[patch 3/6] mempolicy: add MPOL_F_STATIC_NODES flag

[David Rientjes]

Add an optional mempolicy mode flag, MPOL_F_STATIC_NODES, that suppresses
the node remap when the policy is rebound.

Adds another member to struct mempolicy, nodemask_t user_nodemask, as
part of a union with cpuset_mems_allowed:

	struct mempolicy {
		...
		union {
			nodemask_t cpuset_mems_allowed;
			nodemask_t user_nodemask;
		} w;
	}

that stores the the nodemask that the user passed when he or she created
the mempolicy via set_mempolicy() or mbind().  When using
MPOL_F_STATIC_NODES, which is passed with any mempolicy mode, the user's
passed nodemask intersected with the VMA or task's allowed nodes is always
used when determining the preferred node, building the MPOL_BIND zonelist,
or creating the interleave nodemask.  This happens whenever the policy is 
rebound, including when a task's cpuset assignment changes or the cpuset's
mems are changed.

This creates an interesting side-effect in that it allows the mempolicy
"intent" to lie dormant and uneffected until it has access to the node(s)
that it desires.  For example, if you currently ask for an interleaved
policy over a set of nodes that you do not have access to, the mempolicy
is not created and the task continues to use the previous policy.  With
this change, however, it is possible to create the same mempolicy; it is
only effected when access to nodes in the nodemask is acquired.

It is also possible to mount tmpfs with the static nodemask behavior when
specifying a node or nodemask.  To do this, simply add "=static"
immediately following the mempolicy mode at mount time:

	mount -o remount mpol=interleave=static:1-3

Also removes mpol_check_policy() and folds its logic into mpol_new() since
it is now obsoleted.  The unused vma_mpol_equal() is also removed.

Cc: Paul Jackson <pj@xxxxxxx>
Cc: Christoph Lameter <clameter@xxxxxxx>
Cc: Lee Schermerhorn <Lee.Schermerhorn@xxxxxx>
Cc: Andi Kleen <ak@xxxxxxx>
Signed-off-by: David Rientjes <rientjes@xxxxxxxxxx>
---
 include/linux/mempolicy.h |   11 ++-
 mm/mempolicy.c            |  208 ++++++++++++++++++++++++---------------------
 mm/shmem.c                |    2 +
 3 files changed, 119 insertions(+), 102 deletions(-)

diff --git a/include/linux/mempolicy.h b/include/linux/mempolicy.h
--- a/include/linux/mempolicy.h
+++ b/include/linux/mempolicy.h
@@ -24,11 +24,13 @@ enum {
 };
 
 /* Flags for set_mempolicy */
+#define MPOL_F_STATIC_NODES	(1 << 15)
+
 /*
  * MPOL_MODE_FLAGS is the union of all possible optional mode flags passed to
  * either set_mempolicy() or mbind().
  */
-#define MPOL_MODE_FLAGS		(0)
+#define MPOL_MODE_FLAGS		(MPOL_F_STATIC_NODES)
 
 /* Flags for get_mempolicy */
 #define MPOL_F_NODE	(1<<0)	/* return next IL mode instead of node mask */
@@ -84,7 +86,10 @@ struct mempolicy {
 		nodemask_t	 nodes;		/* interleave */
 		/* undefined for default */
 	} v;
-	nodemask_t cpuset_mems_allowed;	/* mempolicy relative to these nodes */
+	union {
+		nodemask_t cpuset_mems_allowed;	/* relative to these nodes */
+		nodemask_t user_nodemask;	/* nodemask passed by user */
+	} w;
 };
 
 /*
@@ -123,7 +128,6 @@ static inline int mpol_equal(struct mempolicy *a, struct mempolicy *b)
 		return 1;
 	return __mpol_equal(a, b);
 }
-#define vma_mpol_equal(a,b) mpol_equal(vma_policy(a), vma_policy(b))
 
 /* Could later add inheritance of the process policy here. */
 
@@ -188,7 +192,6 @@ static inline int mpol_equal(struct mempolicy *a, struct mempolicy *b)
 {
 	return 1;
 }
-#define vma_mpol_equal(a,b) 1
 
 #define mpol_set_vma_default(vma) do {} while(0)
 
diff --git a/mm/mempolicy.c b/mm/mempolicy.c
--- a/mm/mempolicy.c
+++ b/mm/mempolicy.c
@@ -113,58 +113,6 @@ struct mempolicy default_policy = {
 static void mpol_rebind_policy(struct mempolicy *pol,
                                const nodemask_t *newmask);
 
