[KVM PATCH v6 0/2] iosignalfd

[KVM PATCH v6 0/2] iosignalfd

[Gregory Haskins]

[
  Applies to kvm.git/master:0281e88f as a base, plus Marcelo's irq locking
  rework v4, which you can find here:

  http://www.mail-archive.com/kvm@vger.kernel.org/msg16494.html

  You can find the entire series here in my git tree: 

  http://git.kernel.org/?p=linux/kernel/git/ghaskins/linux-2.6-hacks.git;a=shortlog;h=for-avi

  Note: I didn't find any conflicts or things that needed addressing when
  rebasing to Marcelo's series, so the two may in fact be applied
  indepenently if need be.

]

This is v6 of the series.  For more details, please see the header to
patch 2/2.

This series has been tested against the kvm-eventfd unit test, and
appears to be functioning properly.  You can download this test here:

ftp://ftp.novell.com/dev/ghaskins/kvm-eventfd.tar.bz2

[
   Todo before "consider for inclusion" ready:
       *) There is still some confusion whether I am doing the "kvm-shutdown"
          path properly from a private discussion I had with Marcelo, so
          this needs to be cleared up before it can be accepted.
       *) As agreed, Marcelo's code need to get committed first
]

[
   Changelog:

      v6:
           *) Removed "FIXME?" comment on choice over RCU vs SRCU after
              discussion/numbers from Paul.  I think RCU is fine to use for
              now based on the conversation.  We can always convert it later
              if need be.
           *) Fixed the "group" free path to eliminate an RCU related race
           *) Fixed a memory/eventfd leak on shutdown for any iosignalfd's
              which were still active at the time the guest shuts down.
           *) Beefed up comments
           *) Rebased to kvm.git/master:0281e88f + irq locking rework and
              verified that kvm-eventfd unit test still passes.

      v5:
           *) Removed "cookie" field, which was a misunderstanding on my
              part on what Avi wanted for a data-match feature
	   *) Added a new "trigger" data-match feature which I think is
              much closer to what we need.
	   *) We retain the dev_count field in the io_bus infrastructure
	      and instead back-fill the array on removal.
	   *) Various minor cleanups
	   *) Rebased to kvm.git/master:25deed73

      v4:
           *) Fixed a bug in the original 2/4 where the PIT failure case
              would potentially leave the io_bus components registered.
           *) Condensed the v3 2/4 and 3/4 into one patch (2/2) since
              the patches became interdependent with the fix described above
           *) Rebased to kvm.git/master:74dfca0a

      v3:
           *) fixed patch 2/4 to handle error cases instead of BUG_ON
           *) implemented same HAVE_EVENTFD protection mechanism as
              irqfd to prevent compilation errors on unsupported arches
           *) completed testing
           *) rebased to kvm.git/master:7391a6d5

      v2:
           *) added optional data-matching capability (via cookie field)
           *) changed name from iofd to iosignalfd
           *) added io_bus unregister function
           *) implemented deassign feature

      v1:
           *) original release (integrated into irqfd v7 series as "iofd")
]


---

Gregory Haskins (2):
      KVM: add iosignalfd support
      KVM: make io_bus interface more robust


 arch/x86/kvm/i8254.c      |   22 ++-
 arch/x86/kvm/i8259.c      |    9 +
 arch/x86/kvm/x86.c        |    1 
 include/linux/kvm.h       |   15 ++
 include/linux/kvm_host.h  |   16 +-
 virt/kvm/coalesced_mmio.c |    8 +
 virt/kvm/eventfd.c        |  389 +++++++++++++++++++++++++++++++++++++++++++++
 virt/kvm/ioapic.c         |    9 +
 virt/kvm/kvm_main.c       |   41 ++++-
 9 files changed, 495 insertions(+), 15 deletions(-)

-- 
Signature

[KVM PATCH v6 1/2] KVM: make io_bus interface more robust

[Gregory Haskins]

Today kvm_io_bus_regsiter_dev() returns void and will internally BUG_ON if it
fails.  We want to create dynamic MMIO/PIO entries driven from userspace later
in the series, so we need to enhance the code to be more robust with the
following changes:

   1) Add a return value to the registration function
   2) Fix up all the callsites to check the return code, handle any
      failures, and percolate the error up to the caller.
   3) Add an unregister function that collapses holes in the array

Signed-off-by: Gregory Haskins <ghaskins@novell.com>
---

 arch/x86/kvm/i8254.c      |   22 ++++++++++++++++++++--
 arch/x86/kvm/i8259.c      |    9 ++++++++-
 include/linux/kvm_host.h  |    6 ++++--
 virt/kvm/coalesced_mmio.c |    8 ++++++--
 virt/kvm/ioapic.c         |    9 +++++++--
 virt/kvm/kvm_main.c       |   30 ++++++++++++++++++++++++++++--
 6 files changed, 73 insertions(+), 11 deletions(-)

diff --git a/arch/x86/kvm/i8254.c b/arch/x86/kvm/i8254.c
index 9749ec3..54372b1 100644
--- a/arch/x86/kvm/i8254.c
+++ b/arch/x86/kvm/i8254.c
@@ -586,6 +586,7 @@ struct kvm_pit *kvm_create_pit(struct kvm *kvm, u32 flags)
 {
 	struct kvm_pit *pit;
 	struct kvm_kpit_state *pit_state;
+	int ret;
 
 	pit = kzalloc(sizeof(struct kvm_pit), GFP_KERNEL);
 	if (!pit)
@@ -620,14 +621,31 @@ struct kvm_pit *kvm_create_pit(struct kvm *kvm, u32 flags)
 	kvm_register_irq_mask_notifier(kvm, 0, &pit->mask_notifier);
 
 	kvm_iodevice_init(&pit->dev, &pit_dev_ops);
-	kvm_io_bus_register_dev(&kvm->pio_bus, &pit->dev);
+	ret = kvm_io_bus_register_dev(&kvm->pio_bus, &pit->dev);
+	if (ret < 0)
+		goto fail;
 
 	if (flags & KVM_PIT_SPEAKER_DUMMY) {
 		kvm_iodevice_init(&pit->speaker_dev, &speaker_dev_ops);
-		kvm_io_bus_register_dev(&kvm->pio_bus, &pit->speaker_dev);
+		ret = kvm_io_bus_register_dev(&kvm->pio_bus,
+					      &pit->speaker_dev);
+		if (ret < 0)
+			goto fail;
 	}
 
 	return pit;
+
+fail:
+	if (flags & KVM_PIT_SPEAKER_DUMMY)
+		kvm_io_bus_unregister_dev(&kvm->pio_bus, &pit->speaker_dev);
+
+	kvm_io_bus_unregister_dev(&kvm->pio_bus, &pit->dev);
+
+	if (pit->irq_source_id >= 0)
+		kvm_free_irq_source_id(kvm, pit->irq_source_id);
+
+	kfree(pit);
+	return NULL;
 }
 
