[PATCH 0/2] Cpuset hardwall flag: Introduction

[PATCH 0/2] Cpuset hardwall flag: Introduction

[menage]

Currently the cpusets mem_exclusive flag is overloaded to mean both
"no-overlapping" and "no GFP_KERNEL allocations outside this cpuset".

These patches add a new mem_hardwall flag with just the allocation
restriction part of the mem_exclusive semantics, without breaking
backwards-compatibility for those who continue to use just
mem_exclusive. Additionally, the cgroup control file registration for
cpusets is cleaned up to reduce boilerplate.

Signed-off-by: Paul Menage <menage@google.com>

--

[PATCH 1/2] Cpuset hardwall flag: Switch cpusets to use the bulk cgroup_add_files() API

[menage]

[-- Attachment #1: cpuset_add_files.patch --]
[-- Type: text/plain, Size: 4831 bytes --]

This change tidies up the cpusets control file definitions, and
reduces the amount of boilerplate required to add/change control files
in the future.

Signed-off-by: Paul Menage <menage@google.com>

---
 kernel/cpuset.c |  149 +++++++++++++++++++++++++-------------------------------
 1 file changed, 68 insertions(+), 81 deletions(-)

Index: hardwall-2.6.25-rc3-mm1/kernel/cpuset.c
===================================================================
--- hardwall-2.6.25-rc3-mm1.orig/kernel/cpuset.c
+++ hardwall-2.6.25-rc3-mm1/kernel/cpuset.c
@@ -1397,46 +1397,69 @@ static u64 cpuset_read_u64(struct cgroup
  * for the common functions, 'private' gives the type of file
  */
 
-static struct cftype cft_cpus = {
-	.name = "cpus",
-	.read = cpuset_common_file_read,
-	.write = cpuset_common_file_write,
-	.private = FILE_CPULIST,
-};
-
-static struct cftype cft_mems = {
-	.name = "mems",
-	.read = cpuset_common_file_read,
-	.write = cpuset_common_file_write,
-	.private = FILE_MEMLIST,
-};
-
-static struct cftype cft_cpu_exclusive = {
-	.name = "cpu_exclusive",
-	.read_u64 = cpuset_read_u64,
-	.write_u64 = cpuset_write_u64,
-	.private = FILE_CPU_EXCLUSIVE,
-};
-
-static struct cftype cft_mem_exclusive = {
-	.name = "mem_exclusive",
-	.read_u64 = cpuset_read_u64,
-	.write_u64 = cpuset_write_u64,
-	.private = FILE_MEM_EXCLUSIVE,
-};
-
-static struct cftype cft_sched_load_balance = {
-	.name = "sched_load_balance",
-	.read_u64 = cpuset_read_u64,
-	.write_u64 = cpuset_write_u64,
-	.private = FILE_SCHED_LOAD_BALANCE,
-};
-
-static struct cftype cft_memory_migrate = {
-	.name = "memory_migrate",
-	.read_u64 = cpuset_read_u64,
-	.write_u64 = cpuset_write_u64,
-	.private = FILE_MEMORY_MIGRATE,
+static struct cftype files[] = {
+	{
+		.name = "cpus",
+		.read = cpuset_common_file_read,
+		.write = cpuset_common_file_write,
+		.private = FILE_CPULIST,
+	},
+
+	{
+		.name = "mems",
+		.read = cpuset_common_file_read,
+		.write = cpuset_common_file_write,
+		.private = FILE_MEMLIST,
+	},
+
+	{
+		.name = "cpu_exclusive",
+		.read_u64 = cpuset_read_u64,
+		.write_u64 = cpuset_write_u64,
+		.private = FILE_CPU_EXCLUSIVE,
+	},
+
+	{
+		.name = "mem_exclusive",
+		.read_u64 = cpuset_read_u64,
+		.write_u64 = cpuset_write_u64,
+		.private = FILE_MEM_EXCLUSIVE,
+	},
+
+	{
+		.name = "sched_load_balance",
+		.read_u64 = cpuset_read_u64,
+		.write_u64 = cpuset_write_u64,
+		.private = FILE_SCHED_LOAD_BALANCE,
+	},
+
+	{
+		.name = "memory_migrate",
+		.read_u64 = cpuset_read_u64,
+		.write_u64 = cpuset_write_u64,
+		.private = FILE_MEMORY_MIGRATE,
+	},
+
+	{
+		.name = "memory_pressure",
+		.read_u64 = cpuset_read_u64,
+		.write_u64 = cpuset_write_u64,
+		.private = FILE_MEMORY_PRESSURE,
+	},
+
+	{
+		.name = "memory_spread_page",
+		.read_u64 = cpuset_read_u64,
+		.write_u64 = cpuset_write_u64,
+		.private = FILE_SPREAD_PAGE,
+	},
+
+	{
+		.name = "memory_spread_slab",
+		.read_u64 = cpuset_read_u64,
+		.write_u64 = cpuset_write_u64,
+		.private = FILE_SPREAD_SLAB,
+	},
 };
 
