct cacheinfo *this_leaf)
 {
 	pal_cache_shared_info_t	csi;
-	int num_shared, i = 0;
-	unsigned int j;
+	int num_shared, i = 0, j;
+	enum cache_type type = this_leaf->type;
 
 	if (cpu_data(cpu)->threads_per_core <= 1 &&
 		cpu_data(cpu)->cores_per_socket <= 1) {
 		cpu_set(cpu, this_leaf->shared_cpu_map);
-		return;
+		return 0;
 	}
 
-	if (ia64_pal_cache_shared_info(this_leaf->level,
-					this_leaf->type,
-					0,
-					&csi) != PAL_STATUS_SUCCESS)
-		return;
+	if (type == CACHE_TYPE_UNIFIED)
+		type = CACHE_TYPE_DATA;
+
+	if (ia64_pal_cache_shared_info(this_leaf->level, type, 0,
+				       &csi) != PAL_STATUS_SUCCESS)
+		return -EIO;
 
 	num_shared = (int) csi.num_shared;
 	do {
@@ -168,305 +133,141 @@ static void cache_shared_cpu_map_setup(unsigned int cpu,
 
 		i++;
 	} while (i < num_shared &&
-		ia64_pal_cache_shared_info(this_leaf->level,
-				this_leaf->type,
-				i,
-				&csi) == PAL_STATUS_SUCCESS);
-}
-#else
-static void cache_shared_cpu_map_setup(unsigned int cpu,
-		struct cache_info * this_leaf)
-{
-	cpu_set(cpu, this_leaf->shared_cpu_map);
-	return;
-}
-#endif
-
-static ssize_t show_coherency_line_size(struct cache_info *this_leaf,
-					char *buf)
-{
-	return sprintf(buf, "%u\n", 1 << this_leaf->cci.pcci_line_size);
+		 ia64_pal_cache_shared_info(this_leaf->level, type, i,
+					    &csi) == PAL_STATUS_SUCCESS);
+	return 0;
 }
 
-static ssize_t show_ways_of_associativity(struct cache_info *this_leaf,
-					char *buf)
+static int cache_shared_cpu_map_setup(unsigned int cpu)
 {
-	return sprintf(buf, "%u\n", this_leaf->cci.pcci_assoc);
+	unsigned int idx;
+	int ret = 0;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++, this_leaf++) {
+		ret = __cache_cpumap_setup(cpu, this_leaf);
+		if (ret)
+			break;
+	}
+	return ret;
 }
-
-static ssize_t show_attributes(struct cache_info *this_leaf, char *buf)
+#else
+static int cache_shared_cpu_map_setup(unsigned int cpu)
 {
-	return sprintf(buf,
-			"%s\n",
-			cache_mattrib[this_leaf->cci.pcci_cache_attr]);
-}
+	int idx;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
 
-static ssize_t show_size(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf, "%uK\n", this_leaf->cci.pcci_cache_size / 1024);
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++, this_leaf++)
+		cpu_set(cpu, this_leaf->shared_cpu_map);
+	return 0;
 }
+#endif
 
-static ssize_t show_number_of_sets(struct cache_info *this_leaf, char *buf)
+static ssize_t attributes_show(struct device *dev,
+			       struct device_attribute *attr, char *buf)
 {
-	unsigned number_of_sets = this_leaf->cci.pcci_cache_size;
-	number_of_sets /= this_leaf->cci.pcci_assoc;
-	number_of_sets /= 1 << this_leaf->cci.pcci_line_size;
-
-	return sprintf(buf, "%u\n", number_of_sets);
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	unsigned int ci_attr = this_leaf->attributes;
+	int n = 0;
+
+	if (ci_attr & CACHE_WRITE_THROUGH)
+		n = sprintf(buf, "WriteThrough\n");
+	else if (ci_attr & CACHE_WRITE_BACK)
+		n = sprintf(buf, "WriteBack\n");
+	return n;
 }
 
-static ssize_t show_shared_cpu_map(struct cache_info *this_leaf, char *buf)
-{
-	ssize_t	len;
-	cpumask_t shared_cpu_map;
-
-	cpumask_and(&shared_cpu_map,
-				&this_leaf->shared_cpu_map, cpu_online_mask);
-	len = cpumask_scnprintf(buf, NR_CPUS+1, &shared_cpu_map);
-	len += sprintf(buf+len, "\n");
-	return len;
-}
+static DEVICE_ATTR_RO(attributes);
 
-static ssize_t show_type(struct cache_info *this_leaf, char *buf)
+static umode_t
+cache_private_attrs_is_visible(struct kobject *kobj,
+			       struct attribute *attr, int unused)
 {
-	int type = this_leaf->type + this_leaf->cci.pcci_unified;
-	return sprintf(buf, "%s\n", cache_types[type]);
-}
+	struct device *dev = kobj_to_dev(kobj);
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
 
-static ssize_t show_level(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf, "%u\n", this_leaf->level);
+	if ((attr == &dev_attr_attributes.attr) && this_leaf->attributes)
+		return attr->mode;
+	return 0;
 }
 
-struct cache_attr {
-	struct attribute attr;
-	ssize_t (*show)(struct cache_info *, char *);
-	ssize_t (*store)(struct cache_info *, const char *, size_t count);
-};
-
-#ifdef define_one_ro
-	#undef define_one_ro
-#endif
-#define define_one_ro(_name) \
-	static struct cache_attr _name = \
-__ATTR(_name, 0444, show_##_name, NULL)
-
-define_one_ro(level);
-define_one_ro(type);
-define_one_ro(coherency_line_size);
-define_one_ro(ways_of_associativity);
-define_one_ro(size);
-define_one_ro(number_of_sets);
-define_one_ro(shared_cpu_map);
-define_one_ro(attributes);
-
-static struct attribute * cache_default_attrs[] = {
-	&type.attr,
-	&level.attr,
-	&coherency_line_size.attr,
-	&ways_of_associativity.attr,
-	&attributes.attr,
-	&size.attr,
-	&number_of_sets.attr,
-	&shared_cpu_map.attr,
+static struct attribute *cache_priv_attrs[] = {
+	&dev_attr_attributes.attr,
 	NULL
 };
 
-#define to_object(k) container_of(k, struct cache_info, kobj)
-#define to_attr(a) container_of(a, struct cache_attr, attr)
-
-static ssize_t ia64_cache_show(struct kobject * kobj, struct attribute * attr, char * buf)
-{
-	struct cache_attr *fattr = to_attr(attr);
-	struct cache_info *this_leaf = to_object(kobj);
-	ssize_t ret;
-
-	ret = fattr->show ? fattr->show(this_leaf, buf) : 0;
-	return ret;
-}
-
-static const struct sysfs_ops cache_sysfs_ops = {
-	.show   = ia64_cache_show
-};
-
-static struct kobj_type cache_ktype = {
-	.sysfs_ops	= &cache_sysfs_ops,
-	.default_attrs	= cache_default_attrs,
+static struct attribute_group cache_private_group = {
+	.attrs = cache_priv_attrs,
+	.is_visible = cache_private_attrs_is_visible,
 };
 
-static struct kobj_type cache_ktype_percpu_entry = {
-	.sysfs_ops	= &cache_sysfs_ops,
-};
-
-static void cpu_cache_sysfs_exit(unsigned int cpu)
+const struct attribute_group *
+cache_get_priv_group(struct cacheinfo *this_leaf)
 {
-	kfree(all_cpu_cache_info[cpu].cache_leaves);
-	all_cpu_cache_info[cpu].cache_leaves = NULL;
-	all_cpu_cache_info[cpu].num_cache_leaves = 0;
-	memset(&all_cpu_cache_info[cpu].kobj, 0, sizeof(struct kobject));
-	return;
-}
-
-static int cpu_cache_sysfs_init(unsigned int cpu)
-{
-	unsigned long i, levels, unique_caches;
-	pal_cache_config_info_t cci;
-	int j;
-	long status;
-	struct cache_info *this_cache;
-	int num_cache_leaves = 0;
-
-	if ((status = ia64_pal_cache_summary(&levels, &unique_caches)) != 0) {
-		printk(KERN_ERR "ia64_pal_cache_summary=%ld\n", status);
-		return -1;
-	}
-
-	this_cache=kzalloc(sizeof(struct cache_info)*unique_caches,
-			GFP_KERNEL);
-	if (this_cache == NULL)
-		return -ENOMEM;
-
-	for (i=0; i < levels; i++) {
-		for (j=2; j >0 ; j--) {
-			if ((status=ia64_pal_cache_config_info(i,j, &cci)) !=
-					PAL_STATUS_SUCCESS)
-				continue;
-
-			this_cache[num_cache_leaves].cci = cci;
-			this_cache[num_cache_leaves].level = i + 1;
-			this_cache[num_cache_leaves].type = j;
-
-			cache_shared_cpu_map_setup(cpu,
-					&this_cache[num_cache_leaves]);
-			num_cache_leaves ++;
-		}
-	}
-
-	all_cpu_cache_info[cpu].cache_leaves = this_cache;
-	all_cpu_cache_info[cpu].num_cache_leaves = num_cache_leaves;
-
-	memset(&all_cpu_cache_info[cpu].kobj, 0, sizeof(struct kobject));
-
-	return 0;
+	return &cache_private_group;
 }
 
-/* Add cache interface for CPU device */
-static int cache_add_dev(struct device *sys_dev)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 pal_cache_config_info_t *cci,
+			 enum cache_type type, unsigned int level)
 {
-	unsigned int cpu = sys_dev->id;
-	unsigned long i, j;
-	struct cache_info *this_object;
-	int retval = 0;
-	cpumask_t oldmask;
-
-	if (all_cpu_cache_info[cpu].kobj.parent)
-		return 0;
-
-	oldmask = current->cpus_allowed;
-	retval = set_cpus_allowed_ptr(current, cpumask_of(cpu));
-	if (unlikely(retval))
-		return retval;
-
-	retval = cpu_cache_sysfs_init(cpu);
-	set_cpus_allowed_ptr(current, &oldmask);
-	if (unlikely(retval < 0))
-		return retval;
-
-	retval = kobject_init_and_add(&all_cpu_cache_info[cpu].kobj,
-				      &cache_ktype_percpu_entry, &sys_dev->kobj,
-				      "%s", "cache");
-	if (unlikely(retval < 0)) {
-		cpu_cache_sysfs_exit(cpu);
-		return retval;
-	}
-
-	for (i = 0; i < all_cpu_cache_info[cpu].num_cache_leaves; i++) {
-		this_object = LEAF_KOBJECT_PTR(cpu,i);
-		retval = kobject_init_and_add(&(this_object->kobj),
-					      &cache_ktype,
-					      &all_cpu_cache_info[cpu].kobj,
-					      "index%1lu", i);
-		if (unlikely(retval)) {
-			for (j = 0; j < i; j++) {
-				kobject_put(&(LEAF_KOBJECT_PTR(cpu,j)->kobj));
-			}
-			kobject_put(&all_cpu_cache_info[cpu].kobj);
-			cpu_cache_sysfs_exit(cpu);
-			return retval;
-		}
-		kobject_uevent(&(this_object->kobj), KOBJ_ADD);
-	}
-	kobject_uevent(&all_cpu_cache_info[cpu].kobj, KOBJ_ADD);
-	return retval;
+	unsigned number_of_sets = cci->pcci_cache_size;
+
+	number_of_sets /= cci->pcci_assoc;
+	number_of_sets /= 1 << cci->pcci_line_size;
+	this_leaf->level = level;
+	this_leaf->type = cci->pcci_unified ? CACHE_TYPE_UNIFIED : type;
+	this_leaf->coherency_line_size = cci->pcci_line_size;
+	this_leaf->ways_of_associativity = cci->pcci_assoc;
+	this_leaf->size = cci->pcci_cache_size;
+	this_leaf->attributes = cci->pcci_cache_attr;
+	this_leaf->number_of_sets = number_of_sets;
 }
 
-/* Remove cache interface for CPU device */
-static int cache_remove_dev(struct device *sys_dev)
+int init_cache_level(unsigned int cpu)
 {
-	unsigned int cpu = sys_dev->id;
-	unsigned long i;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned long levels, unique_caches;
+	long status;
 
-	for (i = 0; i < all_cpu_cache_info[cpu].num_cache_leaves; i++)
-		kobject_put(&(LEAF_KOBJECT_PTR(cpu,i)->kobj));
+	if (!this_cpu_ci)
+		return -EINVAL;
 
-	if (all_cpu_cache_info[cpu].kobj.parent) {
-		kobject_put(&all_cpu_cache_info[cpu].kobj);
-		memset(&all_cpu_cache_info[cpu].kobj,
-			0,
-			sizeof(struct kobject));
+	status = ia64_pal_cache_summary(&levels, &unique_caches);
+	if (status != PAL_STATUS_SUCCESS) {
+		pr_err("ia64_pal_cache_summary = %ld\n", status);
+		return -EIO;
 	}
-
-	cpu_cache_sysfs_exit(cpu);
+	this_cpu_ci->num_levels = levels;
+	this_cpu_ci->num_leaves = unique_caches;
 
 	return 0;
 }
 
-/*
- * When a cpu is hot-plugged, do a check and initiate
- * cache kobject if necessary
- */
-static int cache_cpu_callback(struct notifier_block *nfb,
-		unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (unsigned long)hcpu;
-	struct device *sys_dev;
-
-	sys_dev = get_cpu_device(cpu);
-	switch (action) {
-	case CPU_ONLINE:
-	case CPU_ONLINE_FROZEN:
-		cache_add_dev(sys_dev);
-		break;
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		cache_remove_dev(sys_dev);
-		break;
-	}
-	return NOTIFY_OK;
-}
-
-static struct notifier_block cache_cpu_notifier =
-{
-	.notifier_call = cache_cpu_callback
-};
-
-static int __init cache_sysfs_init(void)
+int populate_cache_leaves(unsigned int cpu)
 {
-	int i;
-
-	cpu_notifier_register_begin();
-
-	for_each_online_cpu(i) {
-		struct device *sys_dev = get_cpu_device((unsigned int)i);
-		cache_add_dev(sys_dev);
+	unsigned int level, idx;
+	s64 status;
+	pal_cache_config_info_t cci;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		type = CACHE_TYPE_INST;
+		status = ia64_pal_cache_config_info(level - 1, type, &cci);
+		if (status == PAL_STATUS_SUCCESS)
+			ci_leaf_init(this_leaf++, &cci, type, level);
+		type = CACHE_TYPE_DATA;
+		status = ia64_pal_cache_config_info(level - 1, type, &cci);
+		if (status == PAL_STATUS_SUCCESS)
+			ci_leaf_init(this_leaf++, &cci, type, level);
 	}
-
-	__register_hotcpu_notifier(&cache_cpu_notifier);
-
-	cpu_notifier_register_done();
-
-	return 0;
+	return cache_shared_cpu_map_setup(cpu);
 }
-
-device_initcall(cache_sysfs_init);
-
-- 
1.8.3.2

[PATCH v3 05/11] ia64: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: linux-ia64@vger.kernel.org
---
 arch/ia64/kernel/topology.c | 421 ++++++++++++--------------------------------
 1 file changed, 111 insertions(+), 310 deletions(-)

diff --git a/arch/ia64/kernel/topology.c b/arch/ia64/kernel/topology.c
index f295f9abba4b..3d6749cb53ec 100644
--- a/arch/ia64/kernel/topology.c
+++ b/arch/ia64/kernel/topology.c
@@ -13,6 +13,7 @@
  *	Populate cpu cache entries in sysfs for cpu cache info
  */
 
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
 #include <linux/kernel.h>
 #include <linux/mm.h>
@@ -21,7 +22,6 @@
 #include <linux/init.h>
 #include <linux/bootmem.h>
 #include <linux/nodemask.h>
-#include <linux/notifier.h>
 #include <linux/export.h>
 #include <asm/mmzone.h>
 #include <asm/numa.h>
@@ -103,60 +103,25 @@ subsys_initcall(topology_init);
 /*
  * Export cpu cache information through sysfs
  */
-
-/*
- *  A bunch of string array to get pretty printing
- */
-static const char *cache_types[] = {
-	"",			/* not used */
-	"Instruction",
-	"Data",
-	"Unified"	/* unified */
-};
-
-static const char *cache_mattrib[]={
-	"WriteThrough",
-	"WriteBack",
-	"",		/* reserved */
-	""		/* reserved */
-};
-
-struct cache_info {
-	pal_cache_config_info_t	cci;
-	cpumask_t shared_cpu_map;
-	int level;
-	int type;
-	struct kobject kobj;
-};
-
-struct cpu_cache_info {
-	struct cache_info *cache_leaves;
-	int	num_cache_leaves;
-	struct kobject kobj;
-};
-
-static struct cpu_cache_info	all_cpu_cache_info[NR_CPUS];
-#define LEAF_KOBJECT_PTR(x,y)    (&all_cpu_cache_info[x].cache_leaves[y])
-
 #ifdef CONFIG_SMP
-static void cache_shared_cpu_map_setup(unsigned int cpu,
-		struct cache_info * this_leaf)
+static int __cache_cpumap_setup(unsigned int cpu, struct cacheinfo *this_leaf)
 {
 	pal_cache_shared_info_t	csi;
-	int num_shared, i = 0;
-	unsigned int j;
+	int num_shared, i = 0, j;
+	enum cache_type type = this_leaf->type;
 
 	if (cpu_data(cpu)->threads_per_core <= 1 &&
 		cpu_data(cpu)->cores_per_socket <= 1) {
 		cpu_set(cpu, this_leaf->shared_cpu_map);
-		return;
+		return 0;
 	}
 
-	if (ia64_pal_cache_shared_info(this_leaf->level,
-					this_leaf->type,
-					0,
-					&csi) != PAL_STATUS_SUCCESS)
-		return;
+	if (type = CACHE_TYPE_UNIFIED)
+		type = CACHE_TYPE_DATA;
+
+	if (ia64_pal_cache_shared_info(this_leaf->level, type, 0,
+				       &csi) != PAL_STATUS_SUCCESS)
+		return -EIO;
 
 	num_shared = (int) csi.num_shared;
 	do {
@@ -168,305 +133,141 @@ static void cache_shared_cpu_map_setup(unsigned int cpu,
 
 		i++;
 	} while (i < num_shared &&
-		ia64_pal_cache_shared_info(this_leaf->level,
-				this_leaf->type,
-				i,
-				&csi) = PAL_STATUS_SUCCESS);
-}
-#else
-static void cache_shared_cpu_map_setup(unsigned int cpu,
-		struct cache_info * this_leaf)
-{
-	cpu_set(cpu, this_leaf->shared_cpu_map);
-	return;
-}
-#endif
-
-static ssize_t show_coherency_line_size(struct cache_info *this_leaf,
-					char *buf)
-{
-	return sprintf(buf, "%u\n", 1 << this_leaf->cci.pcci_line_size);
+		 ia64_pal_cache_shared_info(this_leaf->level, type, i,
+					    &csi) = PAL_STATUS_SUCCESS);
+	return 0;
 }
 
-static ssize_t show_ways_of_associativity(struct cache_info *this_leaf,
-					char *buf)
+static int cache_shared_cpu_map_setup(unsigned int cpu)
 {
-	return sprintf(buf, "%u\n", this_leaf->cci.pcci_assoc);
+	unsigned int idx;
+	int ret = 0;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++, this_leaf++) {
+		ret = __cache_cpumap_setup(cpu, this_leaf);
+		if (ret)
+			break;
+	}
+	return ret;
 }
-
-static ssize_t show_attributes(struct cache_info *this_leaf, char *buf)
+#else
+static int cache_shared_cpu_map_setup(unsigned int cpu)
 {
-	return sprintf(buf,
-			"%s\n",
-			cache_mattrib[this_leaf->cci.pcci_cache_attr]);
-}
+	int idx;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
 
-static ssize_t show_size(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf, "%uK\n", this_leaf->cci.pcci_cache_size / 1024);
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++, this_leaf++)
+		cpu_set(cpu, this_leaf->shared_cpu_map);
+	return 0;
 }
+#endif
 
-static ssize_t show_number_of_sets(struct cache_info *this_leaf, char *buf)
+static ssize_t attributes_show(struct device *dev,
+			       struct device_attribute *attr, char *buf)
 {
-	unsigned number_of_sets = this_leaf->cci.pcci_cache_size;
-	number_of_sets /= this_leaf->cci.pcci_assoc;
-	number_of_sets /= 1 << this_leaf->cci.pcci_line_size;
-
-	return sprintf(buf, "%u\n", number_of_sets);
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	unsigned int ci_attr = this_leaf->attributes;
+	int n = 0;
+
+	if (ci_attr & CACHE_WRITE_THROUGH)
+		n = sprintf(buf, "WriteThrough\n");
+	else if (ci_attr & CACHE_WRITE_BACK)
+		n = sprintf(buf, "WriteBack\n");
+	return n;
 }
 
-static ssize_t show_shared_cpu_map(struct cache_info *this_leaf, char *buf)
-{
-	ssize_t	len;
-	cpumask_t shared_cpu_map;
-
-	cpumask_and(&shared_cpu_map,
-				&this_leaf->shared_cpu_map, cpu_online_mask);
-	len = cpumask_scnprintf(buf, NR_CPUS+1, &shared_cpu_map);
-	len += sprintf(buf+len, "\n");
-	return len;
-}
+static DEVICE_ATTR_RO(attributes);
 
-static ssize_t show_type(struct cache_info *this_leaf, char *buf)
+static umode_t
+cache_private_attrs_is_visible(struct kobject *kobj,
+			       struct attribute *attr, int unused)
 {
-	int type = this_leaf->type + this_leaf->cci.pcci_unified;
-	return sprintf(buf, "%s\n", cache_types[type]);
-}
+	struct device *dev = kobj_to_dev(kobj);
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
 
-static ssize_t show_level(struct cache_info *this_leaf, char *buf)
-{
-	return sprintf(buf, "%u\n", this_leaf->level);
+	if ((attr = &dev_attr_attributes.attr) && this_leaf->attributes)
+		return attr->mode;
+	return 0;
 }
 
-struct cache_attr {
-	struct attribute attr;
-	ssize_t (*show)(struct cache_info *, char *);
-	ssize_t (*store)(struct cache_info *, const char *, size_t count);
-};
-
-#ifdef define_one_ro
-	#undef define_one_ro
-#endif
-#define define_one_ro(_name) \
-	static struct cache_attr _name = \
-__ATTR(_name, 0444, show_##_name, NULL)
-
-define_one_ro(level);
-define_one_ro(type);
-define_one_ro(coherency_line_size);
-define_one_ro(ways_of_associativity);
-define_one_ro(size);
-define_one_ro(number_of_sets);
-define_one_ro(shared_cpu_map);
-define_one_ro(attributes);
-
-static struct attribute * cache_default_attrs[] = {
-	&type.attr,
-	&level.attr,
-	&coherency_line_size.attr,
-	&ways_of_associativity.attr,
-	&attributes.attr,
-	&size.attr,
-	&number_of_sets.attr,
-	&shared_cpu_map.attr,
+static struct attribute *cache_priv_attrs[] = {
+	&dev_attr_attributes.attr,
 	NULL
 };
 
-#define to_object(k) container_of(k, struct cache_info, kobj)
-#define to_attr(a) container_of(a, struct cache_attr, attr)
-
-static ssize_t ia64_cache_show(struct kobject * kobj, struct attribute * attr, char * buf)
-{
-	struct cache_attr *fattr = to_attr(attr);
-	struct cache_info *this_leaf = to_object(kobj);
-	ssize_t ret;
-
-	ret = fattr->show ? fattr->show(this_leaf, buf) : 0;
-	return ret;
-}
-
-static const struct sysfs_ops cache_sysfs_ops = {
-	.show   = ia64_cache_show
-};
-
-static struct kobj_type cache_ktype = {
-	.sysfs_ops	= &cache_sysfs_ops,
-	.default_attrs	= cache_default_attrs,
+static struct attribute_group cache_private_group = {
+	.attrs = cache_priv_attrs,
+	.is_visible = cache_private_attrs_is_visible,
 };
 
-static struct kobj_type cache_ktype_percpu_entry = {
-	.sysfs_ops	= &cache_sysfs_ops,
-};
-
-static void cpu_cache_sysfs_exit(unsigned int cpu)
+const struct attribute_group *
+cache_get_priv_group(struct cacheinfo *this_leaf)
 {
-	kfree(all_cpu_cache_info[cpu].cache_leaves);
-	all_cpu_cache_info[cpu].cache_leaves = NULL;
-	all_cpu_cache_info[cpu].num_cache_leaves = 0;
-	memset(&all_cpu_cache_info[cpu].kobj, 0, sizeof(struct kobject));
-	return;
-}
-
-static int cpu_cache_sysfs_init(unsigned int cpu)
-{
-	unsigned long i, levels, unique_caches;
-	pal_cache_config_info_t cci;
-	int j;
-	long status;
-	struct cache_info *this_cache;
-	int num_cache_leaves = 0;
-
-	if ((status = ia64_pal_cache_summary(&levels, &unique_caches)) != 0) {
-		printk(KERN_ERR "ia64_pal_cache_summary=%ld\n", status);
-		return -1;
-	}
-
-	this_cache=kzalloc(sizeof(struct cache_info)*unique_caches,
-			GFP_KERNEL);
-	if (this_cache = NULL)
-		return -ENOMEM;
-
-	for (i=0; i < levels; i++) {
-		for (j=2; j >0 ; j--) {
-			if ((status=ia64_pal_cache_config_info(i,j, &cci)) !-					PAL_STATUS_SUCCESS)
-				continue;
-
-			this_cache[num_cache_leaves].cci = cci;
-			this_cache[num_cache_leaves].level = i + 1;
-			this_cache[num_cache_leaves].type = j;
-
-			cache_shared_cpu_map_setup(cpu,
-					&this_cache[num_cache_leaves]);
-			num_cache_leaves ++;
-		}
-	}
-
-	all_cpu_cache_info[cpu].cache_leaves = this_cache;
-	all_cpu_cache_info[cpu].num_cache_leaves = num_cache_leaves;
-
-	memset(&all_cpu_cache_info[cpu].kobj, 0, sizeof(struct kobject));
-
-	return 0;
+	return &cache_private_group;
 }
 