 void kvm_free_pit(struct kvm *kvm)
diff --git a/arch/x86/kvm/i8259.c b/arch/x86/kvm/i8259.c
index bf94a45..8228a15 100644
--- a/arch/x86/kvm/i8259.c
+++ b/arch/x86/kvm/i8259.c
@@ -532,6 +532,8 @@ static const struct kvm_io_device_ops picdev_ops = {
 struct kvm_pic *kvm_create_pic(struct kvm *kvm)
 {
 	struct kvm_pic *s;
+	int ret;
+
 	s = kzalloc(sizeof(struct kvm_pic), GFP_KERNEL);
 	if (!s)
 		return NULL;
@@ -548,6 +550,11 @@ struct kvm_pic *kvm_create_pic(struct kvm *kvm)
 	 * Initialize PIO device
 	 */
 	kvm_iodevice_init(&s->dev, &picdev_ops);
-	kvm_io_bus_register_dev(&kvm->pio_bus, &s->dev);
+	ret = kvm_io_bus_register_dev(&kvm->pio_bus, &s->dev);
+	if (ret < 0) {
+		kfree(s);
+		return NULL;
+	}
+
 	return s;
 }
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index cdf1279..2dbc3ec 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -61,8 +61,10 @@ void kvm_io_bus_init(struct kvm_io_bus *bus);
 void kvm_io_bus_destroy(struct kvm_io_bus *bus);
 struct kvm_io_device *kvm_io_bus_find_dev(struct kvm_io_bus *bus,
 					  gpa_t addr, int len, int is_write);
-void kvm_io_bus_register_dev(struct kvm_io_bus *bus,
-			     struct kvm_io_device *dev);
+int kvm_io_bus_register_dev(struct kvm_io_bus *bus,
+			    struct kvm_io_device *dev);
+void kvm_io_bus_unregister_dev(struct kvm_io_bus *bus,
+			    struct kvm_io_device *dev);
 
 struct kvm_vcpu {
 	struct kvm *kvm;
diff --git a/virt/kvm/coalesced_mmio.c b/virt/kvm/coalesced_mmio.c
index 1a6c8d0..7c4811a 100644
--- a/virt/kvm/coalesced_mmio.c
+++ b/virt/kvm/coalesced_mmio.c
@@ -94,6 +94,7 @@ static const struct kvm_io_device_ops coalesced_mmio_ops = {
 int kvm_coalesced_mmio_init(struct kvm *kvm)
 {
 	struct kvm_coalesced_mmio_dev *dev;
+	int ret;
 
 	dev = kzalloc(sizeof(struct kvm_coalesced_mmio_dev), GFP_KERNEL);
 	if (!dev)
@@ -102,9 +103,12 @@ int kvm_coalesced_mmio_init(struct kvm *kvm)
 	kvm_iodevice_init(&dev->dev, &coalesced_mmio_ops);
 	dev->kvm = kvm;
 	kvm->coalesced_mmio_dev = dev;
-	kvm_io_bus_register_dev(&kvm->mmio_bus, &dev->dev);
 
-	return 0;
+	ret = kvm_io_bus_register_dev(&kvm->mmio_bus, &dev->dev);
+	if (ret < 0)
+		kfree(dev);
+
+	return ret;
 }
 
 int kvm_vm_ioctl_register_coalesced_mmio(struct kvm *kvm,
diff --git a/virt/kvm/ioapic.c b/virt/kvm/ioapic.c
index 1665961..ac4fece 100644
--- a/virt/kvm/ioapic.c
+++ b/virt/kvm/ioapic.c
@@ -333,6 +333,7 @@ static const struct kvm_io_device_ops ioapic_mmio_ops = {
 int kvm_ioapic_init(struct kvm *kvm)
 {
 	struct kvm_ioapic *ioapic;
+	int ret;
 
 	ioapic = kzalloc(sizeof(struct kvm_ioapic), GFP_KERNEL);
 	if (!ioapic)
@@ -341,7 +342,11 @@ int kvm_ioapic_init(struct kvm *kvm)
 	kvm_ioapic_reset(ioapic);
 	kvm_iodevice_init(&ioapic->dev, &ioapic_mmio_ops);
 	ioapic->kvm = kvm;
-	kvm_io_bus_register_dev(&kvm->mmio_bus, &ioapic->dev);
-	return 0;
+
+	ret = kvm_io_bus_register_dev(&kvm->mmio_bus, &ioapic->dev);
+	if (ret < 0)
+		kfree(ioapic);
+
+	return ret;
 }
 
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 813d7b8..5980bec 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -2461,11 +2461,37 @@ struct kvm_io_device *kvm_io_bus_find_dev(struct kvm_io_bus *bus,
 	return NULL;
 }
 
-void kvm_io_bus_register_dev(struct kvm_io_bus *bus, struct kvm_io_device *dev)
+/* assumes kvm->lock held */
+int kvm_io_bus_register_dev(struct kvm_io_bus *bus, struct kvm_io_device *dev)
 {
-	BUG_ON(bus->dev_count > (NR_IOBUS_DEVS-1));
+	if (bus->dev_count > (NR_IOBUS_DEVS-1))
+		return -ENOSPC;
 
 	bus->devs[bus->dev_count++] = dev;
+
+	return 0;
+}
+
+/* assumes kvm->lock held */
+void kvm_io_bus_unregister_dev(struct kvm_io_bus *bus,
+			       struct kvm_io_device *dev)
+{
+	int i;
+
+	for (i = 0; i < bus->dev_count; i++) {
+
+		if (bus->devs[i] == dev) {
+			int j;
+
+			/* backfill the hole */
+			for (j = i; j < bus->dev_count-1; j++)
+				bus->devs[j] = bus->devs[j+1];
+
+			bus->dev_count--;
+
+			break;
+		}
+	}
 }
 
 static struct notifier_block kvm_cpu_notifier = {

[KVM PATCH v6 2/2] KVM: add iosignalfd support

[Gregory Haskins]

iosignalfd is a mechanism to register PIO/MMIO regions to trigger an eventfd
signal when written to by a guest.  Host userspace can register any arbitrary
IO address with a corresponding eventfd and then pass the eventfd to a
specific end-point of interest for handling.

Normal IO requires a blocking round-trip since the operation may cause
side-effects in the emulated model or may return data to the caller.
Therefore, an IO in KVM traps from the guest to the host, causes a VMX/SVM
"heavy-weight" exit back to userspace, and is ultimately serviced by qemu's
device model synchronously before returning control back to the vcpu.

However, there is a subclass of IO which acts purely as a trigger for
other IO (such as to kick off an out-of-band DMA request, etc).  For these
patterns, the synchronous call is particularly expensive since we really
only want to simply get our notification transmitted asychronously and
return as quickly as possible.  All the sychronous infrastructure to ensure
proper data-dependencies are met in the normal IO case are just unecessary
overhead for signalling.  This adds additional computational load on the
system, as well as latency to the signalling path.

Therefore, we provide a mechanism for registration of an in-kernel trigger
point that allows the VCPU to only require a very brief, lightweight
exit just long enough to signal an eventfd.  This also means that any
clients compatible with the eventfd interface (which includes userspace
and kernelspace equally well) can now register to be notified. The end
result should be a more flexible and higher performance notification API
for the backend KVM hypervisor and perhipheral components.

To test this theory, we built a test-harness called "doorbell".  This
module has a function called "doorbell_ring()" which simply increments a
counter for each time the doorbell is signaled.  It supports signalling
from either an eventfd, or an ioctl().

We then wired up two paths to the doorbell: One via QEMU via a registered
io region and through the doorbell ioctl().  The other is direct via
iosignalfd.

You can download this test harness here:

ftp://ftp.novell.com/dev/ghaskins/doorbell.tar.bz2

The measured results are as follows:

qemu-mmio:       110000 iops, 9.09us rtt
iosignalfd-mmio: 200100 iops, 5.00us rtt
iosignalfd-pio:  367300 iops, 2.72us rtt

I didn't measure qemu-pio, because I have to figure out how to register a
PIO region with qemu's device model, and I got lazy.  However, for now we
can extrapolate based on the data from the NULLIO runs of +2.56us for MMIO,
and -350ns for HC, we get:

qemu-pio:      153139 iops, 6.53us rtt
iosignalfd-hc: 412585 iops, 2.37us rtt

these are just for fun, for now, until I can gather more data.

