2 * Simple NUMA memory policy for the Linux kernel.
4 * Copyright 2003,2004 Andi Kleen, SuSE Labs.
5 * (C) Copyright 2005 Christoph Lameter, Silicon Graphics, Inc.
6 * Subject to the GNU Public License, version 2.
8 * NUMA policy allows the user to give hints in which node(s) memory should
11 * Support four policies per VMA and per process:
13 * The VMA policy has priority over the process policy for a page fault.
15 * interleave Allocate memory interleaved over a set of nodes,
16 * with normal fallback if it fails.
17 * For VMA based allocations this interleaves based on the
18 * offset into the backing object or offset into the mapping
19 * for anonymous memory. For process policy an process counter
22 * bind Only allocate memory on a specific set of nodes,
24 * FIXME: memory is allocated starting with the first node
25 * to the last. It would be better if bind would truly restrict
26 * the allocation to memory nodes instead
28 * preferred Try a specific node first before normal fallback.
29 * As a special case node -1 here means do the allocation
30 * on the local CPU. This is normally identical to default,
31 * but useful to set in a VMA when you have a non default
34 * default Allocate on the local node first, or when on a VMA
35 * use the process policy. This is what Linux always did
36 * in a NUMA aware kernel and still does by, ahem, default.
38 * The process policy is applied for most non interrupt memory allocations
39 * in that process' context. Interrupts ignore the policies and always
40 * try to allocate on the local CPU. The VMA policy is only applied for memory
41 * allocations for a VMA in the VM.
43 * Currently there are a few corner cases in swapping where the policy
44 * is not applied, but the majority should be handled. When process policy
45 * is used it is not remembered over swap outs/swap ins.
47 * Only the highest zone in the zone hierarchy gets policied. Allocations
48 * requesting a lower zone just use default policy. This implies that
49 * on systems with highmem kernel lowmem allocation don't get policied.
50 * Same with GFP_DMA allocations.
52 * For shmfs/tmpfs/hugetlbfs shared memory the policy is shared between
53 * all users and remembered even when nobody has memory mapped.
57 fix mmap readahead to honour policy and enable policy for any page cache
59 statistics for bigpages
60 global policy for page cache? currently it uses process policy. Requires
62 handle mremap for shared memory (currently ignored for the policy)
64 make bind policy root only? It can trigger oom much faster and the
65 kernel is not always grateful with that.
66 could replace all the switch()es with a mempolicy_ops structure.
69 #include <linux/mempolicy.h>
71 #include <linux/highmem.h>
72 #include <linux/hugetlb.h>
73 #include <linux/kernel.h>
74 #include <linux/sched.h>
76 #include <linux/nodemask.h>
77 #include <linux/cpuset.h>
78 #include <linux/gfp.h>
79 #include <linux/slab.h>
80 #include <linux/string.h>
81 #include <linux/module.h>
82 #include <linux/interrupt.h>
83 #include <linux/init.h>
84 #include <linux/compat.h>
85 #include <linux/mempolicy.h>
86 #include <linux/swap.h>
87 #include <linux/seq_file.h>
88 #include <linux/proc_fs.h>
89 #include <linux/vs_cvirt.h>
91 #include <asm/tlbflush.h>
92 #include <asm/uaccess.h>
95 #define MPOL_MF_DISCONTIG_OK (MPOL_MF_INTERNAL << 0) /* Skip checks for continuous vmas */
96 #define MPOL_MF_INVERT (MPOL_MF_INTERNAL << 1) /* Invert check for nodemask */
97 #define MPOL_MF_STATS (MPOL_MF_INTERNAL << 2) /* Gather statistics */
99 /* The number of pages to migrate per call to migrate_pages() */
100 #define MIGRATE_CHUNK_SIZE 256
102 static kmem_cache_t *policy_cache;
103 static kmem_cache_t *sn_cache;
105 #define PDprintk(fmt...)