-/* Add cache interface for CPU device */
-static int cache_add_dev(struct device *sys_dev)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 pal_cache_config_info_t *cci,
+			 enum cache_type type, unsigned int level)
 {
-	unsigned int cpu = sys_dev->id;
-	unsigned long i, j;
-	struct cache_info *this_object;
-	int retval = 0;
-	cpumask_t oldmask;
-
-	if (all_cpu_cache_info[cpu].kobj.parent)
-		return 0;
-
-	oldmask = current->cpus_allowed;
-	retval = set_cpus_allowed_ptr(current, cpumask_of(cpu));
-	if (unlikely(retval))
-		return retval;
-
-	retval = cpu_cache_sysfs_init(cpu);
-	set_cpus_allowed_ptr(current, &oldmask);
-	if (unlikely(retval < 0))
-		return retval;
-
-	retval = kobject_init_and_add(&all_cpu_cache_info[cpu].kobj,
-				      &cache_ktype_percpu_entry, &sys_dev->kobj,
-				      "%s", "cache");
-	if (unlikely(retval < 0)) {
-		cpu_cache_sysfs_exit(cpu);
-		return retval;
-	}
-
-	for (i = 0; i < all_cpu_cache_info[cpu].num_cache_leaves; i++) {
-		this_object = LEAF_KOBJECT_PTR(cpu,i);
-		retval = kobject_init_and_add(&(this_object->kobj),
-					      &cache_ktype,
-					      &all_cpu_cache_info[cpu].kobj,
-					      "index%1lu", i);
-		if (unlikely(retval)) {
-			for (j = 0; j < i; j++) {
-				kobject_put(&(LEAF_KOBJECT_PTR(cpu,j)->kobj));
-			}
-			kobject_put(&all_cpu_cache_info[cpu].kobj);
-			cpu_cache_sysfs_exit(cpu);
-			return retval;
-		}
-		kobject_uevent(&(this_object->kobj), KOBJ_ADD);
-	}
-	kobject_uevent(&all_cpu_cache_info[cpu].kobj, KOBJ_ADD);
-	return retval;
+	unsigned number_of_sets = cci->pcci_cache_size;
+
+	number_of_sets /= cci->pcci_assoc;
+	number_of_sets /= 1 << cci->pcci_line_size;
+	this_leaf->level = level;
+	this_leaf->type = cci->pcci_unified ? CACHE_TYPE_UNIFIED : type;
+	this_leaf->coherency_line_size = cci->pcci_line_size;
+	this_leaf->ways_of_associativity = cci->pcci_assoc;
+	this_leaf->size = cci->pcci_cache_size;
+	this_leaf->attributes = cci->pcci_cache_attr;
+	this_leaf->number_of_sets = number_of_sets;
 }
 
-/* Remove cache interface for CPU device */
-static int cache_remove_dev(struct device *sys_dev)
+int init_cache_level(unsigned int cpu)
 {
-	unsigned int cpu = sys_dev->id;
-	unsigned long i;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned long levels, unique_caches;
+	long status;
 
-	for (i = 0; i < all_cpu_cache_info[cpu].num_cache_leaves; i++)
-		kobject_put(&(LEAF_KOBJECT_PTR(cpu,i)->kobj));
+	if (!this_cpu_ci)
+		return -EINVAL;
 
-	if (all_cpu_cache_info[cpu].kobj.parent) {
-		kobject_put(&all_cpu_cache_info[cpu].kobj);
-		memset(&all_cpu_cache_info[cpu].kobj,
-			0,
-			sizeof(struct kobject));
+	status = ia64_pal_cache_summary(&levels, &unique_caches);
+	if (status != PAL_STATUS_SUCCESS) {
+		pr_err("ia64_pal_cache_summary = %ld\n", status);
+		return -EIO;
 	}
-
-	cpu_cache_sysfs_exit(cpu);
+	this_cpu_ci->num_levels = levels;
+	this_cpu_ci->num_leaves = unique_caches;
 
 	return 0;
 }
 
-/*
- * When a cpu is hot-plugged, do a check and initiate
- * cache kobject if necessary
- */
-static int cache_cpu_callback(struct notifier_block *nfb,
-		unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (unsigned long)hcpu;
-	struct device *sys_dev;
-
-	sys_dev = get_cpu_device(cpu);
-	switch (action) {
-	case CPU_ONLINE:
-	case CPU_ONLINE_FROZEN:
-		cache_add_dev(sys_dev);
-		break;
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		cache_remove_dev(sys_dev);
-		break;
-	}
-	return NOTIFY_OK;
-}
-
-static struct notifier_block cache_cpu_notifier -{
-	.notifier_call = cache_cpu_callback
-};
-
-static int __init cache_sysfs_init(void)
+int populate_cache_leaves(unsigned int cpu)
 {
-	int i;
-
-	cpu_notifier_register_begin();
-
-	for_each_online_cpu(i) {
-		struct device *sys_dev = get_cpu_device((unsigned int)i);
-		cache_add_dev(sys_dev);
+	unsigned int level, idx;
+	s64 status;
+	pal_cache_config_info_t cci;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		type = CACHE_TYPE_INST;
+		status = ia64_pal_cache_config_info(level - 1, type, &cci);
+		if (status = PAL_STATUS_SUCCESS)
+			ci_leaf_init(this_leaf++, &cci, type, level);
+		type = CACHE_TYPE_DATA;
+		status = ia64_pal_cache_config_info(level - 1, type, &cci);
+		if (status = PAL_STATUS_SUCCESS)
+			ci_leaf_init(this_leaf++, &cci, type, level);
 	}
-
-	__register_hotcpu_notifier(&cache_cpu_notifier);
-
-	cpu_notifier_register_done();
-
-	return 0;
+	return cache_shared_cpu_map_setup(cpu);
 }
-
-device_initcall(cache_sysfs_init);
-
-- 
1.8.3.2

[PATCH v3 06/11] s390: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: linux390@de.ibm.com
Cc: linux-s390@vger.kernel.org
---
 arch/s390/kernel/cache.c | 388 +++++++++++------------------------------------
 1 file changed, 92 insertions(+), 296 deletions(-)

diff --git a/arch/s390/kernel/cache.c b/arch/s390/kernel/cache.c
index c0b03c28d157..fe21f074cf9f 100644
--- a/arch/s390/kernel/cache.c
+++ b/arch/s390/kernel/cache.c
@@ -5,37 +5,11 @@
  *    Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>
  */
 
-#include <linux/notifier.h>
 #include <linux/seq_file.h>
-#include <linux/init.h>
-#include <linux/list.h>
-#include <linux/slab.h>
 #include <linux/cpu.h>
+#include <linux/cacheinfo.h>
 #include <asm/facility.h>
 
-struct cache {
-	unsigned long size;
-	unsigned int line_size;
-	unsigned int associativity;
-	unsigned int nr_sets;
-	unsigned int level   : 3;
-	unsigned int type    : 2;
-	unsigned int private : 1;
-	struct list_head list;
-};
-
-struct cache_dir {
-	struct kobject *kobj;
-	struct cache_index_dir *index;
-};
-
-struct cache_index_dir {
-	struct kobject kobj;
-	int cpu;
-	struct cache *cache;
-	struct cache_index_dir *next;
-};
-
 enum {
 	CACHE_SCOPE_NOTEXISTS,
 	CACHE_SCOPE_PRIVATE,
@@ -44,10 +18,10 @@ enum {
 };
 
 enum {
-	CACHE_TYPE_SEPARATE,
-	CACHE_TYPE_DATA,
-	CACHE_TYPE_INSTRUCTION,
-	CACHE_TYPE_UNIFIED,
+	CTYPE_SEPARATE,
+	CTYPE_DATA,
+	CTYPE_INSTRUCTION,
+	CTYPE_UNIFIED,
 };
 
 enum {
@@ -70,39 +44,59 @@ struct cache_info {
 };
 
 #define CACHE_MAX_LEVEL 8
-
 union cache_topology {
 	struct cache_info ci[CACHE_MAX_LEVEL];
 	unsigned long long raw;
 };
 
 static const char * const cache_type_string[] = {
-	"Data",
+	"",
 	"Instruction",
+	"Data",
+	"",
 	"Unified",
 };
 
-static struct cache_dir *cache_dir_cpu[NR_CPUS];
-static LIST_HEAD(cache_list);
+static const enum cache_type cache_type_map[] = {
+	[CTYPE_SEPARATE] = CACHE_TYPE_SEPARATE,
+	[CTYPE_DATA] = CACHE_TYPE_DATA,
+	[CTYPE_INSTRUCTION] = CACHE_TYPE_INST,
+	[CTYPE_UNIFIED] = CACHE_TYPE_UNIFIED,
+};
 
 void show_cacheinfo(struct seq_file *m)
 {
-	struct cache *cache;
-	int index = 0;
+	int cpu = smp_processor_id(), idx;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *cache;
 
-	list_for_each_entry(cache, &cache_list, list) {
-		seq_printf(m, "cache%-11d: ", index);
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++) {
+		cache = this_cpu_ci->info_list + idx;
+		seq_printf(m, "cache%-11d: ", idx);
 		seq_printf(m, "level=%d ", cache->level);
 		seq_printf(m, "type=%s ", cache_type_string[cache->type]);
-		seq_printf(m, "scope=%s ", cache->private ? "Private" : "Shared");
-		seq_printf(m, "size=%luK ", cache->size >> 10);
-		seq_printf(m, "line_size=%u ", cache->line_size);
-		seq_printf(m, "associativity=%d", cache->associativity);
+		seq_printf(m, "scope=%s ",
+			   cache->disable_sysfs ? "Shared" : "Private");
+		seq_printf(m, "size=%dK ", cache->size >> 10);
+		seq_printf(m, "line_size=%u ", cache->coherency_line_size);
+		seq_printf(m, "associativity=%d", cache->ways_of_associativity);
 		seq_puts(m, "\n");
-		index++;
 	}
 }
 
+static inline enum cache_type get_cache_type(struct cache_info *ci, int level)
+{
+	if (level >= CACHE_MAX_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+
+	ci += level;
+
+	if (ci->scope != CACHE_SCOPE_SHARED && ci->scope != CACHE_SCOPE_PRIVATE)
+		return CACHE_TYPE_NOCACHE;
+
+	return cache_type_map[ci->type];
+}
+
 static inline unsigned long ecag(int ai, int li, int ti)
 {
 	unsigned long cmd, val;
@@ -113,277 +107,79 @@ static inline unsigned long ecag(int ai, int li, int ti)
 	return val;
 }
 
-static int __init cache_add(int level, int private, int type)
+static void ci_leaf_init(struct cacheinfo *this_leaf, int private,
+			 enum cache_type type, unsigned int level)
 {
-	struct cache *cache;
-	int ti;
+	int ti, num_sets;
+	int cpu = smp_processor_id();
 
-	cache = kzalloc(sizeof(*cache), GFP_KERNEL);
-	if (!cache)
-		return -ENOMEM;
-	if (type == CACHE_TYPE_INSTRUCTION)
+	if (type == CACHE_TYPE_INST)
 		ti = CACHE_TI_INSTRUCTION;
 	else
 		ti = CACHE_TI_UNIFIED;
-	cache->size = ecag(EXTRACT_SIZE, level, ti);
-	cache->line_size = ecag(EXTRACT_LINE_SIZE, level, ti);
-	cache->associativity = ecag(EXTRACT_ASSOCIATIVITY, level, ti);
-	cache->nr_sets = cache->size / cache->associativity;
-	cache->nr_sets /= cache->line_size;
-	cache->private = private;
-	cache->level = level + 1;
-	cache->type = type - 1;
-	list_add_tail(&cache->list, &cache_list);
-	return 0;
-}
-
-static void __init cache_build_info(void)
-{
-	struct cache *cache, *next;
-	union cache_topology ct;
-	int level, private, rc;
-
-	ct.raw = ecag(EXTRACT_TOPOLOGY, 0, 0);
-	for (level = 0; level < CACHE_MAX_LEVEL; level++) {
-		switch (ct.ci[level].scope) {
-		case CACHE_SCOPE_SHARED:
-			private = 0;
-			break;
-		case CACHE_SCOPE_PRIVATE:
-			private = 1;
-			break;
-		default:
-			return;
-		}
-		if (ct.ci[level].type == CACHE_TYPE_SEPARATE) {
-			rc  = cache_add(level, private, CACHE_TYPE_DATA);
-			rc |= cache_add(level, private, CACHE_TYPE_INSTRUCTION);
-		} else {
-			rc = cache_add(level, private, ct.ci[level].type);
-		}
-		if (rc)
-			goto error;
-	}
-	return;
-error:
-	list_for_each_entry_safe(cache, next, &cache_list, list) {
-		list_del(&cache->list);
-		kfree(cache);
-	}
-}
-
-static struct cache_dir *cache_create_cache_dir(int cpu)
-{
-	struct cache_dir *cache_dir;
-	struct kobject *kobj = NULL;
-	struct device *dev;
-
-	dev = get_cpu_device(cpu);
-	if (!dev)
-		goto out;
-	kobj = kobject_create_and_add("cache", &dev->kobj);
-	if (!kobj)
-		goto out;
-	cache_dir = kzalloc(sizeof(*cache_dir), GFP_KERNEL);
-	if (!cache_dir)
-		goto out;
-	cache_dir->kobj = kobj;
-	cache_dir_cpu[cpu] = cache_dir;
-	return cache_dir;
-out:
-	kobject_put(kobj);
-	return NULL;
-}
-
-static struct cache_index_dir *kobj_to_cache_index_dir(struct kobject *kobj)
-{
-	return container_of(kobj, struct cache_index_dir, kobj);
-}
-
-static void cache_index_release(struct kobject *kobj)
-{
-	struct cache_index_dir *index;
-
-	index = kobj_to_cache_index_dir(kobj);
-	kfree(index);
-}
-
-static ssize_t cache_index_show(struct kobject *kobj,
-				struct attribute *attr, char *buf)
-{
-	struct kobj_attribute *kobj_attr;
-
-	kobj_attr = container_of(attr, struct kobj_attribute, attr);
-	return kobj_attr->show(kobj, kobj_attr, buf);
-}
-
-#define DEFINE_CACHE_ATTR(_name, _format, _value)			\
-static ssize_t cache_##_name##_show(struct kobject *kobj,		\
-				    struct kobj_attribute *attr,	\
-				    char *buf)				\
-{									\
-	struct cache_index_dir *index;					\
-									\
-	index = kobj_to_cache_index_dir(kobj);				\
-	return sprintf(buf, _format, _value);				\
-}									\
-static struct kobj_attribute cache_##_name##_attr =			\
-	__ATTR(_name, 0444, cache_##_name##_show, NULL);
 
-DEFINE_CACHE_ATTR(size, "%luK\n", index->cache->size >> 10);
-DEFINE_CACHE_ATTR(coherency_line_size, "%u\n", index->cache->line_size);
-DEFINE_CACHE_ATTR(number_of_sets, "%u\n", index->cache->nr_sets);
-DEFINE_CACHE_ATTR(ways_of_associativity, "%u\n", index->cache->associativity);
-DEFINE_CACHE_ATTR(type, "%s\n", cache_type_string[index->cache->type]);
-DEFINE_CACHE_ATTR(level, "%d\n", index->cache->level);
+	this_leaf->level = level + 1;
+	this_leaf->type = type;
+	this_leaf->coherency_line_size = ecag(EXTRACT_LINE_SIZE, level, ti);
+	this_leaf->ways_of_associativity = ecag(EXTRACT_ASSOCIATIVITY,
+						level, ti);
+	this_leaf->size = ecag(EXTRACT_SIZE, level, ti);
 
-static ssize_t shared_cpu_map_func(struct kobject *kobj, int type, char *buf)
-{
-	struct cache_index_dir *index;
-	int len;
-
-	index = kobj_to_cache_index_dir(kobj);
-	len = type ?
-		cpulist_scnprintf(buf, PAGE_SIZE - 2, cpumask_of(index->cpu)) :
-		cpumask_scnprintf(buf, PAGE_SIZE - 2, cpumask_of(index->cpu));
-	len += sprintf(&buf[len], "\n");
-	return len;
-}
-
-static ssize_t shared_cpu_map_show(struct kobject *kobj,
-				   struct kobj_attribute *attr, char *buf)
-{
-	return shared_cpu_map_func(kobj, 0, buf);
+	num_sets = this_leaf->size / this_leaf->coherency_line_size;
+	num_sets /= this_leaf->ways_of_associativity;
+	this_leaf->number_of_sets = num_sets;
+	cpumask_set_cpu(cpu, &this_leaf->shared_cpu_map);
+	if (!private)
+		this_leaf->disable_sysfs = true;
 }
-static struct kobj_attribute cache_shared_cpu_map_attr =
-	__ATTR(shared_cpu_map, 0444, shared_cpu_map_show, NULL);
 
-static ssize_t shared_cpu_list_show(struct kobject *kobj,
-				    struct kobj_attribute *attr, char *buf)
+int init_cache_level(unsigned int cpu)
 {
-	return shared_cpu_map_func(kobj, 1, buf);
-}
-static struct kobj_attribute cache_shared_cpu_list_attr =
-	__ATTR(shared_cpu_list, 0444, shared_cpu_list_show, NULL);
-
-static struct attribute *cache_index_default_attrs[] = {
-	&cache_type_attr.attr,
-	&cache_size_attr.attr,
-	&cache_number_of_sets_attr.attr,
-	&cache_ways_of_associativity_attr.attr,
-	&cache_level_attr.attr,
-	&cache_coherency_line_size_attr.attr,
-	&cache_shared_cpu_map_attr.attr,
-	&cache_shared_cpu_list_attr.attr,
-	NULL,
-};
-
-static const struct sysfs_ops cache_index_ops = {
-	.show = cache_index_show,
-};
-
-static struct kobj_type cache_index_type = {
-	.sysfs_ops = &cache_index_ops,
-	.release = cache_index_release,
-	.default_attrs = cache_index_default_attrs,
-};
-
-static int cache_create_index_dir(struct cache_dir *cache_dir,
-				  struct cache *cache, int index, int cpu)
-{
-	struct cache_index_dir *index_dir;
-	int rc;
-
-	index_dir = kzalloc(sizeof(*index_dir), GFP_KERNEL);
-	if (!index_dir)
-		return -ENOMEM;
-	index_dir->cache = cache;
-	index_dir->cpu = cpu;
-	rc = kobject_init_and_add(&index_dir->kobj, &cache_index_type,
-				  cache_dir->kobj, "index%d", index);
-	if (rc)
-		goto out;
-	index_dir->next = cache_dir->index;
-	cache_dir->index = index_dir;
-	return 0;
-out:
-	kfree(index_dir);
-	return rc;
-}
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int level = 0, leaves = 0;
+	union cache_topology ct;
+	enum cache_type ctype;
 
-static int cache_add_cpu(int cpu)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-	int rc, index = 0;
+	if (!this_cpu_ci)
+		return -EINVAL;
 
-	if (list_empty(&cache_list))
-		return 0;
-	cache_dir = cache_create_cache_dir(cpu);
-	if (!cache_dir)
-		return -ENOMEM;
-	list_for_each_entry(cache, &cache_list, list) {
-		if (!cache->private)
+	ct.raw = ecag(EXTRACT_TOPOLOGY, 0, 0);
+	do {
+		ctype = get_cache_type(&ct.ci[0], level);
+		if (ctype == CACHE_TYPE_NOCACHE)
 			break;
-		rc = cache_create_index_dir(cache_dir, cache, index, cpu);
-		if (rc)
-			return rc;
-		index++;
-	}
-	return 0;
-}
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	} while (++level < CACHE_MAX_LEVEL);
 
-static void cache_remove_cpu(int cpu)
-{
-	struct cache_index_dir *index, *next;
-	struct cache_dir *cache_dir;
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
 
-	cache_dir = cache_dir_cpu[cpu];
-	if (!cache_dir)
-		return;
-	index = cache_dir->index;
-	while (index) {
-		next = index->next;
-		kobject_put(&index->kobj);
-		index = next;
-	}
-	kobject_put(cache_dir->kobj);
-	kfree(cache_dir);
-	cache_dir_cpu[cpu] = NULL;
+	return 0;
 }
 
-static int cache_hotplug(struct notifier_block *nfb, unsigned long action,
-			 void *hcpu)
+int populate_cache_leaves(unsigned int cpu)
 {
-	int cpu = (long)hcpu;
-	int rc = 0;
+	unsigned int level, idx, pvt;
+	union cache_topology ct;
+	enum cache_type ctype;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_ONLINE:
-		rc = cache_add_cpu(cpu);
-		if (rc)
-			cache_remove_cpu(cpu);
-		break;
-	case CPU_DEAD:
-		cache_remove_cpu(cpu);
-		break;
+	ct.raw = ecag(EXTRACT_TOPOLOGY, 0, 0);
+	for (idx = 0, level = 0; level < this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		pvt = (ct.ci[level].scope == CACHE_SCOPE_PRIVATE) ? 1 : 0;
+		ctype = get_cache_type(&ct.ci[0], level);
+		if (ctype == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, pvt, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, pvt, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, pvt, ctype, level);
+		}
 	}
-	return rc ? NOTIFY_BAD : NOTIFY_OK;
-}
-
-static int __init cache_init(void)
-{
-	int cpu;
-
-	if (!test_facility(34))
-		return 0;
-	cache_build_info();
-
-	cpu_notifier_register_begin();
-	for_each_online_cpu(cpu)
-		cache_add_cpu(cpu);
-	__hotcpu_notifier(cache_hotplug, 0);
-	cpu_notifier_register_done();
 	return 0;
 }
-device_initcall(cache_init);
-- 
1.8.3.2

[PATCH v3 07/11] x86: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

The private pointer provided by the cacheinfo is used to implement
the AMD L3 cache specific attributes.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: x86@kernel.org
---
 arch/x86/kernel/cpu/intel_cacheinfo.c | 709 +++++++++-------------------------
 1 file changed, 189 insertions(+), 520 deletions(-)

diff --git a/arch/x86/kernel/cpu/intel_cacheinfo.c b/arch/x86/kernel/cpu/intel_cacheinfo.c
index c7035073dfc1..5746aea7e9b8 100644
--- a/arch/x86/kernel/cpu/intel_cacheinfo.c
+++ b/arch/x86/kernel/cpu/intel_cacheinfo.c
@@ -7,16 +7,14 @@
  *	Andi Kleen / Andreas Herrmann	: CPUID4 emulation on AMD.
  */
 
-#include <linux/init.h>
 #include <linux/slab.h>
-#include <linux/device.h>
-#include <linux/compiler.h>
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
 #include <linux/sched.h>
+#include <linux/sysfs.h>
 #include <linux/pci.h>
 
 #include <asm/processor.h>
-#include <linux/smp.h>
 #include <asm/amd_nb.h>
 #include <asm/smp.h>
 
@@ -116,10 +114,10 @@ static const struct _cache_table cache_table[] =
 
 
 enum _cache_type {
-	CACHE_TYPE_NULL	= 0,
-	CACHE_TYPE_DATA = 1,
-	CACHE_TYPE_INST = 2,
-	CACHE_TYPE_UNIFIED = 3
+	CTYPE_NULL = 0,
+	CTYPE_DATA = 1,
+	CTYPE_INST = 2,
+	CTYPE_UNIFIED = 3
 };
 
 union _cpuid4_leaf_eax {
@@ -159,11 +157,6 @@ struct _cpuid4_info_regs {
 	struct amd_northbridge *nb;
 };
 
-struct _cpuid4_info {
-	struct _cpuid4_info_regs base;
-	DECLARE_BITMAP(shared_cpu_map, NR_CPUS);
-};
-
 unsigned short			num_cache_leaves;
 
 /* AMD doesn't have CPUID4. Emulate it here to report the same
@@ -220,6 +213,13 @@ static const unsigned short assocs[] = {
 static const unsigned char levels[] = { 1, 1, 2, 3 };
 static const unsigned char types[] = { 1, 2, 3, 3 };
 
+static const enum cache_type cache_type_map[] = {
+	[CTYPE_NULL] = CACHE_TYPE_NOCACHE,
+	[CTYPE_DATA] = CACHE_TYPE_DATA,
+	[CTYPE_INST] = CACHE_TYPE_INST,
+	[CTYPE_UNIFIED] = CACHE_TYPE_UNIFIED,
+};
+
 static void
 amd_cpuid4(int leaf, union _cpuid4_leaf_eax *eax,
 		     union _cpuid4_leaf_ebx *ebx,
@@ -291,14 +291,8 @@ amd_cpuid4(int leaf, union _cpuid4_leaf_eax *eax,
 		(ebx->split.ways_of_associativity + 1) - 1;
 }
 
-struct _cache_attr {
-	struct attribute attr;
-	ssize_t (*show)(struct _cpuid4_info *, char *, unsigned int);
-	ssize_t (*store)(struct _cpuid4_info *, const char *, size_t count,
-			 unsigned int);
-};
-
 #if defined(CONFIG_AMD_NB) && defined(CONFIG_SYSFS)
+
 /*
  * L3 cache descriptors
  */
@@ -325,20 +319,6 @@ static void amd_calc_l3_indices(struct amd_northbridge *nb)
 	l3->indices = (max(max3(sc0, sc1, sc2), sc3) << 10) - 1;
 }
 
-static void amd_init_l3_cache(struct _cpuid4_info_regs *this_leaf, int index)
-{
-	int node;
-
-	/* only for L3, and not in virtualized environments */
-	if (index < 3)
-		return;
-
-	node = amd_get_nb_id(smp_processor_id());
-	this_leaf->nb = node_to_amd_nb(node);
-	if (this_leaf->nb && !this_leaf->nb->l3_cache.indices)
-		amd_calc_l3_indices(this_leaf->nb);
-}
-
 /*
  * check whether a slot used for disabling an L3 index is occupied.
  * @l3: L3 cache descriptor
@@ -359,15 +339,13 @@ int amd_get_l3_disable_slot(struct amd_northbridge *nb, unsigned slot)
 	return -1;
 }
 