Here is a graph for your convenience:

http://developer.novell.com/wiki/images/7/76/Iofd-chart.png

The conclusion to draw is that we save about 4us by skipping the userspace
hop.

--------------------

Signed-off-by: Gregory Haskins <ghaskins@novell.com>
---

 arch/x86/kvm/x86.c       |    1 
 include/linux/kvm.h      |   15 ++
 include/linux/kvm_host.h |   10 +
 virt/kvm/eventfd.c       |  389 ++++++++++++++++++++++++++++++++++++++++++++++
 virt/kvm/kvm_main.c      |   11 +
 5 files changed, 422 insertions(+), 4 deletions(-)

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 272e2e8..ef6c1d8 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -1097,6 +1097,7 @@ int kvm_dev_ioctl_check_extension(long ext)
 	case KVM_CAP_IRQ_INJECT_STATUS:
 	case KVM_CAP_ASSIGN_DEV_IRQ:
 	case KVM_CAP_IRQFD:
+	case KVM_CAP_IOSIGNALFD:
 	case KVM_CAP_PIT2:
 		r = 1;
 		break;
diff --git a/include/linux/kvm.h b/include/linux/kvm.h
index 29b62cc..5863a67 100644
--- a/include/linux/kvm.h
+++ b/include/linux/kvm.h
@@ -300,6 +300,19 @@ struct kvm_guest_debug {
 	struct kvm_guest_debug_arch arch;
 };
 
+#define KVM_IOSIGNALFD_FLAG_TRIGGER   (1 << 0)
+#define KVM_IOSIGNALFD_FLAG_PIO       (1 << 1)
+#define KVM_IOSIGNALFD_FLAG_DEASSIGN  (1 << 2)
+
+struct kvm_iosignalfd {
+	__u64 trigger;
+	__u64 addr;
+	__u32 len;
+	__u32 fd;
+	__u32 flags;
+	__u8  pad[36];
+};
+
 #define KVM_TRC_SHIFT           16
 /*
  * kvm trace categories
@@ -430,6 +443,7 @@ struct kvm_trace_rec {
 #ifdef __KVM_HAVE_PIT
 #define KVM_CAP_PIT2 33
 #endif
+#define KVM_CAP_IOSIGNALFD 34
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
@@ -535,6 +549,7 @@ struct kvm_irqfd {
 #define KVM_DEASSIGN_DEV_IRQ       _IOW(KVMIO, 0x75, struct kvm_assigned_irq)
 #define KVM_IRQFD                  _IOW(KVMIO, 0x76, struct kvm_irqfd)
 #define KVM_CREATE_PIT2		   _IOW(KVMIO, 0x77, struct kvm_pit_config)
+#define KVM_IOSIGNALFD             _IOW(KVMIO, 0x78, struct kvm_iosignalfd)
 
 /*
  * ioctls for vcpu fds
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index 2dbc3ec..adc351e 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -138,6 +138,7 @@ struct kvm {
 	struct kvm_io_bus pio_bus;
 #ifdef CONFIG_HAVE_KVM_EVENTFD
 	struct list_head irqfds;
+	struct list_head iosignalfds;
 #endif
 	struct kvm_vm_stat stat;
 	struct kvm_arch arch;
@@ -535,19 +536,24 @@ static inline void kvm_free_irq_routing(struct kvm *kvm) {}
 
 #ifdef CONFIG_HAVE_KVM_EVENTFD
 
-void kvm_irqfd_init(struct kvm *kvm);
+void kvm_eventfd_init(struct kvm *kvm);
 int kvm_irqfd(struct kvm *kvm, int fd, int gsi, int flags);
 void kvm_irqfd_release(struct kvm *kvm);
+int kvm_iosignalfd(struct kvm *kvm, struct kvm_iosignalfd *args);
 
 #else
 
-static inline void kvm_irqfd_init(struct kvm *kvm) {}
+static inline void kvm_eventfd_init(struct kvm *kvm) {}
 static inline int kvm_irqfd(struct kvm *kvm, int fd, int gsi, int flags)
 {
 	return -EINVAL;
 }
 
 static inline void kvm_irqfd_release(struct kvm *kvm) {}
+static inline int kvm_iosignalfd(struct kvm *kvm, struct kvm_iosignalfd *args)
+{
+	return -EINVAL;
+}
 
 #endif /* CONFIG_HAVE_KVM_EVENTFD */
 
diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c
index b3ed846..e4fceab 100644
--- a/virt/kvm/eventfd.c
+++ b/virt/kvm/eventfd.c
@@ -21,6 +21,7 @@
  */
 
 #include <linux/kvm_host.h>
+#include <linux/kvm.h>
 #include <linux/workqueue.h>
 #include <linux/syscalls.h>
 #include <linux/wait.h>
@@ -29,6 +30,8 @@
 #include <linux/list.h>
 #include <linux/eventfd.h>
 
+#include "iodev.h"
+
 /*
  * --------------------------------------------------------------------
  * irqfd: Allows an fd to be used to inject an interrupt to the guest
@@ -201,9 +204,10 @@ fail:
 }
 
 void
-kvm_irqfd_init(struct kvm *kvm)
+kvm_eventfd_init(struct kvm *kvm)
 {
 	INIT_LIST_HEAD(&kvm->irqfds);
+	INIT_LIST_HEAD(&kvm->iosignalfds);
 }
 