 static struct cftype cft_memory_pressure_enabled = {
@@ -1446,54 +1469,18 @@ static struct cftype cft_memory_pressure
 	.private = FILE_MEMORY_PRESSURE_ENABLED,
 };
 
-static struct cftype cft_memory_pressure = {
-	.name = "memory_pressure",
-	.read_u64 = cpuset_read_u64,
-	.write_u64 = cpuset_write_u64,
-	.private = FILE_MEMORY_PRESSURE,
-};
-
-static struct cftype cft_spread_page = {
-	.name = "memory_spread_page",
-	.read_u64 = cpuset_read_u64,
-	.write_u64 = cpuset_write_u64,
-	.private = FILE_SPREAD_PAGE,
-};
-
-static struct cftype cft_spread_slab = {
-	.name = "memory_spread_slab",
-	.read_u64 = cpuset_read_u64,
-	.write_u64 = cpuset_write_u64,
-	.private = FILE_SPREAD_SLAB,
-};
-
 static int cpuset_populate(struct cgroup_subsys *ss, struct cgroup *cont)
 {
 	int err;
 
-	if ((err = cgroup_add_file(cont, ss, &cft_cpus)) < 0)
-		return err;
-	if ((err = cgroup_add_file(cont, ss, &cft_mems)) < 0)
-		return err;
-	if ((err = cgroup_add_file(cont, ss, &cft_cpu_exclusive)) < 0)
-		return err;
-	if ((err = cgroup_add_file(cont, ss, &cft_mem_exclusive)) < 0)
-		return err;
-	if ((err = cgroup_add_file(cont, ss, &cft_memory_migrate)) < 0)
-		return err;
-	if ((err = cgroup_add_file(cont, ss, &cft_sched_load_balance)) < 0)
-		return err;
-	if ((err = cgroup_add_file(cont, ss, &cft_memory_pressure)) < 0)
-		return err;
-	if ((err = cgroup_add_file(cont, ss, &cft_spread_page)) < 0)
-		return err;
-	if ((err = cgroup_add_file(cont, ss, &cft_spread_slab)) < 0)
+	err = cgroup_add_files(cont, ss, files, ARRAY_SIZE(files));
+	if (err)
 		return err;
 	/* memory_pressure_enabled is in root cpuset only */
-	if (err == 0 && !cont->parent)
+	if (!cont->parent)
 		err = cgroup_add_file(cont, ss,
-					 &cft_memory_pressure_enabled);
-	return 0;
+				      &cft_memory_pressure_enabled);
+	return err;
 }
 
 /*

--

[PATCH 2/2] Cpuset hardwall flag: Add a mem_hardwall flag to cpusets

[menage]

[-- Attachment #1: hardwall.patch --]
[-- Type: text/plain, Size: 8254 bytes --]

This flag provides the hardwalling properties of mem_exclusive,
without enforcing the exclusivity. Either mem_hardwall or
mem_exclusive is sufficient to prevent GFP_KERNEL allocations from
passing outside the cpuset's assigned nodes.