-static ssize_t show_cache_disable(struct _cpuid4_info *this_leaf, char *buf,
+static ssize_t show_cache_disable(struct cacheinfo *this_leaf, char *buf,
 				  unsigned int slot)
 {
 	int index;
+	struct amd_northbridge *nb = this_leaf->priv;
 
-	if (!this_leaf->base.nb || !amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
-		return -EINVAL;
-
-	index = amd_get_l3_disable_slot(this_leaf->base.nb, slot);
+	index = amd_get_l3_disable_slot(nb, slot);
 	if (index >= 0)
 		return sprintf(buf, "%d\n", index);
 
@@ -376,9 +354,10 @@ static ssize_t show_cache_disable(struct _cpuid4_info *this_leaf, char *buf,
 
 #define SHOW_CACHE_DISABLE(slot)					\
 static ssize_t								\
-show_cache_disable_##slot(struct _cpuid4_info *this_leaf, char *buf,	\
-			  unsigned int cpu)				\
+cache_disable_##slot##_show(struct device *dev,				\
+			    struct device_attribute *attr, char *buf)	\
 {									\
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);		\
 	return show_cache_disable(this_leaf, buf, slot);		\
 }
 SHOW_CACHE_DISABLE(0)
@@ -446,25 +425,23 @@ int amd_set_l3_disable_slot(struct amd_northbridge *nb, int cpu, unsigned slot,
 	return 0;
 }
 
-static ssize_t store_cache_disable(struct _cpuid4_info *this_leaf,
-				  const char *buf, size_t count,
-				  unsigned int slot)
+static ssize_t store_cache_disable(struct cacheinfo *this_leaf,
+				   const char *buf, size_t count,
+				   unsigned int slot)
 {
 	unsigned long val = 0;
 	int cpu, err = 0;
+	struct amd_northbridge *nb = this_leaf->priv;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
-	if (!this_leaf->base.nb || !amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
-		return -EINVAL;
-
-	cpu = cpumask_first(to_cpumask(this_leaf->shared_cpu_map));
+	cpu = cpumask_first(&this_leaf->shared_cpu_map);
 
 	if (kstrtoul(buf, 10, &val) < 0)
 		return -EINVAL;
 
-	err = amd_set_l3_disable_slot(this_leaf->base.nb, cpu, slot, val);
+	err = amd_set_l3_disable_slot(nb, cpu, slot, val);
 	if (err) {
 		if (err == -EEXIST)
 			pr_warning("L3 slot %d in use/index already disabled!\n",
@@ -476,41 +453,36 @@ static ssize_t store_cache_disable(struct _cpuid4_info *this_leaf,
 
 #define STORE_CACHE_DISABLE(slot)					\
 static ssize_t								\
-store_cache_disable_##slot(struct _cpuid4_info *this_leaf,		\
-			   const char *buf, size_t count,		\
-			   unsigned int cpu)				\
+cache_disable_##slot##_store(struct device *dev,			\
+			     struct device_attribute *attr,		\
+			     const char *buf, size_t count)		\
 {									\
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);		\
 	return store_cache_disable(this_leaf, buf, count, slot);	\
 }
 STORE_CACHE_DISABLE(0)
 STORE_CACHE_DISABLE(1)
 
-static struct _cache_attr cache_disable_0 = __ATTR(cache_disable_0, 0644,
-		show_cache_disable_0, store_cache_disable_0);
-static struct _cache_attr cache_disable_1 = __ATTR(cache_disable_1, 0644,
-		show_cache_disable_1, store_cache_disable_1);
-
-static ssize_t
-show_subcaches(struct _cpuid4_info *this_leaf, char *buf, unsigned int cpu)
+static ssize_t subcaches_show(struct device *dev,
+			      struct device_attribute *attr, char *buf)
 {
-	if (!this_leaf->base.nb || !amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
-		return -EINVAL;
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	int cpu = cpumask_first(&this_leaf->shared_cpu_map);
 
 	return sprintf(buf, "%x\n", amd_get_subcaches(cpu));
 }
 
-static ssize_t
-store_subcaches(struct _cpuid4_info *this_leaf, const char *buf, size_t count,
-		unsigned int cpu)
+static ssize_t subcaches_store(struct device *dev,
+			       struct device_attribute *attr,
+			       const char *buf, size_t count)
 {
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	int cpu = cpumask_first(&this_leaf->shared_cpu_map);
 	unsigned long val;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
-	if (!this_leaf->base.nb || !amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
-		return -EINVAL;
-
 	if (kstrtoul(buf, 16, &val) < 0)
 		return -EINVAL;
 
@@ -520,9 +492,88 @@ store_subcaches(struct _cpuid4_info *this_leaf, const char *buf, size_t count,
 	return count;
 }
 
-static struct _cache_attr subcaches =
-	__ATTR(subcaches, 0644, show_subcaches, store_subcaches);
+static DEVICE_ATTR_RW(cache_disable_0);
+static DEVICE_ATTR_RW(cache_disable_1);
+static DEVICE_ATTR_RW(subcaches);
+
+static umode_t
+cache_private_attrs_is_visible(struct kobject *kobj,
+			       struct attribute *attr, int unused)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct cacheinfo *this_leaf = dev_get_drvdata(dev);
+	umode_t mode = attr->mode;
+
+	if (!this_leaf->priv)
+		return 0;
+
+	if ((attr == &dev_attr_subcaches.attr) &&
+	    amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
+		return mode;
+
+	if ((attr == &dev_attr_cache_disable_0.attr ||
+	     attr == &dev_attr_cache_disable_1.attr) &&
+	    amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
+		return mode;
+
+	return 0;
+}
+
+static struct attribute_group cache_private_group = {
+	.is_visible = cache_private_attrs_is_visible,
+};
+
+static void init_amd_l3_attrs(void)
+{
+	int n = 1;
+	static struct attribute **amd_l3_attrs;
+
+	if (amd_l3_attrs) /* already initialized */
+		return;
+
+	if (amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
+		n += 2;
+	if (amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
+		n += 1;
+
+	amd_l3_attrs = kcalloc(n, sizeof(*amd_l3_attrs), GFP_KERNEL);
+	if (!amd_l3_attrs)
+		return;
+
+	n = 0;
+	if (amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE)) {
+		amd_l3_attrs[n++] = &dev_attr_cache_disable_0.attr;
+		amd_l3_attrs[n++] = &dev_attr_cache_disable_1.attr;
+	}
+	if (amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
+		amd_l3_attrs[n++] = &dev_attr_subcaches.attr;
+
+	cache_private_group.attrs = amd_l3_attrs;
+}
+
+const struct attribute_group *
+cache_get_priv_group(struct cacheinfo *this_leaf)
+{
+	if (!this_leaf->priv || !cache_private_group.attrs)
+		return NULL;
+	return &cache_private_group;
+}
+
+static void amd_init_l3_cache(struct _cpuid4_info_regs *this_leaf, int index)
+{
+	int node;
+
+	/* only for L3, and not in virtualized environments */
+	if (index < 3)
+		return;
 
+	node = amd_get_nb_id(smp_processor_id());
+	this_leaf->nb = node_to_amd_nb(node);
+	if (this_leaf->nb && !this_leaf->nb->l3_cache.indices) {
+		amd_calc_l3_indices(this_leaf->nb);
+		init_amd_l3_attrs();
+	}
+}
 #else
 #define amd_init_l3_cache(x, y)
 #endif  /* CONFIG_AMD_NB && CONFIG_SYSFS */
@@ -546,7 +597,7 @@ cpuid4_cache_lookup_regs(int index, struct _cpuid4_info_regs *this_leaf)
 		cpuid_count(4, index, &eax.full, &ebx.full, &ecx.full, &edx);
 	}
 
-	if (eax.split.type == CACHE_TYPE_NULL)
+	if (eax.split.type == CTYPE_NULL)
 		return -EIO; /* better error ? */
 
 	this_leaf->eax = eax;
@@ -575,7 +626,7 @@ static int find_num_cache_leaves(struct cpuinfo_x86 *c)
 		/* Do cpuid(op) loop to find out num_cache_leaves */
 		cpuid_count(op, i, &eax, &ebx, &ecx, &edx);
 		cache_eax.full = eax;
-	} while (cache_eax.split.type != CACHE_TYPE_NULL);
+	} while (cache_eax.split.type != CTYPE_NULL);
 	return i;
 }
 
@@ -626,9 +677,9 @@ unsigned int init_intel_cacheinfo(struct cpuinfo_x86 *c)
 
 			switch (this_leaf.eax.split.level) {
 			case 1:
-				if (this_leaf.eax.split.type == CACHE_TYPE_DATA)
+				if (this_leaf.eax.split.type == CTYPE_DATA)
 					new_l1d = this_leaf.size/1024;
-				else if (this_leaf.eax.split.type == CACHE_TYPE_INST)
+				else if (this_leaf.eax.split.type == CTYPE_INST)
 					new_l1i = this_leaf.size/1024;
 				break;
 			case 2:
@@ -747,55 +798,46 @@ unsigned int init_intel_cacheinfo(struct cpuinfo_x86 *c)
 	return l2;
 }
 
-#ifdef CONFIG_SYSFS
-
-/* pointer to _cpuid4_info array (for each cache leaf) */
-static DEFINE_PER_CPU(struct _cpuid4_info *, ici_cpuid4_info);
-#define CPUID4_INFO_IDX(x, y)	(&((per_cpu(ici_cpuid4_info, x))[y]))
-
-#ifdef CONFIG_SMP
-
-static int cache_shared_amd_cpu_map_setup(unsigned int cpu, int index)
+static int __cache_amd_cpumap_setup(unsigned int cpu, int index,
+				    struct _cpuid4_info_regs *base)
 {
-	struct _cpuid4_info *this_leaf;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf;
 	int i, sibling;
 
 	if (cpu_has_topoext) {
 		unsigned int apicid, nshared, first, last;
 
-		if (!per_cpu(ici_cpuid4_info, cpu))
-			return 0;
-
-		this_leaf = CPUID4_INFO_IDX(cpu, index);
-		nshared = this_leaf->base.eax.split.num_threads_sharing + 1;
+		this_leaf = this_cpu_ci->info_list + index;
+		nshared = base->eax.split.num_threads_sharing + 1;
 		apicid = cpu_data(cpu).apicid;
 		first = apicid - (apicid % nshared);
 		last = first + nshared - 1;
 
 		for_each_online_cpu(i) {
+			this_cpu_ci = get_cpu_cacheinfo(i);
 			apicid = cpu_data(i).apicid;
 			if ((apicid < first) || (apicid > last))
 				continue;
-			if (!per_cpu(ici_cpuid4_info, i))
-				continue;
-			this_leaf = CPUID4_INFO_IDX(i, index);
+			this_leaf = this_cpu_ci->info_list + index;
 
 			for_each_online_cpu(sibling) {
 				apicid = cpu_data(sibling).apicid;
 				if ((apicid < first) || (apicid > last))
 					continue;
-				set_bit(sibling, this_leaf->shared_cpu_map);
+				cpumask_set_cpu(sibling,
+						&this_leaf->shared_cpu_map);
 			}
 		}
 	} else if (index == 3) {
 		for_each_cpu(i, cpu_llc_shared_mask(cpu)) {
-			if (!per_cpu(ici_cpuid4_info, i))
-				continue;
-			this_leaf = CPUID4_INFO_IDX(i, index);
+			this_cpu_ci = get_cpu_cacheinfo(i);
+			this_leaf = this_cpu_ci->info_list + index;
 			for_each_cpu(sibling, cpu_llc_shared_mask(cpu)) {
 				if (!cpu_online(sibling))
 					continue;
-				set_bit(sibling, this_leaf->shared_cpu_map);
+				cpumask_set_cpu(sibling,
+						&this_leaf->shared_cpu_map);
 			}
 		}
 	} else
@@ -804,459 +846,86 @@ static int cache_shared_amd_cpu_map_setup(unsigned int cpu, int index)
 	return 1;
 }
 
-static void cache_shared_cpu_map_setup(unsigned int cpu, int index)
+static void __cache_cpumap_setup(unsigned int cpu, int index,
+				 struct _cpuid4_info_regs *base)
 {
-	struct _cpuid4_info *this_leaf, *sibling_leaf;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf, *sibling_leaf;
 	unsigned long num_threads_sharing;
 	int index_msb, i;
 	struct cpuinfo_x86 *c = &cpu_data(cpu);
 
 	if (c->x86_vendor == X86_VENDOR_AMD) {
-		if (cache_shared_amd_cpu_map_setup(cpu, index))
+		if (__cache_amd_cpumap_setup(cpu, index, base))
 			return;
 	}
 
-	this_leaf = CPUID4_INFO_IDX(cpu, index);
-	num_threads_sharing = 1 + this_leaf->base.eax.split.num_threads_sharing;
+	this_leaf = this_cpu_ci->info_list + index;
+	num_threads_sharing = 1 + base->eax.split.num_threads_sharing;
 
+	cpumask_set_cpu(cpu, &this_leaf->shared_cpu_map);
 	if (num_threads_sharing == 1)
-		cpumask_set_cpu(cpu, to_cpumask(this_leaf->shared_cpu_map));
-	else {
-		index_msb = get_count_order(num_threads_sharing);
-
-		for_each_online_cpu(i) {
-			if (cpu_data(i).apicid >> index_msb ==
-			    c->apicid >> index_msb) {
-				cpumask_set_cpu(i,
-					to_cpumask(this_leaf->shared_cpu_map));
-				if (i != cpu && per_cpu(ici_cpuid4_info, i))  {
-					sibling_leaf =
-						CPUID4_INFO_IDX(i, index);
-					cpumask_set_cpu(cpu, to_cpumask(
-						sibling_leaf->shared_cpu_map));
-				}
-			}
-		}
-	}
-}
-static void cache_remove_shared_cpu_map(unsigned int cpu, int index)
-{
-	struct _cpuid4_info	*this_leaf, *sibling_leaf;
-	int sibling;
-
-	this_leaf = CPUID4_INFO_IDX(cpu, index);
-	for_each_cpu(sibling, to_cpumask(this_leaf->shared_cpu_map)) {
-		sibling_leaf = CPUID4_INFO_IDX(sibling, index);
-		cpumask_clear_cpu(cpu,
-				  to_cpumask(sibling_leaf->shared_cpu_map));
-	}
-}
-#else
-static void cache_shared_cpu_map_setup(unsigned int cpu, int index)
-{
-}
-
-static void cache_remove_shared_cpu_map(unsigned int cpu, int index)
-{
-}
-#endif
-
-static void free_cache_attributes(unsigned int cpu)
-{
-	int i;
-
-	for (i = 0; i < num_cache_leaves; i++)
-		cache_remove_shared_cpu_map(cpu, i);
-
-	kfree(per_cpu(ici_cpuid4_info, cpu));
-	per_cpu(ici_cpuid4_info, cpu) = NULL;
-}
-
-static void get_cpu_leaves(void *_retval)
-{
-	int j, *retval = _retval, cpu = smp_processor_id();
+		return;
 
-	/* Do cpuid and store the results */
-	for (j = 0; j < num_cache_leaves; j++) {
-		struct _cpuid4_info *this_leaf = CPUID4_INFO_IDX(cpu, j);
+	index_msb = get_count_order(num_threads_sharing);
 
-		*retval = cpuid4_cache_lookup_regs(j, &this_leaf->base);
-		if (unlikely(*retval < 0)) {
-			int i;
+	for_each_online_cpu(i)
+		if (cpu_data(i).apicid >> index_msb == c->apicid >> index_msb) {
+			struct cpu_cacheinfo *sib_cpu_ci = get_cpu_cacheinfo(i);
 
-			for (i = 0; i < j; i++)
-				cache_remove_shared_cpu_map(cpu, i);
-			break;
+			if (i == cpu || !sib_cpu_ci->info_list)
+				continue;/* skip if itself or no cacheinfo */
+			sibling_leaf = sib_cpu_ci->info_list + index;
+			cpumask_set_cpu(i, &this_leaf->shared_cpu_map);
+			cpumask_set_cpu(cpu, &sibling_leaf->shared_cpu_map);
 		}
-		cache_shared_cpu_map_setup(cpu, j);
-	}
 }
 
-static int detect_cache_attributes(unsigned int cpu)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 struct _cpuid4_info_regs *base)
 {
-	int			retval;
-
-	if (num_cache_leaves == 0)
-		return -ENOENT;
-
-	per_cpu(ici_cpuid4_info, cpu) = kzalloc(
-	    sizeof(struct _cpuid4_info) * num_cache_leaves, GFP_KERNEL);
-	if (per_cpu(ici_cpuid4_info, cpu) == NULL)
-		return -ENOMEM;
-
-	smp_call_function_single(cpu, get_cpu_leaves, &retval, true);
-	if (retval) {
-		kfree(per_cpu(ici_cpuid4_info, cpu));
-		per_cpu(ici_cpuid4_info, cpu) = NULL;
-	}
-
-	return retval;
-}
-
-#include <linux/kobject.h>
-#include <linux/sysfs.h>
-#include <linux/cpu.h>
-
-/* pointer to kobject for cpuX/cache */
-static DEFINE_PER_CPU(struct kobject *, ici_cache_kobject);
-
-struct _index_kobject {
-	struct kobject kobj;
-	unsigned int cpu;
-	unsigned short index;
-};
-
-/* pointer to array of kobjects for cpuX/cache/indexY */
-static DEFINE_PER_CPU(struct _index_kobject *, ici_index_kobject);
-#define INDEX_KOBJECT_PTR(x, y)		(&((per_cpu(ici_index_kobject, x))[y]))
-
-#define show_one_plus(file_name, object, val)				\
-static ssize_t show_##file_name(struct _cpuid4_info *this_leaf, char *buf, \
-				unsigned int cpu)			\
-{									\
-	return sprintf(buf, "%lu\n", (unsigned long)this_leaf->object + val); \
-}
-
-show_one_plus(level, base.eax.split.level, 0);
-show_one_plus(coherency_line_size, base.ebx.split.coherency_line_size, 1);
-show_one_plus(physical_line_partition, base.ebx.split.physical_line_partition, 1);
-show_one_plus(ways_of_associativity, base.ebx.split.ways_of_associativity, 1);
-show_one_plus(number_of_sets, base.ecx.split.number_of_sets, 1);
-
-static ssize_t show_size(struct _cpuid4_info *this_leaf, char *buf,
-			 unsigned int cpu)
-{
-	return sprintf(buf, "%luK\n", this_leaf->base.size / 1024);
-}
-
-static ssize_t show_shared_cpu_map_func(struct _cpuid4_info *this_leaf,
-					int type, char *buf)
-{
-	ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf;
-	int n = 0;
-
-	if (len > 1) {
-		const struct cpumask *mask;
-
-		mask = to_cpumask(this_leaf->shared_cpu_map);
-		n = type ?
-			cpulist_scnprintf(buf, len-2, mask) :
-			cpumask_scnprintf(buf, len-2, mask);
-		buf[n++] = '\n';
-		buf[n] = '\0';
-	}
-	return n;
+	this_leaf->level = base->eax.split.level;
+	this_leaf->type = cache_type_map[base->eax.split.type];
+	this_leaf->coherency_line_size =
+				base->ebx.split.coherency_line_size + 1;
+	this_leaf->ways_of_associativity =
+				base->ebx.split.ways_of_associativity + 1;
+	this_leaf->size = base->size;
+	this_leaf->number_of_sets = base->ecx.split.number_of_sets + 1;
+	this_leaf->physical_line_partition =
+				base->ebx.split.physical_line_partition + 1;
+	this_leaf->priv = base->nb;
 }
 
-static inline ssize_t show_shared_cpu_map(struct _cpuid4_info *leaf, char *buf,
-					  unsigned int cpu)
+static int __init_cache_level(unsigned int cpu)
 {
-	return show_shared_cpu_map_func(leaf, 0, buf);
-}
-
-static inline ssize_t show_shared_cpu_list(struct _cpuid4_info *leaf, char *buf,
-					   unsigned int cpu)
-{
-	return show_shared_cpu_map_func(leaf, 1, buf);
-}
-
-static ssize_t show_type(struct _cpuid4_info *this_leaf, char *buf,
-			 unsigned int cpu)
-{
-	switch (this_leaf->base.eax.split.type) {
-	case CACHE_TYPE_DATA:
-		return sprintf(buf, "Data\n");
-	case CACHE_TYPE_INST:
-		return sprintf(buf, "Instruction\n");
-	case CACHE_TYPE_UNIFIED:
-		return sprintf(buf, "Unified\n");
-	default:
-		return sprintf(buf, "Unknown\n");
-	}
-}
-
-#define to_object(k)	container_of(k, struct _index_kobject, kobj)
-#define to_attr(a)	container_of(a, struct _cache_attr, attr)
-
-#define define_one_ro(_name) \
-static struct _cache_attr _name = \
-	__ATTR(_name, 0444, show_##_name, NULL)
-
-define_one_ro(level);
-define_one_ro(type);
-define_one_ro(coherency_line_size);
-define_one_ro(physical_line_partition);
-define_one_ro(ways_of_associativity);
-define_one_ro(number_of_sets);
-define_one_ro(size);
-define_one_ro(shared_cpu_map);
-define_one_ro(shared_cpu_list);
-
-static struct attribute *default_attrs[] = {
-	&type.attr,
-	&level.attr,
-	&coherency_line_size.attr,
-	&physical_line_partition.attr,
-	&ways_of_associativity.attr,
-	&number_of_sets.attr,
-	&size.attr,
-	&shared_cpu_map.attr,
-	&shared_cpu_list.attr,
-	NULL
-};
-
-#ifdef CONFIG_AMD_NB
-static struct attribute **amd_l3_attrs(void)
-{
-	static struct attribute **attrs;
-	int n;
-
-	if (attrs)
-		return attrs;
-
-	n = ARRAY_SIZE(default_attrs);
-
-	if (amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE))
-		n += 2;
-
-	if (amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
-		n += 1;
-
-	attrs = kzalloc(n * sizeof (struct attribute *), GFP_KERNEL);
-	if (attrs == NULL)
-		return attrs = default_attrs;
-
-	for (n = 0; default_attrs[n]; n++)
-		attrs[n] = default_attrs[n];
-
-	if (amd_nb_has_feature(AMD_NB_L3_INDEX_DISABLE)) {
-		attrs[n++] = &cache_disable_0.attr;
-		attrs[n++] = &cache_disable_1.attr;
-	}
-
-	if (amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
-		attrs[n++] = &subcaches.attr;
-
-	return attrs;
-}
-#endif
-
-static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
-{
-	struct _cache_attr *fattr = to_attr(attr);
-	struct _index_kobject *this_leaf = to_object(kobj);
-	ssize_t ret;
-
-	ret = fattr->show ?
-		fattr->show(CPUID4_INFO_IDX(this_leaf->cpu, this_leaf->index),
-			buf, this_leaf->cpu) :
-		0;
-	return ret;
-}
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
 
-static ssize_t store(struct kobject *kobj, struct attribute *attr,
-		     const char *buf, size_t count)
-{
-	struct _cache_attr *fattr = to_attr(attr);
-	struct _index_kobject *this_leaf = to_object(kobj);
-	ssize_t ret;
-
-	ret = fattr->store ?
-		fattr->store(CPUID4_INFO_IDX(this_leaf->cpu, this_leaf->index),
-			buf, count, this_leaf->cpu) :
-		0;
-	return ret;
-}
-
-static const struct sysfs_ops sysfs_ops = {
-	.show   = show,
-	.store  = store,
-};
-
-static struct kobj_type ktype_cache = {
-	.sysfs_ops	= &sysfs_ops,
-	.default_attrs	= default_attrs,
-};
-
-static struct kobj_type ktype_percpu_entry = {
-	.sysfs_ops	= &sysfs_ops,
-};
-
-static void cpuid4_cache_sysfs_exit(unsigned int cpu)
-{
-	kfree(per_cpu(ici_cache_kobject, cpu));
-	kfree(per_cpu(ici_index_kobject, cpu));
-	per_cpu(ici_cache_kobject, cpu) = NULL;
-	per_cpu(ici_index_kobject, cpu) = NULL;
-	free_cache_attributes(cpu);
-}
-
-static int cpuid4_cache_sysfs_init(unsigned int cpu)
-{
-	int err;
-
-	if (num_cache_leaves == 0)
+	if (!num_cache_leaves)
 		return -ENOENT;
-
-	err = detect_cache_attributes(cpu);
-	if (err)
-		return err;
-
-	/* Allocate all required memory */
-	per_cpu(ici_cache_kobject, cpu) =
-		kzalloc(sizeof(struct kobject), GFP_KERNEL);
-	if (unlikely(per_cpu(ici_cache_kobject, cpu) == NULL))
-		goto err_out;
-
-	per_cpu(ici_index_kobject, cpu) = kzalloc(
-	    sizeof(struct _index_kobject) * num_cache_leaves, GFP_KERNEL);
-	if (unlikely(per_cpu(ici_index_kobject, cpu) == NULL))
-		goto err_out;
-
+	if (!this_cpu_ci)
+		return -EINVAL;
+	this_cpu_ci->num_levels = 3;
+	this_cpu_ci->num_leaves = num_cache_leaves;
 	return 0;
-
-err_out:
-	cpuid4_cache_sysfs_exit(cpu);
-	return -ENOMEM;
 }
 
-static DECLARE_BITMAP(cache_dev_map, NR_CPUS);
-
-/* Add/Remove cache interface for CPU device */
-static int cache_add_dev(struct device *dev)
+static int __populate_cache_leaves(unsigned int cpu)
 {
-	unsigned int cpu = dev->id;
-	unsigned long i, j;
-	struct _index_kobject *this_object;
-	struct _cpuid4_info   *this_leaf;
-	int retval;
-
-	retval = cpuid4_cache_sysfs_init(cpu);
-	if (unlikely(retval < 0))
-		return retval;
-
-	retval = kobject_init_and_add(per_cpu(ici_cache_kobject, cpu),
-				      &ktype_percpu_entry,
-				      &dev->kobj, "%s", "cache");
-	if (retval < 0) {
-		cpuid4_cache_sysfs_exit(cpu);
-		return retval;
-	}
-
-	for (i = 0; i < num_cache_leaves; i++) {
-		this_object = INDEX_KOBJECT_PTR(cpu, i);
-		this_object->cpu = cpu;
-		this_object->index = i;
+	unsigned int idx, ret;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+	struct _cpuid4_info_regs id4_regs = {};
 