107 /* Highest zone. An specific allocation for a zone below that is not
109 int policy_zone = ZONE_DMA;
111 struct mempolicy default_policy = {
112 .refcnt = ATOMIC_INIT(1), /* never free it */
113 .policy = MPOL_DEFAULT,
116 /* Do sanity checking on a policy */
117 static int mpol_check_policy(int mode, nodemask_t *nodes)
119 int empty = nodes_empty(*nodes);
127 case MPOL_INTERLEAVE:
128 /* Preferred will only use the first bit, but allow
134 return nodes_subset(*nodes, node_online_map) ? 0 : -EINVAL;
137 /* Generate a custom zonelist for the BIND policy. */
138 static struct zonelist *bind_zonelist(nodemask_t *nodes)
143 max = 1 + MAX_NR_ZONES * nodes_weight(*nodes);
144 zl = kmalloc(sizeof(struct zone *) * max, GFP_KERNEL);
148 /* First put in the highest zones from all nodes, then all the next
149 lower zones etc. Avoid empty zones because the memory allocator
150 doesn't like them. If you implement node hot removal you
152 for (k = policy_zone; k >= 0; k--) {
153 for_each_node_mask(nd, *nodes) {
154 struct zone *z = &NODE_DATA(nd)->node_zones[k];
155 if (z->present_pages > 0)
156 zl->zones[num++] = z;
159 zl->zones[num] = NULL;
163 /* Create a new policy */
164 static struct mempolicy *mpol_new(int mode, nodemask_t *nodes)
166 struct mempolicy *policy;
168 PDprintk("setting mode %d nodes[0] %lx\n", mode, nodes_addr(*nodes)[0]);
169 if (mode == MPOL_DEFAULT)
171 policy = kmem_cache_alloc(policy_cache, GFP_KERNEL);
173 return ERR_PTR(-ENOMEM);
174 atomic_set(&policy->refcnt, 1);
176 case MPOL_INTERLEAVE:
177 policy->v.nodes = *nodes;
178 if (nodes_weight(*nodes) == 0) {
179 kmem_cache_free(policy_cache, policy);
180 return ERR_PTR(-EINVAL);
184 policy->v.preferred_node = first_node(*nodes);
185 if (policy->v.preferred_node >= MAX_NUMNODES)
186 policy->v.preferred_node = -1;
189 policy->v.zonelist = bind_zonelist(nodes);
190 if (policy->v.zonelist == NULL) {
191 kmem_cache_free(policy_cache, policy);
192 return ERR_PTR(-ENOMEM);
196 policy->policy = mode;
197 policy->cpuset_mems_allowed = cpuset_mems_allowed(current);
201 static void gather_stats(struct page *, void *, int pte_dirty);
202 static void migrate_page_add(struct page *page, struct list_head *pagelist,
203 unsigned long flags);
205 /* Scan through pages checking if pages follow certain conditions. */
206 static int check_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
207 unsigned long addr, unsigned long end,
208 const nodemask_t *nodes, unsigned long flags,
215 orig_pte = pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
220 if (!pte_present(*pte))
222 page = vm_normal_page(vma, addr, *pte);
226 * The check for PageReserved here is important to avoid
227 * handling zero pages and other pages that may have been
228 * marked special by the system.
230 * If the PageReserved would not be checked here then f.e.
231 * the location of the zero page could have an influence
232 * on MPOL_MF_STRICT, zero pages would be counted for
233 * the per node stats, and there would be useless attempts
234 * to put zero pages on the migration list.
236 if (PageReserved(page))
238 nid = page_to_nid(page);
239 if (node_isset(nid, *nodes) == !!(flags & MPOL_MF_INVERT))
242 if (flags & MPOL_MF_STATS)
243 gather_stats(page, private, pte_dirty(*pte));
244 else if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL))
245 migrate_page_add(page, private, flags);
248 } while (pte++, addr += PAGE_SIZE, addr != end);
249 pte_unmap_unlock(orig_pte, ptl);
253 static inline int check_pmd_range(struct vm_area_struct *vma, pud_t *pud,
254 unsigned long addr, unsigned long end,
255 const nodemask_t *nodes, unsigned long flags,
261 pmd = pmd_offset(pud, addr);
263 next = pmd_addr_end(addr, end);
264 if (pmd_none_or_clear_bad(pmd))
266 if (check_pte_range(vma, pmd, addr, next, nodes,
269 } while (pmd++, addr = next, addr != end);
273 static inline int check_pud_range(struct vm_area_struct *vma, pgd_t *pgd,
274 unsigned long addr, unsigned long end,
275 const nodemask_t *nodes, unsigned long flags,
281 pud = pud_offset(pgd, addr);
283 next = pud_addr_end(addr, end);
284 if (pud_none_or_clear_bad(pud))
286 if (check_pmd_range(vma, pud, addr, next, nodes,
289 } while (pud++, addr = next, addr != end);
293 static inline int check_pgd_range(struct vm_area_struct *vma,
294 unsigned long addr, unsigned long end,
295 const nodemask_t *nodes, unsigned long flags,
301 pgd = pgd_offset(vma->vm_mm, addr);
303 next = pgd_addr_end(addr, end);
304 if (pgd_none_or_clear_bad(pgd))
306 if (check_pud_range(vma, pgd, addr, next, nodes,
309 } while (pgd++, addr = next, addr != end);
313 /* Check if a vma is migratable */
314 static inline int vma_migratable(struct vm_area_struct *vma)
316 if (vma->vm_flags & (
317 VM_LOCKED|VM_IO|VM_HUGETLB|VM_PFNMAP|VM_RESERVED))
323 * Check if all pages in a range are on a set of nodes.
324 * If pagelist != NULL then isolate pages from the LRU and
325 * put them on the pagelist.
327 static struct vm_area_struct *
328 check_range(struct mm_struct *mm, unsigned long start, unsigned long end,
329 const nodemask_t *nodes, unsigned long flags, void *private)
332 struct vm_area_struct *first, *vma, *prev;
334 if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)) {
335 /* Must have swap device for migration */
336 if (nr_swap_pages <= 0)
337 return ERR_PTR(-ENODEV);
340 * Clear the LRU lists so pages can be isolated.
341 * Note that pages may be moved off the LRU after we have
342 * drained them. Those pages will fail to migrate like other
343 * pages that may be busy.