Signed-off-by: Paul Menage <menage@google.com>

---
 Documentation/cpusets.txt |   26 +++++++++++++-----------
 kernel/cpuset.c           |   48 ++++++++++++++++++++++++++++++----------------
 2 files changed, 46 insertions(+), 28 deletions(-)

Index: hardwall-2.6.25-rc3-mm1/kernel/cpuset.c
===================================================================
--- hardwall-2.6.25-rc3-mm1.orig/kernel/cpuset.c
+++ hardwall-2.6.25-rc3-mm1/kernel/cpuset.c
@@ -124,6 +124,7 @@ struct cpuset_hotplug_scanner {
 typedef enum {
 	CS_CPU_EXCLUSIVE,
 	CS_MEM_EXCLUSIVE,
+	CS_MEM_HARDWALL,
 	CS_MEMORY_MIGRATE,
 	CS_SCHED_LOAD_BALANCE,
 	CS_SPREAD_PAGE,
@@ -141,6 +142,11 @@ static inline int is_mem_exclusive(const
 	return test_bit(CS_MEM_EXCLUSIVE, &cs->flags);
 }
 
+static inline int is_mem_hardwall(const struct cpuset *cs)
+{
+	return test_bit(CS_MEM_HARDWALL, &cs->flags);
+}
+
 static inline int is_sched_load_balance(const struct cpuset *cs)
 {
 	return test_bit(CS_SCHED_LOAD_BALANCE, &cs->flags);
@@ -1002,12 +1008,9 @@ int current_cpuset_is_being_rebound(void
 
 /*
  * update_flag - read a 0 or a 1 in a file and update associated flag
- * bit:	the bit to update (CS_CPU_EXCLUSIVE, CS_MEM_EXCLUSIVE,
- *				CS_SCHED_LOAD_BALANCE,
- *				CS_NOTIFY_ON_RELEASE, CS_MEMORY_MIGRATE,
- *				CS_SPREAD_PAGE, CS_SPREAD_SLAB)
- * cs:	the cpuset to update
- * buf:	the buffer where we read the 0 or 1
+ * bit:		the bit to update (see cpuset_flagbits_t)
+ * cs:		the cpuset to update
+ * turning_on: 	whether the flag is being set or cleared
  *
  * Call with cgroup_mutex held.
  */
@@ -1188,6 +1191,7 @@ typedef enum {
 	FILE_MEMLIST,
 	FILE_CPU_EXCLUSIVE,
 	FILE_MEM_EXCLUSIVE,
+	FILE_MEM_HARDWALL,
 	FILE_SCHED_LOAD_BALANCE,
 	FILE_MEMORY_PRESSURE_ENABLED,
 	FILE_MEMORY_PRESSURE,
@@ -1268,6 +1272,9 @@ static int cpuset_write_u64(struct cgrou
 	case FILE_MEM_EXCLUSIVE:
 		retval = update_flag(CS_MEM_EXCLUSIVE, cs, val);
 		break;
+	case FILE_MEM_HARDWALL:
+		retval = update_flag(CS_MEM_HARDWALL, cs, val);
+		break;
 	case FILE_SCHED_LOAD_BALANCE:
 		retval = update_flag(CS_SCHED_LOAD_BALANCE, cs, val);
 		break;
@@ -1375,6 +1382,8 @@ static u64 cpuset_read_u64(struct cgroup
 		return is_cpu_exclusive(cs);
 	case FILE_MEM_EXCLUSIVE:
 		return is_mem_exclusive(cs);
+	case FILE_MEM_HARDWALL:
+		return is_mem_hardwall(cs);
 	case FILE_SCHED_LOAD_BALANCE:
 		return is_sched_load_balance(cs);
 	case FILE_MEMORY_MIGRATE:
@@ -1427,6 +1436,13 @@ static struct cftype files[] = {
 	},
 
 	{
+		.name = "mem_hardwall",
+		.read_u64 = cpuset_read_u64,
+		.write_u64 = cpuset_write_u64,
+		.private = FILE_MEM_HARDWALL,
+	},
+
+	{
 		.name = "sched_load_balance",
 		.read_u64 = cpuset_read_u64,
 		.write_u64 = cpuset_write_u64,
@@ -1913,14 +1929,14 @@ int cpuset_nodemask_valid_mems_allowed(n
 }
 