-		this_leaf = CPUID4_INFO_IDX(cpu, i);
-
-		ktype_cache.default_attrs = default_attrs;
-#ifdef CONFIG_AMD_NB
-		if (this_leaf->base.nb)
-			ktype_cache.default_attrs = amd_l3_attrs();
-#endif
-		retval = kobject_init_and_add(&(this_object->kobj),
-					      &ktype_cache,
-					      per_cpu(ici_cache_kobject, cpu),
-					      "index%1lu", i);
-		if (unlikely(retval)) {
-			for (j = 0; j < i; j++)
-				kobject_put(&(INDEX_KOBJECT_PTR(cpu, j)->kobj));
-			kobject_put(per_cpu(ici_cache_kobject, cpu));
-			cpuid4_cache_sysfs_exit(cpu);
-			return retval;
-		}
-		kobject_uevent(&(this_object->kobj), KOBJ_ADD);
+	for (idx = 0; idx < this_cpu_ci->num_leaves; idx++) {
+		ret = cpuid4_cache_lookup_regs(idx, &id4_regs);
+		if (ret)
+			return ret;
+		ci_leaf_init(this_leaf++, &id4_regs);
+		__cache_cpumap_setup(cpu, idx, &id4_regs);
 	}
-	cpumask_set_cpu(cpu, to_cpumask(cache_dev_map));
-
-	kobject_uevent(per_cpu(ici_cache_kobject, cpu), KOBJ_ADD);
 	return 0;
 }
 
-static void cache_remove_dev(struct device *dev)
-{
-	unsigned int cpu = dev->id;
-	unsigned long i;
-
-	if (per_cpu(ici_cpuid4_info, cpu) == NULL)
-		return;
-	if (!cpumask_test_cpu(cpu, to_cpumask(cache_dev_map)))
-		return;
-	cpumask_clear_cpu(cpu, to_cpumask(cache_dev_map));
-
-	for (i = 0; i < num_cache_leaves; i++)
-		kobject_put(&(INDEX_KOBJECT_PTR(cpu, i)->kobj));
-	kobject_put(per_cpu(ici_cache_kobject, cpu));
-	cpuid4_cache_sysfs_exit(cpu);
-}
-
-static int cacheinfo_cpu_callback(struct notifier_block *nfb,
-				  unsigned long action, void *hcpu)
-{
-	unsigned int cpu = (unsigned long)hcpu;
-	struct device *dev;
-
-	dev = get_cpu_device(cpu);
-	switch (action) {
-	case CPU_ONLINE:
-	case CPU_ONLINE_FROZEN:
-		cache_add_dev(dev);
-		break;
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		cache_remove_dev(dev);
-		break;
-	}
-	return NOTIFY_OK;
-}
-
-static struct notifier_block cacheinfo_cpu_notifier = {
-	.notifier_call = cacheinfo_cpu_callback,
-};
-
-static int __init cache_sysfs_init(void)
-{
-	int i, err = 0;
-
-	if (num_cache_leaves == 0)
-		return 0;
-
-	cpu_notifier_register_begin();
-	for_each_online_cpu(i) {
-		struct device *dev = get_cpu_device(i);
-
-		err = cache_add_dev(dev);
-		if (err)
-			goto out;
-	}
-	__register_hotcpu_notifier(&cacheinfo_cpu_notifier);
-
-out:
-	cpu_notifier_register_done();
-	return err;
-}
-
-device_initcall(cache_sysfs_init);
-
-#endif
+DEFINE_SMP_CALL_CACHE_FUNCTION(init_cache_level)
+DEFINE_SMP_CALL_CACHE_FUNCTION(populate_cache_leaves)
-- 
1.8.3.2

[PATCH v3 08/11] powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: linuxppc-dev@lists.ozlabs.org
---
 arch/powerpc/kernel/cacheinfo.c | 812 +++++-----------------------------------
 arch/powerpc/kernel/cacheinfo.h |   8 -
 arch/powerpc/kernel/sysfs.c     |  12 +-
 3 files changed, 90 insertions(+), 742 deletions(-)
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h

diff --git a/arch/powerpc/kernel/cacheinfo.c b/arch/powerpc/kernel/cacheinfo.c
index 40198d50b4c2..6845eb7fcc18 100644
--- a/arch/powerpc/kernel/cacheinfo.c
+++ b/arch/powerpc/kernel/cacheinfo.c
@@ -10,38 +10,10 @@
  * 2 as published by the Free Software Foundation.
  */
 
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
-#include <linux/cpumask.h>
 #include <linux/kernel.h>
-#include <linux/kobject.h>
-#include <linux/list.h>
-#include <linux/notifier.h>
 #include <linux/of.h>
-#include <linux/percpu.h>
-#include <linux/slab.h>
-#include <asm/prom.h>
-
-#include "cacheinfo.h"
-
-/* per-cpu object for tracking:
- * - a "cache" kobject for the top-level directory
- * - a list of "index" objects representing the cpu's local cache hierarchy
- */
-struct cache_dir {
-	struct kobject *kobj; /* bare (not embedded) kobject for cache
-			       * directory */
-	struct cache_index_dir *index; /* list of index objects */
-};
-
-/* "index" object: each cpu's cache directory has an index
- * subdirectory corresponding to a cache object associated with the
- * cpu.  This object's lifetime is managed via the embedded kobject.
- */
-struct cache_index_dir {
-	struct kobject kobj;
-	struct cache_index_dir *next; /* next index in parent directory */
-	struct cache *cache;
-};
 
 /* Template for determining which OF properties to query for a given
  * cache type */
@@ -60,11 +32,6 @@ struct cache_type_info {
 	const char *nr_sets_prop;
 };
 
-/* These are used to index the cache_type_info array. */
-#define CACHE_TYPE_UNIFIED     0
-#define CACHE_TYPE_INSTRUCTION 1
-#define CACHE_TYPE_DATA        2
-
 static const struct cache_type_info cache_type_info[] = {
 	{
 		/* PowerPC Processor binding says the [di]-cache-*
@@ -92,231 +59,82 @@ static const struct cache_type_info cache_type_info[] = {
 	},
 };
 
-/* Cache object: each instance of this corresponds to a distinct cache
- * in the system.  There are separate objects for Harvard caches: one
- * each for instruction and data, and each refers to the same OF node.
- * The refcount of the OF node is elevated for the lifetime of the
- * cache object.  A cache object is released when its shared_cpu_map
- * is cleared (see cache_cpu_clear).
- *
- * A cache object is on two lists: an unsorted global list
- * (cache_list) of cache objects; and a singly-linked list
- * representing the local cache hierarchy, which is ordered by level
- * (e.g. L1d -> L1i -> L2 -> L3).
- */
-struct cache {
-	struct device_node *ofnode;    /* OF node for this cache, may be cpu */
-	struct cpumask shared_cpu_map; /* online CPUs using this cache */
-	int type;                      /* split cache disambiguation */
-	int level;                     /* level not explicit in device tree */
-	struct list_head list;         /* global list of cache objects */
-	struct cache *next_local;      /* next cache of >= level */
-};
-
-static DEFINE_PER_CPU(struct cache_dir *, cache_dir_pcpu);
-
-/* traversal/modification of this list occurs only at cpu hotplug time;
- * access is serialized by cpu hotplug locking
- */
-static LIST_HEAD(cache_list);
-
-static struct cache_index_dir *kobj_to_cache_index_dir(struct kobject *k)
-{
-	return container_of(k, struct cache_index_dir, kobj);
-}
-
-static const char *cache_type_string(const struct cache *cache)
+static inline int get_cacheinfo_idx(enum cache_type type)
 {
-	return cache_type_info[cache->type].name;
-}
-
-static void cache_init(struct cache *cache, int type, int level,
-		       struct device_node *ofnode)
-{
-	cache->type = type;
-	cache->level = level;
-	cache->ofnode = of_node_get(ofnode);
-	INIT_LIST_HEAD(&cache->list);
-	list_add(&cache->list, &cache_list);
-}
-
-static struct cache *new_cache(int type, int level, struct device_node *ofnode)
-{
-	struct cache *cache;
-
-	cache = kzalloc(sizeof(*cache), GFP_KERNEL);
-	if (cache)
-		cache_init(cache, type, level, ofnode);
-
-	return cache;
-}
-
-static void release_cache_debugcheck(struct cache *cache)
-{
-	struct cache *iter;
-
-	list_for_each_entry(iter, &cache_list, list)
-		WARN_ONCE(iter->next_local == cache,
-			  "cache for %s(%s) refers to cache for %s(%s)\n",
-			  iter->ofnode->full_name,
-			  cache_type_string(iter),
-			  cache->ofnode->full_name,
-			  cache_type_string(cache));
-}
-
-static void release_cache(struct cache *cache)
-{
-	if (!cache)
-		return;
-
-	pr_debug("freeing L%d %s cache for %s\n", cache->level,
-		 cache_type_string(cache), cache->ofnode->full_name);
-
-	release_cache_debugcheck(cache);
-	list_del(&cache->list);
-	of_node_put(cache->ofnode);
-	kfree(cache);
-}
-
-static void cache_cpu_set(struct cache *cache, int cpu)
-{
-	struct cache *next = cache;
-
-	while (next) {
-		WARN_ONCE(cpumask_test_cpu(cpu, &next->shared_cpu_map),
-			  "CPU %i already accounted in %s(%s)\n",
-			  cpu, next->ofnode->full_name,
-			  cache_type_string(next));
-		cpumask_set_cpu(cpu, &next->shared_cpu_map);
-		next = next->next_local;
-	}
+	if (type == CACHE_TYPE_UNIFIED)
+		return 0;
+	return type;
 }
 
-static int cache_size(const struct cache *cache, unsigned int *ret)
+static void cache_size(struct cacheinfo *this_leaf)
 {
 	const char *propname;
 	const __be32 *cache_size;
+	int ct_idx;
 
-	propname = cache_type_info[cache->type].size_prop;
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	propname = cache_type_info[ct_idx].size_prop;
 
-	cache_size = of_get_property(cache->ofnode, propname, NULL);
+	cache_size = of_get_property(this_leaf->of_node, propname, NULL);
 	if (!cache_size)
-		return -ENODEV;
-
-	*ret = of_read_number(cache_size, 1);
-	return 0;
-}
-
-static int cache_size_kb(const struct cache *cache, unsigned int *ret)
-{
-	unsigned int size;
-
-	if (cache_size(cache, &size))
-		return -ENODEV;
-
-	*ret = size / 1024;
-	return 0;
+		this_leaf->size = 0;
+	else
+		this_leaf->size = of_read_number(cache_size, 1);
 }
 
 /* not cache_line_size() because that's a macro in include/linux/cache.h */
-static int cache_get_line_size(const struct cache *cache, unsigned int *ret)
+static void cache_get_line_size(struct cacheinfo *this_leaf)
 {
 	const __be32 *line_size;
-	int i, lim;
+	int i, lim, ct_idx;
 
-	lim = ARRAY_SIZE(cache_type_info[cache->type].line_size_props);
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	lim = ARRAY_SIZE(cache_type_info[ct_idx].line_size_props);
 
 	for (i = 0; i < lim; i++) {
 		const char *propname;
 
-		propname = cache_type_info[cache->type].line_size_props[i];
-		line_size = of_get_property(cache->ofnode, propname, NULL);
+		propname = cache_type_info[ct_idx].line_size_props[i];
+		line_size = of_get_property(this_leaf->of_node, propname, NULL);
 		if (line_size)
 			break;
 	}
 
 	if (!line_size)
-		return -ENODEV;
-
-	*ret = of_read_number(line_size, 1);
-	return 0;
+		this_leaf->coherency_line_size = 0;
+	else
+		this_leaf->coherency_line_size = of_read_number(line_size, 1);
 }
 
-static int cache_nr_sets(const struct cache *cache, unsigned int *ret)
+static void cache_nr_sets(struct cacheinfo *this_leaf)
 {
 	const char *propname;
 	const __be32 *nr_sets;
+	int ct_idx;
 
-	propname = cache_type_info[cache->type].nr_sets_prop;
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	propname = cache_type_info[ct_idx].nr_sets_prop;
 
-	nr_sets = of_get_property(cache->ofnode, propname, NULL);
+	nr_sets = of_get_property(this_leaf->of_node, propname, NULL);
 	if (!nr_sets)
-		return -ENODEV;
-
-	*ret = of_read_number(nr_sets, 1);
-	return 0;
+		this_leaf->number_of_sets = 0;
+	else
+		this_leaf->number_of_sets = of_read_number(nr_sets, 1);
 }
 
-static int cache_associativity(const struct cache *cache, unsigned int *ret)
+static void cache_associativity(struct cacheinfo *this_leaf)
 {
-	unsigned int line_size;
-	unsigned int nr_sets;
-	unsigned int size;
-
-	if (cache_nr_sets(cache, &nr_sets))
-		goto err;
+	unsigned int line_size = this_leaf->coherency_line_size;
+	unsigned int nr_sets = this_leaf->number_of_sets;
+	unsigned int size = this_leaf->size;
 
 	/* If the cache is fully associative, there is no need to
 	 * check the other properties.
 	 */
-	if (nr_sets == 1) {
-		*ret = 0;
-		return 0;
-	}
-
-	if (cache_get_line_size(cache, &line_size))
-		goto err;
-	if (cache_size(cache, &size))
-		goto err;
-
-	if (!(nr_sets > 0 && size > 0 && line_size > 0))
-		goto err;
-
-	*ret = (size / nr_sets) / line_size;
-	return 0;
-err:
-	return -ENODEV;
-}
-
-/* helper for dealing with split caches */
-static struct cache *cache_find_first_sibling(struct cache *cache)
-{
-	struct cache *iter;
-
-	if (cache->type == CACHE_TYPE_UNIFIED)
-		return cache;
-
-	list_for_each_entry(iter, &cache_list, list)
-		if (iter->ofnode == cache->ofnode && iter->next_local == cache)
-			return iter;
-
-	return cache;
-}
-
-/* return the first cache on a local list matching node */
-static struct cache *cache_lookup_by_node(const struct device_node *node)
-{
-	struct cache *cache = NULL;
-	struct cache *iter;
-
-	list_for_each_entry(iter, &cache_list, list) {
-		if (iter->ofnode != node)
-			continue;
-		cache = cache_find_first_sibling(iter);
-		break;
-	}
-
-	return cache;
+	if ((nr_sets == 1) || !(nr_sets > 0 && size > 0 && line_size > 0))
+		this_leaf->ways_of_associativity = 0;
+	else
+		this_leaf->ways_of_associativity = (size / nr_sets) / line_size;
 }
 
 static bool cache_node_is_unified(const struct device_node *np)
@@ -324,526 +142,74 @@ static bool cache_node_is_unified(const struct device_node *np)
 	return of_get_property(np, "cache-unified", NULL);
 }
 
-static struct cache *cache_do_one_devnode_unified(struct device_node *node,
-						  int level)
-{
-	struct cache *cache;
-
-	pr_debug("creating L%d ucache for %s\n", level, node->full_name);
-
-	cache = new_cache(CACHE_TYPE_UNIFIED, level, node);
-
-	return cache;
-}
-
-static struct cache *cache_do_one_devnode_split(struct device_node *node,
-						int level)
-{
-	struct cache *dcache, *icache;
-
-	pr_debug("creating L%d dcache and icache for %s\n", level,
-		 node->full_name);
-
-	dcache = new_cache(CACHE_TYPE_DATA, level, node);
-	icache = new_cache(CACHE_TYPE_INSTRUCTION, level, node);
-
-	if (!dcache || !icache)
-		goto err;
-
-	dcache->next_local = icache;
-
-	return dcache;
-err:
-	release_cache(dcache);
-	release_cache(icache);
-	return NULL;
-}
-
-static struct cache *cache_do_one_devnode(struct device_node *node, int level)
-{
-	struct cache *cache;
-
-	if (cache_node_is_unified(node))
-		cache = cache_do_one_devnode_unified(node, level);
-	else
-		cache = cache_do_one_devnode_split(node, level);
-
-	return cache;
-}
-
-static struct cache *cache_lookup_or_instantiate(struct device_node *node,
-						 int level)
-{
-	struct cache *cache;
-
-	cache = cache_lookup_by_node(node);
-
-	WARN_ONCE(cache && cache->level != level,
-		  "cache level mismatch on lookup (got %d, expected %d)\n",
-		  cache->level, level);
-
-	if (!cache)
-		cache = cache_do_one_devnode(node, level);
-
-	return cache;
-}
-
-static void link_cache_lists(struct cache *smaller, struct cache *bigger)
-{
-	while (smaller->next_local) {
-		if (smaller->next_local == bigger)
-			return; /* already linked */
-		smaller = smaller->next_local;
-	}
-
-	smaller->next_local = bigger;
-}
-
-static void do_subsidiary_caches_debugcheck(struct cache *cache)
-{
-	WARN_ON_ONCE(cache->level != 1);
-	WARN_ON_ONCE(strcmp(cache->ofnode->type, "cpu"));
-}
-
-static void do_subsidiary_caches(struct cache *cache)
-{
-	struct device_node *subcache_node;
-	int level = cache->level;
-
-	do_subsidiary_caches_debugcheck(cache);
-
-	while ((subcache_node = of_find_next_cache_node(cache->ofnode))) {
-		struct cache *subcache;
-
-		level++;
-		subcache = cache_lookup_or_instantiate(subcache_node, level);
-		of_node_put(subcache_node);
-		if (!subcache)
-			break;
-
-		link_cache_lists(cache, subcache);
-		cache = subcache;
-	}
-}
-
-static struct cache *cache_chain_instantiate(unsigned int cpu_id)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
 {
-	struct device_node *cpu_node;
-	struct cache *cpu_cache = NULL;
-
-	pr_debug("creating cache object(s) for CPU %i\n", cpu_id);
-
-	cpu_node = of_get_cpu_node(cpu_id, NULL);
-	WARN_ONCE(!cpu_node, "no OF node found for CPU %i\n", cpu_id);
-	if (!cpu_node)
-		goto out;
-
-	cpu_cache = cache_lookup_or_instantiate(cpu_node, 1);
-	if (!cpu_cache)
-		goto out;
-
-	do_subsidiary_caches(cpu_cache);
-
-	cache_cpu_set(cpu_cache, cpu_id);
-out:
-	of_node_put(cpu_node);
-
-	return cpu_cache;
+	this_leaf->level = level;
+	this_leaf->type = type;
+	cache_size(this_leaf);
+	cache_get_line_size(this_leaf);
+	cache_nr_sets(this_leaf);
+	cache_associativity(this_leaf);
 }
 
-static struct cache_dir *cacheinfo_create_cache_dir(unsigned int cpu_id)
+int init_cache_level(unsigned int cpu)
 {
-	struct cache_dir *cache_dir;
-	struct device *dev;
-	struct kobject *kobj = NULL;
-
-	dev = get_cpu_device(cpu_id);
-	WARN_ONCE(!dev, "no dev for CPU %i\n", cpu_id);
-	if (!dev)
-		goto err;
-
-	kobj = kobject_create_and_add("cache", &dev->kobj);
-	if (!kobj)
-		goto err;
-
-	cache_dir = kzalloc(sizeof(*cache_dir), GFP_KERNEL);
-	if (!cache_dir)
-		goto err;
-
-	cache_dir->kobj = kobj;
-
-	WARN_ON_ONCE(per_cpu(cache_dir_pcpu, cpu_id) != NULL);
-
-	per_cpu(cache_dir_pcpu, cpu_id) = cache_dir;
+	struct device_node *np;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int level = 0, leaves = 0;
 
-	return cache_dir;
-err:
-	kobject_put(kobj);
-	return NULL;
-}
-
-static void cache_index_release(struct kobject *kobj)
-{
-	struct cache_index_dir *index;
-
-	index = kobj_to_cache_index_dir(kobj);
-
-	pr_debug("freeing index directory for L%d %s cache\n",
-		 index->cache->level, cache_type_string(index->cache));
-
-	kfree(index);
-}
-
-static ssize_t cache_index_show(struct kobject *k, struct attribute *attr, char *buf)
-{
-	struct kobj_attribute *kobj_attr;
-
-	kobj_attr = container_of(attr, struct kobj_attribute, attr);
-
-	return kobj_attr->show(k, kobj_attr, buf);
-}
-
-static struct cache *index_kobj_to_cache(struct kobject *k)
-{
-	struct cache_index_dir *index;
-
-	index = kobj_to_cache_index_dir(k);
-
-	return index->cache;
-}
-
-static ssize_t size_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int size_kb;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_size_kb(cache, &size_kb))
-		return -ENODEV;
-
-	return sprintf(buf, "%uK\n", size_kb);
-}
-
-static struct kobj_attribute cache_size_attr =
-	__ATTR(size, 0444, size_show, NULL);
-
-
-static ssize_t line_size_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int line_size;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_get_line_size(cache, &line_size))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", line_size);
-}
-
-static struct kobj_attribute cache_line_size_attr =
-	__ATTR(coherency_line_size, 0444, line_size_show, NULL);
-
-static ssize_t nr_sets_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int nr_sets;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_nr_sets(cache, &nr_sets))
+	if (!cpu_dev) {
+		pr_err("No cpu device for CPU %d\n", cpu);
 		return -ENODEV;
-
-	return sprintf(buf, "%u\n", nr_sets);
-}
-
-static struct kobj_attribute cache_nr_sets_attr =
-	__ATTR(number_of_sets, 0444, nr_sets_show, NULL);
-
-static ssize_t associativity_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int associativity;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_associativity(cache, &associativity))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", associativity);
-}
-
-static struct kobj_attribute cache_assoc_attr =
-	__ATTR(ways_of_associativity, 0444, associativity_show, NULL);
-
-static ssize_t type_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	return sprintf(buf, "%s\n", cache_type_string(cache));
-}
-
-static struct kobj_attribute cache_type_attr =
-	__ATTR(type, 0444, type_show, NULL);
-
-static ssize_t level_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache_index_dir *index;
-	struct cache *cache;
-
-	index = kobj_to_cache_index_dir(k);
-	cache = index->cache;
-
-	return sprintf(buf, "%d\n", cache->level);
-}
-
-static struct kobj_attribute cache_level_attr =
-	__ATTR(level, 0444, level_show, NULL);
-
-static ssize_t shared_cpu_map_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache_index_dir *index;
-	struct cache *cache;
-	int len;
-	int n = 0;
-
-	index = kobj_to_cache_index_dir(k);
-	cache = index->cache;
-	len = PAGE_SIZE - 2;
-
-	if (len > 1) {
-		n = cpumask_scnprintf(buf, len, &cache->shared_cpu_map);
-		buf[n++] = '\n';
-		buf[n] = '\0';
 	}
-	return n;
-}
-
-static struct kobj_attribute cache_shared_cpu_map_attr =
-	__ATTR(shared_cpu_map, 0444, shared_cpu_map_show, NULL);
-
-/* Attributes which should always be created -- the kobject/sysfs core
- * does this automatically via kobj_type->default_attrs.  This is the
- * minimum data required to uniquely identify a cache.
- */
-static struct attribute *cache_index_default_attrs[] = {
-	&cache_type_attr.attr,
-	&cache_level_attr.attr,
-	&cache_shared_cpu_map_attr.attr,
-	NULL,
-};
-
-/* Attributes which should be created if the cache device node has the
- * right properties -- see cacheinfo_create_index_opt_attrs
- */
-static struct kobj_attribute *cache_index_opt_attrs[] = {
-	&cache_size_attr,
-	&cache_line_size_attr,
-	&cache_nr_sets_attr,
-	&cache_assoc_attr,
-};
-
-static const struct sysfs_ops cache_index_ops = {
-	.show = cache_index_show,
-};
-
-static struct kobj_type cache_index_type = {
-	.release = cache_index_release,
-	.sysfs_ops = &cache_index_ops,
-	.default_attrs = cache_index_default_attrs,
-};
-
-static void cacheinfo_create_index_opt_attrs(struct cache_index_dir *dir)
-{
-	const char *cache_name;
-	const char *cache_type;
-	struct cache *cache;
-	char *buf;
-	int i;
-
-	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
-	if (!buf)
-		return;
-
-	cache = dir->cache;
-	cache_name = cache->ofnode->full_name;
-	cache_type = cache_type_string(cache);
-
-	/* We don't want to create an attribute that can't provide a
-	 * meaningful value.  Check the return value of each optional
-	 * attribute's ->show method before registering the
-	 * attribute.
-	 */
-	for (i = 0; i < ARRAY_SIZE(cache_index_opt_attrs); i++) {
-		struct kobj_attribute *attr;
-		ssize_t rc;
-
-		attr = cache_index_opt_attrs[i];
-
-		rc = attr->show(&dir->kobj, attr, buf);
-		if (rc <= 0) {
-			pr_debug("not creating %s attribute for "
-				 "%s(%s) (rc = %zd)\n",
-				 attr->attr.name, cache_name,
-				 cache_type, rc);
-			continue;
-		}
-		if (sysfs_create_file(&dir->kobj, &attr->attr))
-			pr_debug("could not create %s attribute for %s(%s)\n",
-				 attr->attr.name, cache_name, cache_type);
+	np = cpu_dev->of_node;
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
 	}
 