 void
@@ -214,3 +218,386 @@ kvm_irqfd_release(struct kvm *kvm)
 	list_for_each_entry_safe(irqfd, tmp, &kvm->irqfds, list)
 		irqfd_disconnect(irqfd);
 }
+
+/*
+ * --------------------------------------------------------------------
+ * iosignalfd: translate a PIO/MMIO memory write to an eventfd signal.
+ *
+ * userspace can register a PIO/MMIO address with an eventfd for recieving
+ * notification when the memory has been touched.
+ * --------------------------------------------------------------------
+ */
+
+/*
+ * Design note: We create one PIO/MMIO device (iosignalfd_group) which
+ * aggregates  one or more iosignalfd_items.  Each item points to exactly one
+ * eventfd, and can be registered to trigger on any write to the group
+ * (wildcard), or to a write of a specific value.  If more than one item is to
+ * be supported, the addr/len ranges must all be identical in the group.  If a
+ * trigger value is to be supported on a particular item, the group range must
+ * be exactly the width of the trigger.
+ */
+
+struct _iosignalfd_item {
+	struct list_head     list;
+	struct file         *file;
+	unsigned char       *match;
+	struct rcu_head      rcu;
+};
+
+struct _iosignalfd_group {
+	struct list_head     list;
+	u64                  addr;
+	size_t               length;
+	struct list_head     items;
+	struct kvm_io_device dev;
+	struct rcu_head      rcu;
+};
+
+static inline struct _iosignalfd_group *
+to_group(struct kvm_io_device *dev)
+{
+	return container_of(dev, struct _iosignalfd_group, dev);
+}
+
+static void
+iosignalfd_item_free(struct _iosignalfd_item *item)
+{
+	fput(item->file);
+	kfree(item->match);
+	kfree(item);
+}
+
+static void
+iosignalfd_item_deferred_free(struct rcu_head *rhp)
+{
+	struct _iosignalfd_item *item;
+
+	item = container_of(rhp, struct _iosignalfd_item, rcu);
+
+	iosignalfd_item_free(item);
+}
+
+static void
+iosignalfd_group_deferred_free(struct rcu_head *rhp)
+{
+	struct _iosignalfd_group *group;
+
+	group = container_of(rhp, struct _iosignalfd_group, rcu);
+
+	kfree(group);
+}
+
+static int
+iosignalfd_group_in_range(struct kvm_io_device *this, gpa_t addr, int len,
+			  int is_write)
+{
+	struct _iosignalfd_group *p = to_group(this);
+
+	return ((addr >= p->addr && (addr < p->addr + p->length)));
+}
+
+static int
+iosignalfd_is_match(struct _iosignalfd_group *group,
+		    struct _iosignalfd_item *item,
+		    const void *val,
+		    int len)
+{
+	if (!item->match)
+		/* wildcard is a hit */
+		return true;
+
+	if (len != group->length)
+		/* mis-matched length is a miss */
+		return false;
+
+	/* otherwise, we have to actually compare the data */
+	return !memcmp(item->match, val, len) ? true : false;
+}
+
+/*
+ * MMIO/PIO writes trigger an event (if the data matches).
+ *
+ * This is invoked by the io_bus subsystem in response to an address match
+ * against the group.  We must then walk the list of individual items to check
+ * for a match and, if applicable, to send the appropriate signal. If the item
+ * in question does not have a "match" pointer, it is considered a wildcard
+ * and will always generate a signal.  There can be an arbitrary number
+ * of distinct matches or wildcards per group.
+ */
+static void
+iosignalfd_group_write(struct kvm_io_device *this, gpa_t addr, int len,
+		       const void *val)
+{
+	struct _iosignalfd_group *group = to_group(this);
+	struct _iosignalfd_item *item;
+
+	rcu_read_lock();
+
+	list_for_each_entry_rcu(item, &group->items, list) {
+		if (iosignalfd_is_match(group, item, val, len))
+			eventfd_signal(item->file, 1);
+	}
+
+	rcu_read_unlock();
+}
+
+/*
+ * MMIO/PIO reads against the group indiscriminately return all zeros
+ */
+static void
+iosignalfd_group_read(struct kvm_io_device *this, gpa_t addr, int len,
+		      void *val)
+{
+	memset(val, 0, len);
+}
+
+/*
+ * This function is called as KVM is completely shutting down.  We do not
+ * need to worry about locking or careful RCU dancing...just nuke anything
+ * we have as quickly as possible
+ */
+static void
+iosignalfd_group_destructor(struct kvm_io_device *this)
+{
+	struct _iosignalfd_group *group = to_group(this);
+	struct _iosignalfd_item *item, *tmp;
+
+	list_for_each_entry_safe(item, tmp, &group->items, list) {
+		list_del(&item->list);
+		iosignalfd_item_free(item);
+	}
+
+	list_del(&group->list);
+	kfree(group);
+}
+
+static const struct kvm_io_device_ops iosignalfd_ops = {
+	.read       = iosignalfd_group_read,
+	.write      = iosignalfd_group_write,
+	.in_range   = iosignalfd_group_in_range,
+	.destructor = iosignalfd_group_destructor,
+};
+
+/* assumes kvm->lock held */
+static struct _iosignalfd_group *
+iosignalfd_group_find(struct kvm *kvm, u64 addr)
+{
+	struct _iosignalfd_group *group;
+
+	list_for_each_entry(group, &kvm->iosignalfds, list) {
+		if (group->addr == addr)
+			return group;
+	}
+
+	return NULL;
+}
+
+/*
+ * Atomically find an existing group, or create a new one if it doesn't already
+ * exist.
+ *
+ * assumes kvm->lock is held
+ */
+static struct _iosignalfd_group *
+iosignalfd_group_get(struct kvm *kvm, struct kvm_io_bus *bus,
+		      u64 addr, size_t len)
+{
+	struct _iosignalfd_group *group;
+
+	group = iosignalfd_group_find(kvm, addr);
+	if (!group) {
+		int ret;
+
+		group = kzalloc(sizeof(*group), GFP_KERNEL);
+		if (!group)
+			return ERR_PTR(-ENOMEM);
+
+		INIT_LIST_HEAD(&group->list);
+		INIT_LIST_HEAD(&group->items);
+		group->addr   = addr;
+		group->length = len;
+		kvm_iodevice_init(&group->dev, &iosignalfd_ops);
+
+		ret = kvm_io_bus_register_dev(bus, &group->dev);
+		if (ret < 0) {
+			kfree(group);
+			return ERR_PTR(ret);
+		}
+
+		list_add_tail(&group->list, &kvm->iosignalfds);
+
+	} else if (group->length != len)
+		/*
+		 * Existing groups must have the same addr/len tuple or we
+		 * reject the request
+		 */
+		return ERR_PTR(-EINVAL);
+
+	return group;
+}
+
+static int
+kvm_assign_iosignalfd(struct kvm *kvm, struct kvm_iosignalfd *args)
+{
+	int                       pio = args->flags & KVM_IOSIGNALFD_FLAG_PIO;
+	struct kvm_io_bus        *bus = pio ? &kvm->pio_bus : &kvm->mmio_bus;
+	struct _iosignalfd_group *group = NULL;
+	struct _iosignalfd_item  *item = NULL;
+	struct file              *file;
+	int                       ret;
+
+	file = eventfd_fget(args->fd);
+	if (IS_ERR(file))
+		return PTR_ERR(file);
+
+	item = kzalloc(sizeof(*item), GFP_KERNEL);
+	if (!item) {
+		ret = -ENOMEM;
+		goto fail;
+	}
+
+	INIT_LIST_HEAD(&item->list);
+	item->file = file;
+
+	/*
+	 * Registering a "trigger" address is optional.  If this flag
+	 * is not specified, we leave the item->match pointer NULL, which
+	 * indicates a wildcard
+	 */
+	if (args->flags & KVM_IOSIGNALFD_FLAG_TRIGGER) {
+		if (args->len > sizeof(u64)) {
+			ret = -EINVAL;
+			goto fail;
+		}
+
+		item->match = kzalloc(args->len, GFP_KERNEL);
+		if (!item->match) {
+			ret = -ENOMEM;
+			goto fail;
+		}
+
+		if (copy_from_user(item->match,
+				   (void *)args->trigger,
+				   args->len)) {
+			ret = -EFAULT;
+			goto fail;
+		}
+	}
+
+	mutex_lock(&kvm->lock);
+
+	group = iosignalfd_group_get(kvm, bus, args->addr, args->len);
+	if (IS_ERR(group)) {
+		ret = PTR_ERR(group);
+		mutex_unlock(&kvm->lock);
+		goto fail;
+	}
+
+	/*
+	 * Note: We are committed to succeed at this point since we have
+	 * (potentially) published a new group-device.  Any failure handling
+	 * added in the future after this point will need to be carefully
+	 * considered.
+	 */
+
+	list_add_tail_rcu(&item->list, &group->items);
+
+	mutex_unlock(&kvm->lock);
+
+	return 0;
+
+fail:
+	if (item) {
+		/*
+		 * it would have never made it to the group->items list
+		 * in the failure path, so we dont need to worry about removing
+		 * it
+		 */
+		kfree(item->match);
+		kfree(item);
+	}
+
+	fput(file);
+
+	return ret;
+}
+
+
+static int
+kvm_deassign_iosignalfd(struct kvm *kvm, struct kvm_iosignalfd *args)
+{
+	int                       pio = args->flags & KVM_IOSIGNALFD_FLAG_PIO;
+	struct kvm_io_bus        *bus = pio ? &kvm->pio_bus : &kvm->mmio_bus;
+	struct _iosignalfd_group *group;
+	struct _iosignalfd_item  *item, *tmp;
+	struct file              *file;
+	int                       ret = 0;
+
+	file = eventfd_fget(args->fd);
+	if (IS_ERR(file))
+		return PTR_ERR(file);
+
+	mutex_lock(&kvm->lock);
+
+	group = iosignalfd_group_find(kvm, args->addr);
+	if (!group) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	/*
+	 * Exhaustively search our group->items list for any items that might
+	 * match the specified fd, and (carefully) remove each one found.
+	 */
+	list_for_each_entry_safe(item, tmp, &group->items, list) {
+
+		if (item->file != file)
+			continue;
+
+		list_del_rcu(&item->list);
+
+		/*
+		 * The item may be still referenced inside our group->write()
+		 * path's RCU read-side CS, so defer the actual free to the
+		 * next grace
+		 */
+		call_rcu(&item->rcu, iosignalfd_item_deferred_free);
+	}
+
+	/*
+	 * Check if the group is now completely vacated as a result of
+	 * removing the items.  If so, unregister/delete it
+	 */
+	if (list_empty(&group->items)) {
+
+		kvm_io_bus_unregister_dev(bus, &group->dev);
+
+		/*
+		 * Like the item, the group may also still be referenced as
+		 * per above.  However, the kvm->iosignalfds list is not
+		 * RCU protected (its protected by kvm->lock instead) so
+		 * we can just plain-vanilla remove it.  What needs to be
+		 * done carefully is the actual freeing of the group pointer
+		 * since we walk the group->items list within the RCU CS.
+		 */
+		list_del(&group->list);
+		call_rcu(&group->rcu, iosignalfd_group_deferred_free);
+	}
+
+out:
+	mutex_unlock(&kvm->lock);
+
+	fput(file);
+
+	return ret;
+}
+
+int
+kvm_iosignalfd(struct kvm *kvm, struct kvm_iosignalfd *args)
+{
+	if (args->flags & KVM_IOSIGNALFD_FLAG_DEASSIGN)
+		return kvm_deassign_iosignalfd(kvm, args);
+
+	return kvm_assign_iosignalfd(kvm, args);
+}
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 5980bec..5bf6988 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -979,7 +979,7 @@ static struct kvm *kvm_create_vm(void)
 	atomic_inc(&kvm->mm->mm_count);
 	spin_lock_init(&kvm->mmu_lock);
 	kvm_io_bus_init(&kvm->pio_bus);
-	kvm_irqfd_init(kvm);
+	kvm_eventfd_init(kvm);
 	mutex_init(&kvm->lock);
 	mutex_init(&kvm->irq_lock);
 	kvm_io_bus_init(&kvm->mmio_bus);
@@ -2219,6 +2219,15 @@ static long kvm_vm_ioctl(struct file *filp,
 		r = kvm_irqfd(kvm, data.fd, data.gsi, data.flags);
 		break;
 	}
+	case KVM_IOSIGNALFD: {
+		struct kvm_iosignalfd entry;
+
+		r = -EFAULT;
+		if (copy_from_user(&entry, argp, sizeof entry))
+			goto out;
+		r = kvm_iosignalfd(kvm, &entry);
+		break;
+	}
 	default:
 		r = kvm_arch_vm_ioctl(filp, ioctl, arg);
 	}

