Merge branch 'master'

6 files changed, 190 insertions, 54 deletions

diff --git a/mm/mempolicy.c b/mm/mempolicy.c
index bedfa4f09c80..67af4cea1e23 100644
--- a/mm/mempolicy.c
+++ b/mm/mempolicy.c
@@ -552,7 +552,7 @@ static void migrate_page_add(struct page *page, struct list_head *pagelist,
          */
         if ((flags & MPOL_MF_MOVE_ALL) || page_mapcount(page) == 1) {
                 if (isolate_lru_page(page))
-                        list_add(&page->lru, pagelist);
+                        list_add_tail(&page->lru, pagelist);
         }
 }
 
@@ -569,6 +569,7 @@ static int migrate_pages_to(struct list_head *pagelist,
         LIST_HEAD(moved);
         LIST_HEAD(failed);
         int err = 0;
+        unsigned long offset = 0;
         int nr_pages;
         struct page *page;
         struct list_head *p;
@@ -576,8 +577,21 @@ static int migrate_pages_to(struct list_head *pagelist,
 redo:
         nr_pages = 0;
         list_for_each(p, pagelist) {
-                if (vma)
-                        page = alloc_page_vma(GFP_HIGHUSER, vma, vma->vm_start);
+                if (vma) {
+                        /*
+                         * The address passed to alloc_page_vma is used to
+                         * generate the proper interleave behavior. We fake
+                         * the address here by an increasing offset in order
+                         * to get the proper distribution of pages.
+                         *
+                         * No decision has been made as to which page
+                         * a certain old page is moved to so we cannot
+                         * specify the correct address.
+                         */
+                        page = alloc_page_vma(GFP_HIGHUSER, vma,
+                                        offset + vma->vm_start);
+                        offset += PAGE_SIZE;
+                }
                 else
                         page = alloc_pages_node(dest, GFP_HIGHUSER, 0);
 
@@ -585,9 +599,9 @@ redo:
                         err = -ENOMEM;
                         goto out;
                 }
-                list_add(&page->lru, &newlist);
+                list_add_tail(&page->lru, &newlist);
                 nr_pages++;
-                if (nr_pages > MIGRATE_CHUNK_SIZE);
+                if (nr_pages > MIGRATE_CHUNK_SIZE)
                         break;
         }
         err = migrate_pages(pagelist, &newlist, &moved, &failed);
@@ -808,7 +822,7 @@ static int get_nodes(nodemask_t *nodes, const unsigned long __user *nmask,
         nodes_clear(*nodes);
         if (maxnode == 0 || !nmask)
                 return 0;
-        if (maxnode > PAGE_SIZE)
+        if (maxnode > PAGE_SIZE*BITS_PER_BYTE)
                 return -EINVAL;
 
         nlongs = BITS_TO_LONGS(maxnode);
diff --git a/mm/nommu.c b/mm/nommu.c
index c10262d68232..99d21020ec9d 100644
--- a/mm/nommu.c
+++ b/mm/nommu.c
@@ -57,6 +57,8 @@ EXPORT_SYMBOL(vmalloc);
 EXPORT_SYMBOL(vfree);
 EXPORT_SYMBOL(vmalloc_to_page);
 EXPORT_SYMBOL(vmalloc_32);
+EXPORT_SYMBOL(vmap);
+EXPORT_SYMBOL(vunmap);
 
 /*
  * Handle all mappings that got truncated by a "truncate()"
diff --git a/mm/oom_kill.c b/mm/oom_kill.c
index b05ab8f2a562..8123fad5a485 100644
--- a/mm/oom_kill.c
+++ b/mm/oom_kill.c
@@ -58,15 +58,17 @@ unsigned long badness(struct task_struct *p, unsigned long uptime)
 
         /*
          * Processes which fork a lot of child processes are likely
-         * a good choice. We add the vmsize of the children if they
+         * a good choice. We add half the vmsize of the children if they
          * have an own mm. This prevents forking servers to flood the
-         * machine with an endless amount of children
+         * machine with an endless amount of children. In case a single
+         * child is eating the vast majority of memory, adding only half
+         * to the parents will make the child our kill candidate of choice.
          */
         list_for_each(tsk, &p->children) {
                 struct task_struct *chld;
                 chld = list_entry(tsk, struct task_struct, sibling);
                 if (chld->mm != p->mm && chld->mm)
-                        points += chld->mm->total_vm;
+                        points += chld->mm->total_vm/2 + 1;
         }
 