348 first = find_vma(mm, start);
350 return ERR_PTR(-EFAULT);
352 for (vma = first; vma && vma->vm_start < end; vma = vma->vm_next) {
353 if (!(flags & MPOL_MF_DISCONTIG_OK)) {
354 if (!vma->vm_next && vma->vm_end < end)
355 return ERR_PTR(-EFAULT);
356 if (prev && prev->vm_end < vma->vm_start)
357 return ERR_PTR(-EFAULT);
359 if (!is_vm_hugetlb_page(vma) &&
360 ((flags & MPOL_MF_STRICT) ||
361 ((flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)) &&
362 vma_migratable(vma)))) {
363 unsigned long endvma = vma->vm_end;
367 if (vma->vm_start > start)
368 start = vma->vm_start;
369 err = check_pgd_range(vma, start, endvma, nodes,
372 first = ERR_PTR(err);
381 /* Apply policy to a single VMA */
382 static int policy_vma(struct vm_area_struct *vma, struct mempolicy *new)
385 struct mempolicy *old = vma->vm_policy;
387 PDprintk("vma %lx-%lx/%lx vm_ops %p vm_file %p set_policy %p\n",
388 vma->vm_start, vma->vm_end, vma->vm_pgoff,
389 vma->vm_ops, vma->vm_file,
390 vma->vm_ops ? vma->vm_ops->set_policy : NULL);
392 if (vma->vm_ops && vma->vm_ops->set_policy)
393 err = vma->vm_ops->set_policy(vma, new);
396 vma->vm_policy = new;
402 /* Step 2: apply policy to a range and do splits. */
403 static int mbind_range(struct vm_area_struct *vma, unsigned long start,
404 unsigned long end, struct mempolicy *new)
406 struct vm_area_struct *next;
410 for (; vma && vma->vm_start < end; vma = next) {
412 if (vma->vm_start < start)
413 err = split_vma(vma->vm_mm, vma, start, 1);
414 if (!err && vma->vm_end > end)
415 err = split_vma(vma->vm_mm, vma, end, 0);
417 err = policy_vma(vma, new);
424 static int contextualize_policy(int mode, nodemask_t *nodes)
429 cpuset_update_task_memory_state();
430 if (!cpuset_nodes_subset_current_mems_allowed(*nodes))
432 return mpol_check_policy(mode, nodes);
435 /* Set the process memory policy */
436 long do_set_mempolicy(int mode, nodemask_t *nodes)
438 struct mempolicy *new;
440 if (contextualize_policy(mode, nodes))
442 new = mpol_new(mode, nodes);
445 mpol_free(current->mempolicy);
446 current->mempolicy = new;
447 if (new && new->policy == MPOL_INTERLEAVE)
448 current->il_next = first_node(new->v.nodes);
452 /* Fill a zone bitmap for a policy */
453 static void get_zonemask(struct mempolicy *p, nodemask_t *nodes)
460 for (i = 0; p->v.zonelist->zones[i]; i++)
461 node_set(p->v.zonelist->zones[i]->zone_pgdat->node_id,
466 case MPOL_INTERLEAVE:
470 /* or use current node instead of online map? */
471 if (p->v.preferred_node < 0)
472 *nodes = node_online_map;
474 node_set(p->v.preferred_node, *nodes);
481 static int lookup_node(struct mm_struct *mm, unsigned long addr)
486 err = get_user_pages(current, mm, addr & PAGE_MASK, 1, 0, 0, &p, NULL);
488 err = page_to_nid(p);
494 /* Retrieve NUMA policy */
495 long do_get_mempolicy(int *policy, nodemask_t *nmask,
496 unsigned long addr, unsigned long flags)
499 struct mm_struct *mm = current->mm;
500 struct vm_area_struct *vma = NULL;
501 struct mempolicy *pol = current->mempolicy;
503 cpuset_update_task_memory_state();
504 if (flags & ~(unsigned long)(MPOL_F_NODE|MPOL_F_ADDR))
506 if (flags & MPOL_F_ADDR) {
507 down_read(&mm->mmap_sem);
508 vma = find_vma_intersection(mm, addr, addr+1);
510 up_read(&mm->mmap_sem);
513 if (vma->vm_ops && vma->vm_ops->get_policy)
514 pol = vma->vm_ops->get_policy(vma, addr);
516 pol = vma->vm_policy;
521 pol = &default_policy;
523 if (flags & MPOL_F_NODE) {
524 if (flags & MPOL_F_ADDR) {
525 err = lookup_node(mm, addr);
529 } else if (pol == current->mempolicy &&
530 pol->policy == MPOL_INTERLEAVE) {
531 *policy = current->il_next;
537 *policy = pol->policy;
540 up_read(¤t->mm->mmap_sem);
546 get_zonemask(pol, nmask);
550 up_read(¤t->mm->mmap_sem);
558 static void migrate_page_add(struct page *page, struct list_head *pagelist,
562 * Avoid migrating a page that is shared with others.
564 if ((flags & MPOL_MF_MOVE_ALL) || page_mapcount(page) == 1) {
565 if (isolate_lru_page(page))
566 list_add_tail(&page->lru, pagelist);
571 * Migrate the list 'pagelist' of pages to a certain destination.