Re: [KVM PATCH v6 2/2] KVM: add iosignalfd support

[Gregory Haskins]

[-- Attachment #1: Type: text/plain, Size: 20050 bytes --]

Hi Marcelo!

Comments about the shutdown path ambiguity are in-line

Gregory Haskins wrote:
> iosignalfd is a mechanism to register PIO/MMIO regions to trigger an eventfd
> signal when written to by a guest.  Host userspace can register any arbitrary
> IO address with a corresponding eventfd and then pass the eventfd to a
> specific end-point of interest for handling.
>
> Normal IO requires a blocking round-trip since the operation may cause
> side-effects in the emulated model or may return data to the caller.
> Therefore, an IO in KVM traps from the guest to the host, causes a VMX/SVM
> "heavy-weight" exit back to userspace, and is ultimately serviced by qemu's
> device model synchronously before returning control back to the vcpu.
>
> However, there is a subclass of IO which acts purely as a trigger for
> other IO (such as to kick off an out-of-band DMA request, etc).  For these
> patterns, the synchronous call is particularly expensive since we really
> only want to simply get our notification transmitted asychronously and
> return as quickly as possible.  All the sychronous infrastructure to ensure
> proper data-dependencies are met in the normal IO case are just unecessary
> overhead for signalling.  This adds additional computational load on the
> system, as well as latency to the signalling path.
>
> Therefore, we provide a mechanism for registration of an in-kernel trigger
> point that allows the VCPU to only require a very brief, lightweight
> exit just long enough to signal an eventfd.  This also means that any
> clients compatible with the eventfd interface (which includes userspace
> and kernelspace equally well) can now register to be notified. The end
> result should be a more flexible and higher performance notification API
> for the backend KVM hypervisor and perhipheral components.
>
> To test this theory, we built a test-harness called "doorbell".  This
> module has a function called "doorbell_ring()" which simply increments a
> counter for each time the doorbell is signaled.  It supports signalling
> from either an eventfd, or an ioctl().
>
> We then wired up two paths to the doorbell: One via QEMU via a registered
> io region and through the doorbell ioctl().  The other is direct via
> iosignalfd.
>
> You can download this test harness here:
>
> ftp://ftp.novell.com/dev/ghaskins/doorbell.tar.bz2
>
> The measured results are as follows:
>
> qemu-mmio:       110000 iops, 9.09us rtt
> iosignalfd-mmio: 200100 iops, 5.00us rtt
> iosignalfd-pio:  367300 iops, 2.72us rtt
>
> I didn't measure qemu-pio, because I have to figure out how to register a
> PIO region with qemu's device model, and I got lazy.  However, for now we
> can extrapolate based on the data from the NULLIO runs of +2.56us for MMIO,
> and -350ns for HC, we get:
>
> qemu-pio:      153139 iops, 6.53us rtt
> iosignalfd-hc: 412585 iops, 2.37us rtt
>
> these are just for fun, for now, until I can gather more data.
>
> Here is a graph for your convenience:
>
> http://developer.novell.com/wiki/images/7/76/Iofd-chart.png
>
> The conclusion to draw is that we save about 4us by skipping the userspace
> hop.
>
> --------------------
>
> Signed-off-by: Gregory Haskins <ghaskins@novell.com>
> ---
>
>  arch/x86/kvm/x86.c       |    1 
>  include/linux/kvm.h      |   15 ++
>  include/linux/kvm_host.h |   10 +
>  virt/kvm/eventfd.c       |  389 ++++++++++++++++++++++++++++++++++++++++++++++
>  virt/kvm/kvm_main.c      |   11 +
>  5 files changed, 422 insertions(+), 4 deletions(-)
>
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 272e2e8..ef6c1d8 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -1097,6 +1097,7 @@ int kvm_dev_ioctl_check_extension(long ext)
>  	case KVM_CAP_IRQ_INJECT_STATUS:
>  	case KVM_CAP_ASSIGN_DEV_IRQ:
>  	case KVM_CAP_IRQFD:
> +	case KVM_CAP_IOSIGNALFD:
>  	case KVM_CAP_PIT2:
>  		r = 1;
>  		break;
> diff --git a/include/linux/kvm.h b/include/linux/kvm.h
> index 29b62cc..5863a67 100644
> --- a/include/linux/kvm.h
> +++ b/include/linux/kvm.h
> @@ -300,6 +300,19 @@ struct kvm_guest_debug {
>  	struct kvm_guest_debug_arch arch;
>  };
>  
> +#define KVM_IOSIGNALFD_FLAG_TRIGGER   (1 << 0)
> +#define KVM_IOSIGNALFD_FLAG_PIO       (1 << 1)
> +#define KVM_IOSIGNALFD_FLAG_DEASSIGN  (1 << 2)
> +
> +struct kvm_iosignalfd {
> +	__u64 trigger;
> +	__u64 addr;
> +	__u32 len;
> +	__u32 fd;
> +	__u32 flags;
> +	__u8  pad[36];
> +};
> +
>  #define KVM_TRC_SHIFT           16
>  /*
>   * kvm trace categories
> @@ -430,6 +443,7 @@ struct kvm_trace_rec {
>  #ifdef __KVM_HAVE_PIT
>  #define KVM_CAP_PIT2 33
>  #endif
> +#define KVM_CAP_IOSIGNALFD 34
>  
>  #ifdef KVM_CAP_IRQ_ROUTING
>  
> @@ -535,6 +549,7 @@ struct kvm_irqfd {
>  #define KVM_DEASSIGN_DEV_IRQ       _IOW(KVMIO, 0x75, struct kvm_assigned_irq)
>  #define KVM_IRQFD                  _IOW(KVMIO, 0x76, struct kvm_irqfd)
>  #define KVM_CREATE_PIT2		   _IOW(KVMIO, 0x77, struct kvm_pit_config)
> +#define KVM_IOSIGNALFD             _IOW(KVMIO, 0x78, struct kvm_iosignalfd)
>  
>  /*
>   * ioctls for vcpu fds
> diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
> index 2dbc3ec..adc351e 100644
> --- a/include/linux/kvm_host.h
> +++ b/include/linux/kvm_host.h
> @@ -138,6 +138,7 @@ struct kvm {
>  	struct kvm_io_bus pio_bus;
>  #ifdef CONFIG_HAVE_KVM_EVENTFD
>  	struct list_head irqfds;
> +	struct list_head iosignalfds;
>  #endif
>  	struct kvm_vm_stat stat;
>  	struct kvm_arch arch;
> @@ -535,19 +536,24 @@ static inline void kvm_free_irq_routing(struct kvm *kvm) {}
>  
>  #ifdef CONFIG_HAVE_KVM_EVENTFD
>  
> -void kvm_irqfd_init(struct kvm *kvm);
> +void kvm_eventfd_init(struct kvm *kvm);
>  int kvm_irqfd(struct kvm *kvm, int fd, int gsi, int flags);
>  void kvm_irqfd_release(struct kvm *kvm);
> +int kvm_iosignalfd(struct kvm *kvm, struct kvm_iosignalfd *args);
