[PATCHv11 0/3] rdmacg: IB/core: rdma controller support

[PATCHv11 0/3] rdmacg: IB/core: rdma controller support

[Parav Pandit]

rdmacg: IB/core: rdma controller support

Patch is generated and tested against below Doug's linux-rdma
git tree.

URL: git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma.git
Branch: master

Patchset is also compiled and tested against below Tejun's cgroup tree
using cgroup v2 mode.
URL: git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup.git
Branch: master

Overview:
Currently user space applications can easily take away all the rdma
device specific resources such as AH, CQ, QP, MR etc. Due to which other
applications in other cgroup or kernel space ULPs may not even get chance
to allocate any rdma resources. This results into service unavailibility.

RDMA cgroup addresses this issue by allowing resource accounting,
limit enforcement on per cgroup, per rdma device basis.

RDMA uverbs layer will enforce limits on well defined RDMA verb
resources without any HCA vendor device driver involvement.

RDMA uverbs layer will not do limit enforcement of HCA hw vendor
specific resources. Instead rdma cgroup provides set of APIs
through which vendor specific drivers can do resource accounting
by making use of rdma cgroup.

Resource limit enforcement is hierarchical.

When process is migrated with active RDMA resources, rdma cgroup
continues to uncharge original cgroup for allocated resource. New resource
is charged to current process's cgroup, which means if the process is
migrated with active resources, for new resources it will be charged to
new cgroup and old resources will be correctly uncharged from old cgroup.

Changes from v10:
  * (To address comments from Tejun, Christoph)
   1. Removed unused rpool_list_lock from rdma_cgroup structure.
   2. Moved rdma resource definition to rdma cgroup instead of IB stack
   3. Added prefix rdmacg to static instances
   4. Simplified locking with single mutex for all operations
   5. Following approach of atomically allocating object and
      charging resource in hirerchy
   6. Code simplification due to single lock
   7. Using for_each_set_bit API for bit operation
   8. Renamed list heads as Objects instead of _head
   9. Renamed list entries as _node instead of _list.
  10. Made usage_num to 64 bit to avoid overflow and to avoid 
      additional code to track non zero number of usage counts.
  * (To address comments from Doug)
   1. Added copyright and GPLv2 license

Changes from v9:
  * (To address comments from Tejun)
   1. Included clear documentation of resources.
   2. Fixed issue of race condition of process migration during
      charging stage.
   3. Fixed comments and code to adhere to CodingStyle.
   4. Simplified and removed support to charge/uncharge multiple
      resource.
   5. Fixed replaced refcnt with usage_num that tracks how many
      resources are unused to trigger freeing the object.
   6. Simplified locking scheme to use single spin lock for whole
      subsystem.

Changes from v8:
 * Fixed compilation error.
 * Fixed warning reported by checkpatch script.

Changes from v7:
 * (To address comments from Haggai)
   1. Removed max_limit from query_limit function as it is
      unnecessary.
   2. Kept existing printk as it is to instead of replacing all
      with pr_warn except newly added printk.

Changes from v6:
 * (To address comments from Haggai)
   1. Made functions as void wherever necessary.
   2. Code cleanup related to correting few spelling mistakes
      in comments, correcting comments to reflect the code.
   3. Removed max_count parameter from query_limit as its not
      necessary.
   4. Fixed printk to pr_warn.
   5. Removed dependency on pd, instead relying on ib_dev.
   6. Added more documentation to reflect that IB stack honors
      configured limit during query_device operation.
   7. Added pr_warn and avoided system crash in case of
      IB stack or rdma cgroup bug.
 * (To address comments from Leon)
   1. Removed #ifdef CONFIG_CGROUP_RDMA from .c files and added
      necessary dummy functions in header file.
   2. Removed unwanted forward declaration.
 * Fixed uncharing to rdma controller after resource is released
   from verb layer, instead of uncharing first. This ensures that
   uncharging doesn't complete while resource is still allocated.
 
Changes from v5:
 * (To address comments from Tejun)
   1. Removed two type of resource pool, made is single type (as Tejun
      described in past comment)
   2. Removed match tokens and have array definition like "qp", "mr",
      "cq" etc.
   3. Wrote small parser and avoided match_token API as that won't work
      due to different array definitions
   4. Removed one-off remove API to unconfigure cgroup, instead all
      resource should be set to max.
   5. Removed resource pool type (user/default), instead having
      max_num_cnt, when ref_cnt drops to zero and
      max_num_cnt = total_rescource_cnt, pool is freed.
   6. Resource definition ownership is now only with IB stack at single
      header file, no longer in each low level driver.
      This goes through IB maintainer and other reviewers eyes.
      This continue to give flexibility to not force kernel upgrade for
      few enums additions for new resource type.
   7. Wherever possible pool lock is pushed out, except for hierarchical
      charging/unchanging points, as it not possible to do so, due to
      iterative process involves blocking allocations of rpool. Coming up
      more levels up to release locks doesn't make any sense either.
      This is anyway slow path where rpool is not allocated. Except for
      typical first resource allocation, this is less travelled path.
   8. Avoided %d manipulation due to removal of match_token and replaced
      with seq_putc etc friend functions.
 * Other minor cleanups.
 * Fixed rdmacg_register_device to return error in case of IB stack
   tries to register for than 64 resources.
 * Fixed not allowing negative value on resource setting.
 * Fixed cleaning up resource pools during device removal.
 * Simplfied and rename table length field to use ARRAY_SIZE macro.
 * Updated documentation to reflect single resource pool and shorter
   file names.

Changes from v4:
 * Fixed compilation errors for lockdep_assert_held reported by kbuild
   test robot
 * Fixed compilation warning reported by coccinelle for extra
   semicolon.
 * Fixed compilation error for inclusion of linux/parser.h which
   cannot be included in any header file, as that triggers multiple
   inclusion error. parser.h is included in C files which intent to
   use it.
 * Removed unused header file inclusion in cgroup_rdma.c

Changes from v3:
 * (To address comments from Tejun)
   1. Renamed cg_resource to rdmacg_resource
   2. Merged dealloc_cg_rpool and _dealloc_cg_rpool to single function
   3. Renamed _find_cg_rpool to find_cg_rpool_locked()
   5. Removed RDMACG_MAX_RESOURCE_INDEX limitation
   6. Fixed few alignments.
   7. Improved description for RDMA cgroup configuration menu
   8. Renamed cg_list_lock to rpool_list_lock to reflect the lock
      is for rpools.
   9. Renamed _get_cg_rpool to find_cg_rpool.
   10. Made creator as int variable, instead of atomic as its not 
      required to be atomic.
 * Fixed freeing right rpool during query_limit error path
 * Added copywrite for cgroup.c
 * Removed including parser.h from cgroup.c as its included by
   cgroup_rdma.h
 * Reduced try_charge functions to single function and removed duplicate
   comments.

Changes from v2:
 * Fixed compilation error reported by 0-DAY kernel test infrastructure
   for m68k architecture where CONFIG_CGROUP is also not defined.
 * Fixed comment in patch to refer to legacy mode of cgroup, changed to 
   refer to v1 and v2 version.
 * Added more information in commit log for rdma controller patch.

Changes from v1:
 * (To address comments from Tejun)
   a. reduces 3 patches to single patch
   b. removed resource word from the cgroup configuration files
   c. changed cgroup configuration file names to match other cgroups
   d. removed .list file and merged functionality with .max file
 * Based on comment to merge to single patch for rdma controller;
   IB/core patches are reduced to single patch.
 * Removed pid cgroup map and simplified design -
   Charge/Uncharge caller stack keeps track of the rdmacg for
   given resource. This removes the need to maintain and perform
   hash lookup. This also allows little more accurate resource
   charging/uncharging when process moved from one to other cgroup
   with active resources and continue to allocate more.
 * Critical fix: Removed rdma cgroup's dependency on the kernel module
   header files to avoid crashes when modules are upgraded without kernel
   upgrade, which is very common due to high amount of changes in IB stack
   and it is also shipped as individual kernel modules.
 * uboject extended to keep track of the owner rdma cgroup, so that same
   rdmacg can be used while uncharging.
 * Added support functions to hide details of rdmacg device in uverbs
   modules for cases of cgroup enabled/disabled at compile time. This
   avoids multiple ifdefs for every API in uverbs layer.
 * Removed failure counters in first patch, which will be added once
   initial feature is merged.
 * Fixed stale rpool access which is getting freed, while doing
   configuration to rdma.verb.max file.
 * Fixed rpool resource leak while querying max, current values.

Changes from v0:
(To address comments from Haggai, Doug, Liran, Tejun, Sean, Jason)
 * Redesigned to support per device per cgroup limit settings by bringing
   concept of resource pool.
 * Redesigned to let IB stack define the resources instead of rdma
   controller using resource template.
 * Redesigned to support hw vendor specific limits setting
   (optional to drivers).
 * Created new rdma controller instead of piggyback on device cgroup.
 * Fixed race conditions for multiple tasks sharing rdma resources.
 * Removed dependency on the task_struct.


Parav Pandit (3):
  rdmacg: Added rdma cgroup controller
  IB/core: added support to use rdma cgroup controller
  rdmacg: Added documentation for rdmacg

 Documentation/cgroup-v1/rdma.txt      | 117 ++++++
 Documentation/cgroup-v2.txt           |  44 +++
 drivers/infiniband/core/Makefile      |   1 +
 drivers/infiniband/core/cgroup.c      |  69 ++++
 drivers/infiniband/core/core_priv.h   |  41 +++
 drivers/infiniband/core/device.c      |  10 +
 drivers/infiniband/core/uverbs_cmd.c  | 136 ++++++-
 drivers/infiniband/core/uverbs_main.c |  19 +
 include/linux/cgroup_rdma.h           |  66 ++++
 include/linux/cgroup_subsys.h         |   4 +
 include/rdma/ib_verbs.h               |  13 +
 init/Kconfig                          |  10 +
 kernel/Makefile                       |   1 +
 kernel/cgroup_rdma.c                  | 664 ++++++++++++++++++++++++++++++++++
 14 files changed, 1179 insertions(+), 16 deletions(-)
 create mode 100644 Documentation/cgroup-v1/rdma.txt
 create mode 100644 drivers/infiniband/core/cgroup.c
 create mode 100644 include/linux/cgroup_rdma.h
 create mode 100644 kernel/cgroup_rdma.c

-- 
1.8.3.1

[PATCHv11 1/3] rdmacg: Added rdma cgroup controller

[Parav Pandit]

Added rdma cgroup controller that does accounting, limit enforcement
on rdma/IB verbs and hw resources.