573 * Specify destination with either non-NULL vma or dest_node >= 0
574 * Return the number of pages not migrated or error code
576 static int migrate_pages_to(struct list_head *pagelist,
577 struct vm_area_struct *vma, int dest)
583 unsigned long offset = 0;
590 list_for_each(p, pagelist) {
593 * The address passed to alloc_page_vma is used to
594 * generate the proper interleave behavior. We fake
595 * the address here by an increasing offset in order
596 * to get the proper distribution of pages.
598 * No decision has been made as to which page
599 * a certain old page is moved to so we cannot
600 * specify the correct address.
602 page = alloc_page_vma(GFP_HIGHUSER, vma,
603 offset + vma->vm_start);
607 page = alloc_pages_node(dest, GFP_HIGHUSER, 0);
613 list_add_tail(&page->lru, &newlist);
615 if (nr_pages > MIGRATE_CHUNK_SIZE)
618 err = migrate_pages(pagelist, &newlist, &moved, &failed);
620 putback_lru_pages(&moved); /* Call release pages instead ?? */
622 if (err >= 0 && list_empty(&newlist) && !list_empty(pagelist))
625 /* Return leftover allocated pages */
626 while (!list_empty(&newlist)) {
627 page = list_entry(newlist.next, struct page, lru);
628 list_del(&page->lru);
631 list_splice(&failed, pagelist);
635 /* Calculate number of leftover pages */
637 list_for_each(p, pagelist)
643 * Migrate pages from one node to a target node.
644 * Returns error or the number of pages not migrated.
646 int migrate_to_node(struct mm_struct *mm, int source, int dest, int flags)
653 node_set(source, nmask);
655 check_range(mm, mm->mmap->vm_start, TASK_SIZE, &nmask,
656 flags | MPOL_MF_DISCONTIG_OK, &pagelist);
658 if (!list_empty(&pagelist)) {
659 err = migrate_pages_to(&pagelist, NULL, dest);
660 if (!list_empty(&pagelist))
661 putback_lru_pages(&pagelist);
667 * Move pages between the two nodesets so as to preserve the physical
668 * layout as much as possible.
670 * Returns the number of page that could not be moved.
672 int do_migrate_pages(struct mm_struct *mm,
673 const nodemask_t *from_nodes, const nodemask_t *to_nodes, int flags)
680 down_read(&mm->mmap_sem);
683 * Find a 'source' bit set in 'tmp' whose corresponding 'dest'
684 * bit in 'to' is not also set in 'tmp'. Clear the found 'source'
685 * bit in 'tmp', and return that <source, dest> pair for migration.
686 * The pair of nodemasks 'to' and 'from' define the map.
688 * If no pair of bits is found that way, fallback to picking some
689 * pair of 'source' and 'dest' bits that are not the same. If the
690 * 'source' and 'dest' bits are the same, this represents a node
691 * that will be migrating to itself, so no pages need move.
693 * If no bits are left in 'tmp', or if all remaining bits left
694 * in 'tmp' correspond to the same bit in 'to', return false
695 * (nothing left to migrate).
697 * This lets us pick a pair of nodes to migrate between, such that
698 * if possible the dest node is not already occupied by some other
699 * source node, minimizing the risk of overloading the memory on a
700 * node that would happen if we migrated incoming memory to a node
701 * before migrating outgoing memory source that same node.
703 * A single scan of tmp is sufficient. As we go, we remember the
704 * most recent <s, d> pair that moved (s != d). If we find a pair
705 * that not only moved, but what's better, moved to an empty slot
706 * (d is not set in tmp), then we break out then, with that pair.
707 * Otherwise when we finish scannng from_tmp, we at least have the
708 * most recent <s, d> pair that moved. If we get all the way through
709 * the scan of tmp without finding any node that moved, much less
710 * moved to an empty node, then there is nothing left worth migrating.
714 while (!nodes_empty(tmp)) {
719 for_each_node_mask(s, tmp) {
720 d = node_remap(s, *from_nodes, *to_nodes);
724 source = s; /* Node moved. Memorize */
727 /* dest not in remaining from nodes? */
728 if (!node_isset(dest, tmp))
734 node_clear(source, tmp);
735 err = migrate_to_node(mm, source, dest, flags);
742 up_read(&mm->mmap_sem);
748 long do_mbind(unsigned long start, unsigned long len,
749 unsigned long mode, nodemask_t *nmask, unsigned long flags)
751 struct vm_area_struct *vma;
752 struct mm_struct *mm = current->mm;
753 struct mempolicy *new;
758 if ((flags & ~(unsigned long)(MPOL_MF_STRICT |
759 MPOL_MF_MOVE | MPOL_MF_MOVE_ALL))
762 if ((flags & MPOL_MF_MOVE_ALL) && !capable(CAP_SYS_NICE))
765 if (start & ~PAGE_MASK)
768 if (mode == MPOL_DEFAULT)
769 flags &= ~MPOL_MF_STRICT;
771 len = (len + PAGE_SIZE - 1) & PAGE_MASK;
779 if (mpol_check_policy(mode, nmask))
782 new = mpol_new(mode, nmask);
787 * If we are using the default policy then operation
788 * on discontinuous address spaces is okay after all
791 flags |= MPOL_MF_DISCONTIG_OK;
793 PDprintk("mbind %lx-%lx mode:%ld nodes:%lx\n",start,start+len,
794 mode,nodes_addr(nodes)[0]);
796 down_write(&mm->mmap_sem);
797 vma = check_range(mm, start, end, nmask,
798 flags | MPOL_MF_INVERT, &pagelist);
804 err = mbind_range(vma, start, end, new);
806 if (!list_empty(&pagelist))
807 nr_failed = migrate_pages_to(&pagelist, vma, -1);
809 if (!err && nr_failed && (flags & MPOL_MF_STRICT))
812 if (!list_empty(&pagelist))
813 putback_lru_pages(&pagelist);
815 up_write(&mm->mmap_sem);
821 * User space interface with variable sized bitmaps for nodelists.