-/* Do sanity checking on a policy */
-static int mpol_check_policy(unsigned short mode, nodemask_t *nodes)
-{
-	int was_empty, is_empty;
-
-	if (!nodes)
-		return 0;
-
-	/*
-	 * "Contextualize" the in-coming nodemast for cpusets:
-	 * Remember whether in-coming nodemask was empty,  If not,
-	 * restrict the nodes to the allowed nodes in the cpuset.
-	 * This is guaranteed to be a subset of nodes with memory.
-	 */
-	cpuset_update_task_memory_state();
-	is_empty = was_empty = nodes_empty(*nodes);
-	if (!was_empty) {
-		nodes_and(*nodes, *nodes, cpuset_current_mems_allowed);
-		is_empty = nodes_empty(*nodes);	/* after "contextualization" */
-	}
-
-	switch (mode) {
-	case MPOL_DEFAULT:
-		/*
-		 * require caller to specify an empty nodemask
-		 * before "contextualization"
-		 */
-		if (!was_empty)
-			return -EINVAL;
-		break;
-	case MPOL_BIND:
-	case MPOL_INTERLEAVE:
-		/*
-		 * require at least 1 valid node after "contextualization"
-		 */
-		if (is_empty)
-			return -EINVAL;
-		break;
-	case MPOL_PREFERRED:
-		/*
-		 * Did caller specify invalid nodes?
-		 * Don't silently accept this as "local allocation".
-		 */
-		if (!was_empty && is_empty)
-			return -EINVAL;
-		break;
-	default:
-		BUG();
-	}
-	return 0;
-}
-
 /* Generate a custom zonelist for the BIND policy. */
 static struct zonelist *bind_zonelist(nodemask_t *nodes)
 {
@@ -202,40 +150,48 @@ static struct zonelist *bind_zonelist(nodemask_t *nodes)
 	return zl;
 }
 
+static inline int mpol_store_user_nodemask(const struct mempolicy *pol)
+{
+	return pol->flags & MPOL_F_STATIC_NODES;
+}
+
 /* Create a new policy */
 static struct mempolicy *mpol_new(unsigned short mode, unsigned short flags,
 				  nodemask_t *nodes)
 {
 	struct mempolicy *policy;
+	nodemask_t cpuset_context_nmask;
+	void *error = ERR_PTR(-EINVAL);
 
 	pr_debug("setting mode %d flags %d nodes[0] %lx\n",
 		 mode, flags, nodes ? nodes_addr(*nodes)[0] : -1);
 
 	if (mode == MPOL_DEFAULT)
-		return NULL;
+		return (nodes && nodes_weight(*nodes)) ? error : NULL;
 	policy = kmem_cache_alloc(policy_cache, GFP_KERNEL);
 	if (!policy)
 		return ERR_PTR(-ENOMEM);
 	atomic_set(&policy->refcnt, 1);
+	cpuset_update_task_memory_state();
+	nodes_and(cpuset_context_nmask, *nodes, cpuset_current_mems_allowed);
 	switch (mode) {
 	case MPOL_INTERLEAVE:
-		policy->v.nodes = *nodes;
-		if (nodes_weight(policy->v.nodes) == 0) {
-			kmem_cache_free(policy_cache, policy);
-			return ERR_PTR(-EINVAL);
-		}
+		if (nodes_empty(*nodes) || nodes_empty(cpuset_context_nmask))
+			goto free;
+		policy->v.nodes = cpuset_context_nmask;
 		break;
 	case MPOL_PREFERRED:
-		policy->v.preferred_node = first_node(*nodes);
+		policy->v.preferred_node = first_node(cpuset_context_nmask);
 		if (policy->v.preferred_node >= MAX_NUMNODES)
-			policy->v.preferred_node = -1;
+			goto free;
 		break;
 	case MPOL_BIND:
-		policy->v.zonelist = bind_zonelist(nodes);
+		if (nodes_empty(*nodes))
+			goto free;
+		policy->v.zonelist = bind_zonelist(&cpuset_context_nmask);
 		if (IS_ERR(policy->v.zonelist)) {
-			void *error_code = policy->v.zonelist;
-			kmem_cache_free(policy_cache, policy);
-			return error_code;
+			error = policy->v.zonelist;
+			goto free;
 		}
 		break;
 	default:
@@ -243,8 +199,15 @@ static struct mempolicy *mpol_new(unsigned short mode, unsigned short flags,
 	}
 	policy->policy = mode;
 	policy->flags = flags;
-	policy->cpuset_mems_allowed = cpuset_mems_allowed(current);
+	if (mpol_store_user_nodemask(policy))
+		policy->w.user_nodemask = *nodes;
+	else
+		policy->w.cpuset_mems_allowed = cpuset_mems_allowed(current);
 	return policy;
+
+free:
+	kmem_cache_free(policy_cache, policy);
+	return error;
 }
 
 static void gather_stats(struct page *, void *, int pte_dirty);
@@ -490,15 +453,14 @@ static long do_set_mempolicy(unsigned short mode, unsigned short flags,
 {
 	struct mempolicy *new;
 