-	kfree(buf);
-}
-
-static void cacheinfo_create_index_dir(struct cache *cache, int index,
-				       struct cache_dir *cache_dir)
-{
-	struct cache_index_dir *index_dir;
-	int rc;
-
-	index_dir = kzalloc(sizeof(*index_dir), GFP_KERNEL);
-	if (!index_dir)
-		goto err;
-
-	index_dir->cache = cache;
-
-	rc = kobject_init_and_add(&index_dir->kobj, &cache_index_type,
-				  cache_dir->kobj, "index%d", index);
-	if (rc)
-		goto err;
-
-	index_dir->next = cache_dir->index;
-	cache_dir->index = index_dir;
-
-	cacheinfo_create_index_opt_attrs(index_dir);
-
-	return;
-err:
-	kfree(index_dir);
-}
-
-static void cacheinfo_sysfs_populate(unsigned int cpu_id,
-				     struct cache *cache_list)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-	int index = 0;
-
-	cache_dir = cacheinfo_create_cache_dir(cpu_id);
-	if (!cache_dir)
-		return;
-
-	cache = cache_list;
-	while (cache) {
-		cacheinfo_create_index_dir(cache, index, cache_dir);
-		index++;
-		cache = cache->next_local;
+	while (np) {
+		leaves += cache_node_is_unified(np) ? 1 : 2;
+		level++;
+		of_node_put(np);
+		np = of_find_next_cache_node(np);
 	}
-}
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
 
-void cacheinfo_cpu_online(unsigned int cpu_id)
-{
-	struct cache *cache;
-
-	cache = cache_chain_instantiate(cpu_id);
-	if (!cache)
-		return;
-
-	cacheinfo_sysfs_populate(cpu_id, cache);
-}
-
-/* functions needed to remove cache entry for cpu offline or suspend/resume */
-
-#if (defined(CONFIG_PPC_PSERIES) && defined(CONFIG_SUSPEND)) || \
-    defined(CONFIG_HOTPLUG_CPU)
-
-static struct cache *cache_lookup_by_cpu(unsigned int cpu_id)
-{
-	struct device_node *cpu_node;
-	struct cache *cache;
-
-	cpu_node = of_get_cpu_node(cpu_id, NULL);
-	WARN_ONCE(!cpu_node, "no OF node found for CPU %i\n", cpu_id);
-	if (!cpu_node)
-		return NULL;
-
-	cache = cache_lookup_by_node(cpu_node);
-	of_node_put(cpu_node);
-
-	return cache;
+	return 0;
 }
 
-static void remove_index_dirs(struct cache_dir *cache_dir)
+int populate_cache_leaves(unsigned int cpu)
 {
-	struct cache_index_dir *index;
-
-	index = cache_dir->index;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct device_node *np;
+	unsigned int level, idx;
 
-	while (index) {
-		struct cache_index_dir *next;
-
-		next = index->next;
-		kobject_put(&index->kobj);
-		index = next;
+	np = of_node_get(cpu_dev->of_node);
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
 	}
-}
-
-static void remove_cache_dir(struct cache_dir *cache_dir)
-{
-	remove_index_dirs(cache_dir);
 
-	/* Remove cache dir from sysfs */
-	kobject_del(cache_dir->kobj);
-
-	kobject_put(cache_dir->kobj);
-
-	kfree(cache_dir);
-}
-
-static void cache_cpu_clear(struct cache *cache, int cpu)
-{
-	while (cache) {
-		struct cache *next = cache->next_local;
-
-		WARN_ONCE(!cpumask_test_cpu(cpu, &cache->shared_cpu_map),
-			  "CPU %i not accounted in %s(%s)\n",
-			  cpu, cache->ofnode->full_name,
-			  cache_type_string(cache));
-
-		cpumask_clear_cpu(cpu, &cache->shared_cpu_map);
-
-		/* Release the cache object if all the cpus using it
-		 * are offline */
-		if (cpumask_empty(&cache->shared_cpu_map))
-			release_cache(cache);
-
-		cache = next;
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		this_leaf->of_node = np;
+		if (cache_node_is_unified(np)) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_UNIFIED, level);
+		} else {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		}
+		np = of_find_next_cache_node(np);
 	}
+	return 0;
 }
 
-void cacheinfo_cpu_offline(unsigned int cpu_id)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-
-	/* Prevent userspace from seeing inconsistent state - remove
-	 * the sysfs hierarchy first */
-	cache_dir = per_cpu(cache_dir_pcpu, cpu_id);
-
-	/* careful, sysfs population may have failed */
-	if (cache_dir)
-		remove_cache_dir(cache_dir);
-
-	per_cpu(cache_dir_pcpu, cpu_id) = NULL;
-
-	/* clear the CPU's bit in its cache chain, possibly freeing
-	 * cache objects */
-	cache = cache_lookup_by_cpu(cpu_id);
-	if (cache)
-		cache_cpu_clear(cache, cpu_id);
-}
-#endif /* (CONFIG_PPC_PSERIES && CONFIG_SUSPEND) || CONFIG_HOTPLUG_CPU */
diff --git a/arch/powerpc/kernel/cacheinfo.h b/arch/powerpc/kernel/cacheinfo.h
deleted file mode 100644
index a7b74d36acd7..000000000000
--- a/arch/powerpc/kernel/cacheinfo.h
+++ /dev/null
@@ -1,8 +0,0 @@
-#ifndef _PPC_CACHEINFO_H
-#define _PPC_CACHEINFO_H
-
-/* These are just hooks for sysfs.c to use. */
-extern void cacheinfo_cpu_online(unsigned int cpu_id);
-extern void cacheinfo_cpu_offline(unsigned int cpu_id);
-
-#endif /* _PPC_CACHEINFO_H */
diff --git a/arch/powerpc/kernel/sysfs.c b/arch/powerpc/kernel/sysfs.c
index 67fd2fd2620a..6e9c5a8141bb 100644
--- a/arch/powerpc/kernel/sysfs.c
+++ b/arch/powerpc/kernel/sysfs.c
@@ -19,8 +19,6 @@
 #include <asm/pmc.h>
 #include <asm/firmware.h>
 
-#include "cacheinfo.h"
-
 #ifdef CONFIG_PPC64
 #include <asm/paca.h>
 #include <asm/lppaca.h>
@@ -743,7 +741,6 @@ static void register_cpu_online(unsigned int cpu)
 		device_create_file(s, &dev_attr_altivec_idle_wait_time);
 	}
 #endif
-	cacheinfo_cpu_online(cpu);
 }
 
 #ifdef CONFIG_HOTPLUG_CPU
@@ -824,7 +821,6 @@ static void unregister_cpu_online(unsigned int cpu)
 		device_remove_file(s, &dev_attr_altivec_idle_wait_time);
 	}
 #endif
-	cacheinfo_cpu_offline(cpu);
 }
 
 #ifdef CONFIG_ARCH_CPU_PROBE_RELEASE
@@ -988,8 +984,7 @@ static int __init topology_init(void)
 	int cpu;
 
 	register_nodes();
-
-	cpu_notifier_register_begin();
+	register_cpu_notifier(&sysfs_cpu_nb);
 
 	for_each_possible_cpu(cpu) {
 		struct cpu *c = &per_cpu(cpu_devices, cpu);
@@ -1013,11 +1008,6 @@ static int __init topology_init(void)
 		if (cpu_online(cpu))
 			register_cpu_online(cpu);
 	}
-
-	__register_cpu_notifier(&sysfs_cpu_nb);
-
-	cpu_notifier_register_done();
-
 #ifdef CONFIG_PPC64
 	sysfs_create_dscr_default();
 #endif /* CONFIG_PPC64 */
-- 
1.8.3.2

[PATCH v3 08/11] powerpc: move cacheinfo sysfs to generic cacheinfo infrastructure

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch removes the redundant sysfs cacheinfo code by making use of
the newly introduced generic cacheinfo infrastructure.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: linuxppc-dev@lists.ozlabs.org
---
 arch/powerpc/kernel/cacheinfo.c | 812 +++++-----------------------------------
 arch/powerpc/kernel/cacheinfo.h |   8 -
 arch/powerpc/kernel/sysfs.c     |  12 +-
 3 files changed, 90 insertions(+), 742 deletions(-)
 delete mode 100644 arch/powerpc/kernel/cacheinfo.h

diff --git a/arch/powerpc/kernel/cacheinfo.c b/arch/powerpc/kernel/cacheinfo.c
index 40198d50b4c2..6845eb7fcc18 100644
--- a/arch/powerpc/kernel/cacheinfo.c
+++ b/arch/powerpc/kernel/cacheinfo.c
@@ -10,38 +10,10 @@
  * 2 as published by the Free Software Foundation.
  */
 
+#include <linux/cacheinfo.h>
 #include <linux/cpu.h>
-#include <linux/cpumask.h>
 #include <linux/kernel.h>
-#include <linux/kobject.h>
-#include <linux/list.h>
-#include <linux/notifier.h>
 #include <linux/of.h>
-#include <linux/percpu.h>
-#include <linux/slab.h>
-#include <asm/prom.h>
-
-#include "cacheinfo.h"
-
-/* per-cpu object for tracking:
- * - a "cache" kobject for the top-level directory
- * - a list of "index" objects representing the cpu's local cache hierarchy
- */
-struct cache_dir {
-	struct kobject *kobj; /* bare (not embedded) kobject for cache
-			       * directory */
-	struct cache_index_dir *index; /* list of index objects */
-};
-
-/* "index" object: each cpu's cache directory has an index
- * subdirectory corresponding to a cache object associated with the
- * cpu.  This object's lifetime is managed via the embedded kobject.
- */
-struct cache_index_dir {
-	struct kobject kobj;
-	struct cache_index_dir *next; /* next index in parent directory */
-	struct cache *cache;
-};
 
 /* Template for determining which OF properties to query for a given
  * cache type */
@@ -60,11 +32,6 @@ struct cache_type_info {
 	const char *nr_sets_prop;
 };
 
-/* These are used to index the cache_type_info array. */
-#define CACHE_TYPE_UNIFIED     0
-#define CACHE_TYPE_INSTRUCTION 1
-#define CACHE_TYPE_DATA        2
-
 static const struct cache_type_info cache_type_info[] = {
 	{
 		/* PowerPC Processor binding says the [di]-cache-*
@@ -92,231 +59,82 @@ static const struct cache_type_info cache_type_info[] = {
 	},
 };
 
-/* Cache object: each instance of this corresponds to a distinct cache
- * in the system.  There are separate objects for Harvard caches: one
- * each for instruction and data, and each refers to the same OF node.
- * The refcount of the OF node is elevated for the lifetime of the
- * cache object.  A cache object is released when its shared_cpu_map
- * is cleared (see cache_cpu_clear).
- *
- * A cache object is on two lists: an unsorted global list
- * (cache_list) of cache objects; and a singly-linked list
- * representing the local cache hierarchy, which is ordered by level
- * (e.g. L1d -> L1i -> L2 -> L3).
- */
-struct cache {
-	struct device_node *ofnode;    /* OF node for this cache, may be cpu */
-	struct cpumask shared_cpu_map; /* online CPUs using this cache */
-	int type;                      /* split cache disambiguation */
-	int level;                     /* level not explicit in device tree */
-	struct list_head list;         /* global list of cache objects */
-	struct cache *next_local;      /* next cache of >= level */
-};
-
-static DEFINE_PER_CPU(struct cache_dir *, cache_dir_pcpu);
-
-/* traversal/modification of this list occurs only at cpu hotplug time;
- * access is serialized by cpu hotplug locking
- */
-static LIST_HEAD(cache_list);
-
-static struct cache_index_dir *kobj_to_cache_index_dir(struct kobject *k)
-{
-	return container_of(k, struct cache_index_dir, kobj);
-}
-
-static const char *cache_type_string(const struct cache *cache)
+static inline int get_cacheinfo_idx(enum cache_type type)
 {
-	return cache_type_info[cache->type].name;
-}
-
-static void cache_init(struct cache *cache, int type, int level,
-		       struct device_node *ofnode)
-{
-	cache->type = type;
-	cache->level = level;
-	cache->ofnode = of_node_get(ofnode);
-	INIT_LIST_HEAD(&cache->list);
-	list_add(&cache->list, &cache_list);
-}
-
-static struct cache *new_cache(int type, int level, struct device_node *ofnode)
-{
-	struct cache *cache;
-
-	cache = kzalloc(sizeof(*cache), GFP_KERNEL);
-	if (cache)
-		cache_init(cache, type, level, ofnode);
-
-	return cache;
-}
-
-static void release_cache_debugcheck(struct cache *cache)
-{
-	struct cache *iter;
-
-	list_for_each_entry(iter, &cache_list, list)
-		WARN_ONCE(iter->next_local == cache,
-			  "cache for %s(%s) refers to cache for %s(%s)\n",
-			  iter->ofnode->full_name,
-			  cache_type_string(iter),
-			  cache->ofnode->full_name,
-			  cache_type_string(cache));
-}
-
-static void release_cache(struct cache *cache)
-{
-	if (!cache)
-		return;
-
-	pr_debug("freeing L%d %s cache for %s\n", cache->level,
-		 cache_type_string(cache), cache->ofnode->full_name);
-
-	release_cache_debugcheck(cache);
-	list_del(&cache->list);
-	of_node_put(cache->ofnode);
-	kfree(cache);
-}
-
-static void cache_cpu_set(struct cache *cache, int cpu)
-{
-	struct cache *next = cache;
-
-	while (next) {
-		WARN_ONCE(cpumask_test_cpu(cpu, &next->shared_cpu_map),
-			  "CPU %i already accounted in %s(%s)\n",
-			  cpu, next->ofnode->full_name,
-			  cache_type_string(next));
-		cpumask_set_cpu(cpu, &next->shared_cpu_map);
-		next = next->next_local;
-	}
+	if (type == CACHE_TYPE_UNIFIED)
+		return 0;
+	return type;
 }
 
-static int cache_size(const struct cache *cache, unsigned int *ret)
+static void cache_size(struct cacheinfo *this_leaf)
 {
 	const char *propname;
 	const __be32 *cache_size;
+	int ct_idx;
 
-	propname = cache_type_info[cache->type].size_prop;
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	propname = cache_type_info[ct_idx].size_prop;
 
-	cache_size = of_get_property(cache->ofnode, propname, NULL);
+	cache_size = of_get_property(this_leaf->of_node, propname, NULL);
 	if (!cache_size)
-		return -ENODEV;
-
-	*ret = of_read_number(cache_size, 1);
-	return 0;
-}
-
-static int cache_size_kb(const struct cache *cache, unsigned int *ret)
-{
-	unsigned int size;
-
-	if (cache_size(cache, &size))
-		return -ENODEV;
-
-	*ret = size / 1024;
-	return 0;
+		this_leaf->size = 0;
+	else
+		this_leaf->size = of_read_number(cache_size, 1);
 }
 
 /* not cache_line_size() because that's a macro in include/linux/cache.h */
-static int cache_get_line_size(const struct cache *cache, unsigned int *ret)
+static void cache_get_line_size(struct cacheinfo *this_leaf)
 {
 	const __be32 *line_size;
-	int i, lim;
+	int i, lim, ct_idx;
 
-	lim = ARRAY_SIZE(cache_type_info[cache->type].line_size_props);
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	lim = ARRAY_SIZE(cache_type_info[ct_idx].line_size_props);
 
 	for (i = 0; i < lim; i++) {
 		const char *propname;
 
-		propname = cache_type_info[cache->type].line_size_props[i];
-		line_size = of_get_property(cache->ofnode, propname, NULL);
+		propname = cache_type_info[ct_idx].line_size_props[i];
+		line_size = of_get_property(this_leaf->of_node, propname, NULL);
 		if (line_size)
 			break;
 	}
 
 	if (!line_size)
-		return -ENODEV;
-
-	*ret = of_read_number(line_size, 1);
-	return 0;
+		this_leaf->coherency_line_size = 0;
+	else
+		this_leaf->coherency_line_size = of_read_number(line_size, 1);
 }
 
-static int cache_nr_sets(const struct cache *cache, unsigned int *ret)
+static void cache_nr_sets(struct cacheinfo *this_leaf)
 {
 	const char *propname;
 	const __be32 *nr_sets;
+	int ct_idx;
 
-	propname = cache_type_info[cache->type].nr_sets_prop;
+	ct_idx = get_cacheinfo_idx(this_leaf->type);
+	propname = cache_type_info[ct_idx].nr_sets_prop;
 
-	nr_sets = of_get_property(cache->ofnode, propname, NULL);
+	nr_sets = of_get_property(this_leaf->of_node, propname, NULL);
 	if (!nr_sets)
-		return -ENODEV;
-
-	*ret = of_read_number(nr_sets, 1);
-	return 0;
+		this_leaf->number_of_sets = 0;
+	else
+		this_leaf->number_of_sets = of_read_number(nr_sets, 1);
 }
 
-static int cache_associativity(const struct cache *cache, unsigned int *ret)
+static void cache_associativity(struct cacheinfo *this_leaf)
 {
-	unsigned int line_size;
-	unsigned int nr_sets;
-	unsigned int size;
-
-	if (cache_nr_sets(cache, &nr_sets))
-		goto err;
+	unsigned int line_size = this_leaf->coherency_line_size;
+	unsigned int nr_sets = this_leaf->number_of_sets;
+	unsigned int size = this_leaf->size;
 
 	/* If the cache is fully associative, there is no need to
 	 * check the other properties.
 	 */
-	if (nr_sets == 1) {
-		*ret = 0;
-		return 0;
-	}
-
-	if (cache_get_line_size(cache, &line_size))
-		goto err;
-	if (cache_size(cache, &size))
-		goto err;
-
-	if (!(nr_sets > 0 && size > 0 && line_size > 0))
-		goto err;
-
-	*ret = (size / nr_sets) / line_size;
-	return 0;
-err:
-	return -ENODEV;
-}
-
-/* helper for dealing with split caches */
-static struct cache *cache_find_first_sibling(struct cache *cache)
-{
-	struct cache *iter;
-
-	if (cache->type == CACHE_TYPE_UNIFIED)
-		return cache;
-
-	list_for_each_entry(iter, &cache_list, list)
-		if (iter->ofnode == cache->ofnode && iter->next_local == cache)
-			return iter;
-
-	return cache;
-}
-
-/* return the first cache on a local list matching node */
-static struct cache *cache_lookup_by_node(const struct device_node *node)
-{
-	struct cache *cache = NULL;
-	struct cache *iter;
-
-	list_for_each_entry(iter, &cache_list, list) {
-		if (iter->ofnode != node)
-			continue;
-		cache = cache_find_first_sibling(iter);
-		break;
-	}
-
-	return cache;
+	if ((nr_sets == 1) || !(nr_sets > 0 && size > 0 && line_size > 0))
+		this_leaf->ways_of_associativity = 0;
+	else
+		this_leaf->ways_of_associativity = (size / nr_sets) / line_size;
 }
 
 static bool cache_node_is_unified(const struct device_node *np)
@@ -324,526 +142,74 @@ static bool cache_node_is_unified(const struct device_node *np)
 	return of_get_property(np, "cache-unified", NULL);
 }
 
-static struct cache *cache_do_one_devnode_unified(struct device_node *node,
-						  int level)
-{
-	struct cache *cache;
-
-	pr_debug("creating L%d ucache for %s\n", level, node->full_name);
-
-	cache = new_cache(CACHE_TYPE_UNIFIED, level, node);
-
-	return cache;
-}
-
-static struct cache *cache_do_one_devnode_split(struct device_node *node,
-						int level)
-{
-	struct cache *dcache, *icache;
-
-	pr_debug("creating L%d dcache and icache for %s\n", level,
-		 node->full_name);
-
-	dcache = new_cache(CACHE_TYPE_DATA, level, node);
-	icache = new_cache(CACHE_TYPE_INSTRUCTION, level, node);
-
-	if (!dcache || !icache)
-		goto err;
-
-	dcache->next_local = icache;
-
-	return dcache;
-err:
-	release_cache(dcache);
-	release_cache(icache);
-	return NULL;
-}
-
-static struct cache *cache_do_one_devnode(struct device_node *node, int level)
-{
-	struct cache *cache;
-
-	if (cache_node_is_unified(node))
-		cache = cache_do_one_devnode_unified(node, level);
-	else
-		cache = cache_do_one_devnode_split(node, level);
-
-	return cache;
-}
-
-static struct cache *cache_lookup_or_instantiate(struct device_node *node,
-						 int level)
-{
-	struct cache *cache;
-
-	cache = cache_lookup_by_node(node);
-
-	WARN_ONCE(cache && cache->level != level,
-		  "cache level mismatch on lookup (got %d, expected %d)\n",
-		  cache->level, level);
-
-	if (!cache)
-		cache = cache_do_one_devnode(node, level);
-
-	return cache;
-}
-
-static void link_cache_lists(struct cache *smaller, struct cache *bigger)
-{
-	while (smaller->next_local) {
-		if (smaller->next_local == bigger)
-			return; /* already linked */
-		smaller = smaller->next_local;
-	}
-
-	smaller->next_local = bigger;
-}
-
-static void do_subsidiary_caches_debugcheck(struct cache *cache)
-{
-	WARN_ON_ONCE(cache->level != 1);
-	WARN_ON_ONCE(strcmp(cache->ofnode->type, "cpu"));
-}
-
-static void do_subsidiary_caches(struct cache *cache)
-{
-	struct device_node *subcache_node;
-	int level = cache->level;
-
-	do_subsidiary_caches_debugcheck(cache);
-
-	while ((subcache_node = of_find_next_cache_node(cache->ofnode))) {
-		struct cache *subcache;
-
-		level++;
-		subcache = cache_lookup_or_instantiate(subcache_node, level);
-		of_node_put(subcache_node);
-		if (!subcache)
-			break;
-
-		link_cache_lists(cache, subcache);
-		cache = subcache;
-	}
-}
-
-static struct cache *cache_chain_instantiate(unsigned int cpu_id)
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
 {
-	struct device_node *cpu_node;
-	struct cache *cpu_cache = NULL;
-
-	pr_debug("creating cache object(s) for CPU %i\n", cpu_id);
-
-	cpu_node = of_get_cpu_node(cpu_id, NULL);
-	WARN_ONCE(!cpu_node, "no OF node found for CPU %i\n", cpu_id);
-	if (!cpu_node)
-		goto out;
-
-	cpu_cache = cache_lookup_or_instantiate(cpu_node, 1);
-	if (!cpu_cache)
-		goto out;
-
-	do_subsidiary_caches(cpu_cache);
-
-	cache_cpu_set(cpu_cache, cpu_id);
-out:
-	of_node_put(cpu_node);
-
-	return cpu_cache;
+	this_leaf->level = level;
+	this_leaf->type = type;
+	cache_size(this_leaf);
+	cache_get_line_size(this_leaf);
+	cache_nr_sets(this_leaf);
+	cache_associativity(this_leaf);
 }
 
-static struct cache_dir *cacheinfo_create_cache_dir(unsigned int cpu_id)
+int init_cache_level(unsigned int cpu)
 {
-	struct cache_dir *cache_dir;
-	struct device *dev;
-	struct kobject *kobj = NULL;
-
-	dev = get_cpu_device(cpu_id);
-	WARN_ONCE(!dev, "no dev for CPU %i\n", cpu_id);
-	if (!dev)
-		goto err;
-
-	kobj = kobject_create_and_add("cache", &dev->kobj);
-	if (!kobj)
-		goto err;
-
-	cache_dir = kzalloc(sizeof(*cache_dir), GFP_KERNEL);
-	if (!cache_dir)
-		goto err;
-
-	cache_dir->kobj = kobj;
-
-	WARN_ON_ONCE(per_cpu(cache_dir_pcpu, cpu_id) != NULL);
-
-	per_cpu(cache_dir_pcpu, cpu_id) = cache_dir;
+	struct device_node *np;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	unsigned int level = 0, leaves = 0;
 