824 /* Copy a node mask from user space. */
825 static int get_nodes(nodemask_t *nodes, const unsigned long __user *nmask,
826 unsigned long maxnode)
829 unsigned long nlongs;
830 unsigned long endmask;
834 if (maxnode == 0 || !nmask)
836 if (maxnode > PAGE_SIZE*BITS_PER_BYTE)
839 nlongs = BITS_TO_LONGS(maxnode);
840 if ((maxnode % BITS_PER_LONG) == 0)
843 endmask = (1UL << (maxnode % BITS_PER_LONG)) - 1;
845 /* When the user specified more nodes than supported just check
846 if the non supported part is all zero. */
847 if (nlongs > BITS_TO_LONGS(MAX_NUMNODES)) {
848 if (nlongs > PAGE_SIZE/sizeof(long))
850 for (k = BITS_TO_LONGS(MAX_NUMNODES); k < nlongs; k++) {
852 if (get_user(t, nmask + k))
854 if (k == nlongs - 1) {
860 nlongs = BITS_TO_LONGS(MAX_NUMNODES);
864 if (copy_from_user(nodes_addr(*nodes), nmask, nlongs*sizeof(unsigned long)))
866 nodes_addr(*nodes)[nlongs-1] &= endmask;
870 /* Copy a kernel node mask to user space */
871 static int copy_nodes_to_user(unsigned long __user *mask, unsigned long maxnode,
874 unsigned long copy = ALIGN(maxnode-1, 64) / 8;
875 const int nbytes = BITS_TO_LONGS(MAX_NUMNODES) * sizeof(long);
878 if (copy > PAGE_SIZE)
880 if (clear_user((char __user *)mask + nbytes, copy - nbytes))
884 return copy_to_user(mask, nodes_addr(*nodes), copy) ? -EFAULT : 0;
887 asmlinkage long sys_mbind(unsigned long start, unsigned long len,
889 unsigned long __user *nmask, unsigned long maxnode,
895 err = get_nodes(&nodes, nmask, maxnode);
898 return do_mbind(start, len, mode, &nodes, flags);
901 /* Set the process memory policy */
902 asmlinkage long sys_set_mempolicy(int mode, unsigned long __user *nmask,
903 unsigned long maxnode)
908 if (mode < 0 || mode > MPOL_MAX)
910 err = get_nodes(&nodes, nmask, maxnode);
913 return do_set_mempolicy(mode, &nodes);
916 asmlinkage long sys_migrate_pages(pid_t pid, unsigned long maxnode,
917 const unsigned long __user *old_nodes,
918 const unsigned long __user *new_nodes)
920 struct mm_struct *mm;
921 struct task_struct *task;
924 nodemask_t task_nodes;
927 err = get_nodes(&old, old_nodes, maxnode);
931 err = get_nodes(&new, new_nodes, maxnode);
935 /* Find the mm_struct */
936 read_lock(&tasklist_lock);
937 task = pid ? find_task_by_pid(pid) : current;
939 read_unlock(&tasklist_lock);
942 mm = get_task_mm(task);
943 read_unlock(&tasklist_lock);
949 * Check if this process has the right to modify the specified
950 * process. The right exists if the process has administrative
951 * capabilities, superuser priviledges or the same
952 * userid as the target process.