Added rdma cgroup header file which defines its APIs to perform
charing/uncharing functionality. It also defined APIs for RDMA/IB
stack for device registration. Devices which are registered will
participate in controller functions of accounting and limit
enforcements. It define rdmacg_device structure to bind IB stack
and RDMA cgroup controller.

RDMA resources are tracked using resource pool. Resource pool is per
device, per cgroup entity which allows setting up accounting limits
on per device basis.

Currently resources are defined by the RDMA cgroup.

Resource pool is created/destroyed dynamically whenever
charging/uncharging occurs respectively and whenever user
configuration is done. Its a tradeoff of memory vs little more code
space that creates resource pool object whenever necessary, instead of
creating them during cgroup creation and device registration time.

Signed-off-by: Parav Pandit <pandit.parav@gmail.com>
---
 include/linux/cgroup_rdma.h   |  66 +++++
 include/linux/cgroup_subsys.h |   4 +
 init/Kconfig                  |  10 +
 kernel/Makefile               |   1 +
 kernel/cgroup_rdma.c          | 664 ++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 745 insertions(+)
 create mode 100644 include/linux/cgroup_rdma.h
 create mode 100644 kernel/cgroup_rdma.c

diff --git a/include/linux/cgroup_rdma.h b/include/linux/cgroup_rdma.h
new file mode 100644
index 0000000..6710e28
--- /dev/null
+++ b/include/linux/cgroup_rdma.h
@@ -0,0 +1,66 @@
+/*
+ * Copyright (C) 2016 Parav Pandit <pandit.parav@gmail.com>
+ *
+ * This file is subject to the terms and conditions of version 2 of the GNU
+ * General Public License. See the file COPYING in the main directory of the
+ * Linux distribution for more details.
+ */
+
+#ifndef _CGROUP_RDMA_H
+#define _CGROUP_RDMA_H
+
+#include <linux/cgroup.h>
+
+enum rdmacg_resource_type {
+	RDMACG_VERB_RESOURCE_UCTX,
+	RDMACG_VERB_RESOURCE_AH,
+	RDMACG_VERB_RESOURCE_PD,
+	RDMACG_VERB_RESOURCE_CQ,
+	RDMACG_VERB_RESOURCE_MR,
+	RDMACG_VERB_RESOURCE_MW,
+	RDMACG_VERB_RESOURCE_SRQ,
+	RDMACG_VERB_RESOURCE_QP,
+	RDMACG_VERB_RESOURCE_FLOW,
+	/*
+	 * add any hw specific resource here as RDMA_HW_RESOURCE_NAME
+	 */
+	RDMACG_RESOURCE_MAX,
+};
+
+#ifdef CONFIG_CGROUP_RDMA
+
+struct rdma_cgroup {
+	struct cgroup_subsys_state	css;
+
+	/*
+	 * head to keep track of all resource pools
+	 * that belongs to this cgroup.
+	 */
+	struct list_head		rpools;
+};
+
+struct rdmacg_device {
+	struct list_head	dev_node;
+	struct list_head	rpools;
+	char			*name;
+};
+
+/*
+ * APIs for RDMA/IB stack to publish when a device wants to
+ * participate in resource accounting
+ */
+int rdmacg_register_device(struct rdmacg_device *device);
+void rdmacg_unregister_device(struct rdmacg_device *device);
+
+/* APIs for RDMA/IB stack to charge/uncharge pool specific resources */
+int rdmacg_try_charge(struct rdma_cgroup **rdmacg,
+		      struct rdmacg_device *device,
+		      enum rdmacg_resource_type index);
+void rdmacg_uncharge(struct rdma_cgroup *cg,
+		     struct rdmacg_device *device,
+		     enum rdmacg_resource_type index);
+void rdmacg_query_limit(struct rdmacg_device *device,
+			int *limits);
+
+#endif	/* CONFIG_CGROUP_RDMA */
+#endif	/* _CGROUP_RDMA_H */
diff --git a/include/linux/cgroup_subsys.h b/include/linux/cgroup_subsys.h
index 0df0336a..d0e597c 100644
--- a/include/linux/cgroup_subsys.h
+++ b/include/linux/cgroup_subsys.h
@@ -56,6 +56,10 @@ SUBSYS(hugetlb)
 SUBSYS(pids)
 #endif
 
+#if IS_ENABLED(CONFIG_CGROUP_RDMA)
+SUBSYS(rdma)
+#endif
+
 /*
  * The following subsystems are not supported on the default hierarchy.
  */
diff --git a/init/Kconfig b/init/Kconfig
index 6988649..a8acbe6 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -1080,6 +1080,16 @@ config CGROUP_PIDS
 	  since the PIDs limit only affects a process's ability to fork, not to
 	  attach to a cgroup.
 