>  
>  #else
>  
> -static inline void kvm_irqfd_init(struct kvm *kvm) {}
> +static inline void kvm_eventfd_init(struct kvm *kvm) {}
>  static inline int kvm_irqfd(struct kvm *kvm, int fd, int gsi, int flags)
>  {
>  	return -EINVAL;
>  }
>  
>  static inline void kvm_irqfd_release(struct kvm *kvm) {}
> +static inline int kvm_iosignalfd(struct kvm *kvm, struct kvm_iosignalfd *args)
> +{
> +	return -EINVAL;
> +}
>  
>  #endif /* CONFIG_HAVE_KVM_EVENTFD */
>  
> diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c
> index b3ed846..e4fceab 100644
> --- a/virt/kvm/eventfd.c
> +++ b/virt/kvm/eventfd.c
> @@ -21,6 +21,7 @@
>   */
>  
>  #include <linux/kvm_host.h>
> +#include <linux/kvm.h>
>  #include <linux/workqueue.h>
>  #include <linux/syscalls.h>
>  #include <linux/wait.h>
> @@ -29,6 +30,8 @@
>  #include <linux/list.h>
>  #include <linux/eventfd.h>
>  
> +#include "iodev.h"
> +
>  /*
>   * --------------------------------------------------------------------
>   * irqfd: Allows an fd to be used to inject an interrupt to the guest
> @@ -201,9 +204,10 @@ fail:
>  }
>  
>  void
> -kvm_irqfd_init(struct kvm *kvm)
> +kvm_eventfd_init(struct kvm *kvm)
>  {
>  	INIT_LIST_HEAD(&kvm->irqfds);
> +	INIT_LIST_HEAD(&kvm->iosignalfds);
>  }
>  
>  void
> @@ -214,3 +218,386 @@ kvm_irqfd_release(struct kvm *kvm)
>  	list_for_each_entry_safe(irqfd, tmp, &kvm->irqfds, list)
>  		irqfd_disconnect(irqfd);
>  }
> +
> +/*
> + * --------------------------------------------------------------------
> + * iosignalfd: translate a PIO/MMIO memory write to an eventfd signal.
> + *
> + * userspace can register a PIO/MMIO address with an eventfd for recieving
> + * notification when the memory has been touched.
> + * --------------------------------------------------------------------
> + */
> +
> +/*
> + * Design note: We create one PIO/MMIO device (iosignalfd_group) which
> + * aggregates  one or more iosignalfd_items.  Each item points to exactly one
> + * eventfd, and can be registered to trigger on any write to the group
> + * (wildcard), or to a write of a specific value.  If more than one item is to
> + * be supported, the addr/len ranges must all be identical in the group.  If a
> + * trigger value is to be supported on a particular item, the group range must
> + * be exactly the width of the trigger.
> + */
> +
> +struct _iosignalfd_item {
> +	struct list_head     list;
> +	struct file         *file;
> +	unsigned char       *match;
> +	struct rcu_head      rcu;
> +};
> +
> +struct _iosignalfd_group {
> +	struct list_head     list;
> +	u64                  addr;
> +	size_t               length;
> +	struct list_head     items;
> +	struct kvm_io_device dev;
> +	struct rcu_head      rcu;
> +};
> +
> +static inline struct _iosignalfd_group *
> +to_group(struct kvm_io_device *dev)
> +{
> +	return container_of(dev, struct _iosignalfd_group, dev);
> +}
> +
> +static void
> +iosignalfd_item_free(struct _iosignalfd_item *item)
> +{
> +	fput(item->file);
> +	kfree(item->match);
> +	kfree(item);
> +}
> +
> +static void
> +iosignalfd_item_deferred_free(struct rcu_head *rhp)
> +{
> +	struct _iosignalfd_item *item;
> +
> +	item = container_of(rhp, struct _iosignalfd_item, rcu);
> +
> +	iosignalfd_item_free(item);
> +}
> +
> +static void
> +iosignalfd_group_deferred_free(struct rcu_head *rhp)
> +{
> +	struct _iosignalfd_group *group;
> +
> +	group = container_of(rhp, struct _iosignalfd_group, rcu);
> +
> +	kfree(group);
> +}
> +
> +static int
> +iosignalfd_group_in_range(struct kvm_io_device *this, gpa_t addr, int len,
> +			  int is_write)
> +{
> +	struct _iosignalfd_group *p = to_group(this);
> +
> +	return ((addr >= p->addr && (addr < p->addr + p->length)));
> +}
> +
> +static int
> +iosignalfd_is_match(struct _iosignalfd_group *group,
> +		    struct _iosignalfd_item *item,
> +		    const void *val,
> +		    int len)
> +{
> +	if (!item->match)
> +		/* wildcard is a hit */
> +		return true;
> +
> +	if (len != group->length)
> +		/* mis-matched length is a miss */
> +		return false;
> +
> +	/* otherwise, we have to actually compare the data */
> +	return !memcmp(item->match, val, len) ? true : false;
> +}
> +
> +/*
> + * MMIO/PIO writes trigger an event (if the data matches).
> + *
> + * This is invoked by the io_bus subsystem in response to an address match
> + * against the group.  We must then walk the list of individual items to check
> + * for a match and, if applicable, to send the appropriate signal. If the item
> + * in question does not have a "match" pointer, it is considered a wildcard
> + * and will always generate a signal.  There can be an arbitrary number
> + * of distinct matches or wildcards per group.
> + */
> +static void
> +iosignalfd_group_write(struct kvm_io_device *this, gpa_t addr, int len,
> +		       const void *val)
> +{
> +	struct _iosignalfd_group *group = to_group(this);
> +	struct _iosignalfd_item *item;
> +
> +	rcu_read_lock();
> +
> +	list_for_each_entry_rcu(item, &group->items, list) {
> +		if (iosignalfd_is_match(group, item, val, len))
> +			eventfd_signal(item->file, 1);
> +	}
> +
> +	rcu_read_unlock();
> +}
> +
> +/*
> + * MMIO/PIO reads against the group indiscriminately return all zeros
> + */
> +static void
> +iosignalfd_group_read(struct kvm_io_device *this, gpa_t addr, int len,
> +		      void *val)
> +{
> +	memset(val, 0, len);
> +}
> +
> +/*
> + * This function is called as KVM is completely shutting down.  We do not
> + * need to worry about locking or careful RCU dancing...just nuke anything
> + * we have as quickly as possible
> + */
> +static void
> +iosignalfd_group_destructor(struct kvm_io_device *this)
> +{
> +	struct _iosignalfd_group *group = to_group(this);
> +	struct _iosignalfd_item *item, *tmp;
> +
> +	list_for_each_entry_safe(item, tmp, &group->items, list) {
> +		list_del(&item->list);
> +		iosignalfd_item_free(item);
> +	}
> +
> +	list_del(&group->list);
> +	kfree(group);
> +}
>   

