[PATCH] Update comments in cpuset.c

From: Paul Menage
Date: Tue Jan 29 2008 - 19:40:54 EST


Update comments in cpuset.c

Some of the comments in kernel/cpuset.c were stale following the
transition to control groups; this patch updates them to more closely
match reality.

Signed-off-by: Paul Menage <menage@xxxxxxxxxx>
Acked-by: Paul Jackson <pj@xxxxxxx>

---
kernel/cpuset.c | 128 ++++++++++++++++++--------------------------------------
1 file changed, 43 insertions(+), 85 deletions(-)

Index: linux-2.6.24-rc6-mm1/kernel/cpuset.c
===================================================================
--- linux-2.6.24-rc6-mm1.orig/kernel/cpuset.c
+++ linux-2.6.24-rc6-mm1/kernel/cpuset.c
@@ -64,7 +64,7 @@
*/
int number_of_cpusets __read_mostly;

-/* Retrieve the cpuset from a cgroup */
+/* Forward declare cgroup structures */
struct cgroup_subsys cpuset_subsys;
struct cpuset;

@@ -160,17 +160,17 @@ static inline int is_spread_slab(const s
* number, and avoid having to lock and reload mems_allowed unless
* the cpuset they're using changes generation.
*
- * A single, global generation is needed because attach_task() could
+ * A single, global generation is needed because cpuset_attach_task() could
* reattach a task to a different cpuset, which must not have its
* generation numbers aliased with those of that tasks previous cpuset.
*
* Generations are needed for mems_allowed because one task cannot
- * modify anothers memory placement. So we must enable every task,
+ * modify another's memory placement. So we must enable every task,
* on every visit to __alloc_pages(), to efficiently check whether
* its current->cpuset->mems_allowed has changed, requiring an update
* of its current->mems_allowed.
*
- * Since cpuset_mems_generation is guarded by manage_mutex,
+ * Since writes to cpuset_mems_generation are guarded by the cgroup lock
* there is no need to mark it atomic.
*/
static int cpuset_mems_generation;
@@ -182,17 +182,20 @@ static struct cpuset top_cpuset = {
};

/*
- * We have two global cpuset mutexes below. They can nest.
- * It is ok to first take manage_mutex, then nest callback_mutex. We also
- * require taking task_lock() when dereferencing a tasks cpuset pointer.
- * See "The task_lock() exception", at the end of this comment.
+ * There are two global mutexes guarding cpuset structures. The first
+ * is the main control groups cgroup_mutex, accessed via
+ * cgroup_lock()/cgroup_unlock(). The second is the cpuset-specific
+ * callback_mutex, below. They can nest. It is ok to first take
+ * cgroup_mutex, then nest callback_mutex. We also require taking
+ * task_lock() when dereferencing a task's cpuset pointer. See "The
+ * task_lock() exception", at the end of this comment.
*
* A task must hold both mutexes to modify cpusets. If a task
- * holds manage_mutex, then it blocks others wanting that mutex,
+ * holds cgroup_mutex, then it blocks others wanting that mutex,
* ensuring that it is the only task able to also acquire callback_mutex
* and be able to modify cpusets. It can perform various checks on
* the cpuset structure first, knowing nothing will change. It can
- * also allocate memory while just holding manage_mutex. While it is
+ * also allocate memory while just holding cgroup_mutex. While it is
* performing these checks, various callback routines can briefly
* acquire callback_mutex to query cpusets. Once it is ready to make
* the changes, it takes callback_mutex, blocking everyone else.
@@ -208,60 +211,16 @@ static struct cpuset top_cpuset = {
* The task_struct fields mems_allowed and mems_generation may only
* be accessed in the context of that task, so require no locks.
*
- * Any task can increment and decrement the count field without lock.
- * So in general, code holding manage_mutex or callback_mutex can't rely
- * on the count field not changing. However, if the count goes to
- * zero, then only attach_task(), which holds both mutexes, can
- * increment it again. Because a count of zero means that no tasks
- * are currently attached, therefore there is no way a task attached
- * to that cpuset can fork (the other way to increment the count).
- * So code holding manage_mutex or callback_mutex can safely assume that
- * if the count is zero, it will stay zero. Similarly, if a task
- * holds manage_mutex or callback_mutex on a cpuset with zero count, it
- * knows that the cpuset won't be removed, as cpuset_rmdir() needs
- * both of those mutexes.
- *
* The cpuset_common_file_write handler for operations that modify
- * the cpuset hierarchy holds manage_mutex across the entire operation,
+ * the cpuset hierarchy holds cgroup_mutex across the entire operation,
* single threading all such cpuset modifications across the system.
*
* The cpuset_common_file_read() handlers only hold callback_mutex across
* small pieces of code, such as when reading out possibly multi-word
* cpumasks and nodemasks.
*
- * The fork and exit callbacks cpuset_fork() and cpuset_exit(), don't
- * (usually) take either mutex. These are the two most performance
- * critical pieces of code here. The exception occurs on cpuset_exit(),
- * when a task in a notify_on_release cpuset exits. Then manage_mutex
- * is taken, and if the cpuset count is zero, a usermode call made
- * to /sbin/cpuset_release_agent with the name of the cpuset (path
- * relative to the root of cpuset file system) as the argument.
- *
- * A cpuset can only be deleted if both its 'count' of using tasks
- * is zero, and its list of 'children' cpusets is empty. Since all
- * tasks in the system use _some_ cpuset, and since there is always at
- * least one task in the system (init), therefore, top_cpuset
- * always has either children cpusets and/or using tasks. So we don't
- * need a special hack to ensure that top_cpuset cannot be deleted.
- *
- * The above "Tale of Two Semaphores" would be complete, but for:
- *
- * The task_lock() exception
- *
- * The need for this exception arises from the action of attach_task(),
- * which overwrites one tasks cpuset pointer with another. It does
- * so using both mutexes, however there are several performance
- * critical places that need to reference task->cpuset without the
- * expense of grabbing a system global mutex. Therefore except as
- * noted below, when dereferencing or, as in attach_task(), modifying
- * a tasks cpuset pointer we use task_lock(), which acts on a spinlock
- * (task->alloc_lock) already in the task_struct routinely used for
- * such matters.
- *
- * P.S. One more locking exception. RCU is used to guard the
- * update of a tasks cpuset pointer by attach_task() and the
- * access of task->cpuset->mems_generation via that pointer in
- * the routine cpuset_update_task_memory_state().
+ * Accessing a task's cpuset should be done in accordance with the
+ * guidelines for accessing subsystem state in kernel/cgroup.c
*/

