* Processor and Memory placement constraints for sets of tasks.
*
* Copyright (C) 2003 BULL SA.
- * Copyright (C) 2004 Silicon Graphics, Inc.
+ * Copyright (C) 2004-2006 Silicon Graphics, Inc.
*
* Portions derived from Patrick Mochel's sysfs code.
* sysfs is Copyright (c) 2001-3 Patrick Mochel
- * Portions Copyright (c) 2004 Silicon Graphics, Inc.
*
- * 2003-10-10 Written by Simon Derr <simon.derr@bull.net>
+ * 2003-10-10 Written by Simon Derr.
* 2003-10-22 Updates by Stephen Hemminger.
- * 2004 May-July Rework by Paul Jackson <pj@sgi.com>
+ * 2004 May-July Rework by Paul Jackson.
*
* This file is subject to the terms and conditions of the GNU General Public
* License. See the file COPYING in the main directory of the Linux
CS_MEM_EXCLUSIVE,
CS_MEMORY_MIGRATE,
CS_REMOVED,
- CS_NOTIFY_ON_RELEASE
+ CS_NOTIFY_ON_RELEASE,
+ CS_SPREAD_PAGE,
+ CS_SPREAD_SLAB,
} cpuset_flagbits_t;
/* convenient tests for these bits */
return test_bit(CS_MEMORY_MIGRATE, &cs->flags);
}
+static inline int is_spread_page(const struct cpuset *cs)
+{
+ return test_bit(CS_SPREAD_PAGE, &cs->flags);
+}
+
+static inline int is_spread_slab(const struct cpuset *cs)
+{
+ return test_bit(CS_SPREAD_SLAB, &cs->flags);
+}
+
/*
- * Increment this atomic integer everytime any cpuset changes its
+ * Increment this integer everytime any cpuset changes its
* mems_allowed value. Users of cpusets can track this generation
* number, and avoid having to lock and reload mems_allowed unless
* the cpuset they're using changes generation.
* 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,
+ * there is no need to mark it atomic.
*/
-static atomic_t cpuset_mems_generation = ATOMIC_INIT(1);
+static int cpuset_mems_generation;
static struct cpuset top_cpuset = {
.flags = ((1 << CS_CPU_EXCLUSIVE) | (1 << CS_MEM_EXCLUSIVE)),
* current->cpuset if a task has its memory placement changed.
* 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. Doesn't need task_lock to guard
- * against another task changing a non-NULL cpuset pointer to NULL,
- * as that is only done by a task on itself, and if the current task
- * is here, it is not simultaneously in the exit code NULL'ing its
- * cpuset pointer. This routine also might acquire callback_mutex and
+ * 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
+ * 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
cs = tsk->cpuset; /* Maybe changed when task not locked */
guarantee_online_mems(cs, &tsk->mems_allowed);
tsk->cpuset_mems_generation = cs->mems_generation;
+ if (is_spread_page(cs))
+ tsk->flags |= PF_SPREAD_PAGE;
+ else
+ tsk->flags &= ~PF_SPREAD_PAGE;
+ if (is_spread_slab(cs))
+ tsk->flags |= PF_SPREAD_SLAB;
+ else
+ tsk->flags &= ~PF_SPREAD_SLAB;
task_unlock(tsk);
mutex_unlock(&callback_mutex);
mpol_rebind_task(tsk, &tsk->mems_allowed);
return 0;
}
+/*
+ * cpuset_migrate_mm
+ *
+ * Migrate memory region from one set of nodes to another.
+ *
+ * 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()
+ * 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.
+ *
+ * Hold callback_mutex around the two modifications of our tasks
+ * mems_allowed to synchronize with cpuset_mems_allowed().
+ *
+ * While the mm_struct we are migrating is typically from some
+ * other task, the task_struct mems_allowed that we are hacking
+ * is for our current task, which must allocate new pages for that
+ * migrating memory region.
+ *
+ * We call cpuset_update_task_memory_state() before hacking
+ * our tasks mems_allowed, so that we are assured of being in
+ * sync with our tasks cpuset, and in particular, callbacks to
+ * cpuset_update_task_memory_state() from nested page allocations
+ * won't see any mismatch of our cpuset and task mems_generation
+ * values, so won't overwrite our hacked tasks mems_allowed
+ * nodemask.