-	return cache_dir;
-err:
-	kobject_put(kobj);
-	return NULL;
-}
-
-static void cache_index_release(struct kobject *kobj)
-{
-	struct cache_index_dir *index;
-
-	index = kobj_to_cache_index_dir(kobj);
-
-	pr_debug("freeing index directory for L%d %s cache\n",
-		 index->cache->level, cache_type_string(index->cache));
-
-	kfree(index);
-}
-
-static ssize_t cache_index_show(struct kobject *k, struct attribute *attr, char *buf)
-{
-	struct kobj_attribute *kobj_attr;
-
-	kobj_attr = container_of(attr, struct kobj_attribute, attr);
-
-	return kobj_attr->show(k, kobj_attr, buf);
-}
-
-static struct cache *index_kobj_to_cache(struct kobject *k)
-{
-	struct cache_index_dir *index;
-
-	index = kobj_to_cache_index_dir(k);
-
-	return index->cache;
-}
-
-static ssize_t size_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int size_kb;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_size_kb(cache, &size_kb))
-		return -ENODEV;
-
-	return sprintf(buf, "%uK\n", size_kb);
-}
-
-static struct kobj_attribute cache_size_attr =
-	__ATTR(size, 0444, size_show, NULL);
-
-
-static ssize_t line_size_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int line_size;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_get_line_size(cache, &line_size))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", line_size);
-}
-
-static struct kobj_attribute cache_line_size_attr =
-	__ATTR(coherency_line_size, 0444, line_size_show, NULL);
-
-static ssize_t nr_sets_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int nr_sets;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_nr_sets(cache, &nr_sets))
+	if (!cpu_dev) {
+		pr_err("No cpu device for CPU %d\n", cpu);
 		return -ENODEV;
-
-	return sprintf(buf, "%u\n", nr_sets);
-}
-
-static struct kobj_attribute cache_nr_sets_attr =
-	__ATTR(number_of_sets, 0444, nr_sets_show, NULL);
-
-static ssize_t associativity_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	unsigned int associativity;
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	if (cache_associativity(cache, &associativity))
-		return -ENODEV;
-
-	return sprintf(buf, "%u\n", associativity);
-}
-
-static struct kobj_attribute cache_assoc_attr =
-	__ATTR(ways_of_associativity, 0444, associativity_show, NULL);
-
-static ssize_t type_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache *cache;
-
-	cache = index_kobj_to_cache(k);
-
-	return sprintf(buf, "%s\n", cache_type_string(cache));
-}
-
-static struct kobj_attribute cache_type_attr =
-	__ATTR(type, 0444, type_show, NULL);
-
-static ssize_t level_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache_index_dir *index;
-	struct cache *cache;
-
-	index = kobj_to_cache_index_dir(k);
-	cache = index->cache;
-
-	return sprintf(buf, "%d\n", cache->level);
-}
-
-static struct kobj_attribute cache_level_attr =
-	__ATTR(level, 0444, level_show, NULL);
-
-static ssize_t shared_cpu_map_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
-	struct cache_index_dir *index;
-	struct cache *cache;
-	int len;
-	int n = 0;
-
-	index = kobj_to_cache_index_dir(k);
-	cache = index->cache;
-	len = PAGE_SIZE - 2;
-
-	if (len > 1) {
-		n = cpumask_scnprintf(buf, len, &cache->shared_cpu_map);
-		buf[n++] = '\n';
-		buf[n] = '\0';
 	}
-	return n;
-}
-
-static struct kobj_attribute cache_shared_cpu_map_attr =
-	__ATTR(shared_cpu_map, 0444, shared_cpu_map_show, NULL);
-
-/* Attributes which should always be created -- the kobject/sysfs core
- * does this automatically via kobj_type->default_attrs.  This is the
- * minimum data required to uniquely identify a cache.
- */
-static struct attribute *cache_index_default_attrs[] = {
-	&cache_type_attr.attr,
-	&cache_level_attr.attr,
-	&cache_shared_cpu_map_attr.attr,
-	NULL,
-};
-
-/* Attributes which should be created if the cache device node has the
- * right properties -- see cacheinfo_create_index_opt_attrs
- */
-static struct kobj_attribute *cache_index_opt_attrs[] = {
-	&cache_size_attr,
-	&cache_line_size_attr,
-	&cache_nr_sets_attr,
-	&cache_assoc_attr,
-};
-
-static const struct sysfs_ops cache_index_ops = {
-	.show = cache_index_show,
-};
-
-static struct kobj_type cache_index_type = {
-	.release = cache_index_release,
-	.sysfs_ops = &cache_index_ops,
-	.default_attrs = cache_index_default_attrs,
-};
-
-static void cacheinfo_create_index_opt_attrs(struct cache_index_dir *dir)
-{
-	const char *cache_name;
-	const char *cache_type;
-	struct cache *cache;
-	char *buf;
-	int i;
-
-	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
-	if (!buf)
-		return;
-
-	cache = dir->cache;
-	cache_name = cache->ofnode->full_name;
-	cache_type = cache_type_string(cache);
-
-	/* We don't want to create an attribute that can't provide a
-	 * meaningful value.  Check the return value of each optional
-	 * attribute's ->show method before registering the
-	 * attribute.
-	 */
-	for (i = 0; i < ARRAY_SIZE(cache_index_opt_attrs); i++) {
-		struct kobj_attribute *attr;
-		ssize_t rc;
-
-		attr = cache_index_opt_attrs[i];
-
-		rc = attr->show(&dir->kobj, attr, buf);
-		if (rc <= 0) {
-			pr_debug("not creating %s attribute for "
-				 "%s(%s) (rc = %zd)\n",
-				 attr->attr.name, cache_name,
-				 cache_type, rc);
-			continue;
-		}
-		if (sysfs_create_file(&dir->kobj, &attr->attr))
-			pr_debug("could not create %s attribute for %s(%s)\n",
-				 attr->attr.name, cache_name, cache_type);
+	np = cpu_dev->of_node;
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
 	}
 
-	kfree(buf);
-}
-
-static void cacheinfo_create_index_dir(struct cache *cache, int index,
-				       struct cache_dir *cache_dir)
-{
-	struct cache_index_dir *index_dir;
-	int rc;
-
-	index_dir = kzalloc(sizeof(*index_dir), GFP_KERNEL);
-	if (!index_dir)
-		goto err;
-
-	index_dir->cache = cache;
-
-	rc = kobject_init_and_add(&index_dir->kobj, &cache_index_type,
-				  cache_dir->kobj, "index%d", index);
-	if (rc)
-		goto err;
-
-	index_dir->next = cache_dir->index;
-	cache_dir->index = index_dir;
-
-	cacheinfo_create_index_opt_attrs(index_dir);
-
-	return;
-err:
-	kfree(index_dir);
-}
-
-static void cacheinfo_sysfs_populate(unsigned int cpu_id,
-				     struct cache *cache_list)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-	int index = 0;
-
-	cache_dir = cacheinfo_create_cache_dir(cpu_id);
-	if (!cache_dir)
-		return;
-
-	cache = cache_list;
-	while (cache) {
-		cacheinfo_create_index_dir(cache, index, cache_dir);
-		index++;
-		cache = cache->next_local;
+	while (np) {
+		leaves += cache_node_is_unified(np) ? 1 : 2;
+		level++;
+		of_node_put(np);
+		np = of_find_next_cache_node(np);
 	}
-}
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
 
-void cacheinfo_cpu_online(unsigned int cpu_id)
-{
-	struct cache *cache;
-
-	cache = cache_chain_instantiate(cpu_id);
-	if (!cache)
-		return;
-
-	cacheinfo_sysfs_populate(cpu_id, cache);
-}
-
-/* functions needed to remove cache entry for cpu offline or suspend/resume */
-
-#if (defined(CONFIG_PPC_PSERIES) && defined(CONFIG_SUSPEND)) || \
-    defined(CONFIG_HOTPLUG_CPU)
-
-static struct cache *cache_lookup_by_cpu(unsigned int cpu_id)
-{
-	struct device_node *cpu_node;
-	struct cache *cache;
-
-	cpu_node = of_get_cpu_node(cpu_id, NULL);
-	WARN_ONCE(!cpu_node, "no OF node found for CPU %i\n", cpu_id);
-	if (!cpu_node)
-		return NULL;
-
-	cache = cache_lookup_by_node(cpu_node);
-	of_node_put(cpu_node);
-
-	return cache;
+	return 0;
 }
 
-static void remove_index_dirs(struct cache_dir *cache_dir)
+int populate_cache_leaves(unsigned int cpu)
 {
-	struct cache_index_dir *index;
-
-	index = cache_dir->index;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+	struct device *cpu_dev = get_cpu_device(cpu);
+	struct device_node *np;
+	unsigned int level, idx;
 
-	while (index) {
-		struct cache_index_dir *next;
-
-		next = index->next;
-		kobject_put(&index->kobj);
-		index = next;
+	np = of_node_get(cpu_dev->of_node);
+	if (!np) {
+		pr_err("Failed to find cpu%d device node\n", cpu);
+		return -ENOENT;
 	}
-}
-
-static void remove_cache_dir(struct cache_dir *cache_dir)
-{
-	remove_index_dirs(cache_dir);
 
-	/* Remove cache dir from sysfs */
-	kobject_del(cache_dir->kobj);
-
-	kobject_put(cache_dir->kobj);
-
-	kfree(cache_dir);
-}
-
-static void cache_cpu_clear(struct cache *cache, int cpu)
-{
-	while (cache) {
-		struct cache *next = cache->next_local;
-
-		WARN_ONCE(!cpumask_test_cpu(cpu, &cache->shared_cpu_map),
-			  "CPU %i not accounted in %s(%s)\n",
-			  cpu, cache->ofnode->full_name,
-			  cache_type_string(cache));
-
-		cpumask_clear_cpu(cpu, &cache->shared_cpu_map);
-
-		/* Release the cache object if all the cpus using it
-		 * are offline */
-		if (cpumask_empty(&cache->shared_cpu_map))
-			release_cache(cache);
-
-		cache = next;
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		if (!this_leaf)
+			return -EINVAL;
+
+		this_leaf->of_node = np;
+		if (cache_node_is_unified(np)) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_UNIFIED, level);
+		} else {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		}
+		np = of_find_next_cache_node(np);
 	}
+	return 0;
 }
 
-void cacheinfo_cpu_offline(unsigned int cpu_id)
-{
-	struct cache_dir *cache_dir;
-	struct cache *cache;
-
-	/* Prevent userspace from seeing inconsistent state - remove
-	 * the sysfs hierarchy first */
-	cache_dir = per_cpu(cache_dir_pcpu, cpu_id);
-
-	/* careful, sysfs population may have failed */
-	if (cache_dir)
-		remove_cache_dir(cache_dir);
-
-	per_cpu(cache_dir_pcpu, cpu_id) = NULL;
-
-	/* clear the CPU's bit in its cache chain, possibly freeing
-	 * cache objects */
-	cache = cache_lookup_by_cpu(cpu_id);
-	if (cache)
-		cache_cpu_clear(cache, cpu_id);
-}
-#endif /* (CONFIG_PPC_PSERIES && CONFIG_SUSPEND) || CONFIG_HOTPLUG_CPU */
diff --git a/arch/powerpc/kernel/cacheinfo.h b/arch/powerpc/kernel/cacheinfo.h
deleted file mode 100644
index a7b74d36acd7..000000000000
--- a/arch/powerpc/kernel/cacheinfo.h
+++ /dev/null
@@ -1,8 +0,0 @@
-#ifndef _PPC_CACHEINFO_H
-#define _PPC_CACHEINFO_H
-
-/* These are just hooks for sysfs.c to use. */
-extern void cacheinfo_cpu_online(unsigned int cpu_id);
-extern void cacheinfo_cpu_offline(unsigned int cpu_id);
-
-#endif /* _PPC_CACHEINFO_H */
diff --git a/arch/powerpc/kernel/sysfs.c b/arch/powerpc/kernel/sysfs.c
index 67fd2fd2620a..6e9c5a8141bb 100644
--- a/arch/powerpc/kernel/sysfs.c
+++ b/arch/powerpc/kernel/sysfs.c
@@ -19,8 +19,6 @@
 #include <asm/pmc.h>
 #include <asm/firmware.h>
 
-#include "cacheinfo.h"
-
 #ifdef CONFIG_PPC64
 #include <asm/paca.h>
 #include <asm/lppaca.h>
@@ -743,7 +741,6 @@ static void register_cpu_online(unsigned int cpu)
 		device_create_file(s, &dev_attr_altivec_idle_wait_time);
 	}
 #endif
-	cacheinfo_cpu_online(cpu);
 }
 
 #ifdef CONFIG_HOTPLUG_CPU
@@ -824,7 +821,6 @@ static void unregister_cpu_online(unsigned int cpu)
 		device_remove_file(s, &dev_attr_altivec_idle_wait_time);
 	}
 #endif
-	cacheinfo_cpu_offline(cpu);
 }
 
 #ifdef CONFIG_ARCH_CPU_PROBE_RELEASE
@@ -988,8 +984,7 @@ static int __init topology_init(void)
 	int cpu;
 
 	register_nodes();
-
-	cpu_notifier_register_begin();
+	register_cpu_notifier(&sysfs_cpu_nb);
 
 	for_each_possible_cpu(cpu) {
 		struct cpu *c = &per_cpu(cpu_devices, cpu);
@@ -1013,11 +1008,6 @@ static int __init topology_init(void)
 		if (cpu_online(cpu))
 			register_cpu_online(cpu);
 	}
-
-	__register_cpu_notifier(&sysfs_cpu_nb);
-
-	cpu_notifier_register_done();
-
 #ifdef CONFIG_PPC64
 	sysfs_create_dscr_default();
 #endif /* CONFIG_PPC64 */
-- 
1.8.3.2

[PATCH v3 09/11] ARM64: kernel: add support for cpu cache information

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds support for cacheinfo on ARM64.

On ARMv8, the cache hierarchy can be identified through Cache Level ID
(CLIDR) register while the cache geometry is provided by Cache Size ID
(CCSIDR) register.

Since the architecture doesn't provide any way of detecting the cpus
sharing particular cache, device tree is used for the same purpose.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
---
 arch/arm64/kernel/Makefile    |   2 +-
 arch/arm64/kernel/cacheinfo.c | 142 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 143 insertions(+), 1 deletion(-)
 create mode 100644 arch/arm64/kernel/cacheinfo.c

diff --git a/arch/arm64/kernel/Makefile b/arch/arm64/kernel/Makefile
index df7ef8768fc2..285cd88c1e37 100644
--- a/arch/arm64/kernel/Makefile
+++ b/arch/arm64/kernel/Makefile
@@ -15,7 +15,7 @@ arm64-obj-y		:= cputable.o debug-monitors.o entry.o irq.o fpsimd.o	\
 			   entry-fpsimd.o process.o ptrace.o setup.o signal.o	\
 			   sys.o stacktrace.o time.o traps.o io.o vdso.o	\
 			   hyp-stub.o psci.o cpu_ops.o insn.o return_address.o	\
-			   cpuinfo.o
+			   cpuinfo.o cacheinfo.o
 
 arm64-obj-$(CONFIG_COMPAT)		+= sys32.o kuser32.o signal32.o 	\
 					   sys_compat.o
diff --git a/arch/arm64/kernel/cacheinfo.c b/arch/arm64/kernel/cacheinfo.c
new file mode 100644
index 000000000000..a9cbf3b40a1f
--- /dev/null
+++ b/arch/arm64/kernel/cacheinfo.c
@@ -0,0 +1,142 @@
+/*
+ *  ARM64 cacheinfo support
+ *
+ *  Copyright (C) 2014 ARM Ltd.
+ *  All Rights Reserved
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/cpu.h>
+#include <linux/compiler.h>
+#include <linux/of.h>
+
+#include <asm/processor.h>
+
+#define MAX_CACHE_LEVEL			7	/* Max 7 level supported */
+/* Ctypen, bits[3(n - 1) + 2 : 3(n - 1)], for n = 1 to 7 */
+#define CLIDR_CTYPE_SHIFT(level)	(3 * (level - 1))
+#define CLIDR_CTYPE_MASK(level)		(7 << CLIDR_CTYPE_SHIFT(level))
+#define CLIDR_CTYPE(clidr, level)	\
+	(((clidr) & CLIDR_CTYPE_MASK(level)) >> CLIDR_CTYPE_SHIFT(level))
+
+static inline enum cache_type get_cache_type(int level)
+{
+	u64 clidr;
+
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	asm volatile ("mrs     %x0, clidr_el1" : "=r" (clidr));
+	return CLIDR_CTYPE(clidr, level);
+}
+
+/*
+ * NumSets, bits[27:13] - (Number of sets in cache) - 1
+ * Associativity, bits[12:3] - (Associativity of cache) - 1
+ * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
+ */
+#define CCSIDR_WRITE_THROUGH		BIT(31)
+#define CCSIDR_WRITE_BACK		BIT(30)
+#define CCSIDR_READ_ALLOCATE		BIT(29)
+#define CCSIDR_WRITE_ALLOCATE		BIT(28)
+#define CCSIDR_LINESIZE_MASK		0x7
+#define CCSIDR_ASSOCIATIVITY_SHIFT	3
+#define CCSIDR_ASSOCIATIVITY_MASK	0x3FF
+#define CCSIDR_NUMSETS_SHIFT		13
+#define CCSIDR_NUMSETS_MASK		0x7FF
+
+/*
+ * Which cache CCSIDR represents depends on CSSELR value
+ * Make sure no one else changes CSSELR during this
+ * smp_call_function_single prevents preemption for us
+ */
+static inline u32 get_ccsidr(u64 csselr)
+{
+	u64 ccsidr;
+
+	/* Put value into CSSELR */
+	asm volatile("msr csselr_el1, %x0" : : "r" (csselr));
+	isb();
+	/* Read result out of CCSIDR */
+	asm volatile("mrs %x0, ccsidr_el1" : "=r" (ccsidr));
+
+	return (u32)ccsidr;
+}
+
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
+{
+	bool is_instr_cache = type & CACHE_TYPE_INST;
+	u32 tmp = get_ccsidr((level - 1) << 1 | is_instr_cache);
+
+	this_leaf->level = level;
+	this_leaf->type = type;
+	this_leaf->coherency_line_size =
+	    (1 << ((tmp & CCSIDR_LINESIZE_MASK) + 2)) * 4;
+	this_leaf->number_of_sets =
+	    ((tmp >> CCSIDR_NUMSETS_SHIFT) & CCSIDR_NUMSETS_MASK) + 1;
+	this_leaf->ways_of_associativity = ((tmp >> CCSIDR_ASSOCIATIVITY_SHIFT)
+					    & CCSIDR_ASSOCIATIVITY_MASK) + 1;
+	this_leaf->size = this_leaf->number_of_sets *
+	    this_leaf->coherency_line_size * this_leaf->ways_of_associativity;
+	this_leaf->attributes =
+		((tmp & CCSIDR_WRITE_THROUGH) ? CACHE_WRITE_THROUGH : 0) |
+		((tmp & CCSIDR_WRITE_BACK) ? CACHE_WRITE_BACK : 0) |
+		((tmp & CCSIDR_READ_ALLOCATE) ? CACHE_READ_ALLOCATE : 0) |
+		((tmp & CCSIDR_WRITE_ALLOCATE) ? CACHE_WRITE_ALLOCATE : 0);
+}
+
+static int __init_cache_level(unsigned int cpu)
+{
+	unsigned int ctype, level, leaves;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	for (level = 1, leaves = 0; level <= MAX_CACHE_LEVEL; level++) {
+		ctype = get_cache_type(level);
+		if (ctype == CACHE_TYPE_NOCACHE) {
+			level--;
+			break;
+		}
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	}
+
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
+	return 0;
+}
+
+static int __populate_cache_leaves(unsigned int cpu)
+{
+	unsigned int level, idx;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		type = get_cache_type(level);
+		if (type == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, type, level);
+		}
+	}
+	return 0;
+}
+
+DEFINE_SMP_CALL_CACHE_FUNCTION(init_cache_level)
+DEFINE_SMP_CALL_CACHE_FUNCTION(populate_cache_leaves)
-- 
1.8.3.2

[PATCH v3 09/11] ARM64: kernel: add support for cpu cache information

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds support for cacheinfo on ARM64.

On ARMv8, the cache hierarchy can be identified through Cache Level ID
(CLIDR) register while the cache geometry is provided by Cache Size ID
(CCSIDR) register.

Since the architecture doesn't provide any way of detecting the cpus
sharing particular cache, device tree is used for the same purpose.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel at lists.infradead.org
---
 arch/arm64/kernel/Makefile    |   2 +-
 arch/arm64/kernel/cacheinfo.c | 142 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 143 insertions(+), 1 deletion(-)
 create mode 100644 arch/arm64/kernel/cacheinfo.c

diff --git a/arch/arm64/kernel/Makefile b/arch/arm64/kernel/Makefile
index df7ef8768fc2..285cd88c1e37 100644
--- a/arch/arm64/kernel/Makefile
+++ b/arch/arm64/kernel/Makefile
@@ -15,7 +15,7 @@ arm64-obj-y		:= cputable.o debug-monitors.o entry.o irq.o fpsimd.o	\
 			   entry-fpsimd.o process.o ptrace.o setup.o signal.o	\
 			   sys.o stacktrace.o time.o traps.o io.o vdso.o	\
 			   hyp-stub.o psci.o cpu_ops.o insn.o return_address.o	\
-			   cpuinfo.o
+			   cpuinfo.o cacheinfo.o
 
 arm64-obj-$(CONFIG_COMPAT)		+= sys32.o kuser32.o signal32.o 	\
 					   sys_compat.o
diff --git a/arch/arm64/kernel/cacheinfo.c b/arch/arm64/kernel/cacheinfo.c
new file mode 100644
index 000000000000..a9cbf3b40a1f
--- /dev/null
+++ b/arch/arm64/kernel/cacheinfo.c
@@ -0,0 +1,142 @@
+/*
+ *  ARM64 cacheinfo support
+ *
+ *  Copyright (C) 2014 ARM Ltd.
+ *  All Rights Reserved
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/cpu.h>
+#include <linux/compiler.h>
+#include <linux/of.h>
+
+#include <asm/processor.h>
+
+#define MAX_CACHE_LEVEL			7	/* Max 7 level supported */
+/* Ctypen, bits[3(n - 1) + 2 : 3(n - 1)], for n = 1 to 7 */
+#define CLIDR_CTYPE_SHIFT(level)	(3 * (level - 1))
+#define CLIDR_CTYPE_MASK(level)		(7 << CLIDR_CTYPE_SHIFT(level))
+#define CLIDR_CTYPE(clidr, level)	\
+	(((clidr) & CLIDR_CTYPE_MASK(level)) >> CLIDR_CTYPE_SHIFT(level))
+
+static inline enum cache_type get_cache_type(int level)
+{
+	u64 clidr;
+
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	asm volatile ("mrs     %x0, clidr_el1" : "=r" (clidr));
+	return CLIDR_CTYPE(clidr, level);
+}
+
+/*
+ * NumSets, bits[27:13] - (Number of sets in cache) - 1
+ * Associativity, bits[12:3] - (Associativity of cache) - 1
+ * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
+ */
+#define CCSIDR_WRITE_THROUGH		BIT(31)
+#define CCSIDR_WRITE_BACK		BIT(30)
+#define CCSIDR_READ_ALLOCATE		BIT(29)
+#define CCSIDR_WRITE_ALLOCATE		BIT(28)
+#define CCSIDR_LINESIZE_MASK		0x7
+#define CCSIDR_ASSOCIATIVITY_SHIFT	3
+#define CCSIDR_ASSOCIATIVITY_MASK	0x3FF
+#define CCSIDR_NUMSETS_SHIFT		13
+#define CCSIDR_NUMSETS_MASK		0x7FF
+
+/*
+ * Which cache CCSIDR represents depends on CSSELR value
+ * Make sure no one else changes CSSELR during this
+ * smp_call_function_single prevents preemption for us
+ */
+static inline u32 get_ccsidr(u64 csselr)
+{
+	u64 ccsidr;
+
+	/* Put value into CSSELR */
+	asm volatile("msr csselr_el1, %x0" : : "r" (csselr));
+	isb();
+	/* Read result out of CCSIDR */
+	asm volatile("mrs %x0, ccsidr_el1" : "=r" (ccsidr));
+
+	return (u32)ccsidr;
+}
+
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
+{
+	bool is_instr_cache = type & CACHE_TYPE_INST;
+	u32 tmp = get_ccsidr((level - 1) << 1 | is_instr_cache);
+
+	this_leaf->level = level;
+	this_leaf->type = type;
+	this_leaf->coherency_line_size =
+	    (1 << ((tmp & CCSIDR_LINESIZE_MASK) + 2)) * 4;
+	this_leaf->number_of_sets =
+	    ((tmp >> CCSIDR_NUMSETS_SHIFT) & CCSIDR_NUMSETS_MASK) + 1;
+	this_leaf->ways_of_associativity = ((tmp >> CCSIDR_ASSOCIATIVITY_SHIFT)
+					    & CCSIDR_ASSOCIATIVITY_MASK) + 1;
+	this_leaf->size = this_leaf->number_of_sets *
+	    this_leaf->coherency_line_size * this_leaf->ways_of_associativity;
+	this_leaf->attributes =
+		((tmp & CCSIDR_WRITE_THROUGH) ? CACHE_WRITE_THROUGH : 0) |
+		((tmp & CCSIDR_WRITE_BACK) ? CACHE_WRITE_BACK : 0) |
+		((tmp & CCSIDR_READ_ALLOCATE) ? CACHE_READ_ALLOCATE : 0) |
+		((tmp & CCSIDR_WRITE_ALLOCATE) ? CACHE_WRITE_ALLOCATE : 0);
+}
+
+static int __init_cache_level(unsigned int cpu)
+{
+	unsigned int ctype, level, leaves;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	for (level = 1, leaves = 0; level <= MAX_CACHE_LEVEL; level++) {
+		ctype = get_cache_type(level);
+		if (ctype == CACHE_TYPE_NOCACHE) {
+			level--;
+			break;
+		}
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	}
+
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
+	return 0;
+}
+
+static int __populate_cache_leaves(unsigned int cpu)
+{
+	unsigned int level, idx;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		type = get_cache_type(level);
+		if (type == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, type, level);
+		}
+	}
+	return 0;
+}
+
+DEFINE_SMP_CALL_CACHE_FUNCTION(init_cache_level)
+DEFINE_SMP_CALL_CACHE_FUNCTION(populate_cache_leaves)
-- 
1.8.3.2

[PATCH v3 10/11] ARM: kernel: add support for cpu cache information

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds support for cacheinfo on ARM platforms.

On ARMv7, the cache hierarchy can be identified through Cache Level ID
register(CLIDR) while the cache geometry is provided by Cache Size ID
register(CCSIDR).

On architecture versions before ARMv7, CLIDR and CCSIDR is not
implemented. The cache type register(CTR) provides both cache hierarchy
and geometry if implemented. For implementations that doesn't support
CTR, we need to list the probable value of CTR if it was implemented
along with the cpuid for the sake of simplicity to handle them.