static DEFINE_MUTEX(callback_mutex);
@@ -354,15 +313,14 @@ static void guarantee_online_mems(const * Do not call this routine if in_interrupt().
*
* Call without callback_mutex or task_lock() held. May be
- * called with or without manage_mutex held. Thanks in part to
- * 'the_top_cpuset_hack', the tasks cpuset pointer will never
+ * called with or without cgroup_mutex held. Thanks in part to
+ * 'the_top_cpuset_hack', the task's cpuset pointer will never
* be NULL. This routine also might acquire callback_mutex and
* current->mm->mmap_sem during call.
*
* Reading current->cpuset->mems_generation doesn't need task_lock
* to guard the current->cpuset derefence, because it is guarded
- * from concurrent freeing of current->cpuset by attach_task(),
- * using RCU.
+ * from concurrent freeing of current->cpuset using RCU.
*
* The rcu_dereference() is technically probably not needed,
* as I don't actually mind if I see a new cpuset pointer but
@@ -424,7 +382,7 @@ void cpuset_update_task_memory_state(voi
*
* One cpuset is a subset of another if all its allowed CPUs and
* Memory Nodes are a subset of the other, and its exclusive flags
- * are only set if the other's are set. Call holding manage_mutex.
+ * are only set if the other's are set. Call holding cgroup_mutex.
*/

static int is_cpuset_subset(const struct cpuset *p, const struct cpuset *q)
@@ -442,7 +400,7 @@ static int is_cpuset_subset(const struct
* If we replaced the flag and mask values of the current cpuset
* (cur) with those values in the trial cpuset (trial), would
* our various subset and exclusive rules still be valid? Presumes
- * manage_mutex held.
+ * cgroup_mutex held.
*
* 'cur' is the address of an actual, in-use cpuset. Operations
* such as list traversal that depend on the actual address of the
@@ -476,7 +434,10 @@ static int validate_change(const struct if (!is_cpuset_subset(trial, par))
return -EACCES;

- /* If either I or some sibling (!= me) is exclusive, we can't overlap */
+ /*
+ * If either I or some sibling (!= me) is exclusive, we can't
+ * overlap
+ */
list_for_each_entry(cont, &par->css.cgroup->children, sibling) {
c = cgroup_cs(cont);
if ((is_cpu_exclusive(trial) || is_cpu_exclusive(c)) &&
@@ -733,7 +694,7 @@ static inline int started_after(void *p1
}

/*
- * Call with manage_mutex held. May take callback_mutex during call.
+ * Call with cgroup_mutex held. May take callback_mutex during call.
*/