+ */
+
+static void cpuset_migrate_mm(struct mm_struct *mm, const nodemask_t *from,
+ const nodemask_t *to)
+{
+ struct task_struct *tsk = current;
+
+ cpuset_update_task_memory_state();
+
+ mutex_lock(&callback_mutex);
+ tsk->mems_allowed = *to;
+ mutex_unlock(&callback_mutex);
+
+ do_migrate_pages(mm, from, to, MPOL_MF_MOVE_ALL);
+
+ mutex_lock(&callback_mutex);
+ guarantee_online_mems(tsk->cpuset, &tsk->mems_allowed);
+ mutex_unlock(&callback_mutex);
+}
+
/*
* Handle user request to change the 'mems' memory placement
* of a cpuset. Needs to validate the request, update the
mutex_lock(&callback_mutex);
cs->mems_allowed = trialcs.mems_allowed;
- cs->mems_generation = atomic_inc_return(&cpuset_mems_generation);
+ cs->mems_generation = cpuset_mems_generation++;
mutex_unlock(&callback_mutex);
set_cpuset_being_rebound(cs); /* causes mpol_copy() rebind */
struct mm_struct *mm = mmarray[i];
mpol_rebind_mm(mm, &cs->mems_allowed);
- if (migrate) {
- do_migrate_pages(mm, &oldmem, &cs->mems_allowed,
- MPOL_MF_MOVE_ALL);
- }
+ if (migrate)
+ cpuset_migrate_mm(mm, &oldmem, &cs->mems_allowed);
mmput(mm);
}
/*
* 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_NOTIFY_ON_RELEASE, CS_MEMORY_MIGRATE)
+ * 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
*
mm = get_task_mm(tsk);
if (mm) {
mpol_rebind_mm(mm, &to);
+ if (is_memory_migrate(cs))
+ cpuset_migrate_mm(mm, &from, &to);
mmput(mm);
}
- if (is_memory_migrate(cs))
- do_migrate_pages(tsk->mm, &from, &to, MPOL_MF_MOVE_ALL);
put_task_struct(tsk);
synchronize_rcu();
if (atomic_dec_and_test(&oldcs->count))
FILE_NOTIFY_ON_RELEASE,
FILE_MEMORY_PRESSURE_ENABLED,
FILE_MEMORY_PRESSURE,
+ FILE_SPREAD_PAGE,
+ FILE_SPREAD_SLAB,
FILE_TASKLIST,
} cpuset_filetype_t;
case FILE_MEMORY_PRESSURE:
retval = -EACCES;
break;
+ case FILE_SPREAD_PAGE:
+ retval = update_flag(CS_SPREAD_PAGE, cs, buffer);
+ cs->mems_generation = cpuset_mems_generation++;
+ break;
+ case FILE_SPREAD_SLAB:
+ retval = update_flag(CS_SPREAD_SLAB, cs, buffer);
+ cs->mems_generation = cpuset_mems_generation++;
+ break;
case FILE_TASKLIST:
retval = attach_task(cs, buffer, &pathbuf);
break;
case FILE_MEMORY_PRESSURE:
s += sprintf(s, "%d", fmeter_getrate(&cs->fmeter));
break;
+ case FILE_SPREAD_PAGE:
+ *s++ = is_spread_page(cs) ? '1' : '0';
+ break;
+ case FILE_SPREAD_SLAB:
+ *s++ = is_spread_slab(cs) ? '1' : '0';
+ break;
default:
retval = -EINVAL;
goto out;
.private = FILE_MEMORY_PRESSURE,
};
+static struct cftype cft_spread_page = {
+ .name = "memory_spread_page",
+ .private = FILE_SPREAD_PAGE,
+};
+
+static struct cftype cft_spread_slab = {
+ .name = "memory_spread_slab",
+ .private = FILE_SPREAD_SLAB,
+};
+
static int cpuset_populate_dir(struct dentry *cs_dentry)
{
int err;
return err;
if ((err = cpuset_add_file(cs_dentry, &cft_memory_pressure)) < 0)
return err;
+ if ((err = cpuset_add_file(cs_dentry, &cft_spread_page)) < 0)
+ return err;
+ if ((err = cpuset_add_file(cs_dentry, &cft_spread_slab)) < 0)
+ return err;
if ((err = cpuset_add_file(cs_dentry, &cft_tasks)) < 0)
return err;
return 0;
cs->flags = 0;
if (notify_on_release(parent))
set_bit(CS_NOTIFY_ON_RELEASE, &cs->flags);
+ if (is_spread_page(parent))
+ set_bit(CS_SPREAD_PAGE, &cs->flags);
+ if (is_spread_slab(parent))
+ set_bit(CS_SPREAD_SLAB, &cs->flags);
cs->cpus_allowed = CPU_MASK_NONE;
cs->mems_allowed = NODE_MASK_NONE;
atomic_set(&cs->count, 0);
INIT_LIST_HEAD(&cs->sibling);
INIT_LIST_HEAD(&cs->children);
- cs->mems_generation = atomic_inc_return(&cpuset_mems_generation);
+ cs->mems_generation = cpuset_mems_generation++;
fmeter_init(&cs->fmeter);
cs->parent = parent;
struct task_struct *tsk = current;
tsk->cpuset = &top_cpuset;
- tsk->cpuset->mems_generation = atomic_inc_return(&cpuset_mems_generation);
+ tsk->cpuset->mems_generation = cpuset_mems_generation++;
return 0;
}
top_cpuset.mems_allowed = NODE_MASK_ALL;
fmeter_init(&top_cpuset.fmeter);
- top_cpuset.mems_generation = atomic_inc_return(&cpuset_mems_generation);
+ top_cpuset.mems_generation = cpuset_mems_generation++;
init_task.cpuset = &top_cpuset;
* because tsk is already marked PF_EXITING, so attach_task() won't
* mess with it, or task is a failed fork, never visible to attach_task.
*
- * Hack:
+ * the_top_cpuset_hack:
*
* Set the exiting tasks cpuset to the root cpuset (top_cpuset).