+config CGROUP_RDMA
+	bool "RDMA controller"
+	help
+	  Provides enforcement of RDMA resources defined by IB stack.
+	  It is fairly easy for consumers to exhaust RDMA resources, which
+	  can result into resource unavailability to other consumers.
+	  RDMA controller is designed to stop this from happening.
+	  Attaching processes with active RDMA resources to the cgroup
+	  hierarchy is allowed even if can cross the hierarchy's limit.
+
 config CGROUP_FREEZER
 	bool "Freezer controller"
 	help
diff --git a/kernel/Makefile b/kernel/Makefile
index e2ec54e..d2b76d0 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -67,6 +67,7 @@ obj-$(CONFIG_COMPAT) += compat.o
 obj-$(CONFIG_CGROUPS) += cgroup.o
 obj-$(CONFIG_CGROUP_FREEZER) += cgroup_freezer.o
 obj-$(CONFIG_CGROUP_PIDS) += cgroup_pids.o
+obj-$(CONFIG_CGROUP_RDMA) += cgroup_rdma.o
 obj-$(CONFIG_CPUSETS) += cpuset.o
 obj-$(CONFIG_UTS_NS) += utsname.o
 obj-$(CONFIG_USER_NS) += user_namespace.o
diff --git a/kernel/cgroup_rdma.c b/kernel/cgroup_rdma.c
new file mode 100644
index 0000000..6ab9bd9
--- /dev/null
+++ b/kernel/cgroup_rdma.c
@@ -0,0 +1,664 @@
+/*
+ * RDMA resource limiting controller for cgroups.
+ *
+ * Used to allow a cgroup hierarchy to stop processes from consuming
+ * additional RDMA resources after a certain limit is reached.
+ *
+ * Copyright (C) 2016 Parav Pandit <pandit.parav@gmail.com>
+ *
+ * This file is subject to the terms and conditions of version 2 of the GNU
+ * General Public License. See the file COPYING in the main directory of the
+ * Linux distribution for more details.
+ */
+
+#include <linux/bitops.h>
+#include <linux/slab.h>
+#include <linux/seq_file.h>
+#include <linux/cgroup.h>
+#include <linux/parser.h>
+#include <linux/cgroup_rdma.h>
+
+#define RDMACG_MAX_STR "max"
+
+/*
+ * Protects list of resource pools maintained on per cgroup basis
+ * and rdma device list.
+ */
+static DEFINE_MUTEX(rdmacg_mutex);
+static LIST_HEAD(rdmacg_devices);
+
+enum rdmacg_file_type {
+	RDMACG_RESOURCE_TYPE_MAX,
+	RDMACG_RESOURCE_TYPE_STAT,
+};
+
+/*
+ * resource table definition as to be seen by the user.
+ * Need to add entries to it when more resources are
+ * added/defined at IB verb/core layer.
+ */
+static char const *rdmacg_resource_names[] = {
+	[RDMACG_VERB_RESOURCE_UCTX]	= "uctx",
+	[RDMACG_VERB_RESOURCE_AH]	= "ah",
+	[RDMACG_VERB_RESOURCE_PD]	= "pd",
+	[RDMACG_VERB_RESOURCE_CQ]	= "cq",
+	[RDMACG_VERB_RESOURCE_MR]	= "mr",
+	[RDMACG_VERB_RESOURCE_MW]	= "mw",
+	[RDMACG_VERB_RESOURCE_SRQ]	= "srq",
+	[RDMACG_VERB_RESOURCE_QP]	= "qp",
+	[RDMACG_VERB_RESOURCE_FLOW]	= "flow",
+};
+
+/* resource tracker for each resource of rdma cgroup */
+struct rdmacg_resource {
+	int max;
+	int usage;
+};
+
+/*
+ * resource pool object which represents per cgroup, per device
+ * resources. There are multiple instances of this object per cgroup,
+ * therefore it cannot be embedded within rdma_cgroup structure. It
+ * is maintained as list.
+ */
+struct rdmacg_resource_pool {
+	struct rdmacg_device	*device;
+	struct rdmacg_resource	resources[RDMACG_RESOURCE_MAX];
+
+	struct list_head	cg_node;
+	struct list_head	dev_node;
+
+	/* count active user tasks of this pool */
+	u64			usage_sum;
+	/* total number counts which are set to max */
+	int			num_max_cnt;
+};
+
+static struct rdma_cgroup *css_rdmacg(struct cgroup_subsys_state *css)
+{
+	return container_of(css, struct rdma_cgroup, css);
+}
+
+static struct rdma_cgroup *parent_rdmacg(struct rdma_cgroup *cg)
+{
+	return css_rdmacg(cg->css.parent);
+}
+
+static inline struct rdma_cgroup *get_current_rdmacg(void)
+{
+	return css_rdmacg(task_get_css(current, rdma_cgrp_id));
+}
+
+static void set_resource_limit(struct rdmacg_resource_pool *rpool,
+			       int index, int new_max)
+{
+	if (new_max == S32_MAX) {
+		if (rpool->resources[index].max != S32_MAX)
+			rpool->num_max_cnt++;
+	} else {
+		if (rpool->resources[index].max == S32_MAX)
+			rpool->num_max_cnt--;
+	}
+	rpool->resources[index].max = new_max;
+}
+
+static void set_all_resource_max_limit(struct rdmacg_resource_pool *rpool)
+{
+	int i;
+
+	for (i = 0; i < RDMACG_RESOURCE_MAX; i++)
+		set_resource_limit(rpool, i, S32_MAX);
+}
+
+static void free_cg_rpool_locked(struct rdmacg_resource_pool *rpool)
+{
+	lockdep_assert_held(&rdmacg_mutex);
+
+	list_del(&rpool->cg_node);
+	list_del(&rpool->dev_node);
+	kfree(rpool);
+}
+
+static struct rdmacg_resource_pool *
+find_cg_rpool_locked(struct rdma_cgroup *cg,
+		     struct rdmacg_device *device)
+
+{
+	struct rdmacg_resource_pool *pool;
+
+	lockdep_assert_held(&rdmacg_mutex);
+
+	list_for_each_entry(pool, &cg->rpools, cg_node)
+		if (pool->device == device)
+			return pool;
+
+	return NULL;
+}
+
+static struct rdmacg_resource_pool *
+get_cg_rpool_locked(struct rdma_cgroup *cg, struct rdmacg_device *device)
+{
+	struct rdmacg_resource_pool *rpool;
+
+	rpool = find_cg_rpool_locked(cg, device);
+	if (rpool)
+		return rpool;
+
+	rpool = kzalloc(sizeof(*rpool), GFP_KERNEL);
+	if (!rpool)
+		return ERR_PTR(-ENOMEM);
+
+	rpool->device = device;
+	set_all_resource_max_limit(rpool);
+
+	INIT_LIST_HEAD(&rpool->cg_node);
+	INIT_LIST_HEAD(&rpool->dev_node);
+	list_add_tail(&rpool->cg_node, &cg->rpools);
+	list_add_tail(&rpool->dev_node, &device->rpools);
+	return rpool;
+}
+
+/**
+ * uncharge_cg_locked - uncharge resource for rdma cgroup
+ * @cg: pointer to cg to uncharge and all parents in hierarchy
+ * @device: pointer to rdmacg device
+ * @index: index of the resource to uncharge in cg (resource pool)
+ *
+ * It also frees the resource pool which was created as part of
+ * charging operation when there are no resources attached to
+ * resource pool.
+ */
+static void
+uncharge_cg_locked(struct rdma_cgroup *cg,
+		   struct rdmacg_device *device,
+		   enum rdmacg_resource_type index)
+{
+	struct rdmacg_resource_pool *rpool;
+
+	rpool = find_cg_rpool_locked(cg, device);
+
+	/*
+	 * rpool cannot be null at this stage. Let kernel operate in case
+	 * if there a bug in IB stack or rdma controller, instead of crashing
+	 * the system.
+	 */
+	if (unlikely(!rpool)) {
+		pr_warn("Invalid device %p or rdma cgroup %p\n", cg, device);
+		return;
+	}
+
+	rpool->resources[index].usage--;
+
+	/*
+	 * A negative count (or overflow) is invalid,
+	 * it indicates a bug in the rdma controller.
+	 */
+	WARN_ON_ONCE(rpool->resources[index].usage < 0);
+	rpool->usage_sum--;
+	if (rpool->usage_sum == 0 &&
+	    rpool->num_max_cnt == RDMACG_RESOURCE_MAX) {
+		/*
+		 * No user of the rpool and all entries are set to max, so
+		 * safe to delete this rpool.
+		 */
+		free_cg_rpool_locked(rpool);
+	}
+}
+
+/**
+ * rdmacg_uncharge_hirerchy - hierarchically uncharge rdma resource count
+ * @device: pointer to rdmacg device
+ * @stop_cg: while traversing hirerchy, when meet with stop_cg cgroup
+ *           stop uncharging
+ * @index: index of the resource to uncharge in cg in given resource pool
+ */
+static void rdmacg_uncharge_hirerchy(struct rdma_cgroup *cg,
+				     struct rdmacg_device *device,
+				     struct rdma_cgroup *stop_cg,
+				     enum rdmacg_resource_type index)
+{
+	struct rdma_cgroup *p;
+
+	mutex_lock(&rdmacg_mutex);
+
+	for (p = cg; p != stop_cg; p = parent_rdmacg(p))
+		uncharge_cg_locked(p, device, index);
+
+	mutex_unlock(&rdmacg_mutex);
+
+	css_put(&cg->css);
+}
+
+/**
+ * rdmacg_uncharge - hierarchically uncharge rdma resource count
+ * @device: pointer to rdmacg device
+ * @index: index of the resource to uncharge in cgroup in given resource pool
+ */
+void rdmacg_uncharge(struct rdma_cgroup *cg,
+		     struct rdmacg_device *device,
+		     enum rdmacg_resource_type index)
+{
+	if (index >= RDMACG_RESOURCE_MAX)
+		return;
+
+	rdmacg_uncharge_hirerchy(cg, device, NULL, index);
+}
+EXPORT_SYMBOL(rdmacg_uncharge);
+
+/**
+ * rdmacg_try_charge - hierarchically try to charge the rdma resource
+ * @rdmacg: pointer to rdma cgroup which will own this resource
+ * @device: pointer to rdmacg device
+ * @index: index of the resource to charge in cgroup (resource pool)
+ *
+ * This function follows charging resource in hierarchical way.
+ * It will fail if the charge would cause the new value to exceed the
+ * hierarchical limit.
+ * Returns 0 if the charge succeded, otherwise -EAGAIN, -ENOMEM or -EINVAL.
+ * Returns pointer to rdmacg for this resource when charging is successful.
+ *
+ * Charger needs to account resources on two criteria.
+ * (a) per cgroup & (b) per device resource usage.
+ * Per cgroup resource usage ensures that tasks of cgroup doesn't cross
+ * the configured limits. Per device provides granular configuration
+ * in multi device usage. It allocates resource pool in the hierarchy
+ * for each parent it come across for first resource. Later on resource
+ * pool will be available. Therefore it will be much faster thereon
+ * to charge/uncharge.
+ */
+int rdmacg_try_charge(struct rdma_cgroup **rdmacg,
+		      struct rdmacg_device *device,
+		      enum rdmacg_resource_type index)
+{
+	struct rdma_cgroup *cg, *p;
+	struct rdmacg_resource_pool *rpool;
+	s64 new;
+	int ret = 0;
+
+	if (index >= RDMACG_RESOURCE_MAX)
+		return -EINVAL;
+
+	/*
+	 * hold on to css, as cgroup can be removed but resource
+	 * accounting happens on css.
+	 */
+	cg = get_current_rdmacg();
+
+	mutex_lock(&rdmacg_mutex);
+	for (p = cg; p; p = parent_rdmacg(p)) {
+		rpool = get_cg_rpool_locked(p, device);
+		if (IS_ERR_OR_NULL(rpool)) {
+			ret = PTR_ERR(rpool);
+			goto err;
+		} else {
+			new = rpool->resources[index].usage + 1;
+			if (new > rpool->resources[index].max) {
+				ret = -EAGAIN;
+				goto err;
+			} else {
+				rpool->resources[index].usage = new;
+				rpool->usage_sum++;
+			}
+		}
+	}
+	mutex_unlock(&rdmacg_mutex);
+
+	*rdmacg = cg;
+	return 0;
+
+err:
+	mutex_unlock(&rdmacg_mutex);
+	rdmacg_uncharge_hirerchy(cg, device, p, index);
+	return ret;
+}
+EXPORT_SYMBOL(rdmacg_try_charge);
+
+/**
+ * rdmacg_register_device - register rdmacg device to rdma controller.
+ * @device: pointer to rdmacg device whose resources need to be accounted.
+ *
+ * If IB stack wish a device to participate in rdma cgroup resource
+ * tracking, it must invoke this API to register with rdma cgroup before
+ * any user space application can start using the RDMA resources.
+ * Returns 0 on success or EINVAL when table length given is beyond
+ * supported size.
+ */
+int rdmacg_register_device(struct rdmacg_device *device)
+{
+	INIT_LIST_HEAD(&device->dev_node);
+	INIT_LIST_HEAD(&device->rpools);
+
+	mutex_lock(&rdmacg_mutex);
+	list_add_tail(&device->dev_node, &rdmacg_devices);
+	mutex_unlock(&rdmacg_mutex);
+	return 0;
+}
+EXPORT_SYMBOL(rdmacg_register_device);
+
+/**
+ * rdmacg_unregister_device - unregister rdmacg device from rdma controller.
+ * @device: pointer to rdmacg device which was previously registered with rdma
+ *          controller using rdmacg_register_device().
+ *
+ * IB stack must invoke this after all the resources of the IB device
+ * are destroyed and after ensuring that no more resources will be created
+ * when this API is invoked.
+ */
+void rdmacg_unregister_device(struct rdmacg_device *device)
+{
+	struct rdmacg_resource_pool *rpool, *tmp;
+
+	/*
+	 * Synchronize with any active resource settings,
+	 * usage query happening via configfs.
+	 */
+	mutex_lock(&rdmacg_mutex);
+	list_del_init(&device->dev_node);
+
+	/*
+	 * Now that this device is off the cgroup list, its safe to free
+	 * all the rpool resources.
+	 */
+	list_for_each_entry_safe(rpool, tmp, &device->rpools, dev_node)
+		free_cg_rpool_locked(rpool);
+
+	mutex_unlock(&rdmacg_mutex);
+}
+EXPORT_SYMBOL(rdmacg_unregister_device);
+
+/**
+ * rdmacg_query_limit - query the resource limits that
+ * might have been configured by the user.
+ * @device: pointer to rdmacg device
+ * @limits: pointer to an array of limits where rdma cg will provide
+ *          the configured limits of the cgroup.
+ *
+ * This function honors resource limit in hierarchical way.
+ * In a cgroup hirarchy, it considers the limit of a controller which has
+ * smallest limit configured.
+ */
+void rdmacg_query_limit(struct rdmacg_device *device, int *limits)
+{
+	struct rdma_cgroup *cg, *p;
+	struct rdmacg_resource_pool *rpool;
+	int i;
+
+	for (i = 0; i < RDMACG_RESOURCE_MAX; i++)
+		limits[i] = S32_MAX;
+
+	cg = get_current_rdmacg();
+	/*
+	 * Check in hirerchy which pool get the least amount of
+	 * resource limits.
+	 */
+	mutex_lock(&rdmacg_mutex);
+	for (p = cg; p; p = parent_rdmacg(p)) {
+		rpool = find_cg_rpool_locked(cg, device);
+		if (!rpool)
+			continue;
+
+		for (i = 0; i < RDMACG_RESOURCE_MAX; i++)
+			limits[i] = min_t(int, limits[i],
+					rpool->resources[i].max);
+	}
+	mutex_unlock(&rdmacg_mutex);
+	css_put(&cg->css);
+}
+EXPORT_SYMBOL(rdmacg_query_limit);
+
+static int parse_resource(char *c, int *intval)
+{
+	substring_t argstr;
+	const char **table = &rdmacg_resource_names[0];
+	char *name, *value = c;
+	size_t len;
+	int ret, i = 0;
+
+	name = strsep(&value, "=");
+	if (!name || !value)
+		return -EINVAL;
+
+	len = strlen(value);
+
+	for (i = 0; i < RDMACG_RESOURCE_MAX; i++) {
+		if (strcmp(table[i], name))
+			continue;
+
+		argstr.from = value;
+		argstr.to = value + len;
+
+		ret = match_int(&argstr, intval);
+		if (ret >= 0) {
+			if (*intval < 0)
+				break;
+			return i;
+		}
+		if (strncmp(value, RDMACG_MAX_STR, len) == 0) {
+			*intval = S32_MAX;
+			return i;
+		}
+		break;
+	}
+	return -EINVAL;
+}
+
+static int rdmacg_parse_limits(char *options,
+			       int *new_limits, unsigned long *enables)
+{
+	char *c;
+	int err = -EINVAL;
+
+	/* parse resource options */
+	while ((c = strsep(&options, " ")) != NULL) {
+		int index, intval;
+
+		index = parse_resource(c, &intval);
+		if (index < 0)
+			goto err;
+
+		new_limits[index] = intval;
+		*enables |= BIT(index);
+	}
+	return 0;
+
+err:
+	return err;
+}
+
+static struct rdmacg_device *rdmacg_get_device_locked(const char *name)
+{
+	struct rdmacg_device *device;
+
+	lockdep_assert_held(&rdmacg_mutex);
+
+	list_for_each_entry(device, &rdmacg_devices, dev_node)
+		if (!strcmp(name, device->name))
+			return device;
+
+	return NULL;
+}
+
+static ssize_t rdmacg_resource_set_max(struct kernfs_open_file *of,
+				       char *buf, size_t nbytes, loff_t off)
+{
+	struct rdma_cgroup *cg = css_rdmacg(of_css(of));
+	const char *dev_name;
+	struct rdmacg_resource_pool *rpool;
+	struct rdmacg_device *device;
+	char *options = strstrip(buf);
+	int *new_limits;
+	unsigned long enables = 0;
+	int i = 0, ret = 0;
+
+	/* extract the device name first */
+	dev_name = strsep(&options, " ");
+	if (!dev_name) {
+		ret = -EINVAL;
+		goto err;
+	}
+
+	new_limits = kcalloc(RDMACG_RESOURCE_MAX, sizeof(int), GFP_KERNEL);
+	if (!new_limits) {
+		ret = -ENOMEM;
+		goto err;
+	}
+
+	ret = rdmacg_parse_limits(options, new_limits, &enables);
+	if (ret)
+		goto parse_err;
+
+	/* acquire lock to synchronize with hot plug devices */
+	mutex_lock(&rdmacg_mutex);
+
+	device = rdmacg_get_device_locked(dev_name);
+	if (!device) {
+		ret = -ENODEV;
+		goto dev_err;
+	}
+
+	rpool = get_cg_rpool_locked(cg, device);
+	if (IS_ERR_OR_NULL(rpool)) {
+		ret = PTR_ERR(rpool);
+		goto dev_err;
+	}
+
+	/* now set the new limits of the rpool */
+	for_each_set_bit(i, &enables, RDMACG_RESOURCE_MAX)
+		set_resource_limit(rpool, i, new_limits[i]);
+
+	if (rpool->usage_sum == 0 &&
+	    rpool->num_max_cnt == RDMACG_RESOURCE_MAX) {
+		/*
+		 * No user of the rpool and all entries are set to max, so
+		 * safe to delete this rpool.
+		 */
+		free_cg_rpool_locked(rpool);
+	}
+
+dev_err:
+	mutex_unlock(&rdmacg_mutex);
+
+parse_err:
+	kfree(new_limits);
+
+err:
+	return ret ?: nbytes;
+}
+
+static void print_rpool_values(struct seq_file *sf,
+			       struct rdmacg_resource_pool *rpool)
+{
+	enum rdmacg_file_type sf_type;
+	int i;
+	u32 value;
+
+	sf_type = seq_cft(sf)->private;
+
+	for (i = 0; i < RDMACG_RESOURCE_MAX; i++) {
+		seq_puts(sf, rdmacg_resource_names[i]);
+		seq_putc(sf, '=');
+		if (sf_type == RDMACG_RESOURCE_TYPE_MAX) {
+			if (rpool)
+				value = rpool->resources[i].max;
+			else
+				value = S32_MAX;
+		} else {
+			if (rpool)
+				value = rpool->resources[i].usage;
+		}
+
+		if (value == S32_MAX)
+			seq_puts(sf, RDMACG_MAX_STR);
+		else
+			seq_printf(sf, "%d", value);
+		seq_putc(sf, ' ');
+	}
+}
+
+static int rdmacg_resource_read(struct seq_file *sf, void *v)
+{
+	struct rdmacg_device *device;
+	struct rdmacg_resource_pool *rpool;
+	struct rdma_cgroup *cg = css_rdmacg(seq_css(sf));
+
+	mutex_lock(&rdmacg_mutex);
+
+	list_for_each_entry(device, &rdmacg_devices, dev_node) {
+		seq_printf(sf, "%s ", device->name);
+
+		rpool = find_cg_rpool_locked(cg, device);
+		print_rpool_values(sf, rpool);
+
+		seq_putc(sf, '\n');
+	}
+
+	mutex_unlock(&rdmacg_mutex);
+	return 0;
+}
+
+static struct cftype rdmacg_files[] = {
+	{
+		.name = "max",
+		.write = rdmacg_resource_set_max,
+		.seq_show = rdmacg_resource_read,
+		.private = RDMACG_RESOURCE_TYPE_MAX,
+		.flags = CFTYPE_NOT_ON_ROOT,
+	},
+	{
+		.name = "current",
+		.seq_show = rdmacg_resource_read,
+		.private = RDMACG_RESOURCE_TYPE_STAT,
+		.flags = CFTYPE_NOT_ON_ROOT,
+	},
+	{ }	/* terminate */
+};
+
+static struct cgroup_subsys_state *
+rdmacg_css_alloc(struct cgroup_subsys_state *parent)
+{
+	struct rdma_cgroup *cg;
+
+	cg = kzalloc(sizeof(*cg), GFP_KERNEL);
+	if (!cg)
+		return ERR_PTR(-ENOMEM);
+
+	INIT_LIST_HEAD(&cg->rpools);
+	return &cg->css;
+}
+
+static void rdmacg_css_free(struct cgroup_subsys_state *css)
+{
+	struct rdma_cgroup *cg = css_rdmacg(css);
+
+	kfree(cg);
+}
+
+/**
+ * rdmacg_css_offline - cgroup css_offline callback
+ * @css: css of interest
+ *
+ * This function is called when @css is about to go away and responsible
+ * for shooting down all rdmacg associated with @css. As part of that it
+ * marks all the resource pool entries to max value, so that when resources are
+ * uncharged, associated resource pool can be freed as well.
+ */
+static void rdmacg_css_offline(struct cgroup_subsys_state *css)
+{
+	struct rdma_cgroup *cg = css_rdmacg(css);
+	struct rdmacg_resource_pool *rpool;
+
+	mutex_lock(&rdmacg_mutex);
+
+	list_for_each_entry(rpool, &cg->rpools, cg_node)
+		set_all_resource_max_limit(rpool);
+
+	mutex_unlock(&rdmacg_mutex);
+}
+
+struct cgroup_subsys rdma_cgrp_subsys = {
+	.css_alloc	= rdmacg_css_alloc,
+	.css_free	= rdmacg_css_free,
+	.css_offline	= rdmacg_css_offline,
+	.legacy_cftypes	= rdmacg_files,
+	.dfl_cftypes	= rdmacg_files,
+};
-- 
1.8.3.1