static int update_cpumask(struct cpuset *cs, char *buf)
@@ -854,11 +815,11 @@ static int update_cpumask(struct cpuset * Temporarilly set tasks mems_allowed to target nodes of migration,
* so that the migration code can allocate pages on these nodes.
*
- * Call holding manage_mutex, so our current->cpuset won't change
- * during this call, as manage_mutex holds off any attach_task()
+ * Call holding cgroup_mutex, so current's cpuset won't change
+ * during this call, as cgroup_mutex holds off any attach_task()
* calls. Therefore we don't need to take task_lock around the
* call to guarantee_online_mems(), as we know no one is changing
- * our tasks cpuset.
+ * our task's cpuset.
*
* Hold callback_mutex around the two modifications of our tasks
* mems_allowed to synchronize with cpuset_mems_allowed().
@@ -903,7 +864,7 @@ static void cpuset_migrate_mm(struct mm_
* the cpuset is marked 'memory_migrate', migrate the tasks
* pages to the new memory.
*
- * Call with manage_mutex held. May take callback_mutex during call.
+ * Call with cgroup_mutex held. May take callback_mutex during call.
* Will take tasklist_lock, scan tasklist for tasks in cpuset cs,
* lock each such tasks mm->mmap_sem, scan its vma's and rebind
* their mempolicies to the cpusets new mems_allowed.
@@ -1016,7 +977,7 @@ static int update_nodemask(struct cpuset
* tasklist_lock. Forks can happen again now - the mpol_copy()
* cpuset_being_rebound check will catch such forks, and rebind
* their vma mempolicies too. Because we still hold the global
- * cpuset manage_mutex, we know that no other rebind effort will
+ * cgroup_mutex, we know that no other rebind effort will
* be contending for the global variable cpuset_being_rebound.
* It's ok if we rebind the same mm twice; mpol_rebind_mm()
* is idempotent. Also migrate pages in each mm to new nodes.
@@ -1031,7 +992,7 @@ static int update_nodemask(struct cpuset
mmput(mm);
}

- /* We're done rebinding vma's to this cpusets new mems_allowed. */
+ /* We're done rebinding vmas to this cpuset's new mems_allowed. */
kfree(mmarray);
cpuset_being_rebound = NULL;
retval = 0;
@@ -1045,7 +1006,7 @@ int current_cpuset_is_being_rebound(void
}

/*
- * Call with manage_mutex held.
+ * Call with cgroup_mutex held.
*/

static int update_memory_pressure_enabled(struct cpuset *cs, char *buf)
@@ -1066,7 +1027,7 @@ static int update_memory_pressure_enable
* cs: the cpuset to update
* buf: the buffer where we read the 0 or 1
*
- * Call with manage_mutex held.
+ * Call with cgroup_mutex held.
*/

static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs, char *buf)
@@ -1200,6 +1161,7 @@ static int fmeter_getrate(struct fmeter return val;
}

+/* Called by cgroups to determine if a cpuset is usable; cgroup_mutex held */
static int cpuset_can_attach(struct cgroup_subsys *ss,
struct cgroup *cont, struct task_struct *tsk)
{
@@ -1547,7 +1509,8 @@ static int cpuset_populate(struct cgroup
* If this becomes a problem for some users who wish to
* allow that scenario, then cpuset_post_clone() could be
* changed to grant parent->cpus_allowed-sibling_cpus_exclusive
- * (and likewise for mems) to the new cgroup.
+ * (and likewise for mems) to the new cgroup. Called with cgroup_mutex
+ * held.
*/
static void cpuset_post_clone(struct cgroup_subsys *ss,
struct cgroup *cgroup)
@@ -1571,11 +1534,8 @@ static void cpuset_post_clone(struct cgr

/*
* cpuset_create - create a cpuset
- * parent: cpuset that will be parent of the new cpuset.
- * name: name of the new cpuset. Will be strcpy'ed.
- * mode: mode to set on new inode
- *
- * Must be called with the mutex on the parent inode held
+ * ss: cpuset cgroup subsystem
+ * cont: control group that the new cpuset will be part of
*/

static struct cgroup_subsys_state *cpuset_create(
@@ -1703,7 +1663,7 @@ int __init cpuset_init(void)
* only one of CPUs or nodes needed to be checked on a given call.
* This was done to minimize text size rather than cpu cycles.
*
- * Call with both manage_mutex and callback_mutex held.
+ * Call with both cgroup_mutex and callback_mutex held.
*
* Recursive, on depth of cpuset subtree.
*/
@@ -1826,7 +1786,7 @@ cpumask_t cpuset_cpus_allowed(struct tas

/**
* cpuset_cpus_allowed_locked - return cpus_allowed mask from a tasks cpuset.
- * Must be called with callback_mutex held.
+ * Must be called with callback_mutex held.
**/
cpumask_t cpuset_cpus_allowed_locked(struct task_struct *tsk)
{
@@ -2163,10 +2123,8 @@ void __cpuset_memory_pressure_bump(void)
* - Used for /proc/<pid>/cpuset.
* - No need to task_lock(tsk) on this tsk->cpuset reference, as it
* doesn't really matter if tsk->cpuset changes after we read it,
- * and we take manage_mutex, keeping attach_task() from changing it
- * anyway. No need to check that tsk->cpuset != NULL, thanks to
- * the_top_cpuset_hack in cpuset_exit(), which sets an exiting tasks
- * cpuset to top_cpuset.
+ * and we take cgroup_mutex, keeping attach_task() from changing it
+ * anyway.
*/
static int proc_cpuset_show(struct seq_file *m, void *unused_v)
{
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/