-	if (mpol_check_policy(mode, nodes))
-		return -EINVAL;
 	new = mpol_new(mode, flags, nodes);
 	if (IS_ERR(new))
 		return PTR_ERR(new);
 	mpol_free(current->mempolicy);
 	current->mempolicy = new;
 	mpol_set_task_struct_flag();
-	if (new && new->policy == MPOL_INTERLEAVE)
+	if (new && new->policy == MPOL_INTERLEAVE &&
+	    nodes_weight(new->v.nodes))
 		current->il_next = first_node(new->v.nodes);
 	return 0;
 }
@@ -818,9 +780,6 @@ static long do_mbind(unsigned long start, unsigned long len,
 	if (end == start)
 		return 0;
 
-	if (mpol_check_policy(mode, nmask))
-		return -EINVAL;
-
 	new = mpol_new(mode, mode_flags, nmask);
 	if (IS_ERR(new))
 		return PTR_ERR(new);
@@ -1211,7 +1170,8 @@ static unsigned interleave_nodes(struct mempolicy *policy)
 	next = next_node(nid, policy->v.nodes);
 	if (next >= MAX_NUMNODES)
 		next = first_node(policy->v.nodes);
-	me->il_next = next;
+	if (next < MAX_NUMNODES)
+		me->il_next = next;
 	return nid;
 }
 
@@ -1249,10 +1209,13 @@ static unsigned offset_il_node(struct mempolicy *pol,
 		struct vm_area_struct *vma, unsigned long off)
 {
 	unsigned nnodes = nodes_weight(pol->v.nodes);
-	unsigned target = (unsigned)off % nnodes;
+	unsigned target;
 	int c;
 	int nid = -1;
 
+	if (!nnodes)
+		return numa_node_id();
+	target = (unsigned int)off % nnodes;
 	c = 0;
 	do {
 		nid = next_node(nid, pol->v.nodes);
@@ -1457,6 +1420,16 @@ struct mempolicy *__mpol_copy(struct mempolicy *old)
 	return new;
 }
 
+static int mpol_match_intent(const struct mempolicy *a,
+			     const struct mempolicy *b)
+{
+	if (a->flags != b->flags)
+		return 0;
+	if (!mpol_store_user_nodemask(a))
+		return 1;
+	return nodes_equal(a->w.user_nodemask, b->w.user_nodemask);
+}
+
 /* Slow path of a mempolicy comparison */
 int __mpol_equal(struct mempolicy *a, struct mempolicy *b)
 {
@@ -1464,6 +1437,8 @@ int __mpol_equal(struct mempolicy *a, struct mempolicy *b)
 		return 0;
 	if (a->policy != b->policy)
 		return 0;
+	if (a->policy != MPOL_DEFAULT && !mpol_match_intent(a, b))
+		return 0;
 	switch (a->policy) {
 	case MPOL_DEFAULT:
 		return 1;
@@ -1770,54 +1745,79 @@ void numa_default_policy(void)
 static void mpol_rebind_policy(struct mempolicy *pol,
 			       const nodemask_t *newmask)
 {
-	nodemask_t *mpolmask;
 	nodemask_t tmp;
+	int static_nodes;
 
 	if (!pol)
 		return;
-	mpolmask = &pol->cpuset_mems_allowed;
-	if (nodes_equal(*mpolmask, *newmask))
+	static_nodes = pol->flags & MPOL_F_STATIC_NODES;
+	if (!mpol_store_user_nodemask(pol) &&
+	    nodes_equal(pol->w.cpuset_mems_allowed, *newmask))
 		return;
 