[PATCHv11 2/3] IB/core: added support to use rdma cgroup controller

[Parav Pandit]

Added support APIs for IB core to register/unregister every IB/RDMA
device with rdma cgroup for tracking verbs and hw resources.
IB core registers with rdma cgroup controller.
Added support APIs for uverbs layer to make use of rdma controller.
Added uverbs layer to perform resource charge/uncharge functionality.
Added support during query_device uverb operation to ensure it
returns resource limits by honoring rdma cgroup configured limits.

Signed-off-by: Parav Pandit <pandit.parav@gmail.com>
---
 drivers/infiniband/core/Makefile      |   1 +
 drivers/infiniband/core/cgroup.c      |  69 +++++++++++++++++
 drivers/infiniband/core/core_priv.h   |  41 ++++++++++
 drivers/infiniband/core/device.c      |  10 +++
 drivers/infiniband/core/uverbs_cmd.c  | 136 ++++++++++++++++++++++++++++++----
 drivers/infiniband/core/uverbs_main.c |  19 +++++
 include/rdma/ib_verbs.h               |  13 ++++
 7 files changed, 273 insertions(+), 16 deletions(-)
 create mode 100644 drivers/infiniband/core/cgroup.c

diff --git a/drivers/infiniband/core/Makefile b/drivers/infiniband/core/Makefile
index edaae9f..e426ac8 100644
--- a/drivers/infiniband/core/Makefile
+++ b/drivers/infiniband/core/Makefile
@@ -13,6 +13,7 @@ ib_core-y :=			packer.o ud_header.o verbs.o cq.o rw.o sysfs.o \
 				multicast.o mad.o smi.o agent.o mad_rmpp.o
 ib_core-$(CONFIG_INFINIBAND_USER_MEM) += umem.o
 ib_core-$(CONFIG_INFINIBAND_ON_DEMAND_PAGING) += umem_odp.o umem_rbtree.o
+ib_core-$(CONFIG_CGROUP_RDMA) += cgroup.o
 