         /*
@@ -131,17 +133,47 @@ unsigned long badness(struct task_struct *p, unsigned long uptime)
 }
 
 /*
+ * Types of limitations to the nodes from which allocations may occur
+ */
+#define CONSTRAINT_NONE 1
+#define CONSTRAINT_MEMORY_POLICY 2
+#define CONSTRAINT_CPUSET 3
+
+/*
+ * Determine the type of allocation constraint.
+ */
+static inline int constrained_alloc(struct zonelist *zonelist, gfp_t gfp_mask)
+{
+#ifdef CONFIG_NUMA
+        struct zone **z;
+        nodemask_t nodes = node_online_map;
+
+        for (z = zonelist->zones; *z; z++)
+                if (cpuset_zone_allowed(*z, gfp_mask))
+                        node_clear((*z)->zone_pgdat->node_id,
+                                        nodes);
+                else
+                        return CONSTRAINT_CPUSET;
+
+        if (!nodes_empty(nodes))
+                return CONSTRAINT_MEMORY_POLICY;
+#endif
+
+        return CONSTRAINT_NONE;
+}
+
+/*
  * Simple selection loop. We chose the process with the highest
  * number of 'points'. We expect the caller will lock the tasklist.
  *
  * (not docbooked, we don't want this one cluttering up the manual)
  */
-static struct task_struct * select_bad_process(void)
+static struct task_struct *select_bad_process(unsigned long *ppoints)
 {
-        unsigned long maxpoints = 0;
         struct task_struct *g, *p;
         struct task_struct *chosen = NULL;
         struct timespec uptime;
+        *ppoints = 0;
 
         do_posix_clock_monotonic_gettime(&uptime);
         do_each_thread(g, p) {
@@ -169,9 +201,9 @@ static struct task_struct * select_bad_process(void)
                         return p;
 
                 points = badness(p, uptime.tv_sec);
-                if (points > maxpoints || !chosen) {
+                if (points > *ppoints || !chosen) {
                         chosen = p;
-                        maxpoints = points;
+                        *ppoints = points;
                 }
         } while_each_thread(g, p);
         return chosen;
@@ -182,7 +214,7 @@ static struct task_struct * select_bad_process(void)
  * CAP_SYS_RAW_IO set, send SIGTERM instead (but it's unlikely that
  * we select a process with CAP_SYS_RAW_IO set).
  */
-static void __oom_kill_task(task_t *p)
+static void __oom_kill_task(task_t *p, const char *message)
 {
         if (p->pid == 1) {
                 WARN_ON(1);
@@ -198,8 +230,8 @@ static void __oom_kill_task(task_t *p)
                 return;
         }
         task_unlock(p);
-        printk(KERN_ERR "Out of Memory: Killed process %d (%s).\n",
-                                                        p->pid, p->comm);
+        printk(KERN_ERR "%s: Killed process %d (%s).\n",
+                                message, p->pid, p->comm);
 
         /*
          * We give our sacrificial lamb high priority and access to
@@ -212,7 +244,7 @@ static void __oom_kill_task(task_t *p)
         force_sig(SIGKILL, p);
 }
 
-static struct mm_struct *oom_kill_task(task_t *p)
+static struct mm_struct *oom_kill_task(task_t *p, const char *message)
 {
         struct mm_struct *mm = get_task_mm(p);
         task_t * g, * q;
@@ -224,35 +256,38 @@ static struct mm_struct *oom_kill_task(task_t *p)
                 return NULL;
         }
 
-        __oom_kill_task(p);
+        __oom_kill_task(p, message);
         /*
          * kill all processes that share the ->mm (i.e. all threads),
          * but are in a different thread group
          */
         do_each_thread(g, q)
                 if (q->mm == mm && q->tgid != p->tgid)
-                        __oom_kill_task(q);
+                        __oom_kill_task(q, message);
         while_each_thread(g, q);
 
         return mm;
 }
 
-static struct mm_struct *oom_kill_process(struct task_struct *p)
+static struct mm_struct *oom_kill_process(struct task_struct *p,
+                                unsigned long points, const char *message)
 {
         struct mm_struct *mm;
         struct task_struct *c;
         struct list_head *tsk;
 