954 if ((current->euid != task->suid) && (current->euid != task->uid) &&
955 (current->uid != task->suid) && (current->uid != task->uid) &&
956 !capable(CAP_SYS_NICE)) {
961 task_nodes = cpuset_mems_allowed(task);
962 /* Is the user allowed to access the target nodes? */
963 if (!nodes_subset(new, task_nodes) && !capable(CAP_SYS_NICE)) {
968 err = do_migrate_pages(mm, &old, &new,
969 capable(CAP_SYS_NICE) ? MPOL_MF_MOVE_ALL : MPOL_MF_MOVE);
976 /* Retrieve NUMA policy */
977 asmlinkage long sys_get_mempolicy(int __user *policy,
978 unsigned long __user *nmask,
979 unsigned long maxnode,
980 unsigned long addr, unsigned long flags)
985 if (nmask != NULL && maxnode < MAX_NUMNODES)
988 err = do_get_mempolicy(&pval, &nodes, addr, flags);
993 if (policy && put_user(pval, policy))
997 err = copy_nodes_to_user(nmask, maxnode, &nodes);
1002 #ifdef CONFIG_COMPAT
1004 asmlinkage long compat_sys_get_mempolicy(int __user *policy,
1005 compat_ulong_t __user *nmask,
1006 compat_ulong_t maxnode,
1007 compat_ulong_t addr, compat_ulong_t flags)
1010 unsigned long __user *nm = NULL;
1011 unsigned long nr_bits, alloc_size;
1012 DECLARE_BITMAP(bm, MAX_NUMNODES);
1014 nr_bits = min_t(unsigned long, maxnode-1, MAX_NUMNODES);
1015 alloc_size = ALIGN(nr_bits, BITS_PER_LONG) / 8;
1018 nm = compat_alloc_user_space(alloc_size);
1020 err = sys_get_mempolicy(policy, nm, nr_bits+1, addr, flags);
1022 if (!err && nmask) {
1023 err = copy_from_user(bm, nm, alloc_size);
1024 /* ensure entire bitmap is zeroed */
1025 err |= clear_user(nmask, ALIGN(maxnode-1, 8) / 8);
1026 err |= compat_put_bitmap(nmask, bm, nr_bits);
1032 asmlinkage long compat_sys_set_mempolicy(int mode, compat_ulong_t __user *nmask,
1033 compat_ulong_t maxnode)
1036 unsigned long __user *nm = NULL;
1037 unsigned long nr_bits, alloc_size;
1038 DECLARE_BITMAP(bm, MAX_NUMNODES);
1040 nr_bits = min_t(unsigned long, maxnode-1, MAX_NUMNODES);
1041 alloc_size = ALIGN(nr_bits, BITS_PER_LONG) / 8;
1044 err = compat_get_bitmap(bm, nmask, nr_bits);
1045 nm = compat_alloc_user_space(alloc_size);
1046 err |= copy_to_user(nm, bm, alloc_size);
1052 return sys_set_mempolicy(mode, nm, nr_bits+1);
1055 asmlinkage long compat_sys_mbind(compat_ulong_t start, compat_ulong_t len,
1056 compat_ulong_t mode, compat_ulong_t __user *nmask,
1057 compat_ulong_t maxnode, compat_ulong_t flags)
1060 unsigned long __user *nm = NULL;
1061 unsigned long nr_bits, alloc_size;
1064 nr_bits = min_t(unsigned long, maxnode-1, MAX_NUMNODES);
1065 alloc_size = ALIGN(nr_bits, BITS_PER_LONG) / 8;
1068 err = compat_get_bitmap(nodes_addr(bm), nmask, nr_bits);
1069 nm = compat_alloc_user_space(alloc_size);
1070 err |= copy_to_user(nm, nodes_addr(bm), alloc_size);
1076 return sys_mbind(start, len, mode, nm, nr_bits+1, flags);
1081 /* Return effective policy for a VMA */
1082 static struct mempolicy * get_vma_policy(struct task_struct *task,
1083 struct vm_area_struct *vma, unsigned long addr)
1085 struct mempolicy *pol = task->mempolicy;
1088 if (vma->vm_ops && vma->vm_ops->get_policy)
1089 pol = vma->vm_ops->get_policy(vma, addr);
1090 else if (vma->vm_policy &&
1091 vma->vm_policy->policy != MPOL_DEFAULT)
1092 pol = vma->vm_policy;
1095 pol = &default_policy;
1099 /* Return a zonelist representing a mempolicy */
1100 static struct zonelist *zonelist_policy(gfp_t gfp, struct mempolicy *policy)
1104 switch (policy->policy) {
1105 case MPOL_PREFERRED:
1106 nd = policy->v.preferred_node;
1108 nd = numa_node_id();
1111 /* Lower zones don't get a policy applied */
1112 /* Careful: current->mems_allowed might have moved */
1113 if (gfp_zone(gfp) >= policy_zone)
1114 if (cpuset_zonelist_valid_mems_allowed(policy->v.zonelist))
1115 return policy->v.zonelist;
1117 case MPOL_INTERLEAVE: /* should not happen */
1119 nd = numa_node_id();
1125 return NODE_DATA(nd)->node_zonelists + gfp_zone(gfp);
1128 /* Do dynamic interleaving for a process */
1129 static unsigned interleave_nodes(struct mempolicy *policy)
1132 struct task_struct *me = current;
1135 next = next_node(nid, policy->v.nodes);
1136 if (next >= MAX_NUMNODES)
1137 next = first_node(policy->v.nodes);
1143 * Depending on the memory policy provide a node from which to allocate the
1146 unsigned slab_node(struct mempolicy *policy)
1148 switch (policy->policy) {
1149 case MPOL_INTERLEAVE:
1150 return interleave_nodes(policy);
1154 * Follow bind policy behavior and start allocation at the
1157 return policy->v.zonelist->zones[0]->zone_pgdat->node_id;
1159 case MPOL_PREFERRED:
1160 if (policy->v.preferred_node >= 0)
1161 return policy->v.preferred_node;
1165 return numa_node_id();
1169 /* Do static interleaving for a VMA with known offset. */
1170 static unsigned offset_il_node(struct mempolicy *pol,
1171 struct vm_area_struct *vma, unsigned long off)
1173 unsigned nnodes = nodes_weight(pol->v.nodes);
1174 unsigned target = (unsigned)off % nnodes;
1180 nid = next_node(nid, pol->v.nodes);
1182 } while (c <= target);
1186 /* Determine a node number for interleave */
1187 static inline unsigned interleave_nid(struct mempolicy *pol,
1188 struct vm_area_struct *vma, unsigned long addr, int shift)
1193 off = vma->vm_pgoff;
1194 off += (addr - vma->vm_start) >> shift;
1195 return offset_il_node(pol, vma, off);
1197 return interleave_nodes(pol);
1200 #ifdef CONFIG_HUGETLBFS
1201 /* Return a zonelist suitable for a huge page allocation. */
1202 struct zonelist *huge_zonelist(struct vm_area_struct *vma, unsigned long addr)
1204 struct mempolicy *pol = get_vma_policy(current, vma, addr);
1206 if (pol->policy == MPOL_INTERLEAVE) {
1209 nid = interleave_nid(pol, vma, addr, HPAGE_SHIFT);
1210 return NODE_DATA(nid)->node_zonelists + gfp_zone(GFP_HIGHUSER);
1212 return zonelist_policy(GFP_HIGHUSER, pol);
1216 /* Allocate a page in interleaved policy.