 ib_cm-y :=			cm.o
 
diff --git a/drivers/infiniband/core/cgroup.c b/drivers/infiniband/core/cgroup.c
new file mode 100644
index 0000000..1a85fff
--- /dev/null
+++ b/drivers/infiniband/core/cgroup.c
@@ -0,0 +1,69 @@
+/*
+ * Copyright (C) 2016 Parav Pandit <pandit.parav@gmail.com>
+ *
+ * This file is subject to the terms and conditions of version 2 of the GNU
+ * General Public License. See the file COPYING in the main directory of the
+ * Linux distribution for more details.
+ */
+
+#include "core_priv.h"
+
+/*
+ * resource table definition as to be seen by the user.
+ * Need to add entries to it when more resources are
+ * added/defined at IB verb/core layer.
+ */
+
+/**
+ * ib_device_register_rdmacg - register with rdma cgroup.
+ * @device: device to register to participate in resource
+ *          accounting by rdma cgroup.
+ *
+ * Register with the rdma cgroup. Should be called before
+ * exposing rdma device to user space applications to avoid
+ * resource accounting leak.
+ * Returns 0 on success or otherwise failure code.
+ */
+int ib_device_register_rdmacg(struct ib_device *device)
+{
+	device->cg_device.name = device->name;
+	return rdmacg_register_device(&device->cg_device);
+}
+
+/**
+ * ib_device_unregister_rdmacg - unregister with rdma cgroup.
+ * @device: device to unregister.
+ *
+ * Unregister with the rdma cgroup. Should be called after
+ * all the resources are deallocated, and after a stage when any
+ * other resource allocation by user application cannot be done
+ * for this device to avoid any leak in accounting.
+ */
+void ib_device_unregister_rdmacg(struct ib_device *device)
+{
+	rdmacg_unregister_device(&device->cg_device);
+}
+
+int ib_rdmacg_try_charge(struct ib_rdmacg_object *cg_obj,
+			 struct ib_device *device,
+			 enum rdmacg_resource_type resource_index)
+{
+	return rdmacg_try_charge(&cg_obj->cg, &device->cg_device,
+				 resource_index);
+}
+EXPORT_SYMBOL(ib_rdmacg_try_charge);
+
+void ib_rdmacg_uncharge(struct ib_rdmacg_object *cg_obj,
+			struct ib_device *device,
+			enum rdmacg_resource_type resource_index)
+{
+	rdmacg_uncharge(cg_obj->cg, &device->cg_device,
+			resource_index);
+}
+EXPORT_SYMBOL(ib_rdmacg_uncharge);
+
+void ib_rdmacg_query_limit(struct ib_device *device, int *limits)
+{
+	rdmacg_query_limit(&device->cg_device, limits);
+}
+EXPORT_SYMBOL(ib_rdmacg_query_limit);
diff --git a/drivers/infiniband/core/core_priv.h b/drivers/infiniband/core/core_priv.h
index 19d499d..d1e432e 100644
--- a/drivers/infiniband/core/core_priv.h
+++ b/drivers/infiniband/core/core_priv.h
@@ -35,6 +35,7 @@
 
 #include <linux/list.h>
 #include <linux/spinlock.h>
+#include <linux/cgroup_rdma.h>
 
 #include <rdma/ib_verbs.h>
 
@@ -124,6 +125,46 @@ int ib_cache_setup_one(struct ib_device *device);
 void ib_cache_cleanup_one(struct ib_device *device);
 void ib_cache_release_one(struct ib_device *device);
 
+#ifdef CONFIG_CGROUP_RDMA
+int ib_device_register_rdmacg(struct ib_device *device);
+void ib_device_unregister_rdmacg(struct ib_device *device);
+
+int ib_rdmacg_try_charge(struct ib_rdmacg_object *cg_obj,
+			 struct ib_device *device,
+			 enum rdmacg_resource_type resource_index);
+
+void ib_rdmacg_uncharge(struct ib_rdmacg_object *cg_obj,
+			struct ib_device *device,
+			enum rdmacg_resource_type resource_index);
+
+void ib_rdmacg_query_limit(struct ib_device *device, int *limits);
+#else
+static inline int ib_device_register_rdmacg(struct ib_device *device)
+{ return 0; }
+
+static inline void ib_device_unregister_rdmacg(struct ib_device *device)
+{ }
+
+static inline int ib_rdmacg_try_charge(struct ib_rdmacg_object *cg_obj,
+				       struct ib_device *device,
+				       enum rdmacg_resource_type resource_index)
+{ return 0; }
+
+static inline void ib_rdmacg_uncharge(struct ib_rdmacg_object *cg_obj,
+				      struct ib_device *device,
+				      enum rdmacg_resource_type resource_index)
+{ }
+
+static inline void ib_rdmacg_query_limit(struct ib_device *device,
+					 int *limits)
+{
+	int i;
+
+	for (i = 0; i < RDMACG_RESOURCE_MAX; i++)
+		limits[i] = S32_MAX;
+}
+#endif
+
 static inline bool rdma_is_upper_dev_rcu(struct net_device *dev,
 					 struct net_device *upper)
 {
diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 760ef60..08e3259 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -363,10 +363,18 @@ int ib_register_device(struct ib_device *device,
 		goto out;
 	}
 
+	ret = ib_device_register_rdmacg(device);
+	if (ret) {
+		pr_warn("Couldn't register device with rdma cgroup\n");
+		ib_cache_cleanup_one(device);
+		goto out;
+	}
+
 	memset(&device->attrs, 0, sizeof(device->attrs));
 	ret = device->query_device(device, &device->attrs, &uhw);
 	if (ret) {
 		pr_warn("Couldn't query the device attributes\n");
+		ib_device_unregister_rdmacg(device);
 		ib_cache_cleanup_one(device);
 		goto out;
 	}
@@ -375,6 +383,7 @@ int ib_register_device(struct ib_device *device,
 	if (ret) {
 		pr_warn("Couldn't register device %s with driver model\n",
 			device->name);
+		ib_device_unregister_rdmacg(device);
 		ib_cache_cleanup_one(device);
 		goto out;
 	}
@@ -424,6 +433,7 @@ void ib_unregister_device(struct ib_device *device)
 
 	mutex_unlock(&device_mutex);
 
+	ib_device_unregister_rdmacg(device);
 	ib_device_unregister_sysfs(device);
 	ib_cache_cleanup_one(device);
 
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index f664731..893669b 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -316,6 +316,7 @@ ssize_t ib_uverbs_get_context(struct ib_uverbs_file *file,
 	struct ib_udata                   udata;
 	struct ib_ucontext		 *ucontext;
 	struct file			 *filp;
+	struct ib_rdmacg_object          cg_obj;
 	int ret;
 
 	if (out_len < sizeof resp)
@@ -335,13 +336,18 @@ ssize_t ib_uverbs_get_context(struct ib_uverbs_file *file,
 		   (unsigned long) cmd.response + sizeof resp,
 		   in_len - sizeof cmd, out_len - sizeof resp);
 
+	ret = ib_rdmacg_try_charge(&cg_obj, ib_dev, RDMACG_VERB_RESOURCE_UCTX);
+	if (ret)
+		goto err;
+
 	ucontext = ib_dev->alloc_ucontext(ib_dev, &udata);
 	if (IS_ERR(ucontext)) {
 		ret = PTR_ERR(ucontext);
-		goto err;
+		goto err_alloc;
 	}
 
 	ucontext->device = ib_dev;
+	ucontext->cg_obj = cg_obj;
 	INIT_LIST_HEAD(&ucontext->pd_list);
 	INIT_LIST_HEAD(&ucontext->mr_list);
 	INIT_LIST_HEAD(&ucontext->mw_list);
@@ -407,6 +413,9 @@ err_free:
 	put_pid(ucontext->tgid);
 	ib_dev->dealloc_ucontext(ucontext);
 
+err_alloc:
+	ib_rdmacg_uncharge(&cg_obj, ib_dev, RDMACG_VERB_RESOURCE_UCTX);
+
 err:
 	mutex_unlock(&file->mutex);
 	return ret;
@@ -415,7 +424,8 @@ err:
 static void copy_query_dev_fields(struct ib_uverbs_file *file,
 				  struct ib_device *ib_dev,
 				  struct ib_uverbs_query_device_resp *resp,
-				  struct ib_device_attr *attr)
+				  struct ib_device_attr *attr,
+				  int *limits)
 {
 	resp->fw_ver		= attr->fw_ver;
 	resp->node_guid		= ib_dev->node_guid;
@@ -425,15 +435,19 @@ static void copy_query_dev_fields(struct ib_uverbs_file *file,
 	resp->vendor_id		= attr->vendor_id;
 	resp->vendor_part_id	= attr->vendor_part_id;
 	resp->hw_ver		= attr->hw_ver;
-	resp->max_qp		= attr->max_qp;
+	resp->max_qp		= min_t(int, attr->max_qp,
+					limits[RDMACG_VERB_RESOURCE_QP]);
 	resp->max_qp_wr		= attr->max_qp_wr;
 	resp->device_cap_flags	= lower_32_bits(attr->device_cap_flags);
 	resp->max_sge		= attr->max_sge;
 	resp->max_sge_rd	= attr->max_sge_rd;
-	resp->max_cq		= attr->max_cq;
+	resp->max_cq		= min_t(int, attr->max_cq,
+					limits[RDMACG_VERB_RESOURCE_CQ]);
 	resp->max_cqe		= attr->max_cqe;
-	resp->max_mr		= attr->max_mr;
-	resp->max_pd		= attr->max_pd;
+	resp->max_mr		= min_t(int, attr->max_mr,
+					limits[RDMACG_VERB_RESOURCE_MR]);
+	resp->max_pd		= min_t(int, attr->max_pd,
+					limits[RDMACG_VERB_RESOURCE_PD]);
 	resp->max_qp_rd_atom	= attr->max_qp_rd_atom;
 	resp->max_ee_rd_atom	= attr->max_ee_rd_atom;
 	resp->max_res_rd_atom	= attr->max_res_rd_atom;
@@ -442,16 +456,19 @@ static void copy_query_dev_fields(struct ib_uverbs_file *file,
 	resp->atomic_cap		= attr->atomic_cap;
 	resp->max_ee			= attr->max_ee;
 	resp->max_rdd			= attr->max_rdd;
-	resp->max_mw			= attr->max_mw;
+	resp->max_mw			= min_t(int, attr->max_mw,
+					limits[RDMACG_VERB_RESOURCE_MW]);
 	resp->max_raw_ipv6_qp		= attr->max_raw_ipv6_qp;
 	resp->max_raw_ethy_qp		= attr->max_raw_ethy_qp;
 	resp->max_mcast_grp		= attr->max_mcast_grp;
 	resp->max_mcast_qp_attach	= attr->max_mcast_qp_attach;
 	resp->max_total_mcast_qp_attach	= attr->max_total_mcast_qp_attach;
-	resp->max_ah			= attr->max_ah;
+	resp->max_ah			= min_t(int, attr->max_ah,
+					limits[RDMACG_VERB_RESOURCE_AH]);
 	resp->max_fmr			= attr->max_fmr;
 	resp->max_map_per_fmr		= attr->max_map_per_fmr;
-	resp->max_srq			= attr->max_srq;
+	resp->max_srq			= min_t(int, attr->max_srq,
+					limits[RDMACG_VERB_RESOURCE_SRQ]);
 	resp->max_srq_wr		= attr->max_srq_wr;
 	resp->max_srq_sge		= attr->max_srq_sge;
 	resp->max_pkeys			= attr->max_pkeys;
@@ -466,6 +483,7 @@ ssize_t ib_uverbs_query_device(struct ib_uverbs_file *file,
 {
 	struct ib_uverbs_query_device      cmd;
 	struct ib_uverbs_query_device_resp resp;
+	int                                limits[RDMACG_RESOURCE_MAX];
 
 	if (out_len < sizeof resp)
 		return -ENOSPC;
@@ -473,8 +491,10 @@ ssize_t ib_uverbs_query_device(struct ib_uverbs_file *file,
 	if (copy_from_user(&cmd, buf, sizeof cmd))
 		return -EFAULT;
 