 	switch (pol->policy) {
 	case MPOL_DEFAULT:
 		break;
 	case MPOL_INTERLEAVE:
-		nodes_remap(tmp, pol->v.nodes, *mpolmask, *newmask);
+		if (static_nodes)
+			nodes_and(tmp, pol->w.user_nodemask, *newmask);
+		else {
+			nodes_remap(tmp, pol->v.nodes,
+				    pol->w.cpuset_mems_allowed, *newmask);
+			pol->w.cpuset_mems_allowed = *newmask;
+		}
 		pol->v.nodes = tmp;
-		*mpolmask = *newmask;
-		current->il_next = node_remap(current->il_next,
-						*mpolmask, *newmask);
+		if (!node_isset(current->il_next, tmp)) {
+			current->il_next = next_node(current->il_next, tmp);
+			if (current->il_next >= MAX_NUMNODES)
+				current->il_next = first_node(tmp);
+			if (current->il_next >= MAX_NUMNODES)
+				current->il_next = numa_node_id();
+		}
 		break;
 	case MPOL_PREFERRED:
-		pol->v.preferred_node = node_remap(pol->v.preferred_node,
-						*mpolmask, *newmask);
-		*mpolmask = *newmask;
+		if (static_nodes) {
+			int node = first_node(pol->w.user_nodemask);
+
+			if (node_isset(node, *newmask))
+				pol->v.preferred_node = node;
+			else
+				pol->v.preferred_node = -1;
+		} else {
+			pol->v.preferred_node = node_remap(pol->v.preferred_node,
+					pol->w.cpuset_mems_allowed, *newmask);
+			pol->w.cpuset_mems_allowed = *newmask;
+		}
 		break;
 	case MPOL_BIND: {
-		nodemask_t nodes;
-		struct zone **z;
 		struct zonelist *zonelist;
 
-		nodes_clear(nodes);
-		for (z = pol->v.zonelist->zones; *z; z++)
-			node_set(zone_to_nid(*z), nodes);
-		nodes_remap(tmp, nodes, *mpolmask, *newmask);
-		nodes = tmp;
-
-		zonelist = bind_zonelist(&nodes);
+		if (static_nodes)
+			nodes_and(tmp, pol->w.user_nodemask, *newmask);
+		else {
+			nodemask_t nodes;
+			struct zone **z;
+
+			nodes_clear(nodes);
+			for (z = pol->v.zonelist->zones; *z; z++)
+				node_set(zone_to_nid(*z), nodes);
+			nodes_remap(tmp, nodes, pol->w.cpuset_mems_allowed,
+				    *newmask);
+			pol->w.cpuset_mems_allowed = *newmask;
+		}
 
 		/* If no mem, then zonelist is NULL and we keep old zonelist.
 		 * If that old zonelist has no remaining mems_allowed nodes,
 		 * then zonelist_policy() will "FALL THROUGH" to MPOL_DEFAULT.
 		 */
-
-		if (!IS_ERR(zonelist)) {
-			/* Good - got mem - substitute new zonelist */
-			kfree(pol->v.zonelist);
-			pol->v.zonelist = zonelist;
+		if (nodes_weight(tmp)) {
+			zonelist = bind_zonelist(&tmp);
+			if (!IS_ERR(zonelist)) {
+				/* Good - got mem - substitute new zonelist */
+				kfree(pol->v.zonelist);
+				pol->v.zonelist = zonelist;
+			}
 		}
-		*mpolmask = *newmask;
 		break;
 	}
 	default:
@@ -1870,6 +1870,7 @@ static inline int mpol_to_str(char *buffer, int maxlen, struct mempolicy *pol)
 	int l;
 	nodemask_t nodes;
 	unsigned short mode = pol ? pol->policy : MPOL_DEFAULT;
+	unsigned short flags = pol ? pol->flags : 0;
 
 	switch (mode) {
 	case MPOL_DEFAULT:
@@ -1901,6 +1902,17 @@ static inline int mpol_to_str(char *buffer, int maxlen, struct mempolicy *pol)
 	strcpy(p, policy_types[mode]);
 	p += l;
 
+	if (flags) {
+		int need_bar = 0;
+
+		if (buffer + maxlen < p + 2)
+			return -ENOSPC;
+		*p++ = '=';
+
+		if (flags & MPOL_F_STATIC_NODES)
+			p += sprintf(p, "%sstatic", need_bar++ ? "|" : "");
+	}
+
 	if (!nodes_empty(nodes)) {
 		if (buffer + maxlen < p + 2)
 			return -ENOSPC;
diff --git a/mm/shmem.c b/mm/shmem.c
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -1129,6 +1129,8 @@ static int shmem_parse_mpol(char *value, unsigned short *policy,
 		err = 0;
 	}
 	if (flags) {
+		if (!strcmp(flags, "static"))
+			*mode_flags |= MPOL_F_STATIC_NODES;
 	}
 out:
 	/* Restore string for error message */
--
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/