1217 Own path because it needs to do special accounting. */
1218 static struct page *alloc_page_interleave(gfp_t gfp, unsigned order,
1221 struct zonelist *zl;
1224 zl = NODE_DATA(nid)->node_zonelists + gfp_zone(gfp);
1225 page = __alloc_pages(gfp, order, zl);
1226 if (page && page_zone(page) == zl->zones[0]) {
1227 zone_pcp(zl->zones[0],get_cpu())->interleave_hit++;
1234 * alloc_page_vma - Allocate a page for a VMA.
1237 * %GFP_USER user allocation.
1238 * %GFP_KERNEL kernel allocations,
1239 * %GFP_HIGHMEM highmem/user allocations,
1240 * %GFP_FS allocation should not call back into a file system.
1241 * %GFP_ATOMIC don't sleep.
1243 * @vma: Pointer to VMA or NULL if not available.
1244 * @addr: Virtual Address of the allocation. Must be inside the VMA.
1246 * This function allocates a page from the kernel page pool and applies
1247 * a NUMA policy associated with the VMA or the current process.
1248 * When VMA is not NULL caller must hold down_read on the mmap_sem of the
1249 * mm_struct of the VMA to prevent it from going away. Should be used for
1250 * all allocations for pages that will be mapped into
1251 * user space. Returns NULL when no page can be allocated.
1253 * Should be called with the mm_sem of the vma hold.
1256 alloc_page_vma(gfp_t gfp, struct vm_area_struct *vma, unsigned long addr)
1258 struct mempolicy *pol = get_vma_policy(current, vma, addr);
1260 cpuset_update_task_memory_state();
1262 if (unlikely(pol->policy == MPOL_INTERLEAVE)) {
1265 nid = interleave_nid(pol, vma, addr, PAGE_SHIFT);
1266 return alloc_page_interleave(gfp, 0, nid);
1268 return __alloc_pages(gfp, 0, zonelist_policy(gfp, pol));
1272 * alloc_pages_current - Allocate pages.
1275 * %GFP_USER user allocation,
1276 * %GFP_KERNEL kernel allocation,
1277 * %GFP_HIGHMEM highmem allocation,
1278 * %GFP_FS don't call back into a file system.
1279 * %GFP_ATOMIC don't sleep.
1280 * @order: Power of two of allocation size in pages. 0 is a single page.
1282 * Allocate a page from the kernel page pool. When not in
1283 * interrupt context and apply the current process NUMA policy.
1284 * Returns NULL when no page can be allocated.
1286 * Don't call cpuset_update_task_memory_state() unless
1287 * 1) it's ok to take cpuset_sem (can WAIT), and
1288 * 2) allocating for current task (not interrupt).
1290 struct page *alloc_pages_current(gfp_t gfp, unsigned order)
1292 struct mempolicy *pol = current->mempolicy;
1294 if ((gfp & __GFP_WAIT) && !in_interrupt())
1295 cpuset_update_task_memory_state();
1296 if (!pol || in_interrupt())
1297 pol = &default_policy;
1298 if (pol->policy == MPOL_INTERLEAVE)
1299 return alloc_page_interleave(gfp, order, interleave_nodes(pol));
1300 return __alloc_pages(gfp, order, zonelist_policy(gfp, pol));
1302 EXPORT_SYMBOL(alloc_pages_current);
1305 * If mpol_copy() sees current->cpuset == cpuset_being_rebound, then it
1306 * rebinds the mempolicy its copying by calling mpol_rebind_policy()
1307 * with the mems_allowed returned by cpuset_mems_allowed(). This
1308 * keeps mempolicies cpuset relative after its cpuset moves. See
1309 * further kernel/cpuset.c update_nodemask().