+	ib_rdmacg_query_limit(ib_dev, limits);
+
 	memset(&resp, 0, sizeof resp);
-	copy_query_dev_fields(file, ib_dev, &resp, &ib_dev->attrs);
+	copy_query_dev_fields(file, ib_dev, &resp, &ib_dev->attrs, limits);
 
 	if (copy_to_user((void __user *) (unsigned long) cmd.response,
 			 &resp, sizeof resp))
@@ -560,6 +580,13 @@ ssize_t ib_uverbs_alloc_pd(struct ib_uverbs_file *file,
 	if (!uobj)
 		return -ENOMEM;
 
+	ret = ib_rdmacg_try_charge(&uobj->cg_obj, ib_dev,
+				   RDMACG_VERB_RESOURCE_PD);
+	if (ret) {
+		kfree(uobj);
+		return ret;
+	}
+
 	init_uobj(uobj, 0, file->ucontext, &pd_lock_class);
 	down_write(&uobj->mutex);
 
@@ -605,6 +632,7 @@ err_idr:
 	ib_dealloc_pd(pd);
 
 err:
+	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_VERB_RESOURCE_PD);
 	put_uobj_write(uobj);
 	return ret;
 }
@@ -637,6 +665,8 @@ ssize_t ib_uverbs_dealloc_pd(struct ib_uverbs_file *file,
 	if (ret)
 		goto err_put;
 
+	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_VERB_RESOURCE_PD);
+
 	uobj->live = 0;
 	put_uobj_write(uobj);
 
@@ -1003,10 +1033,15 @@ ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file,
 		      IB_DEVICE_ON_DEMAND_PAGING)) {
 			pr_debug("ODP support not available\n");
 			ret = -EINVAL;
-			goto err_put;
+			goto err_charge;
 		}
 	}
 
+	ret = ib_rdmacg_try_charge(&uobj->cg_obj, ib_dev,
+				   RDMACG_VERB_RESOURCE_MR);
+	if (ret)
+		goto err_charge;
+
 	mr = pd->device->reg_user_mr(pd, cmd.start, cmd.length, cmd.hca_va,
 				     cmd.access_flags, &udata);
 	if (IS_ERR(mr)) {
@@ -1054,6 +1089,9 @@ err_unreg:
 	ib_dereg_mr(mr);
 
 err_put:
+	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_VERB_RESOURCE_MR);
+
+err_charge:
 	put_pd_read(pd);
 
 err_free:
@@ -1178,6 +1216,8 @@ ssize_t ib_uverbs_dereg_mr(struct ib_uverbs_file *file,
 	if (ret)
 		return ret;
 
+	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_VERB_RESOURCE_MR);
+
 	idr_remove_uobj(&ib_uverbs_mr_idr, uobj);
 
 	mutex_lock(&file->mutex);
@@ -1226,6 +1266,11 @@ ssize_t ib_uverbs_alloc_mw(struct ib_uverbs_file *file,
 		   in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr),
 		   out_len - sizeof(resp));
 
+	ret = ib_rdmacg_try_charge(&uobj->cg_obj, ib_dev,
+				   RDMACG_VERB_RESOURCE_MW);
+	if (ret)
+		goto err_charge;
+
 	mw = pd->device->alloc_mw(pd, cmd.mw_type, &udata);
 	if (IS_ERR(mw)) {
 		ret = PTR_ERR(mw);
@@ -1271,6 +1316,9 @@ err_unalloc:
 	uverbs_dealloc_mw(mw);
 
 err_put:
+	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_VERB_RESOURCE_MW);
+
+err_charge:
 	put_pd_read(pd);
 
 err_free:
@@ -1306,6 +1354,8 @@ ssize_t ib_uverbs_dealloc_mw(struct ib_uverbs_file *file,
 	if (ret)
 		return ret;
 
+	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_VERB_RESOURCE_MW);
+
 	idr_remove_uobj(&ib_uverbs_mw_idr, uobj);
 
 	mutex_lock(&file->mutex);
@@ -1405,6 +1455,11 @@ static struct ib_ucq_object *create_cq(struct ib_uverbs_file *file,
 	if (cmd_sz > offsetof(typeof(*cmd), flags) + sizeof(cmd->flags))
 		attr.flags = cmd->flags;
 
+	ret = ib_rdmacg_try_charge(&obj->uobject.cg_obj, ib_dev,
+				   RDMACG_VERB_RESOURCE_CQ);
+	if (ret)
+		goto err_charge;
+
 	cq = ib_dev->create_cq(ib_dev, &attr,
 					     file->ucontext, uhw);
 	if (IS_ERR(cq)) {
@@ -1452,6 +1507,10 @@ err_free:
 	ib_destroy_cq(cq);
 
 err_file:
+	ib_rdmacg_uncharge(&obj->uobject.cg_obj, ib_dev,
+			   RDMACG_VERB_RESOURCE_CQ);
+
+err_charge:
 	if (ev_file)
 		ib_uverbs_release_ucq(file, ev_file, obj);
 
@@ -1732,6 +1791,8 @@ ssize_t ib_uverbs_destroy_cq(struct ib_uverbs_file *file,
 	if (ret)
 		return ret;
 
+	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_VERB_RESOURCE_CQ);
+
 	idr_remove_uobj(&ib_uverbs_cq_idr, uobj);
 
 	mutex_lock(&file->mutex);
@@ -1904,6 +1965,11 @@ static int create_qp(struct ib_uverbs_file *file,
 			goto err_put;
 		}
 
+	ret = ib_rdmacg_try_charge(&obj->uevent.uobject.cg_obj, device,
+				   RDMACG_VERB_RESOURCE_QP);
+	if (ret)
+		goto err_put;
+
 	if (cmd->qp_type == IB_QPT_XRC_TGT)
 		qp = ib_create_qp(pd, &attr);
 	else
@@ -1911,7 +1977,7 @@ static int create_qp(struct ib_uverbs_file *file,
 
 	if (IS_ERR(qp)) {
 		ret = PTR_ERR(qp);
-		goto err_put;
+		goto err_create;
 	}
 
 	if (cmd->qp_type != IB_QPT_XRC_TGT) {
@@ -1992,6 +2058,10 @@ err_cb:
 err_destroy:
 	ib_destroy_qp(qp);
 
+err_create:
+	ib_rdmacg_uncharge(&obj->uevent.uobject.cg_obj, device,
+			   RDMACG_VERB_RESOURCE_QP);
+
 err_put:
 	if (xrcd)
 		put_xrcd_read(xrcd_uobj);
@@ -2462,6 +2532,8 @@ ssize_t ib_uverbs_destroy_qp(struct ib_uverbs_file *file,
 	if (ret)
 		return ret;
 
+	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_VERB_RESOURCE_QP);
+
 	if (obj->uxrcd)
 		atomic_dec(&obj->uxrcd->refcnt);
 
@@ -2908,10 +2980,15 @@ ssize_t ib_uverbs_create_ah(struct ib_uverbs_file *file,
 	memset(&attr.dmac, 0, sizeof(attr.dmac));
 	memcpy(attr.grh.dgid.raw, cmd.attr.grh.dgid, 16);
 
+	ret = ib_rdmacg_try_charge(&uobj->cg_obj, ib_dev,
+				   RDMACG_VERB_RESOURCE_AH);
+	if (ret)
+		goto err_put;
+
 	ah = ib_create_ah(pd, &attr);
 	if (IS_ERR(ah)) {
 		ret = PTR_ERR(ah);
-		goto err_put;
+		goto err_create;
 	}
 
 	ah->uobject  = uobj;
@@ -2947,6 +3024,9 @@ err_copy:
 err_destroy:
 	ib_destroy_ah(ah);
 
+err_create:
+	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_VERB_RESOURCE_AH);
+
 err_put:
 	put_pd_read(pd);
 
@@ -2981,6 +3061,8 @@ ssize_t ib_uverbs_destroy_ah(struct ib_uverbs_file *file,
 	if (ret)
 		return ret;
 
+	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_VERB_RESOURCE_AH);
+
 	idr_remove_uobj(&ib_uverbs_ah_idr, uobj);
 
 	mutex_lock(&file->mutex);
@@ -3688,10 +3770,16 @@ int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file,
 		err = -EINVAL;
 		goto err_free;
 	}
+
+	err = ib_rdmacg_try_charge(&uobj->cg_obj, ib_dev,
+				   RDMACG_VERB_RESOURCE_FLOW);
+	if (err)
+		goto err_free;
+
 	flow_id = ib_create_flow(qp, flow_attr, IB_FLOW_DOMAIN_USER);
 	if (IS_ERR(flow_id)) {
 		err = PTR_ERR(flow_id);
-		goto err_free;
+		goto err_create;
 	}
 	flow_id->qp = qp;
 	flow_id->uobject = uobj;
@@ -3725,6 +3813,8 @@ err_copy:
 	idr_remove_uobj(&ib_uverbs_rule_idr, uobj);
 destroy_flow:
 	ib_destroy_flow(flow_id);
+err_create:
+	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_VERB_RESOURCE_FLOW);
 err_free:
 	kfree(flow_attr);
 err_put:
@@ -3764,8 +3854,11 @@ int ib_uverbs_ex_destroy_flow(struct ib_uverbs_file *file,
 	flow_id = uobj->object;
 
 	ret = ib_destroy_flow(flow_id);
-	if (!ret)
+	if (!ret) {
 		uobj->live = 0;
+		ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev,
+				   RDMACG_VERB_RESOURCE_FLOW);
+	}
 
 	put_uobj_write(uobj);
 
@@ -3833,6 +3926,11 @@ static int __uverbs_create_xsrq(struct ib_uverbs_file *file,
 	obj->uevent.events_reported = 0;
 	INIT_LIST_HEAD(&obj->uevent.event_list);
 
+	ret = ib_rdmacg_try_charge(&obj->uevent.uobject.cg_obj, ib_dev,
+				   RDMACG_VERB_RESOURCE_SRQ);
+	if (ret)
+		goto err_put_cq;
+
 	srq = pd->device->create_srq(pd, &attr, udata);
 	if (IS_ERR(srq)) {
 		ret = PTR_ERR(srq);
@@ -3897,6 +3995,8 @@ err_destroy:
 	ib_destroy_srq(srq);
 
 err_put:
+	ib_rdmacg_uncharge(&obj->uevent.uobject.cg_obj, ib_dev,
+			   RDMACG_VERB_RESOURCE_SRQ);
 	put_pd_read(pd);
 
 err_put_cq:
@@ -4083,6 +4183,8 @@ ssize_t ib_uverbs_destroy_srq(struct ib_uverbs_file *file,
 	if (ret)
 		return ret;
 