Since the architecture doesn't provide any way of detecting the cpus
sharing particular cache, device tree is used fo the same purpose.
On non-DT platforms, first level caches are per-cpu while higher level
caches are assumed system-wide.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Nicolas Pitre <nicolas.pitre@linaro.org>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
---
 arch/arm/kernel/Makefile    |   1 +
 arch/arm/kernel/cacheinfo.c | 275 ++++++++++++++++++++++++++++++++++++++++++++
 arch/arm/mm/Kconfig         |  13 +++
 3 files changed, 289 insertions(+)
 create mode 100644 arch/arm/kernel/cacheinfo.c

diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile
index 38ddd9f83d0e..2c5ff0efb670 100644
--- a/arch/arm/kernel/Makefile
+++ b/arch/arm/kernel/Makefile
@@ -29,6 +29,7 @@ obj-y		+= entry-v7m.o v7m.o
 else
 obj-y		+= entry-armv.o
 endif
+obj-$(CONFIG_CPU_HAS_CACHE) += cacheinfo.o
 
 obj-$(CONFIG_OC_ETM)		+= etm.o
 obj-$(CONFIG_CPU_IDLE)		+= cpuidle.o
diff --git a/arch/arm/kernel/cacheinfo.c b/arch/arm/kernel/cacheinfo.c
new file mode 100644
index 000000000000..d2659563de31
--- /dev/null
+++ b/arch/arm/kernel/cacheinfo.c
@@ -0,0 +1,275 @@
+/*
+ *  ARM cacheinfo support
+ *
+ *  Copyright (C) 2014 ARM Ltd.
+ *  All Rights Reserved
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/cpu.h>
+#include <linux/compiler.h>
+#include <linux/of.h>
+
+#include <asm/cputype.h>
+#include <asm/processor.h>
+#include <asm/system_info.h>
+
+#define cache_is_armv7() \
+	(cpu_architecture() >= CPU_ARCH_ARMv7 && !armv6_extended())
+#define MAX_CACHE_LEVEL		(cache_is_armv7() ? 7 : 1)
+
+#define CTR_CTYPE_SHIFT		24
+#define CTR_CTYPE_MASK		(1 << CTR_CTYPE_SHIFT)
+
+struct ctr_info {
+	unsigned int cpuid_part;
+	unsigned int ctr;
+};
+
+/*
+ *  Cache Type Register
+ *  +---------------------------------+
+ *  | 31 29 | 28 25 |24| 23 12 | 11 0 |
+ *  +---------------------------------+
+ *  | 0 0 0 | Ctype | S| Dsize | Isize|
+ *  +---------------------------------+
+ * The table below encodes only Dsize and Isize
+ */
+static struct ctr_info cache_ctr_list[] = {
+	{0x4400a100, 0x0016A16A }, /* SA-110:  32kB D$, 32kB I$ */
+	{0x4400a110, 0x0012A16A }, /* SA-1100: 16kB D$, 32kB I$ */
+	{0x6900b110, 0x0012A16A }, /* SA-1110: 16kB D$, 32kB I$ */
+};
+
+/*
+ * List of CPUs reported as ARMv7 but don't implement CLIDR,
+ * CSSELR and CCSIDR. Cache information is still available from CTR
+ */
+static int armv6_ext_cpuid_part[] = {
+	0x4100b020, /* ARM11MP */
+	0x4100b760, /* ARM1176 */
+};
+
+static bool armv6_extended(void)
+{
+	int i, cpuid_part = read_cpuid_part();
+
+	for (i = 0; i < ARRAY_SIZE(armv6_ext_cpuid_part); i++)
+		if (armv6_ext_cpuid_part[i] == cpuid_part)
+			return true;
+	return false;
+}
+
+static int get_unimplemented_ctr(unsigned int *ctr)
+{
+	int i, cpuid_part = read_cpuid_part();
+
+	for (i = 0; i < ARRAY_SIZE(cache_ctr_list); i++)
+		if (cache_ctr_list[i].cpuid_part == cpuid_part) {
+			*ctr = cache_ctr_list[i].ctr;
+			return 0;
+		}
+	return -ENOENT;
+}
+
+static unsigned int get_ctr(void)
+{
+	unsigned int ctr;
+
+	if (get_unimplemented_ctr(&ctr))
+		ctr = read_cpuid_cachetype();
+	return ctr;
+}
+
+static enum cache_type __get_cache_type(int level)
+{
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	return get_ctr() & CTR_CTYPE_MASK ?
+		CACHE_TYPE_SEPARATE : CACHE_TYPE_UNIFIED;
+}
+
+/*
+ *  +---------------------------------+
+ *  | 9  8  7  6 | 5  4  3 | 2 | 1  0 |
+ *  +---------------------------------+
+ *  |    size    |  assoc  | m |  len |
+ *  +---------------------------------+
+ * linelen        = 1 << (len + 3)
+ * multiplier     = 2 + m
+ * nsets          = 1 << (size + 6 - assoc - len)
+ * associativity  = multiplier << (assoc - 1)
+ * cache_size     = multiplier << (size + 8)
+ */
+#define CTR_LINESIZE_MASK	0x3
+#define CTR_MULTIPLIER_SHIFT	2
+#define CTR_MULTIPLIER_MASK	0x1
+#define CTR_ASSOCIAT_SHIFT	3
+#define CTR_ASSOCIAT_MASK	0x7
+#define CTR_SIZE_SHIFT		6
+#define CTR_SIZE_MASK		0xF
+#define CTR_DCACHE_SHIFT	12
+
+static void __ci_leaf_init(enum cache_type type, struct cacheinfo *this_leaf)
+{
+	unsigned int size, multiplier, assoc, len, tmp = get_ctr();
+
+	if (type == CACHE_TYPE_DATA)
+		tmp >>= CTR_DCACHE_SHIFT;
+
+	len = tmp & CTR_LINESIZE_MASK;
+	size = (tmp >> CTR_SIZE_SHIFT) & CTR_SIZE_MASK;
+	assoc = (tmp >> CTR_ASSOCIAT_SHIFT) & CTR_ASSOCIAT_MASK;
+	multiplier = ((tmp >> CTR_MULTIPLIER_SHIFT) & CTR_MULTIPLIER_MASK) + 2;
+
+	this_leaf->type = type;
+	this_leaf->coherency_line_size = 1 << (len + 3);
+	this_leaf->number_of_sets = 1 << (size + 6 - assoc - len);
+	this_leaf->ways_of_associativity = multiplier << (assoc - 1);
+	this_leaf->size = multiplier << (size + 8);
+}
+
+/* Ctypen, bits[3(n - 1) + 2 : 3(n - 1)], for n = 1 to 7 */
+#define CLIDR_CTYPE_SHIFT(level)	(3 * (level - 1))
+#define CLIDR_CTYPE_MASK(level)		(7 << CLIDR_CTYPE_SHIFT(level))
+#define CLIDR_CTYPE(clidr, level)	\
+	(((clidr) & CLIDR_CTYPE_MASK(level)) >> CLIDR_CTYPE_SHIFT(level))
+
+static inline enum cache_type __armv7_get_cache_type(int level)
+{
+	unsigned int clidr;
+
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	asm volatile ("mrc p15, 1, %0, c0, c0, 1" : "=r" (clidr));
+	return CLIDR_CTYPE(clidr, level);
+}
+
+/*
+ * NumSets, bits[27:13] - (Number of sets in cache) - 1
+ * Associativity, bits[12:3] - (Associativity of cache) - 1
+ * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
+ */
+#define CCSIDR_WRITE_THROUGH		BIT(31)
+#define CCSIDR_WRITE_BACK		BIT(30)
+#define CCSIDR_READ_ALLOCATE		BIT(29)
+#define CCSIDR_WRITE_ALLOCATE		BIT(28)
+#define CCSIDR_LINESIZE_MASK		0x7
+#define CCSIDR_ASSOCIATIVITY_SHIFT	3
+#define CCSIDR_ASSOCIATIVITY_MASK	0x3FF
+#define CCSIDR_NUMSETS_SHIFT		13
+#define CCSIDR_NUMSETS_MASK		0x7FF
+
+/*
+ * Which cache CCSIDR represents depends on CSSELR value
+ * Make sure no one else changes CSSELR during this
+ * smp_call_function_single prevents preemption for us
+ */
+static inline u32 get_ccsidr(u32 csselr)
+{
+	u32 ccsidr;
+
+	/* Put value into CSSELR */
+	asm volatile ("mcr p15, 2, %0, c0, c0, 0" : : "r" (csselr));
+	isb();
+	/* Read result out of CCSIDR */
+	asm volatile ("mrc p15, 1, %0, c0, c0, 0" : "=r" (ccsidr));
+
+	return ccsidr;
+}
+
+static void __armv7_ci_leaf_init(enum cache_type type,
+				 struct cacheinfo *this_leaf)
+{
+	bool is_instr_cache = type & CACHE_TYPE_INST;
+	u32 tmp = get_ccsidr((this_leaf->level - 1) << 1 | is_instr_cache);
+
+	this_leaf->type = type;
+	this_leaf->coherency_line_size =
+	    (1 << ((tmp & CCSIDR_LINESIZE_MASK) + 2)) * 4;
+	this_leaf->number_of_sets =
+	    ((tmp >> CCSIDR_NUMSETS_SHIFT) & CCSIDR_NUMSETS_MASK) + 1;
+	this_leaf->ways_of_associativity = ((tmp >> CCSIDR_ASSOCIATIVITY_SHIFT)
+					    & CCSIDR_ASSOCIATIVITY_MASK) + 1;
+	this_leaf->size = this_leaf->number_of_sets *
+	    this_leaf->coherency_line_size * this_leaf->ways_of_associativity;
+	this_leaf->attributes =
+		((tmp & CCSIDR_WRITE_THROUGH) ? CACHE_WRITE_THROUGH : 0) |
+		((tmp & CCSIDR_WRITE_BACK) ? CACHE_WRITE_BACK : 0) |
+		((tmp & CCSIDR_READ_ALLOCATE) ? CACHE_READ_ALLOCATE : 0) |
+		((tmp & CCSIDR_WRITE_ALLOCATE) ? CACHE_WRITE_ALLOCATE : 0);
+}
+
+static inline enum cache_type get_cache_type(int level)
+{
+	if (cache_is_armv7())
+		return __armv7_get_cache_type(level);
+	return __get_cache_type(level);
+}
+
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
+{
+	this_leaf->level = level;
+	if (cache_is_armv7())
+		__armv7_ci_leaf_init(type, this_leaf);
+	else
+		__ci_leaf_init(type, this_leaf);
+}
+
+static int __init_cache_level(unsigned int cpu)
+{
+	unsigned int ctype, level, leaves;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	for (level = 1, leaves = 0; level <= MAX_CACHE_LEVEL; level++) {
+		ctype = get_cache_type(level);
+		if (ctype == CACHE_TYPE_NOCACHE) {
+			level--;
+			break;
+		}
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	}
+
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
+
+	return 0;
+}
+
+static int __populate_cache_leaves(unsigned int cpu)
+{
+	unsigned int level, idx;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		type = get_cache_type(level);
+		if (type == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, type, level);
+		}
+	}
+	return 0;
+}
+
+DEFINE_SMP_CALL_CACHE_FUNCTION(init_cache_level)
+DEFINE_SMP_CALL_CACHE_FUNCTION(populate_cache_leaves)
diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig
index ae69809a9e47..9cfbf2fa8bc4 100644
--- a/arch/arm/mm/Kconfig
+++ b/arch/arm/mm/Kconfig
@@ -494,30 +494,42 @@ config CPU_PABRT_V7
 # The cache model
 config CPU_CACHE_V4
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V4WT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V4WB
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V6
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V7
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_NOP
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_VIVT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_VIPT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_FA
 	bool
+	select CPU_HAS_CACHE
+
+config CPU_HAS_CACHE
+	bool
 
 if MMU
 # The copy-page model
@@ -845,6 +857,7 @@ config DMA_CACHE_RWFO
 
 config OUTER_CACHE
 	bool
+	select CPU_HAS_CACHE
 
 config OUTER_CACHE_SYNC
 	bool
-- 
1.8.3.2

[PATCH v3 10/11] ARM: kernel: add support for cpu cache information

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

This patch adds support for cacheinfo on ARM platforms.

On ARMv7, the cache hierarchy can be identified through Cache Level ID
register(CLIDR) while the cache geometry is provided by Cache Size ID
register(CCSIDR).

On architecture versions before ARMv7, CLIDR and CCSIDR is not
implemented. The cache type register(CTR) provides both cache hierarchy
and geometry if implemented. For implementations that doesn't support
CTR, we need to list the probable value of CTR if it was implemented
along with the cpuid for the sake of simplicity to handle them.

Since the architecture doesn't provide any way of detecting the cpus
sharing particular cache, device tree is used fo the same purpose.
On non-DT platforms, first level caches are per-cpu while higher level
caches are assumed system-wide.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Nicolas Pitre <nicolas.pitre@linaro.org>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel at lists.infradead.org
---
 arch/arm/kernel/Makefile    |   1 +
 arch/arm/kernel/cacheinfo.c | 275 ++++++++++++++++++++++++++++++++++++++++++++
 arch/arm/mm/Kconfig         |  13 +++
 3 files changed, 289 insertions(+)
 create mode 100644 arch/arm/kernel/cacheinfo.c

diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile
index 38ddd9f83d0e..2c5ff0efb670 100644
--- a/arch/arm/kernel/Makefile
+++ b/arch/arm/kernel/Makefile
@@ -29,6 +29,7 @@ obj-y		+= entry-v7m.o v7m.o
 else
 obj-y		+= entry-armv.o
 endif
+obj-$(CONFIG_CPU_HAS_CACHE) += cacheinfo.o
 
 obj-$(CONFIG_OC_ETM)		+= etm.o
 obj-$(CONFIG_CPU_IDLE)		+= cpuidle.o
diff --git a/arch/arm/kernel/cacheinfo.c b/arch/arm/kernel/cacheinfo.c
new file mode 100644
index 000000000000..d2659563de31
--- /dev/null
+++ b/arch/arm/kernel/cacheinfo.c
@@ -0,0 +1,275 @@
+/*
+ *  ARM cacheinfo support
+ *
+ *  Copyright (C) 2014 ARM Ltd.
+ *  All Rights Reserved
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/bitops.h>
+#include <linux/cacheinfo.h>
+#include <linux/cpu.h>
+#include <linux/compiler.h>
+#include <linux/of.h>
+
+#include <asm/cputype.h>
+#include <asm/processor.h>
+#include <asm/system_info.h>
+
+#define cache_is_armv7() \
+	(cpu_architecture() >= CPU_ARCH_ARMv7 && !armv6_extended())
+#define MAX_CACHE_LEVEL		(cache_is_armv7() ? 7 : 1)
+
+#define CTR_CTYPE_SHIFT		24
+#define CTR_CTYPE_MASK		(1 << CTR_CTYPE_SHIFT)
+
+struct ctr_info {
+	unsigned int cpuid_part;
+	unsigned int ctr;
+};
+
+/*
+ *  Cache Type Register
+ *  +---------------------------------+
+ *  | 31 29 | 28 25 |24| 23 12 | 11 0 |
+ *  +---------------------------------+
+ *  | 0 0 0 | Ctype | S| Dsize | Isize|
+ *  +---------------------------------+
+ * The table below encodes only Dsize and Isize
+ */
+static struct ctr_info cache_ctr_list[] = {
+	{0x4400a100, 0x0016A16A }, /* SA-110:  32kB D$, 32kB I$ */
+	{0x4400a110, 0x0012A16A }, /* SA-1100: 16kB D$, 32kB I$ */
+	{0x6900b110, 0x0012A16A }, /* SA-1110: 16kB D$, 32kB I$ */
+};
+
+/*
+ * List of CPUs reported as ARMv7 but don't implement CLIDR,
+ * CSSELR and CCSIDR. Cache information is still available from CTR
+ */
+static int armv6_ext_cpuid_part[] = {
+	0x4100b020, /* ARM11MP */
+	0x4100b760, /* ARM1176 */
+};
+
+static bool armv6_extended(void)
+{
+	int i, cpuid_part = read_cpuid_part();
+
+	for (i = 0; i < ARRAY_SIZE(armv6_ext_cpuid_part); i++)
+		if (armv6_ext_cpuid_part[i] == cpuid_part)
+			return true;
+	return false;
+}
+
+static int get_unimplemented_ctr(unsigned int *ctr)
+{
+	int i, cpuid_part = read_cpuid_part();
+
+	for (i = 0; i < ARRAY_SIZE(cache_ctr_list); i++)
+		if (cache_ctr_list[i].cpuid_part == cpuid_part) {
+			*ctr = cache_ctr_list[i].ctr;
+			return 0;
+		}
+	return -ENOENT;
+}
+
+static unsigned int get_ctr(void)
+{
+	unsigned int ctr;
+
+	if (get_unimplemented_ctr(&ctr))
+		ctr = read_cpuid_cachetype();
+	return ctr;
+}
+
+static enum cache_type __get_cache_type(int level)
+{
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	return get_ctr() & CTR_CTYPE_MASK ?
+		CACHE_TYPE_SEPARATE : CACHE_TYPE_UNIFIED;
+}
+
+/*
+ *  +---------------------------------+
+ *  | 9  8  7  6 | 5  4  3 | 2 | 1  0 |
+ *  +---------------------------------+
+ *  |    size    |  assoc  | m |  len |
+ *  +---------------------------------+
+ * linelen        = 1 << (len + 3)
+ * multiplier     = 2 + m
+ * nsets          = 1 << (size + 6 - assoc - len)
+ * associativity  = multiplier << (assoc - 1)
+ * cache_size     = multiplier << (size + 8)
+ */
+#define CTR_LINESIZE_MASK	0x3
+#define CTR_MULTIPLIER_SHIFT	2
+#define CTR_MULTIPLIER_MASK	0x1
+#define CTR_ASSOCIAT_SHIFT	3
+#define CTR_ASSOCIAT_MASK	0x7
+#define CTR_SIZE_SHIFT		6
+#define CTR_SIZE_MASK		0xF
+#define CTR_DCACHE_SHIFT	12
+
+static void __ci_leaf_init(enum cache_type type, struct cacheinfo *this_leaf)
+{
+	unsigned int size, multiplier, assoc, len, tmp = get_ctr();
+
+	if (type == CACHE_TYPE_DATA)
+		tmp >>= CTR_DCACHE_SHIFT;
+
+	len = tmp & CTR_LINESIZE_MASK;
+	size = (tmp >> CTR_SIZE_SHIFT) & CTR_SIZE_MASK;
+	assoc = (tmp >> CTR_ASSOCIAT_SHIFT) & CTR_ASSOCIAT_MASK;
+	multiplier = ((tmp >> CTR_MULTIPLIER_SHIFT) & CTR_MULTIPLIER_MASK) + 2;
+
+	this_leaf->type = type;
+	this_leaf->coherency_line_size = 1 << (len + 3);
+	this_leaf->number_of_sets = 1 << (size + 6 - assoc - len);
+	this_leaf->ways_of_associativity = multiplier << (assoc - 1);
+	this_leaf->size = multiplier << (size + 8);
+}
+
+/* Ctypen, bits[3(n - 1) + 2 : 3(n - 1)], for n = 1 to 7 */
+#define CLIDR_CTYPE_SHIFT(level)	(3 * (level - 1))
+#define CLIDR_CTYPE_MASK(level)		(7 << CLIDR_CTYPE_SHIFT(level))
+#define CLIDR_CTYPE(clidr, level)	\
+	(((clidr) & CLIDR_CTYPE_MASK(level)) >> CLIDR_CTYPE_SHIFT(level))
+
+static inline enum cache_type __armv7_get_cache_type(int level)
+{
+	unsigned int clidr;
+
+	if (level > MAX_CACHE_LEVEL)
+		return CACHE_TYPE_NOCACHE;
+	asm volatile ("mrc p15, 1, %0, c0, c0, 1" : "=r" (clidr));
+	return CLIDR_CTYPE(clidr, level);
+}
+
+/*
+ * NumSets, bits[27:13] - (Number of sets in cache) - 1
+ * Associativity, bits[12:3] - (Associativity of cache) - 1
+ * LineSize, bits[2:0] - (Log2(Number of words in cache line)) - 2
+ */
+#define CCSIDR_WRITE_THROUGH		BIT(31)
+#define CCSIDR_WRITE_BACK		BIT(30)
+#define CCSIDR_READ_ALLOCATE		BIT(29)
+#define CCSIDR_WRITE_ALLOCATE		BIT(28)
+#define CCSIDR_LINESIZE_MASK		0x7
+#define CCSIDR_ASSOCIATIVITY_SHIFT	3
+#define CCSIDR_ASSOCIATIVITY_MASK	0x3FF
+#define CCSIDR_NUMSETS_SHIFT		13
+#define CCSIDR_NUMSETS_MASK		0x7FF
+
+/*
+ * Which cache CCSIDR represents depends on CSSELR value
+ * Make sure no one else changes CSSELR during this
+ * smp_call_function_single prevents preemption for us
+ */
+static inline u32 get_ccsidr(u32 csselr)
+{
+	u32 ccsidr;
+
+	/* Put value into CSSELR */
+	asm volatile ("mcr p15, 2, %0, c0, c0, 0" : : "r" (csselr));
+	isb();
+	/* Read result out of CCSIDR */
+	asm volatile ("mrc p15, 1, %0, c0, c0, 0" : "=r" (ccsidr));
+
+	return ccsidr;
+}
+
+static void __armv7_ci_leaf_init(enum cache_type type,
+				 struct cacheinfo *this_leaf)
+{
+	bool is_instr_cache = type & CACHE_TYPE_INST;
+	u32 tmp = get_ccsidr((this_leaf->level - 1) << 1 | is_instr_cache);
+
+	this_leaf->type = type;
+	this_leaf->coherency_line_size =
+	    (1 << ((tmp & CCSIDR_LINESIZE_MASK) + 2)) * 4;
+	this_leaf->number_of_sets =
+	    ((tmp >> CCSIDR_NUMSETS_SHIFT) & CCSIDR_NUMSETS_MASK) + 1;
+	this_leaf->ways_of_associativity = ((tmp >> CCSIDR_ASSOCIATIVITY_SHIFT)
+					    & CCSIDR_ASSOCIATIVITY_MASK) + 1;
+	this_leaf->size = this_leaf->number_of_sets *
+	    this_leaf->coherency_line_size * this_leaf->ways_of_associativity;
+	this_leaf->attributes =
+		((tmp & CCSIDR_WRITE_THROUGH) ? CACHE_WRITE_THROUGH : 0) |
+		((tmp & CCSIDR_WRITE_BACK) ? CACHE_WRITE_BACK : 0) |
+		((tmp & CCSIDR_READ_ALLOCATE) ? CACHE_READ_ALLOCATE : 0) |
+		((tmp & CCSIDR_WRITE_ALLOCATE) ? CACHE_WRITE_ALLOCATE : 0);
+}
+
+static inline enum cache_type get_cache_type(int level)
+{
+	if (cache_is_armv7())
+		return __armv7_get_cache_type(level);
+	return __get_cache_type(level);
+}
+
+static void ci_leaf_init(struct cacheinfo *this_leaf,
+			 enum cache_type type, unsigned int level)
+{
+	this_leaf->level = level;
+	if (cache_is_armv7())
+		__armv7_ci_leaf_init(type, this_leaf);
+	else
+		__ci_leaf_init(type, this_leaf);
+}
+
+static int __init_cache_level(unsigned int cpu)
+{
+	unsigned int ctype, level, leaves;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+
+	for (level = 1, leaves = 0; level <= MAX_CACHE_LEVEL; level++) {
+		ctype = get_cache_type(level);
+		if (ctype == CACHE_TYPE_NOCACHE) {
+			level--;
+			break;
+		}
+		/* Separate instruction and data caches */
+		leaves += (ctype == CACHE_TYPE_SEPARATE) ? 2 : 1;
+	}
+
+	this_cpu_ci->num_levels = level;
+	this_cpu_ci->num_leaves = leaves;
+
+	return 0;
+}
+
+static int __populate_cache_leaves(unsigned int cpu)
+{
+	unsigned int level, idx;
+	enum cache_type type;
+	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cacheinfo *this_leaf = this_cpu_ci->info_list;
+
+	for (idx = 0, level = 1; level <= this_cpu_ci->num_levels &&
+	     idx < this_cpu_ci->num_leaves; idx++, level++) {
+		type = get_cache_type(level);
+		if (type == CACHE_TYPE_SEPARATE) {
+			ci_leaf_init(this_leaf++, CACHE_TYPE_DATA, level);
+			ci_leaf_init(this_leaf++, CACHE_TYPE_INST, level);
+		} else {
+			ci_leaf_init(this_leaf++, type, level);
+		}
+	}
+	return 0;
+}
+
+DEFINE_SMP_CALL_CACHE_FUNCTION(init_cache_level)
+DEFINE_SMP_CALL_CACHE_FUNCTION(populate_cache_leaves)
diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig
index ae69809a9e47..9cfbf2fa8bc4 100644
--- a/arch/arm/mm/Kconfig
+++ b/arch/arm/mm/Kconfig
@@ -494,30 +494,42 @@ config CPU_PABRT_V7
 # The cache model
 config CPU_CACHE_V4
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V4WT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V4WB
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V6
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_V7
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_NOP
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_VIVT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_VIPT
 	bool
+	select CPU_HAS_CACHE
 
 config CPU_CACHE_FA
 	bool
+	select CPU_HAS_CACHE
+
+config CPU_HAS_CACHE
+	bool
 
 if MMU
 # The copy-page model
@@ -845,6 +857,7 @@ config DMA_CACHE_RWFO
 
 config OUTER_CACHE
 	bool
+	select CPU_HAS_CACHE
 
 config OUTER_CACHE_SYNC
 	bool
-- 
1.8.3.2

[PATCH v3 11/11] ARM: kernel: add outer cache support for cacheinfo implementation

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

In order to support outer cache in the cacheinfo infrastructure, a new
function 'get_info' is added to outer_cache_fns. This function is used
to get the outer cache information namely: line size, number of ways of
associativity and number of sets.