So this function is called by the path that executes as we do the last
kvm_put_kvm().  I do not do any careful RCU wrangling here because I
assume that there cannot possibly be any active MMIO/PIO operations at
this time, or the reference would never have dropped.  Let me know if
anyone sees any holes in that.

An alternative approach is to do this similar to how irqfd_release()
works.   That is: invoke it from the vmfd release() path instead of the
the kvm object destructor.  I currently do not think this is necessary,
but I will throw that out there in case someone likes it better.

-Greg

> +
> +static const struct kvm_io_device_ops iosignalfd_ops = {
> +	.read       = iosignalfd_group_read,
> +	.write      = iosignalfd_group_write,
> +	.in_range   = iosignalfd_group_in_range,
> +	.destructor = iosignalfd_group_destructor,
> +};
> +
> +/* assumes kvm->lock held */
> +static struct _iosignalfd_group *
> +iosignalfd_group_find(struct kvm *kvm, u64 addr)
> +{
> +	struct _iosignalfd_group *group;
> +
> +	list_for_each_entry(group, &kvm->iosignalfds, list) {
> +		if (group->addr == addr)
> +			return group;
> +	}
> +
> +	return NULL;
> +}
> +
> +/*
> + * Atomically find an existing group, or create a new one if it doesn't already
> + * exist.
> + *
> + * assumes kvm->lock is held
> + */
> +static struct _iosignalfd_group *
> +iosignalfd_group_get(struct kvm *kvm, struct kvm_io_bus *bus,
> +		      u64 addr, size_t len)
> +{
> +	struct _iosignalfd_group *group;
> +
> +	group = iosignalfd_group_find(kvm, addr);
> +	if (!group) {
> +		int ret;
> +
> +		group = kzalloc(sizeof(*group), GFP_KERNEL);
> +		if (!group)
> +			return ERR_PTR(-ENOMEM);
> +
> +		INIT_LIST_HEAD(&group->list);
> +		INIT_LIST_HEAD(&group->items);
> +		group->addr   = addr;
> +		group->length = len;
> +		kvm_iodevice_init(&group->dev, &iosignalfd_ops);
> +
> +		ret = kvm_io_bus_register_dev(bus, &group->dev);
> +		if (ret < 0) {
> +			kfree(group);
> +			return ERR_PTR(ret);
> +		}
> +
> +		list_add_tail(&group->list, &kvm->iosignalfds);
> +
> +	} else if (group->length != len)
> +		/*
> +		 * Existing groups must have the same addr/len tuple or we
> +		 * reject the request
> +		 */
> +		return ERR_PTR(-EINVAL);
> +
> +	return group;
> +}
> +
> +static int
> +kvm_assign_iosignalfd(struct kvm *kvm, struct kvm_iosignalfd *args)
> +{
> +	int                       pio = args->flags & KVM_IOSIGNALFD_FLAG_PIO;
> +	struct kvm_io_bus        *bus = pio ? &kvm->pio_bus : &kvm->mmio_bus;
> +	struct _iosignalfd_group *group = NULL;
> +	struct _iosignalfd_item  *item = NULL;
> +	struct file              *file;
> +	int                       ret;
> +
> +	file = eventfd_fget(args->fd);
> +	if (IS_ERR(file))
> +		return PTR_ERR(file);
> +
> +	item = kzalloc(sizeof(*item), GFP_KERNEL);
> +	if (!item) {
> +		ret = -ENOMEM;
> +		goto fail;
> +	}
> +
> +	INIT_LIST_HEAD(&item->list);
> +	item->file = file;
> +
> +	/*
> +	 * Registering a "trigger" address is optional.  If this flag
> +	 * is not specified, we leave the item->match pointer NULL, which
> +	 * indicates a wildcard
> +	 */
> +	if (args->flags & KVM_IOSIGNALFD_FLAG_TRIGGER) {
> +		if (args->len > sizeof(u64)) {
> +			ret = -EINVAL;
> +			goto fail;
> +		}
> +
> +		item->match = kzalloc(args->len, GFP_KERNEL);
> +		if (!item->match) {
> +			ret = -ENOMEM;
> +			goto fail;
> +		}
> +
> +		if (copy_from_user(item->match,
> +				   (void *)args->trigger,
> +				   args->len)) {
> +			ret = -EFAULT;
> +			goto fail;
> +		}
> +	}
> +
> +	mutex_lock(&kvm->lock);
> +
> +	group = iosignalfd_group_get(kvm, bus, args->addr, args->len);
> +	if (IS_ERR(group)) {
> +		ret = PTR_ERR(group);
> +		mutex_unlock(&kvm->lock);
> +		goto fail;
> +	}
> +
> +	/*
> +	 * Note: We are committed to succeed at this point since we have
> +	 * (potentially) published a new group-device.  Any failure handling
> +	 * added in the future after this point will need to be carefully
> +	 * considered.
> +	 */
> +
> +	list_add_tail_rcu(&item->list, &group->items);
> +
> +	mutex_unlock(&kvm->lock);
> +
> +	return 0;
> +
> +fail:
> +	if (item) {
> +		/*
> +		 * it would have never made it to the group->items list
> +		 * in the failure path, so we dont need to worry about removing
> +		 * it
> +		 */
> +		kfree(item->match);
> +		kfree(item);
> +	}
> +
> +	fput(file);
> +
> +	return ret;
> +}
> +
> +
> +static int
> +kvm_deassign_iosignalfd(struct kvm *kvm, struct kvm_iosignalfd *args)
> +{
> +	int                       pio = args->flags & KVM_IOSIGNALFD_FLAG_PIO;
> +	struct kvm_io_bus        *bus = pio ? &kvm->pio_bus : &kvm->mmio_bus;
> +	struct _iosignalfd_group *group;
> +	struct _iosignalfd_item  *item, *tmp;
> +	struct file              *file;
> +	int                       ret = 0;
> +
> +	file = eventfd_fget(args->fd);
> +	if (IS_ERR(file))
> +		return PTR_ERR(file);
> +
> +	mutex_lock(&kvm->lock);
> +
> +	group = iosignalfd_group_find(kvm, args->addr);
> +	if (!group) {
> +		ret = -EINVAL;
> +		goto out;
> +	}
> +
> +	/*
> +	 * Exhaustively search our group->items list for any items that might
> +	 * match the specified fd, and (carefully) remove each one found.
> +	 */
> +	list_for_each_entry_safe(item, tmp, &group->items, list) {
> +
> +		if (item->file != file)
> +			continue;
> +
> +		list_del_rcu(&item->list);
> +
> +		/*
> +		 * The item may be still referenced inside our group->write()
> +		 * path's RCU read-side CS, so defer the actual free to the
> +		 * next grace
> +		 */
> +		call_rcu(&item->rcu, iosignalfd_item_deferred_free);
> +	}
> +
> +	/*
> +	 * Check if the group is now completely vacated as a result of
> +	 * removing the items.  If so, unregister/delete it
> +	 */
> +	if (list_empty(&group->items)) {
> +
> +		kvm_io_bus_unregister_dev(bus, &group->dev);
> +
> +		/*
> +		 * Like the item, the group may also still be referenced as
> +		 * per above.  However, the kvm->iosignalfds list is not
> +		 * RCU protected (its protected by kvm->lock instead) so
> +		 * we can just plain-vanilla remove it.  What needs to be
> +		 * done carefully is the actual freeing of the group pointer
> +		 * since we walk the group->items list within the RCU CS.
> +		 */
> +		list_del(&group->list);
> +		call_rcu(&group->rcu, iosignalfd_group_deferred_free);
> +	}
> +
> +out:
> +	mutex_unlock(&kvm->lock);
> +
> +	fput(file);
> +
> +	return ret;
> +}
> +
> +int
> +kvm_iosignalfd(struct kvm *kvm, struct kvm_iosignalfd *args)
> +{
> +	if (args->flags & KVM_IOSIGNALFD_FLAG_DEASSIGN)
> +		return kvm_deassign_iosignalfd(kvm, args);
> +
> +	return kvm_assign_iosignalfd(kvm, args);
> +}
> diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
> index 5980bec..5bf6988 100644
> --- a/virt/kvm/kvm_main.c
> +++ b/virt/kvm/kvm_main.c
> @@ -979,7 +979,7 @@ static struct kvm *kvm_create_vm(void)
>  	atomic_inc(&kvm->mm->mm_count);
>  	spin_lock_init(&kvm->mmu_lock);
>  	kvm_io_bus_init(&kvm->pio_bus);
> -	kvm_irqfd_init(kvm);
> +	kvm_eventfd_init(kvm);
>  	mutex_init(&kvm->lock);
>  	mutex_init(&kvm->irq_lock);
>  	kvm_io_bus_init(&kvm->mmio_bus);
> @@ -2219,6 +2219,15 @@ static long kvm_vm_ioctl(struct file *filp,
>  		r = kvm_irqfd(kvm, data.fd, data.gsi, data.flags);
>  		break;
>  	}
> +	case KVM_IOSIGNALFD: {
> +		struct kvm_iosignalfd entry;
> +
> +		r = -EFAULT;
> +		if (copy_from_user(&entry, argp, sizeof entry))
> +			goto out;
> +		r = kvm_iosignalfd(kvm, &entry);
> +		break;
> +	}
>  	default:
>  		r = kvm_arch_vm_ioctl(filp, ioctl, arg);
>  	}
>
> --
> To unsubscribe from this list: send the line "unsubscribe kvm" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>   