 /*
- * nearest_exclusive_ancestor() - Returns the nearest mem_exclusive
- * ancestor to the specified cpuset.  Call holding callback_mutex.
- * If no ancestor is mem_exclusive (an unusual configuration), then
- * returns the root cpuset.
+ * nearest_hardwall_ancestor() - Returns the nearest mem_exclusive or
+ * mem_hardwall ancestor to the specified cpuset.  Call holding
+ * callback_mutex.  If no ancestor is mem_exclusive or mem_hardwall
+ * (an unusual configuration), then returns the root cpuset.
  */
-static const struct cpuset *nearest_exclusive_ancestor(const struct cpuset *cs)
+static const struct cpuset *nearest_hardwall_ancestor(const struct cpuset *cs)
 {
-	while (!is_mem_exclusive(cs) && cs->parent)
+	while (!(is_mem_exclusive(cs) || is_mem_hardwall(cs)) && cs->parent)
 		cs = cs->parent;
 	return cs;
 }
@@ -1934,7 +1950,7 @@ static const struct cpuset *nearest_excl
  * __GFP_THISNODE is set, yes, we can always allocate.  If zone
  * z's node is in our tasks mems_allowed, yes.  If it's not a
  * __GFP_HARDWALL request and this zone's nodes is in the nearest
- * mem_exclusive cpuset ancestor to this tasks cpuset, yes.
+ * hardwalled cpuset ancestor to this tasks cpuset, yes.
  * If the task has been OOM killed and has access to memory reserves
  * as specified by the TIF_MEMDIE flag, yes.
  * Otherwise, no.
@@ -1957,7 +1973,7 @@ static const struct cpuset *nearest_excl
  * and do not allow allocations outside the current tasks cpuset
  * unless the task has been OOM killed as is marked TIF_MEMDIE.
  * GFP_KERNEL allocations are not so marked, so can escape to the
- * nearest enclosing mem_exclusive ancestor cpuset.
+ * nearest enclosing hardwalled ancestor cpuset.
  *
  * Scanning up parent cpusets requires callback_mutex.  The
  * __alloc_pages() routine only calls here with __GFP_HARDWALL bit
@@ -1980,7 +1996,7 @@ static const struct cpuset *nearest_excl
  *	in_interrupt - any node ok (current task context irrelevant)
  *	GFP_ATOMIC   - any node ok
  *	TIF_MEMDIE   - any node ok
- *	GFP_KERNEL   - any node in enclosing mem_exclusive cpuset ok
+ *	GFP_KERNEL   - any node in enclosing hardwalled cpuset ok
  *	GFP_USER     - only nodes in current tasks mems allowed ok.
  *
  * Rule:
@@ -2017,7 +2033,7 @@ int __cpuset_zone_allowed_softwall(struc
 	mutex_lock(&callback_mutex);
 
 	task_lock(current);
-	cs = nearest_exclusive_ancestor(task_cs(current));
+	cs = nearest_hardwall_ancestor(task_cs(current));
 	task_unlock(current);
 
 	allowed = node_isset(node, cs->mems_allowed);
Index: hardwall-2.6.25-rc3-mm1/Documentation/cpusets.txt
===================================================================
--- hardwall-2.6.25-rc3-mm1.orig/Documentation/cpusets.txt
+++ hardwall-2.6.25-rc3-mm1/Documentation/cpusets.txt
@@ -169,6 +169,7 @@ files describing that cpuset:
  - memory_migrate flag: if set, move pages to cpusets nodes
  - cpu_exclusive flag: is cpu placement exclusive?
  - mem_exclusive flag: is memory placement exclusive?
+ - mem_hardwall flag:  is memory allocation hardwalled
  - memory_pressure: measure of how much paging pressure in cpuset
 
 In addition, the root cpuset only has the following file:
@@ -220,17 +221,18 @@ If a cpuset is cpu or mem exclusive, no 
 a direct ancestor or descendent, may share any of the same CPUs or
 Memory Nodes.
 
-A cpuset that is mem_exclusive restricts kernel allocations for
-page, buffer and other data commonly shared by the kernel across
-multiple users.  All cpusets, whether mem_exclusive or not, restrict
-allocations of memory for user space.  This enables configuring a
-system so that several independent jobs can share common kernel data,
-such as file system pages, while isolating each jobs user allocation in
-its own cpuset.  To do this, construct a large mem_exclusive cpuset to
-hold all the jobs, and construct child, non-mem_exclusive cpusets for
-each individual job.  Only a small amount of typical kernel memory,
-such as requests from interrupt handlers, is allowed to be taken
-outside even a mem_exclusive cpuset.
+A cpuset that is mem_exclusive *or* mem_hardwall is "hardwalled",
+i.e. it restricts kernel allocations for page, buffer and other data
+commonly shared by the kernel across multiple users.  All cpusets,
+whether hardwalled or not, restrict allocations of memory for user
+space.  This enables configuring a system so that several independent
+jobs can share common kernel data, such as file system pages, while
+isolating each job's user allocation in its own cpuset.  To do this,
+construct a large mem_exclusive cpuset to hold all the jobs, and
+construct child, non-mem_exclusive cpusets for each individual job.
+Only a small amount of typical kernel memory, such as requests from
+interrupt handlers, is allowed to be taken outside even a
+mem_exclusive cpuset.
 