This patch adds 'get_info' supports to all L2 cache implementations on
ARM except Marvell's Feroceon L2 cache.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
---
 arch/arm/include/asm/outercache.h |  9 +++++++++
 arch/arm/kernel/cacheinfo.c       | 14 +++++++++++++-
 arch/arm/mm/cache-l2x0.c          | 35 ++++++++++++++++++++++++++++++++++-
 arch/arm/mm/cache-tauros2.c       | 36 ++++++++++++++++++++++++++++++++++++
 arch/arm/mm/cache-xsc3l2.c        | 17 +++++++++++++++++
 5 files changed, 109 insertions(+), 2 deletions(-)

diff --git a/arch/arm/include/asm/outercache.h b/arch/arm/include/asm/outercache.h
index 891a56b35bcf..e063d8c87077 100644
--- a/arch/arm/include/asm/outercache.h
+++ b/arch/arm/include/asm/outercache.h
@@ -23,7 +23,10 @@
 
 #include <linux/types.h>
 
+struct cacheinfo;
+
 struct outer_cache_fns {
+	void (*get_info)(struct cacheinfo *);
 	void (*inv_range)(unsigned long, unsigned long);
 	void (*clean_range)(unsigned long, unsigned long);
 	void (*flush_range)(unsigned long, unsigned long);
@@ -112,6 +115,11 @@ static inline void outer_resume(void)
 		outer_cache.resume();
 }
 
+static inline void outer_get_info(struct cacheinfo *this_leaf)
+{
+	if (outer_cache.get_info)
+		outer_cache.get_info(this_leaf);
+}
 #else
 
 static inline void outer_inv_range(phys_addr_t start, phys_addr_t end)
@@ -123,6 +131,7 @@ static inline void outer_flush_range(phys_addr_t start, phys_addr_t end)
 static inline void outer_flush_all(void) { }
 static inline void outer_disable(void) { }
 static inline void outer_resume(void) { }
+static inline void outer_get_info(struct cacheinfo *this_leaf) { }
 
 #endif
 
diff --git a/arch/arm/kernel/cacheinfo.c b/arch/arm/kernel/cacheinfo.c
index d2659563de31..8b8043a2c73f 100644
--- a/arch/arm/kernel/cacheinfo.c
+++ b/arch/arm/kernel/cacheinfo.c
@@ -24,6 +24,7 @@
 #include <linux/of.h>
 
 #include <asm/cputype.h>
+#include <asm/outercache.h>
 #include <asm/processor.h>
 #include <asm/system_info.h>
 
@@ -220,11 +221,19 @@ static inline enum cache_type get_cache_type(int level)
 	return __get_cache_type(level);
 }
 
+static inline void __outer_ci_leaf_init(struct cacheinfo *this_leaf)
+{
+	outer_get_info(this_leaf);
+	BUG_ON(this_leaf->type == CACHE_TYPE_SEPARATE);
+}
+
 static void ci_leaf_init(struct cacheinfo *this_leaf,
 			 enum cache_type type, unsigned int level)
 {
 	this_leaf->level = level;
-	if (cache_is_armv7())
+	if (type == CACHE_TYPE_NOCACHE)	/* must be outer cache */
+		__outer_ci_leaf_init(this_leaf);
+	else if (cache_is_armv7())
 		__armv7_ci_leaf_init(type, this_leaf);
 	else
 		__ci_leaf_init(type, this_leaf);
@@ -248,6 +257,9 @@ static int __init_cache_level(unsigned int cpu)
 	this_cpu_ci->num_levels = level;
 	this_cpu_ci->num_leaves = leaves;
 
+	if (IS_ENABLED(CONFIG_OUTER_CACHE) && outer_cache.get_info)
+		this_cpu_ci->num_leaves++, this_cpu_ci->num_levels++;
+
 	return 0;
 }
 
diff --git a/arch/arm/mm/cache-l2x0.c b/arch/arm/mm/cache-l2x0.c
index 5f2c988a06ac..4114b1944807 100644
--- a/arch/arm/mm/cache-l2x0.c
+++ b/arch/arm/mm/cache-l2x0.c
@@ -17,6 +17,7 @@
  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  */
 #include <linux/cpu.h>
+#include <linux/cacheinfo.h>
 #include <linux/err.h>
 #include <linux/init.h>
 #include <linux/smp.h>
@@ -105,6 +106,22 @@ static inline void l2c_unlock(void __iomem *base, unsigned num)
 	}
 }
 
+static void __l2x0_getinfo(struct cacheinfo *this_leaf)
+{
+	unsigned int assoc = get_count_order(l2x0_way_mask);
+
+	this_leaf->size = l2x0_size;
+	this_leaf->coherency_line_size = CACHE_LINE_SIZE;
+	this_leaf->ways_of_associativity = assoc;
+	this_leaf->number_of_sets = l2x0_size / (assoc * CACHE_LINE_SIZE);
+}
+
+static void l2x0_getinfo(struct cacheinfo *this_leaf)
+{
+	this_leaf->type = CACHE_TYPE_UNIFIED;
+	__l2x0_getinfo(this_leaf);
+}
+
 /*
  * Enable the L2 cache controller.  This function must only be
  * called when the cache controller is known to be disabled.
@@ -309,6 +326,7 @@ static const struct l2c_init_data l2c210_data __initconst = {
 		.disable = l2c_disable,
 		.sync = l2c210_sync,
 		.resume = l2c210_resume,
+		.get_info = l2x0_getinfo,
 	},
 };
 
@@ -466,6 +484,7 @@ static const struct l2c_init_data l2c220_data = {
 		.disable = l2c_disable,
 		.sync = l2c220_sync,
 		.resume = l2c210_resume,
+		.get_info = l2x0_getinfo,
 	},
 };
 
@@ -814,6 +833,7 @@ static const struct l2c_init_data l2c310_init_fns __initconst = {
 		.disable = l2c310_disable,
 		.sync = l2c210_sync,
 		.resume = l2c310_resume,
+		.get_info = l2x0_getinfo,
 	},
 };
 
@@ -894,7 +914,6 @@ static void __init __l2c_init(const struct l2c_init_data *data,
 		data->enable(l2x0_base, aux, data->num_lock);
 
 	outer_cache = fns;
-
 	/*
 	 * It is strange to save the register state before initialisation,
 	 * but hey, this is what the DT implementations decided to do.
@@ -994,6 +1013,7 @@ static const struct l2c_init_data of_l2c210_data __initconst = {
 		.disable     = l2c_disable,
 		.sync        = l2c210_sync,
 		.resume      = l2c210_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1012,6 +1032,7 @@ static const struct l2c_init_data of_l2c220_data __initconst = {
 		.disable     = l2c_disable,
 		.sync        = l2c220_sync,
 		.resume      = l2c210_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1065,6 +1086,7 @@ static const struct l2c_init_data of_l2c310_data __initconst = {
 		.disable     = l2c310_disable,
 		.sync        = l2c210_sync,
 		.resume      = l2c310_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1092,6 +1114,7 @@ static const struct l2c_init_data of_l2c310_coherent_data __initconst = {
 		.flush_all   = l2c210_flush_all,
 		.disable     = l2c310_disable,
 		.resume      = l2c310_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1255,6 +1278,12 @@ static void __init aurora_of_parse(const struct device_node *np,
 	*aux_mask &= ~mask;
 }
 
+static void aurora_no_outer_data_getinfo(struct cacheinfo *this_leaf)
+{
+	this_leaf->type = CACHE_TYPE_INST;
+	__l2x0_getinfo(this_leaf);
+}
+
 static const struct l2c_init_data of_aurora_with_outer_data __initconst = {
 	.type = "Aurora",
 	.way_size_0 = SZ_4K,
@@ -1271,6 +1300,7 @@ static const struct l2c_init_data of_aurora_with_outer_data __initconst = {
 		.disable     = l2x0_disable,
 		.sync        = l2x0_cache_sync,
 		.resume      = aurora_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1284,6 +1314,7 @@ static const struct l2c_init_data of_aurora_no_outer_data __initconst = {
 	.save  = aurora_save,
 	.outer_cache = {
 		.resume      = aurora_resume,
+		.get_info    = aurora_no_outer_data_getinfo,
 	},
 };
 
@@ -1439,6 +1470,7 @@ static const struct l2c_init_data of_bcm_l2x0_data __initconst = {
 		.disable     = l2c310_disable,
 		.sync        = l2c210_sync,
 		.resume      = l2c310_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1475,6 +1507,7 @@ static const struct l2c_init_data of_tauros3_data __initconst = {
 	/* Tauros3 broadcasts L1 cache operations to L2 */
 	.outer_cache = {
 		.resume      = tauros3_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
diff --git a/arch/arm/mm/cache-tauros2.c b/arch/arm/mm/cache-tauros2.c
index b273739e6359..0d4dd8ff6a56 100644
--- a/arch/arm/mm/cache-tauros2.c
+++ b/arch/arm/mm/cache-tauros2.c
@@ -14,6 +14,7 @@
  *   Document ID MV-S105190-00, Rev 0.7, March 14 2008.
  */
 
+#include <linux/cacheinfo.h>
 #include <linux/init.h>
 #include <linux/of.h>
 #include <linux/of_address.h>
@@ -60,6 +61,7 @@ static inline void tauros2_inv_pa(unsigned long addr)
  * noninclusive.
  */
 #define CACHE_LINE_SIZE		32
+#define CACHE_LINE_SHIFT	5
 
 static void tauros2_inv_range(unsigned long start, unsigned long end)
 {
@@ -131,6 +133,39 @@ static void tauros2_resume(void)
 	"mcr	p15, 0, %0, c1, c0, 0 @Enable L2 Cache\n\t"
 	: : "r" (0x0));
 }
+
+/*
+ *  +----------------------------------------+
+ *  | 11 10 9  8 | 7  6  5  4  3 | 2 |  1  0 |
+ *  +----------------------------------------+
+ *  |  way size  | associativity | - |line_sz|
+ *  +----------------------------------------+
+ */
+#define L2CTR_ASSOCIAT_SHIFT	3
+#define L2CTR_ASSOCIAT_MASK	0x1F
+#define L2CTR_WAYSIZE_SHIFT	8
+#define L2CTR_WAYSIZE_MASK	0xF
+#define CACHE_WAY_PER_SET(l2ctr)	\
+	(((l2_ctr) >> L2CTR_ASSOCIAT_SHIFT) & L2CTR_ASSOCIAT_MASK)
+#define CACHE_WAY_SIZE(l2ctr)		\
+	(8192 << (((l2ctr) >> L2CTR_WAYSIZE_SHIFT) & L2CTR_WAYSIZE_MASK))
+#define CACHE_SET_SIZE(l2ctr)	(CACHE_WAY_SIZE(l2ctr) >> CACHE_LINE_SHIFT)
+
+static void tauros2_getinfo(struct cacheinfo *this_leaf)
+{
+	unsigned int l2_ctr;
+
+	__asm__("mrc p15, 1, %0, c0, c0, 1" : "=r" (l2_ctr));
+
+	this_leaf->type = CACHE_TYPE_UNIFIED;
+	this_leaf->coherency_line_size = CACHE_LINE_SIZE;
+	this_leaf->ways_of_associativity = CACHE_WAY_PER_SET(l2_ctr);
+	this_leaf->number_of_sets = CACHE_SET_SIZE(l2_ctr);
+	this_leaf->size = this_leaf->coherency_line_size *
+			  this_leaf->number_of_sets *
+			  this_leaf->ways_of_associativity;
+}
+
 #endif
 
 static inline u32 __init read_extra_features(void)
@@ -226,6 +261,7 @@ static void __init tauros2_internal_init(unsigned int features)
 		outer_cache.flush_range = tauros2_flush_range;
 		outer_cache.disable = tauros2_disable;
 		outer_cache.resume = tauros2_resume;
+		outer_cache.get_info = tauros2_getinfo;
 	}
 #endif
 
diff --git a/arch/arm/mm/cache-xsc3l2.c b/arch/arm/mm/cache-xsc3l2.c
index 6c3edeb66e74..175bf44eb039 100644
--- a/arch/arm/mm/cache-xsc3l2.c
+++ b/arch/arm/mm/cache-xsc3l2.c
@@ -16,6 +16,7 @@
  * along with this program; if not, write to the Free Software
  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  */
+#include <linux/cacheinfo.h>
 #include <linux/init.h>
 #include <linux/highmem.h>
 #include <asm/cp15.h>
@@ -201,6 +202,21 @@ static void xsc3_l2_flush_range(unsigned long start, unsigned long end)
 	dsb();
 }
 
+static void xsc3_l2_getinfo(struct cacheinfo *this_leaf)
+{
+	unsigned long l2ctype;
+
+	__asm__("mrc p15, 1, %0, c0, c0, 1" : "=r" (l2ctype));
+
+	this_leaf->type = CACHE_TYPE_UNIFIED;
+	this_leaf->coherency_line_size = CACHE_LINE_SIZE;
+	this_leaf->ways_of_associativity = CACHE_WAY_PER_SET;
+	this_leaf->number_of_sets = CACHE_SET_SIZE(l2ctype);
+	this_leaf->size = this_leaf->coherency_line_size *
+			  this_leaf->number_of_sets *
+			  this_leaf->ways_of_associativity;
+}
+
 static int __init xsc3_l2_init(void)
 {
 	if (!cpu_is_xsc3() || !xsc3_l2_present())
@@ -213,6 +229,7 @@ static int __init xsc3_l2_init(void)
 		outer_cache.inv_range = xsc3_l2_inv_range;
 		outer_cache.clean_range = xsc3_l2_clean_range;
 		outer_cache.flush_range = xsc3_l2_flush_range;
+		outer_cache.get_info    = xsc3_l2_getinfo;
 	}
 
 	return 0;
-- 
1.8.3.2

[PATCH v3 11/11] ARM: kernel: add outer cache support for cacheinfo implementation

[Sudeep Holla]

From: Sudeep Holla <sudeep.holla@arm.com>

In order to support outer cache in the cacheinfo infrastructure, a new
function 'get_info' is added to outer_cache_fns. This function is used
to get the outer cache information namely: line size, number of ways of
associativity and number of sets.

This patch adds 'get_info' supports to all L2 cache implementations on
ARM except Marvell's Feroceon L2 cache.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: linux-arm-kernel at lists.infradead.org
---
 arch/arm/include/asm/outercache.h |  9 +++++++++
 arch/arm/kernel/cacheinfo.c       | 14 +++++++++++++-
 arch/arm/mm/cache-l2x0.c          | 35 ++++++++++++++++++++++++++++++++++-
 arch/arm/mm/cache-tauros2.c       | 36 ++++++++++++++++++++++++++++++++++++
 arch/arm/mm/cache-xsc3l2.c        | 17 +++++++++++++++++
 5 files changed, 109 insertions(+), 2 deletions(-)

diff --git a/arch/arm/include/asm/outercache.h b/arch/arm/include/asm/outercache.h
index 891a56b35bcf..e063d8c87077 100644
--- a/arch/arm/include/asm/outercache.h
+++ b/arch/arm/include/asm/outercache.h
@@ -23,7 +23,10 @@
 
 #include <linux/types.h>
 
+struct cacheinfo;
+
 struct outer_cache_fns {
+	void (*get_info)(struct cacheinfo *);
 	void (*inv_range)(unsigned long, unsigned long);
 	void (*clean_range)(unsigned long, unsigned long);
 	void (*flush_range)(unsigned long, unsigned long);
@@ -112,6 +115,11 @@ static inline void outer_resume(void)
 		outer_cache.resume();
 }
 
+static inline void outer_get_info(struct cacheinfo *this_leaf)
+{
+	if (outer_cache.get_info)
+		outer_cache.get_info(this_leaf);
+}
 #else
 
 static inline void outer_inv_range(phys_addr_t start, phys_addr_t end)
@@ -123,6 +131,7 @@ static inline void outer_flush_range(phys_addr_t start, phys_addr_t end)
 static inline void outer_flush_all(void) { }
 static inline void outer_disable(void) { }
 static inline void outer_resume(void) { }
+static inline void outer_get_info(struct cacheinfo *this_leaf) { }
 
 #endif
 
diff --git a/arch/arm/kernel/cacheinfo.c b/arch/arm/kernel/cacheinfo.c
index d2659563de31..8b8043a2c73f 100644
--- a/arch/arm/kernel/cacheinfo.c
+++ b/arch/arm/kernel/cacheinfo.c
@@ -24,6 +24,7 @@
 #include <linux/of.h>
 
 #include <asm/cputype.h>
+#include <asm/outercache.h>
 #include <asm/processor.h>
 #include <asm/system_info.h>
 
@@ -220,11 +221,19 @@ static inline enum cache_type get_cache_type(int level)
 	return __get_cache_type(level);
 }
 
+static inline void __outer_ci_leaf_init(struct cacheinfo *this_leaf)
+{
+	outer_get_info(this_leaf);
+	BUG_ON(this_leaf->type == CACHE_TYPE_SEPARATE);
+}
+
 static void ci_leaf_init(struct cacheinfo *this_leaf,
 			 enum cache_type type, unsigned int level)
 {
 	this_leaf->level = level;
-	if (cache_is_armv7())
+	if (type == CACHE_TYPE_NOCACHE)	/* must be outer cache */
+		__outer_ci_leaf_init(this_leaf);
+	else if (cache_is_armv7())
 		__armv7_ci_leaf_init(type, this_leaf);
 	else
 		__ci_leaf_init(type, this_leaf);
@@ -248,6 +257,9 @@ static int __init_cache_level(unsigned int cpu)
 	this_cpu_ci->num_levels = level;
 	this_cpu_ci->num_leaves = leaves;
 
+	if (IS_ENABLED(CONFIG_OUTER_CACHE) && outer_cache.get_info)
+		this_cpu_ci->num_leaves++, this_cpu_ci->num_levels++;
+
 	return 0;
 }
 
diff --git a/arch/arm/mm/cache-l2x0.c b/arch/arm/mm/cache-l2x0.c
index 5f2c988a06ac..4114b1944807 100644
--- a/arch/arm/mm/cache-l2x0.c
+++ b/arch/arm/mm/cache-l2x0.c
@@ -17,6 +17,7 @@
  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  */
 #include <linux/cpu.h>
+#include <linux/cacheinfo.h>
 #include <linux/err.h>
 #include <linux/init.h>
 #include <linux/smp.h>
@@ -105,6 +106,22 @@ static inline void l2c_unlock(void __iomem *base, unsigned num)
 	}
 }
 
+static void __l2x0_getinfo(struct cacheinfo *this_leaf)
+{
+	unsigned int assoc = get_count_order(l2x0_way_mask);
+
+	this_leaf->size = l2x0_size;
+	this_leaf->coherency_line_size = CACHE_LINE_SIZE;
+	this_leaf->ways_of_associativity = assoc;
+	this_leaf->number_of_sets = l2x0_size / (assoc * CACHE_LINE_SIZE);
+}
+
+static void l2x0_getinfo(struct cacheinfo *this_leaf)
+{
+	this_leaf->type = CACHE_TYPE_UNIFIED;
+	__l2x0_getinfo(this_leaf);
+}
+
 /*
  * Enable the L2 cache controller.  This function must only be
  * called when the cache controller is known to be disabled.
@@ -309,6 +326,7 @@ static const struct l2c_init_data l2c210_data __initconst = {
 		.disable = l2c_disable,
 		.sync = l2c210_sync,
 		.resume = l2c210_resume,
+		.get_info = l2x0_getinfo,
 	},
 };
 
@@ -466,6 +484,7 @@ static const struct l2c_init_data l2c220_data = {
 		.disable = l2c_disable,
 		.sync = l2c220_sync,
 		.resume = l2c210_resume,
+		.get_info = l2x0_getinfo,
 	},
 };
 
@@ -814,6 +833,7 @@ static const struct l2c_init_data l2c310_init_fns __initconst = {
 		.disable = l2c310_disable,
 		.sync = l2c210_sync,
 		.resume = l2c310_resume,
+		.get_info = l2x0_getinfo,
 	},
 };
 
@@ -894,7 +914,6 @@ static void __init __l2c_init(const struct l2c_init_data *data,
 		data->enable(l2x0_base, aux, data->num_lock);
 
 	outer_cache = fns;
-
 	/*
 	 * It is strange to save the register state before initialisation,
 	 * but hey, this is what the DT implementations decided to do.
@@ -994,6 +1013,7 @@ static const struct l2c_init_data of_l2c210_data __initconst = {
 		.disable     = l2c_disable,
 		.sync        = l2c210_sync,
 		.resume      = l2c210_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1012,6 +1032,7 @@ static const struct l2c_init_data of_l2c220_data __initconst = {
 		.disable     = l2c_disable,
 		.sync        = l2c220_sync,
 		.resume      = l2c210_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1065,6 +1086,7 @@ static const struct l2c_init_data of_l2c310_data __initconst = {
 		.disable     = l2c310_disable,
 		.sync        = l2c210_sync,
 		.resume      = l2c310_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1092,6 +1114,7 @@ static const struct l2c_init_data of_l2c310_coherent_data __initconst = {
 		.flush_all   = l2c210_flush_all,
 		.disable     = l2c310_disable,
 		.resume      = l2c310_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1255,6 +1278,12 @@ static void __init aurora_of_parse(const struct device_node *np,
 	*aux_mask &= ~mask;
 }
 
+static void aurora_no_outer_data_getinfo(struct cacheinfo *this_leaf)
+{
+	this_leaf->type = CACHE_TYPE_INST;
+	__l2x0_getinfo(this_leaf);
+}
+
 static const struct l2c_init_data of_aurora_with_outer_data __initconst = {
 	.type = "Aurora",
 	.way_size_0 = SZ_4K,
@@ -1271,6 +1300,7 @@ static const struct l2c_init_data of_aurora_with_outer_data __initconst = {
 		.disable     = l2x0_disable,
 		.sync        = l2x0_cache_sync,
 		.resume      = aurora_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1284,6 +1314,7 @@ static const struct l2c_init_data of_aurora_no_outer_data __initconst = {
 	.save  = aurora_save,
 	.outer_cache = {
 		.resume      = aurora_resume,
+		.get_info    = aurora_no_outer_data_getinfo,
 	},
 };
 
@@ -1439,6 +1470,7 @@ static const struct l2c_init_data of_bcm_l2x0_data __initconst = {
 		.disable     = l2c310_disable,
 		.sync        = l2c210_sync,
 		.resume      = l2c310_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
@@ -1475,6 +1507,7 @@ static const struct l2c_init_data of_tauros3_data __initconst = {
 	/* Tauros3 broadcasts L1 cache operations to L2 */
 	.outer_cache = {
 		.resume      = tauros3_resume,
+		.get_info    = l2x0_getinfo,
 	},
 };
 
diff --git a/arch/arm/mm/cache-tauros2.c b/arch/arm/mm/cache-tauros2.c
index b273739e6359..0d4dd8ff6a56 100644
--- a/arch/arm/mm/cache-tauros2.c
+++ b/arch/arm/mm/cache-tauros2.c
@@ -14,6 +14,7 @@
  *   Document ID MV-S105190-00, Rev 0.7, March 14 2008.
  */
 
+#include <linux/cacheinfo.h>
 #include <linux/init.h>
 #include <linux/of.h>
 #include <linux/of_address.h>
@@ -60,6 +61,7 @@ static inline void tauros2_inv_pa(unsigned long addr)
  * noninclusive.
  */
 #define CACHE_LINE_SIZE		32
+#define CACHE_LINE_SHIFT	5
 
 static void tauros2_inv_range(unsigned long start, unsigned long end)
 {
@@ -131,6 +133,39 @@ static void tauros2_resume(void)
 	"mcr	p15, 0, %0, c1, c0, 0 @Enable L2 Cache\n\t"
 	: : "r" (0x0));
 }
+
+/*
+ *  +----------------------------------------+
+ *  | 11 10 9  8 | 7  6  5  4  3 | 2 |  1  0 |
+ *  +----------------------------------------+
+ *  |  way size  | associativity | - |line_sz|
+ *  +----------------------------------------+
+ */
+#define L2CTR_ASSOCIAT_SHIFT	3
+#define L2CTR_ASSOCIAT_MASK	0x1F
+#define L2CTR_WAYSIZE_SHIFT	8
+#define L2CTR_WAYSIZE_MASK	0xF
+#define CACHE_WAY_PER_SET(l2ctr)	\
+	(((l2_ctr) >> L2CTR_ASSOCIAT_SHIFT) & L2CTR_ASSOCIAT_MASK)
+#define CACHE_WAY_SIZE(l2ctr)		\
+	(8192 << (((l2ctr) >> L2CTR_WAYSIZE_SHIFT) & L2CTR_WAYSIZE_MASK))
+#define CACHE_SET_SIZE(l2ctr)	(CACHE_WAY_SIZE(l2ctr) >> CACHE_LINE_SHIFT)
+
+static void tauros2_getinfo(struct cacheinfo *this_leaf)
+{
+	unsigned int l2_ctr;
+
+	__asm__("mrc p15, 1, %0, c0, c0, 1" : "=r" (l2_ctr));
+
+	this_leaf->type = CACHE_TYPE_UNIFIED;
+	this_leaf->coherency_line_size = CACHE_LINE_SIZE;
+	this_leaf->ways_of_associativity = CACHE_WAY_PER_SET(l2_ctr);
+	this_leaf->number_of_sets = CACHE_SET_SIZE(l2_ctr);
+	this_leaf->size = this_leaf->coherency_line_size *
+			  this_leaf->number_of_sets *
+			  this_leaf->ways_of_associativity;
+}
+
 #endif
 
 static inline u32 __init read_extra_features(void)
@@ -226,6 +261,7 @@ static void __init tauros2_internal_init(unsigned int features)
 		outer_cache.flush_range = tauros2_flush_range;
 		outer_cache.disable = tauros2_disable;
 		outer_cache.resume = tauros2_resume;
+		outer_cache.get_info = tauros2_getinfo;
 	}
 #endif
 
diff --git a/arch/arm/mm/cache-xsc3l2.c b/arch/arm/mm/cache-xsc3l2.c
index 6c3edeb66e74..175bf44eb039 100644
--- a/arch/arm/mm/cache-xsc3l2.c
+++ b/arch/arm/mm/cache-xsc3l2.c
@@ -16,6 +16,7 @@
  * along with this program; if not, write to the Free Software
  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  */
+#include <linux/cacheinfo.h>
 #include <linux/init.h>
 #include <linux/highmem.h>
 #include <asm/cp15.h>
@@ -201,6 +202,21 @@ static void xsc3_l2_flush_range(unsigned long start, unsigned long end)
 	dsb();
 }
 
+static void xsc3_l2_getinfo(struct cacheinfo *this_leaf)
+{
+	unsigned long l2ctype;
+
+	__asm__("mrc p15, 1, %0, c0, c0, 1" : "=r" (l2ctype));
+
+	this_leaf->type = CACHE_TYPE_UNIFIED;
+	this_leaf->coherency_line_size = CACHE_LINE_SIZE;
+	this_leaf->ways_of_associativity = CACHE_WAY_PER_SET;
+	this_leaf->number_of_sets = CACHE_SET_SIZE(l2ctype);
+	this_leaf->size = this_leaf->coherency_line_size *
+			  this_leaf->number_of_sets *
+			  this_leaf->ways_of_associativity;
+}
+
 static int __init xsc3_l2_init(void)
 {
 	if (!cpu_is_xsc3() || !xsc3_l2_present())
@@ -213,6 +229,7 @@ static int __init xsc3_l2_init(void)
 		outer_cache.inv_range = xsc3_l2_inv_range;
 		outer_cache.clean_range = xsc3_l2_clean_range;
 		outer_cache.flush_range = xsc3_l2_flush_range;
+		outer_cache.get_info    = xsc3_l2_getinfo;
 	}
 
 	return 0;
-- 
1.8.3.2