*
struct cpuset *cs;
cs = tsk->cpuset;
- tsk->cpuset = &top_cpuset; /* Hack - see comment above */
+ tsk->cpuset = &top_cpuset; /* the_top_cpuset_hack - see above */
if (notify_on_release(cs)) {
char *pathbuf = NULL;
* So only GFP_KERNEL allocations, if all nodes in the cpuset are
* short of memory, might require taking the callback_mutex mutex.
*
- * The first loop over the zonelist in mm/page_alloc.c:__alloc_pages()
- * calls here with __GFP_HARDWALL always set in gfp_mask, enforcing
- * hardwall cpusets - no allocation on a node outside the cpuset is
- * allowed (unless in interrupt, of course).
+ * The first call here from mm/page_alloc:get_page_from_freelist()
+ * has __GFP_HARDWALL set in gfp_mask, enforcing hardwall cpusets, so
+ * no allocation on a node outside the cpuset is allowed (unless in
+ * interrupt, of course).
*
- * The second loop doesn't even call here for GFP_ATOMIC requests
- * (if the __alloc_pages() local variable 'wait' is set). That check
- * and the checks below have the combined affect in the second loop of
- * the __alloc_pages() routine that:
+ * The second pass through get_page_from_freelist() doesn't even call
+ * here for GFP_ATOMIC calls. For those calls, the __alloc_pages()
+ * variable 'wait' is not set, and the bit ALLOC_CPUSET is not set
+ * in alloc_flags. That logic and the checks below have the combined
+ * affect that:
* in_interrupt - any node ok (current task context irrelevant)
* GFP_ATOMIC - any node ok
* GFP_KERNEL - any node in enclosing mem_exclusive cpuset ok
* GFP_USER - only nodes in current tasks mems allowed ok.
+ *
+ * Rule:
+ * Don't call cpuset_zone_allowed() if you can't sleep, unless you
+ * pass in the __GFP_HARDWALL flag set in gfp_flag, which disables
+ * the code that might scan up ancestor cpusets and sleep.
**/
int __cpuset_zone_allowed(struct zone *z, gfp_t gfp_mask)
{
int node; /* node that zone z is on */
const struct cpuset *cs; /* current cpuset ancestors */
- int allowed = 1; /* is allocation in zone z allowed? */
+ int allowed; /* is allocation in zone z allowed? */
if (in_interrupt())
return 1;
node = z->zone_pgdat->node_id;
+ might_sleep_if(!(gfp_mask & __GFP_HARDWALL));
if (node_isset(node, current->mems_allowed))
return 1;
if (gfp_mask & __GFP_HARDWALL) /* If hardwall request, stop here */
mutex_unlock(&callback_mutex);
}
+/**
+ * cpuset_mem_spread_node() - On which node to begin search for a page
+ *
+ * If a task is marked PF_SPREAD_PAGE or PF_SPREAD_SLAB (as for
+ * tasks in a cpuset with is_spread_page or is_spread_slab set),
+ * and if the memory allocation used cpuset_mem_spread_node()
+ * to determine on which node to start looking, as it will for
+ * certain page cache or slab cache pages such as used for file
+ * system buffers and inode caches, then instead of starting on the
+ * local node to look for a free page, rather spread the starting
+ * node around the tasks mems_allowed nodes.
+ *
+ * We don't have to worry about the returned node being offline
+ * because "it can't happen", and even if it did, it would be ok.
+ *
+ * The routines calling guarantee_online_mems() are careful to
+ * only set nodes in task->mems_allowed that are online. So it
+ * should not be possible for the following code to return an
+ * offline node. But if it did, that would be ok, as this routine
+ * is not returning the node where the allocation must be, only
+ * the node where the search should start. The zonelist passed to
+ * __alloc_pages() will include all nodes. If the slab allocator
+ * is passed an offline node, it will fall back to the local node.
+ * See kmem_cache_alloc_node().
+ */
+
+int cpuset_mem_spread_node(void)
+{
+ int node;
+
+ node = next_node(current->cpuset_mem_spread_rotor, current->mems_allowed);
+ if (node == MAX_NUMNODES)
+ node = first_node(current->mems_allowed);
+ current->cpuset_mem_spread_rotor = node;
+ return node;
+}
+EXPORT_SYMBOL_GPL(cpuset_mem_spread_node);
+
/**
* cpuset_excl_nodes_overlap - Do we overlap @p's mem_exclusive ancestors?
* @p: pointer to task_struct of some other task.
* - 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.
+ * 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.
*/
-
static int proc_cpuset_show(struct seq_file *m, void *v)
{
- struct cpuset *cs;
struct task_struct *tsk;
char *buf;
int retval = 0;
tsk = m->private;
mutex_lock(&manage_mutex);
- cs = tsk->cpuset;
- if (!cs) {
- retval = -EINVAL;
- goto out;
- }
-
- retval = cpuset_path(cs, buf, PAGE_SIZE);
+ retval = cpuset_path(tsk->cpuset, buf, PAGE_SIZE);
if (retval < 0)
goto out;
seq_puts(m, buf);