+        printk(KERN_ERR "Out of Memory: Kill process %d (%s) score %li and "
+                "children.\n", p->pid, p->comm, points);
         /* Try to kill a child first */
         list_for_each(tsk, &p->children) {
                 c = list_entry(tsk, struct task_struct, sibling);
                 if (c->mm == p->mm)
                         continue;
-                mm = oom_kill_task(c);
+                mm = oom_kill_task(c, message);
                 if (mm)
                         return mm;
         }
-        return oom_kill_task(p);
+        return oom_kill_task(p, message);
 }
 
 /**
@@ -263,10 +298,11 @@ static struct mm_struct *oom_kill_process(struct task_struct *p)
  * OR try to be smart about which process to kill. Note that we
  * don't have to be perfect here, we just have to be good.
  */
-void out_of_memory(gfp_t gfp_mask, int order)
+void out_of_memory(struct zonelist *zonelist, gfp_t gfp_mask, int order)
 {
         struct mm_struct *mm = NULL;
-        task_t * p;
+        task_t *p;
+        unsigned long points;
 
         if (printk_ratelimit()) {
                 printk("oom-killer: gfp_mask=0x%x, order=%d\n",
@@ -277,25 +313,48 @@ void out_of_memory(gfp_t gfp_mask, int order)
 
         cpuset_lock();
         read_lock(&tasklist_lock);
+
+        /*
+         * Check if there were limitations on the allocation (only relevant for
+         * NUMA) that may require different handling.
+         */
+        switch (constrained_alloc(zonelist, gfp_mask)) {
+        case CONSTRAINT_MEMORY_POLICY:
+                mm = oom_kill_process(current, points,
+                                "No available memory (MPOL_BIND)");
+                break;
+
+        case CONSTRAINT_CPUSET:
+                mm = oom_kill_process(current, points,
+                                "No available memory in cpuset");
+                break;
+
+        case CONSTRAINT_NONE:
 retry:
-        p = select_bad_process();
+                /*
+                 * Rambo mode: Shoot down a process and hope it solves whatever
+                 * issues we may have.
+                 */
+                p = select_bad_process(&points);
 
-        if (PTR_ERR(p) == -1UL)
-                goto out;
+                if (PTR_ERR(p) == -1UL)
+                        goto out;
 
-        /* Found nothing?!?! Either we hang forever, or we panic. */
-        if (!p) {
-                read_unlock(&tasklist_lock);
-                cpuset_unlock();
-                panic("Out of memory and no killable processes...\n");
-        }
+                /* Found nothing?!?! Either we hang forever, or we panic. */
+                if (!p) {
+                        read_unlock(&tasklist_lock);
+                        cpuset_unlock();
+                        panic("Out of memory and no killable processes...\n");
+                }
 
-        mm = oom_kill_process(p);
-        if (!mm)
-                goto retry;
+                mm = oom_kill_process(p, points, "Out of memory");
+                if (!mm)
+                        goto retry;
+
+                break;
+        }
 
- out:
-        read_unlock(&tasklist_lock);
+out:
         cpuset_unlock();
         if (mm)
                 mmput(mm);
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 208812b25597..791690d7d3fa 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -1015,7 +1015,7 @@ rebalance:
                 if (page)
                         goto got_pg;
 
-                out_of_memory(gfp_mask, order);
+                out_of_memory(zonelist, gfp_mask, order);
                 goto restart;
         }
 
diff --git a/mm/shmem.c b/mm/shmem.c
index f7ac7b812f92..7c455fbaff7b 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -45,6 +45,7 @@
 #include <linux/swapops.h>
 #include <linux/mempolicy.h>
 #include <linux/namei.h>
+#include <linux/ctype.h>
 #include <asm/uaccess.h>
 #include <asm/div64.h>
 #include <asm/pgtable.h>
@@ -874,6 +875,51 @@ redirty:
 }
 