 
 1.5 What is memory_pressure ?
@@ -639,7 +641,7 @@ Now you want to do something with this c
 
 In this directory you can find several files:
 # ls
-cpus  cpu_exclusive  mems  mem_exclusive  tasks
+cpus  cpu_exclusive  mems  mem_exclusive mem_hardwall  tasks
 
 Reading them will give you information about the state of this cpuset:
 the CPUs and Memory Nodes it can use, the processes that are using

--

Re: [PATCH 1/2] Cpuset hardwall flag: Switch cpusets to use the bulk cgroup_add_files() API

[Li Zefan]

menage@google.com wrote:
> This change tidies up the cpusets control file definitions, and
> reduces the amount of boilerplate required to add/change control files
> in the future.
> 
> Signed-off-by: Paul Menage <menage@google.com>
> 

Actually I've done this cleanup but don't have time to post the
patch. :)

Reviewed-by: Li Zefan <lizf@cn.fujitsu.com>

Re: [PATCH 1/2] Cpuset hardwall flag: Switch cpusets to use the bulk cgroup_add_files() API

[Paul Jackson]

Paul M wrote:
> This change tidies up the cpusets control file definitions,

Acked-by: Paul Jackson <pj@sgi.com>

-- 
                  I won't rest till it's the best ...
                  Programmer, Linux Scalability
                  Paul Jackson <pj@sgi.com> 1.940.382.4214

Re: [PATCH 2/2] Cpuset hardwall flag: Add a mem_hardwall flag to cpusets

[Paul Jackson]

Paul M wrote:
> This flag provides the hardwalling properties of mem_exclusive,
> without enforcing the exclusivity. Either mem_hardwall or ...

Acked-by: Paul Jackson <pj@sgi.com>

-- 
                  I won't rest till it's the best ...
                  Programmer, Linux Scalability
                  Paul Jackson <pj@sgi.com> 1.940.382.4214

Re: [PATCH 0/2] Cpuset hardwall flag: Introduction

[Paul Jackson]

Paul M wrote:
> Currently the cpusets mem_exclusive flag is overloaded to mean both
> "no-overlapping" and "no GFP_KERNEL allocations outside this cpuset".
> 
> These patches add a new mem_hardwall flag with just the allocation
> restriction part of the mem_exclusive semantics, without breaking
> backwards-compatibility for those who continue to use just
> mem_exclusive.

... too bad this nice comment wasn't included in PATCH 2/2, so that
it would automatically make it into the record of history - the source
control log message (as best I understand how Andrew's tools work,
comments off in their own, codeless patch "PATCH 0/N" don't make
it to the source control log, except when Andrew chooses to make a
special effort.)

-- 
                  I won't rest till it's the best ...
                  Programmer, Linux Scalability
                  Paul Jackson <pj@sgi.com> 1.940.382.4214

Re: [PATCH 0/2] Cpuset hardwall flag: Introduction

[Andrew Morton]

On Wed, 5 Mar 2008 06:23:18 -0600 Paul Jackson <pj@sgi.com> wrote:

> Paul M wrote:
> > Currently the cpusets mem_exclusive flag is overloaded to mean both
> > "no-overlapping" and "no GFP_KERNEL allocations outside this cpuset".
> > 
> > These patches add a new mem_hardwall flag with just the allocation
> > restriction part of the mem_exclusive semantics, without breaking
> > backwards-compatibility for those who continue to use just
> > mem_exclusive.
> 
> ... too bad this nice comment wasn't included in PATCH 2/2, so that
> it would automatically make it into the record of history - the source
> control log message (as best I understand how Andrew's tools work,
> comments off in their own, codeless patch "PATCH 0/N" don't make
> it to the source control log, except when Andrew chooses to make a
> special effort.)

I make that special effort almost 100% of the time.  The changelog for the
first patch becomes:


<text from [0/n]>

This patch:

<text from [1/n]>