+	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_VERB_RESOURCE_SRQ);
+
 	if (srq_type == IB_SRQT_XRC) {
 		us = container_of(obj, struct ib_usrq_object, uevent);
 		atomic_dec(&us->uxrcd->refcnt);
@@ -4116,6 +4218,7 @@ int ib_uverbs_ex_query_device(struct ib_uverbs_file *file,
 	struct ib_uverbs_ex_query_device_resp resp = { {0} };
 	struct ib_uverbs_ex_query_device  cmd;
 	struct ib_device_attr attr = {0};
+	int    limits[RDMACG_RESOURCE_MAX];
 	int err;
 
 	if (ucore->inlen < sizeof(cmd))
@@ -4140,7 +4243,8 @@ int ib_uverbs_ex_query_device(struct ib_uverbs_file *file,
 	if (err)
 		return err;
 
-	copy_query_dev_fields(file, ib_dev, &resp.base, &attr);
+	ib_rdmacg_query_limit(ib_dev, limits);
+	copy_query_dev_fields(file, ib_dev, &resp.base, &attr, limits);
 
 	if (ucore->outlen < resp.response_length + sizeof(resp.odp_caps))
 		goto end;
diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index 0012fa5..3414eda 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -51,6 +51,7 @@
 #include <rdma/ib.h>
 
 #include "uverbs.h"
+#include "core_priv.h"
 
 MODULE_AUTHOR("Roland Dreier");
 MODULE_DESCRIPTION("InfiniBand userspace verbs access");
@@ -236,6 +237,8 @@ static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file,
 
 		idr_remove_uobj(&ib_uverbs_ah_idr, uobj);
 		ib_destroy_ah(ah);
+		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
+				   RDMACG_VERB_RESOURCE_AH);
 		kfree(uobj);
 	}
 
@@ -245,6 +248,8 @@ static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file,
 
 		idr_remove_uobj(&ib_uverbs_mw_idr, uobj);
 		uverbs_dealloc_mw(mw);
+		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
+				   RDMACG_VERB_RESOURCE_MW);
 		kfree(uobj);
 	}
 
@@ -253,6 +258,8 @@ static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file,
 
 		idr_remove_uobj(&ib_uverbs_rule_idr, uobj);
 		ib_destroy_flow(flow_id);
+		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
+				   RDMACG_VERB_RESOURCE_FLOW);
 		kfree(uobj);
 	}
 
@@ -267,6 +274,8 @@ static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file,
 		} else {
 			ib_uverbs_detach_umcast(qp, uqp);
 			ib_destroy_qp(qp);
+			ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
+					   RDMACG_VERB_RESOURCE_QP);
 		}
 		ib_uverbs_release_uevent(file, &uqp->uevent);
 		kfree(uqp);
@@ -300,6 +309,8 @@ static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file,
 
 		idr_remove_uobj(&ib_uverbs_srq_idr, uobj);
 		ib_destroy_srq(srq);
+		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
+				   RDMACG_VERB_RESOURCE_SRQ);
 		ib_uverbs_release_uevent(file, uevent);
 		kfree(uevent);
 	}
@@ -312,6 +323,8 @@ static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file,
 
 		idr_remove_uobj(&ib_uverbs_cq_idr, uobj);
 		ib_destroy_cq(cq);
+		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
+				   RDMACG_VERB_RESOURCE_CQ);
 		ib_uverbs_release_ucq(file, ev_file, ucq);
 		kfree(ucq);
 	}
@@ -321,6 +334,8 @@ static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file,
 
 		idr_remove_uobj(&ib_uverbs_mr_idr, uobj);
 		ib_dereg_mr(mr);
+		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
+				   RDMACG_VERB_RESOURCE_MR);
 		kfree(uobj);
 	}
 
@@ -341,9 +356,13 @@ static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file,
 
 		idr_remove_uobj(&ib_uverbs_pd_idr, uobj);
 		ib_dealloc_pd(pd);
+		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
+				   RDMACG_VERB_RESOURCE_PD);
 		kfree(uobj);
 	}
 
+	ib_rdmacg_uncharge(&context->cg_obj, context->device,
+			   RDMACG_VERB_RESOURCE_UCTX);
 	put_pid(context->tgid);
 
 	return context->device->dealloc_ucontext(context);
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 8e90dd2..7178891 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -60,6 +60,7 @@
 #include <linux/atomic.h>
 #include <linux/mmu_notifier.h>
 #include <asm/uaccess.h>
+#include <linux/cgroup_rdma.h>
 
 extern struct workqueue_struct *ib_wq;
 extern struct workqueue_struct *ib_comp_wq;
@@ -1313,6 +1314,12 @@ struct ib_fmr_attr {
 	u8	page_shift;
 };
 
+struct ib_rdmacg_object {
+#ifdef CONFIG_CGROUP_RDMA
+	struct rdma_cgroup	*cg;		/* owner rdma cgroup */
+#endif
+};
+
 struct ib_umem;
 
 struct ib_ucontext {
@@ -1347,12 +1354,14 @@ struct ib_ucontext {
 	struct list_head	no_private_counters;
 	int                     odp_mrs_count;
 #endif
+	struct ib_rdmacg_object cg_obj;
 };
 
 struct ib_uobject {
 	u64			user_handle;	/* handle given to us by userspace */
 	struct ib_ucontext     *context;	/* associated user context */
 	void		       *object;		/* containing object */
+	struct ib_rdmacg_object cg_obj;		/* rdmacg object */
 	struct list_head	list;		/* link to context's list */
 	int			id;		/* index into kernel idr */
 	struct kref		ref;
@@ -2043,6 +2052,10 @@ struct ib_device {
 	struct attribute_group	     *hw_stats_ag;
 	struct rdma_hw_stats         *hw_stats;
 
+#ifdef CONFIG_CGROUP_RDMA
+	struct rdmacg_device	     cg_device;
+#endif
+
 	/**
 	 * The following mandatory functions are used only at device
 	 * registration.  Keep functions such as these at the end of this
-- 
1.8.3.1

[PATCHv11 3/3] rdmacg: Added documentation for rdmacg

[Parav Pandit]

Added documentation for v1 and v2 version describing high
level design and usage examples on using rdma controller.

Signed-off-by: Parav Pandit <pandit.parav@gmail.com>
---
 Documentation/cgroup-v1/rdma.txt | 117 +++++++++++++++++++++++++++++++++++++++
 Documentation/cgroup-v2.txt      |  44 +++++++++++++++
 2 files changed, 161 insertions(+)
 create mode 100644 Documentation/cgroup-v1/rdma.txt

diff --git a/Documentation/cgroup-v1/rdma.txt b/Documentation/cgroup-v1/rdma.txt
new file mode 100644
index 0000000..7b78678
--- /dev/null
+++ b/Documentation/cgroup-v1/rdma.txt
@@ -0,0 +1,117 @@
+				RDMA Controller
+				----------------
+
+Contents
+--------
+
+1. Overview
+  1-1. What is RDMA controller?
+  1-2. Why RDMA controller needed?
+  1-3. How is RDMA controller implemented?
+2. Usage Examples
+
+1. Overview
+
+1-1. What is RDMA controller?
+-----------------------------
+
+RDMA controller allows user to limit RDMA/IB specific resources that a given
+set of processes can use. These processes are grouped using RDMA controller.
+
+RDMA controller defines well defined verb resources which can be limited for
+processes of a cgroup.
+
+1-2. Why RDMA controller needed?
+--------------------------------
+
+Currently user space applications can easily take away all the rdma device
+specific resources such as AH, CQ, QP, MR etc. Due to which other applications
+in other cgroup or kernel space ULPs may not even get chance to allocate any
+rdma resources. This can leads to service unavailability.
+
+Therefore RDMA controller is needed through which resource consumption
+of processes can be limited. Through this controller various different rdma
+resources can be accounted.
+
+1-3. How is RDMA controller implemented?
+----------------------------------------
+
+RDMA cgroup allows limit configuration of resources. Rdma cgroup maintains
+resource accounting per cgroup, per device using resource pool structure.
+Each such resource pool is limited up to 64 resources in given resource pool
+by rdma cgroup, which can be extended later if required.
+
+This resource pool object is linked to the cgroup css. Typically there
+are 0 to 4 resource pool instances per cgroup, per device in most use cases.
+But nothing limits to have it more. At present hundreds of RDMA devices per
+single cgroup may not be handled optimally, however there is no
+known use case or requirement for such configuration either.
+
+Since RDMA resources can be allocated from any process and can be freed by any
+of the child processes which shares the address space, rdma resources are
+always owned by the creator cgroup css. This allows process migration from one
+to other cgroup without major complexity of transferring resource ownership;
+because such ownership is not really present due to shared nature of
+rdma resources. Linking resources around css also ensures that cgroups can be
+deleted after processes migrated. This allow progress migration as well with
+active resources, even though that is not a primary use case.
+
+Whenever RDMA resource charing occurs, owner rdma cgroup is returned to
+the caller. Same rdma cgroup should be passed while uncharging the resource.
+This also allows process migrated with active RDMA resource to charge
+to new owner cgroup for new resource. It also allows to uncharge resource of
+a process from previously charged cgroup which is migrated to new cgroup,
+even though that is not a primary use case.
+
+Resource pool object is created in following situations.
+(a) User sets the limit and no previous resource pool exist for the device
+of interest for the cgroup.
+(b) No resource limits were configured, but IB/RDMA stack tries to
+charge the resource. So that it correctly uncharge them when applications are
+running without limits and later on when limits are enforced during uncharging,
+otherwise usage count will drop to negative.
+
+Resource pool is destroyed if all the resource limits are set to max and
+it is the last resource getting deallocated.
+
+User should set all the limit to max value if it intents to remove/unconfigure
+the resource pool for a particular device.
+
+IB stack honors limits enforced by the rdma controller. When application
+query about maximum resource limits of IB device, it returns minimum of
+what is configured by user for a given cgroup and what is supported by
+IB device.
+
+Following resources can be accounted by rdma controller.
+  uctx		Maximum number of User Contexts
+  pd		Maximum number of Protection domains
+  ah		Maximum number of Address handles
+  mr		Maximum number of Memory Regions
+  mw		Maximum number of Memory Windows
+  cq		Maximum number of Completion Queues
+  srq		Maximum number of Shared Receive Queues
+  qp		Maximum number of Queue Pairs
+  flow		Maximum number of Flows
+
+
+2. Usage Examples
+-----------------
+
+(a) Configure resource limit:
+echo mlx4_0 mr=100 qp=10 ah=2 > /sys/fs/cgroup/rdma/1/rdma.max
+echo ocrdma1 mr=120 qp=20 cq=10 > /sys/fs/cgroup/rdma/2/rdma.max
+
+(b) Query resource limit:
+cat /sys/fs/cgroup/rdma/2/rdma.max
+#Output:
+mlx4_0 uctx=max pd=max ah=2 mr=100 mw=max cq=max srq=max qp=10 flow=max
+ocrdma1 uctx=1 pd=5 ah=1 mr=10 cq=10 srq=max qp=20 flow=max flow=max
+
+(c) Query current usage:
+cat /sys/fs/cgroup/rdma/2/rdma.current
+#Output:
+mlx4_0 uctx=1 pd=2 ah=2 mr=95 mw=0 cq=2 srq=0 qp=8 flow=0
+ocrdma1 uctx=1 pd=6 ah=9 mr=20 mw=0 cq=1 srq=0 qp=2 flow=0
+
+(d) Delete resource limit:
+echo mlx4_0 mr=max qp=max ah=max > /sys/fs/cgroup/rdma/1/rdma.max
diff --git a/Documentation/cgroup-v2.txt b/Documentation/cgroup-v2.txt
index 4cc07ce..cf5a3d3 100644
--- a/Documentation/cgroup-v2.txt
+++ b/Documentation/cgroup-v2.txt
@@ -47,6 +47,8 @@ CONTENTS
   5-3. IO
     5-3-1. IO Interface Files
     5-3-2. Writeback
+  5-4. RDMA
+    5-4-1. RDMA Interface Files
 6. Namespace
   6-1. Basics
   6-2. The Root and Views
@@ -1119,6 +1121,48 @@ writeback as follows.
 	vm.dirty[_background]_ratio.
 