 #ifdef CONFIG_NUMA
+static int shmem_parse_mpol(char *value, int *policy, nodemask_t *policy_nodes)
+{
+        char *nodelist = strchr(value, ':');
+        int err = 1;
+
+        if (nodelist) {
+                /* NUL-terminate policy string */
+                *nodelist++ = '\0';
+                if (nodelist_parse(nodelist, *policy_nodes))
+                        goto out;
+        }
+        if (!strcmp(value, "default")) {
+                *policy = MPOL_DEFAULT;
+                /* Don't allow a nodelist */
+                if (!nodelist)
+                        err = 0;
+        } else if (!strcmp(value, "prefer")) {
+                *policy = MPOL_PREFERRED;
+                /* Insist on a nodelist of one node only */
+                if (nodelist) {
+                        char *rest = nodelist;
+                        while (isdigit(*rest))
+                                rest++;
+                        if (!*rest)
+                                err = 0;
+                }
+        } else if (!strcmp(value, "bind")) {
+                *policy = MPOL_BIND;
+                /* Insist on a nodelist */
+                if (nodelist)
+                        err = 0;
+        } else if (!strcmp(value, "interleave")) {
+                *policy = MPOL_INTERLEAVE;
+                /* Default to nodes online if no nodelist */
+                if (!nodelist)
+                        *policy_nodes = node_online_map;
+                err = 0;
+        }
+out:
+        /* Restore string for error message */
+        if (nodelist)
+                *--nodelist = ':';
+        return err;
+}
+
 static struct page *shmem_swapin_async(struct shared_policy *p,
                                        swp_entry_t entry, unsigned long idx)
 {
@@ -926,6 +972,11 @@ shmem_alloc_page(gfp_t gfp, struct shmem_inode_info *info,
         return page;
 }
 #else
+static inline int shmem_parse_mpol(char *value, int *policy, nodemask_t *policy_nodes)
+{
+        return 1;
+}
+
 static inline struct page *
 shmem_swapin(struct shmem_inode_info *info,swp_entry_t entry,unsigned long idx)
 {
@@ -1859,7 +1910,23 @@ static int shmem_parse_options(char *options, int *mode, uid_t *uid,
 {
         char *this_char, *value, *rest;
 
-        while ((this_char = strsep(&options, ",")) != NULL) {
+        while (options != NULL) {
+                this_char = options;
+                for (;;) {
+                        /*
+                         * NUL-terminate this option: unfortunately,
+                         * mount options form a comma-separated list,
+                         * but mpol's nodelist may also contain commas.
+                         */
+                        options = strchr(options, ',');
+                        if (options == NULL)
+                                break;
+                        options++;
+                        if (!isdigit(*options)) {
+                                options[-1] = '\0';
+                                break;
+                        }
+                }
                 if (!*this_char)
                         continue;
                 if ((value = strchr(this_char,'=')) != NULL) {
@@ -1910,18 +1977,8 @@ static int shmem_parse_options(char *options, int *mode, uid_t *uid,
                         if (*rest)
                                 goto bad_val;
                 } else if (!strcmp(this_char,"mpol")) {
-                        if (!strcmp(value,"default"))
-                                *policy = MPOL_DEFAULT;
-                        else if (!strcmp(value,"preferred"))
-                                *policy = MPOL_PREFERRED;
-                        else if (!strcmp(value,"bind"))
-                                *policy = MPOL_BIND;
-                        else if (!strcmp(value,"interleave"))
-                                *policy = MPOL_INTERLEAVE;
-                        else
+                        if (shmem_parse_mpol(value,policy,policy_nodes))
                                 goto bad_val;
-                } else if (!strcmp(this_char,"mpol_nodelist")) {
-                        nodelist_parse(value, *policy_nodes);
                 } else {
                         printk(KERN_ERR "tmpfs: Bad mount option %s\n",
                                this_char);
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 1838c15ca4fd..b0af7593d01e 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -1908,7 +1908,12 @@ int zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
                 sc.swap_cluster_max = SWAP_CLUSTER_MAX;
 
         cond_resched();
-        p->flags |= PF_MEMALLOC;
+        /*
+         * We need to be able to allocate from the reserves for RECLAIM_SWAP
+         * and we also need to be able to write out pages for RECLAIM_WRITE
+         * and RECLAIM_SWAP.
+         */
+        p->flags |= PF_MEMALLOC | PF_SWAPWRITE;
         reclaim_state.reclaimed_slab = 0;
         p->reclaim_state = &reclaim_state;
 
@@ -1932,11 +1937,10 @@ int zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
                  * a long time.
                  */
                 shrink_slab(sc.nr_scanned, gfp_mask, order);
-                sc.nr_reclaimed = 1;    /* Avoid getting the off node timeout */
         }
 
         p->reclaim_state = NULL;
-        current->flags &= ~PF_MEMALLOC;
+        current->flags &= ~(PF_MEMALLOC | PF_SWAPWRITE);
 
         if (sc.nr_reclaimed == 0)
                 zone->last_unsuccessful_zone_reclaim = jiffies;