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 266 bytes --]

Re: [KVM PATCH v6 2/2] KVM: add iosignalfd support

[Marcelo Tosatti]

On Fri, Jun 05, 2009 at 12:02:02PM -0400, Gregory Haskins wrote:
> Hi Marcelo!
> 
> Comments about the shutdown path ambiguity are in-line
> 
> Gregory Haskins wrote:
> > iosignalfd is a mechanism to register PIO/MMIO regions to trigger an eventfd
> > signal when written to by a guest.  Host userspace can register any arbitrary
> > IO address with a corresponding eventfd and then pass the eventfd to a
> > +		list_del(&item->list);
> > +		iosignalfd_item_free(item);
> > +	}
> > +
> > +	list_del(&group->list);
> > +	kfree(group);
> > +}
> >   
> 
> So this function is called by the path that executes as we do the last
> kvm_put_kvm().  I do not do any careful RCU wrangling here because I
> assume that there cannot possibly be any active MMIO/PIO operations at
> this time, or the reference would never have dropped.  Let me know if
> anyone sees any holes in that.
> 
> An alternative approach is to do this similar to how irqfd_release()
> works.   That is: invoke it from the vmfd release() path instead of the
> the kvm object destructor.  I currently do not think this is necessary,
> but I will throw that out there in case someone likes it better.

Gregory,

Can't see any problems with it. You might want an upper limit
in the number of items per group.

Re: [KVM PATCH v6 2/2] KVM: add iosignalfd support

[Gregory Haskins]

[-- Attachment #1: Type: text/plain, Size: 1556 bytes --]

Hi Marcelo,

Marcelo Tosatti wrote:
> On Fri, Jun 05, 2009 at 12:02:02PM -0400, Gregory Haskins wrote:
>   
>> Hi Marcelo!
>>
>> Comments about the shutdown path ambiguity are in-line
>>
>> Gregory Haskins wrote:
>>     
>>> iosignalfd is a mechanism to register PIO/MMIO regions to trigger an eventfd
>>> signal when written to by a guest.  Host userspace can register any arbitrary
>>> IO address with a corresponding eventfd and then pass the eventfd to a
>>> +		list_del(&item->list);
>>> +		iosignalfd_item_free(item);
>>> +	}
>>> +
>>> +	list_del(&group->list);
>>> +	kfree(group);
>>> +}
>>>   
>>>       
>> So this function is called by the path that executes as we do the last
>> kvm_put_kvm().  I do not do any careful RCU wrangling here because I
>> assume that there cannot possibly be any active MMIO/PIO operations at
>> this time, or the reference would never have dropped.  Let me know if
>> anyone sees any holes in that.
>>
>> An alternative approach is to do this similar to how irqfd_release()
>> works.   That is: invoke it from the vmfd release() path instead of the
>> the kvm object destructor.  I currently do not think this is necessary,
>> but I will throw that out there in case someone likes it better.
>>     
>
> Gregory,
>
> Can't see any problems with it.

Thanks for the review!


>  You might want an upper limit
> in the number of items per group.
>   


Yeah, I agree that is a good idea.  Will fix in v7.

BTW: Did your series get merged while I was away?

-Greg


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 266 bytes --]