 
+5-4. RDMA
+
+The "rdma" controller regulates the distribution and accounting of
+of RDMA resources.
+
+5-4-1. RDMA Interface Files
+
+  rdma.max
+	A readwrite file that exists for all the cgroups except root that
+	describes current configured resource limit for a RDMA/IB device.
+
+	Lines are keyed by device name and are not ordered.
+	Each line contains space separated resource name and its configured
+	limit that can be distributed.
+
+	Following keys are defined.
+
+	  uctx		Maximum number of User Contexts
+	  pd		Maximum number of Protection domains
+	  ah		Maximum number of Address handles
+	  mr		Maximum number of Memory Regions
+	  mw		Maximum number of Memory Windows
+	  cq		Maximum number of Completion Queues
+	  srq		Maximum number of Shared Receive Queues
+	  qp		Maximum number of Queue Pairs
+	  flow		Maximum number of Flows
+
+	An example for mlx4 and ocrdma device follows.
+
+	  mlx4_0 uctx=max pd=4 ah=2 mr=10 mw=max cq=1 srq=1 qp=10 flow=10
+	  ocrdma1 uctx=2 pd=2 ah=2 mr=20 mw=max cq=1 srq=1 qp=10 flow=10
+
+  rdma.current
+	A read-only file that describes current resource usage.
+	It exists for all the cgroup except root.
+
+	An example for mlx4 and ocrdma device follows.
+
+	  mlx4_1 uctx=1 ah=0 pd=1 cq=4 qp=4 mr=100 srq=0 flow=10
+	  ocrdma1 uctx=2 pd=2 ah=2 mr=20 mw=max cq=1 srq=1 qp=10 flow=10
+
+
 6. Namespace
 
 6-1. Basics
-- 
1.8.3.1

Re: [PATCHv11 0/3] rdmacg: IB/core: rdma controller support

[Tejun Heo]

(cc'ing Christoph)

On Mon, Aug 22, 2016 at 06:03:48PM +0530, Parav Pandit wrote:
> rdmacg: IB/core: rdma controller support
> 
> Patch is generated and tested against below Doug's linux-rdma
> git tree.
> 
> URL: git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma.git
> Branch: master
> 
> Patchset is also compiled and tested against below Tejun's cgroup tree
> using cgroup v2 mode.
> URL: git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup.git
> Branch: master
> 
> Overview:
> Currently user space applications can easily take away all the rdma
> device specific resources such as AH, CQ, QP, MR etc. Due to which other
> applications in other cgroup or kernel space ULPs may not even get chance
> to allocate any rdma resources. This results into service unavailibility.
> 
> RDMA cgroup addresses this issue by allowing resource accounting,
> limit enforcement on per cgroup, per rdma device basis.
> 
> RDMA uverbs layer will enforce limits on well defined RDMA verb
> resources without any HCA vendor device driver involvement.
> 
> RDMA uverbs layer will not do limit enforcement of HCA hw vendor
> specific resources. Instead rdma cgroup provides set of APIs
> through which vendor specific drivers can do resource accounting
> by making use of rdma cgroup.
> 
> Resource limit enforcement is hierarchical.
> 
> When process is migrated with active RDMA resources, rdma cgroup
> continues to uncharge original cgroup for allocated resource. New resource
> is charged to current process's cgroup, which means if the process is
> migrated with active resources, for new resources it will be charged to
> new cgroup and old resources will be correctly uncharged from old cgroup.
> 
> Changes from v10:
>   * (To address comments from Tejun, Christoph)
>    1. Removed unused rpool_list_lock from rdma_cgroup structure.
>    2. Moved rdma resource definition to rdma cgroup instead of IB stack
>    3. Added prefix rdmacg to static instances
>    4. Simplified locking with single mutex for all operations
>    5. Following approach of atomically allocating object and
>       charging resource in hirerchy
>    6. Code simplification due to single lock
>    7. Using for_each_set_bit API for bit operation
>    8. Renamed list heads as Objects instead of _head
>    9. Renamed list entries as _node instead of _list.
>   10. Made usage_num to 64 bit to avoid overflow and to avoid 
>       additional code to track non zero number of usage counts.
>   * (To address comments from Doug)
>    1. Added copyright and GPLv2 license

Looks good to me.  I just have a nit in the documentation.  Christoph,
what do you think?

Thanks.

-- 
tejun

Re: [PATCHv11 3/3] rdmacg: Added documentation for rdmacg

[Tejun Heo]

On Mon, Aug 22, 2016 at 06:03:51PM +0530, Parav Pandit wrote:
> +  rdma.max
> +	A readwrite file that exists for all the cgroups except root that

Can you please add that it's a nested-keyed file?

...
> +  rdma.current
> +	A read-only file that describes current resource usage.

Ditto.

Thanks.

-- 
tejun

Re: [PATCHv11 3/3] rdmacg: Added documentation for rdmacg

[Rami Rosen]

Hi,

> +Whenever RDMA resource charing occurs, owner rdma cgroup is returned to
Should be: charging instead of charing

> +(b) Query resource limit:
> +cat /sys/fs/cgroup/rdma/2/rdma.max
> +#Output:
> +mlx4_0 uctx=max pd=max ah=2 mr=100 mw=max cq=max srq=max qp=10 flow=max
> +ocrdma1 uctx=1 pd=5 ah=1 mr=10 cq=10 srq=max qp=20 flow=max flow=max

Is this really so: double"flow=max" at the end of the ocrdma1 line?
(flow=max flow=max)

> +5-4. RDMA
> +
> +The "rdma" controller regulates the distribution and accounting of
> +of RDMA resources.
"of of" should be only a single "of"


> +         mlx4_1 uctx=1 ah=0 pd=1 cq=4 qp=4 mr=100 srq=0 flow=10
> +         ocrdma1 uctx=2 pd=2 ah=2 mr=20 mw=max cq=1 srq=1 qp=10 flow=10

Seems to be inconsistency here: in the first line you have qp=4
*before* srq=0, but in the second line you have qp=10 *after* srq=1.

Keep on the good work!

Regards,
Rami Rosen

Re: [PATCHv11 0/3] rdmacg: IB/core: rdma controller support

[Christoph Hellwig]

On Wed, Aug 24, 2016 at 05:17:47PM -0400, Tejun Heo wrote:
> Looks good to me.  I just have a nit in the documentation.  Christoph,
> what do you think?

Looks reasonable from a quick look, but I didn't do a full review yet.

Re: [PATCHv11 3/3] rdmacg: Added documentation for rdmacg

[Parav Pandit]

Hi Rami Rosen,

On Thu, Aug 25, 2016 at 4:25 AM, Rami Rosen <roszenrami@gmail.com> wrote:
> Hi,
>
>> +Whenever RDMA resource charing occurs, owner rdma cgroup is returned to
> Should be: charging instead of charing
>
>> +(b) Query resource limit:
>> +cat /sys/fs/cgroup/rdma/2/rdma.max
>> +#Output:
>> +mlx4_0 uctx=max pd=max ah=2 mr=100 mw=max cq=max srq=max qp=10 flow=max
>> +ocrdma1 uctx=1 pd=5 ah=1 mr=10 cq=10 srq=max qp=20 flow=max flow=max
>
> Is this really so: double"flow=max" at the end of the ocrdma1 line?
> (flow=max flow=max)
>
>> +5-4. RDMA
>> +
>> +The "rdma" controller regulates the distribution and accounting of
>> +of RDMA resources.
> "of of" should be only a single "of"
>

>
>> +         mlx4_1 uctx=1 ah=0 pd=1 cq=4 qp=4 mr=100 srq=0 flow=10
>> +         ocrdma1 uctx=2 pd=2 ah=2 mr=20 mw=max cq=1 srq=1 qp=10 flow=10
>
> Seems to be inconsistency here: in the first line you have qp=4
> *before* srq=0, but in the second line you have qp=10 *after* srq=1.
>

I will fix above 4 typo errors.
Christoph has done quick review, I will wait for him to complete the
review before spinning v12 for these fixes.

> Keep on the good work!
Thank you for the motivation.

>
> Regards,
> Rami Rosen

Re: [PATCHv11 0/3] rdmacg: IB/core: rdma controller support

[Parav Pandit]

Hi Christoph,

On Thu, Aug 25, 2016 at 1:02 PM, Christoph Hellwig <hch@lst.de> wrote:
> On Wed, Aug 24, 2016 at 05:17:47PM -0400, Tejun Heo wrote:
>> Looks good to me.  I just have a nit in the documentation.  Christoph,
>> what do you think?
>
> Looks reasonable from a quick look, but I didn't do a full review yet.

Thanks for the quick review.
I need to generate new patch for 4 typo errors that Rami pointed out
and documentation addition that Tejun asked.
I will wait for your full review, in case if there are any comments,
so I can cover them in one patch.