1311 void *cpuset_being_rebound;
1313 /* Slow path of a mempolicy copy */
1314 struct mempolicy *__mpol_copy(struct mempolicy *old)
1316 struct mempolicy *new = kmem_cache_alloc(policy_cache, GFP_KERNEL);
1319 return ERR_PTR(-ENOMEM);
1320 if (current_cpuset_is_being_rebound()) {
1321 nodemask_t mems = cpuset_mems_allowed(current);
1322 mpol_rebind_policy(old, &mems);
1325 atomic_set(&new->refcnt, 1);
1326 if (new->policy == MPOL_BIND) {
1327 int sz = ksize(old->v.zonelist);
1328 new->v.zonelist = kmalloc(sz, SLAB_KERNEL);
1329 if (!new->v.zonelist) {
1330 kmem_cache_free(policy_cache, new);
1331 return ERR_PTR(-ENOMEM);
1333 memcpy(new->v.zonelist, old->v.zonelist, sz);
1338 /* Slow path of a mempolicy comparison */
1339 int __mpol_equal(struct mempolicy *a, struct mempolicy *b)
1343 if (a->policy != b->policy)
1345 switch (a->policy) {
1348 case MPOL_INTERLEAVE:
1349 return nodes_equal(a->v.nodes, b->v.nodes);
1350 case MPOL_PREFERRED:
1351 return a->v.preferred_node == b->v.preferred_node;
1354 for (i = 0; a->v.zonelist->zones[i]; i++)
1355 if (a->v.zonelist->zones[i] != b->v.zonelist->zones[i])
1357 return b->v.zonelist->zones[i] == NULL;
1365 /* Slow path of a mpol destructor. */
1366 void __mpol_free(struct mempolicy *p)
1368 if (!atomic_dec_and_test(&p->refcnt))
1370 if (p->policy == MPOL_BIND)
1371 kfree(p->v.zonelist);
1372 p->policy = MPOL_DEFAULT;
1373 kmem_cache_free(policy_cache, p);
1377 * Shared memory backing store policy support.
1379 * Remember policies even when nobody has shared memory mapped.
1380 * The policies are kept in Red-Black tree linked from the inode.
1381 * They are protected by the sp->lock spinlock, which should be held
1382 * for any accesses to the tree.
1385 /* lookup first element intersecting start-end */
1386 /* Caller holds sp->lock */
1387 static struct sp_node *
1388 sp_lookup(struct shared_policy *sp, unsigned long start, unsigned long end)
1390 struct rb_node *n = sp->root.rb_node;
1393 struct sp_node *p = rb_entry(n, struct sp_node, nd);
1395 if (start >= p->end)
1397 else if (end <= p->start)
1405 struct sp_node *w = NULL;
1406 struct rb_node *prev = rb_prev(n);
1409 w = rb_entry(prev, struct sp_node, nd);
1410 if (w->end <= start)
1414 return rb_entry(n, struct sp_node, nd);
1417 /* Insert a new shared policy into the list. */
1418 /* Caller holds sp->lock */
1419 static void sp_insert(struct shared_policy *sp, struct sp_node *new)
1421 struct rb_node **p = &sp->root.rb_node;
1422 struct rb_node *parent = NULL;
1427 nd = rb_entry(parent, struct sp_node, nd);
1428 if (new->start < nd->start)
1430 else if (new->end > nd->end)
1431 p = &(*p)->rb_right;
1435 rb_link_node(&new->nd, parent, p);
1436 rb_insert_color(&new->nd, &sp->root);
1437 PDprintk("inserting %lx-%lx: %d\n", new->start, new->end,
1438 new->policy ? new->policy->policy : 0);
1441 /* Find shared policy intersecting idx */
1443 mpol_shared_policy_lookup(struct shared_policy *sp, unsigned long idx)
1445 struct mempolicy *pol = NULL;
1448 if (!sp->root.rb_node)
1450 spin_lock(&sp->lock);
1451 sn = sp_lookup(sp, idx, idx+1);
1453 mpol_get(sn->policy);
1456 spin_unlock(&sp->lock);
1460 static void sp_delete(struct shared_policy *sp, struct sp_node *n)
1462 PDprintk("deleting %lx-l%x\n", n->start, n->end);
1463 rb_erase(&n->nd, &sp->root);
1464 mpol_free(n->policy);
1465 kmem_cache_free(sn_cache, n);
1469 sp_alloc(unsigned long start, unsigned long end, struct mempolicy *pol)
1471 struct sp_node *n = kmem_cache_alloc(sn_cache, GFP_KERNEL);
1482 /* Replace a policy range. */
1483 static int shared_policy_replace(struct shared_policy *sp, unsigned long start,
1484 unsigned long end, struct sp_node *new)
1486 struct sp_node *n, *new2 = NULL;
1489 spin_lock(&sp->lock);
1490 n = sp_lookup(sp, start, end);
1491 /* Take care of old policies in the same range. */
1492 while (n && n->start < end) {
1493 struct rb_node *next = rb_next(&n->nd);
1494 if (n->start >= start) {
1500 /* Old policy spanning whole new range. */
1503 spin_unlock(&sp->lock);
1504 new2 = sp_alloc(end, n->end, n->policy);
1510 sp_insert(sp, new2);
1518 n = rb_entry(next, struct sp_node, nd);
1522 spin_unlock(&sp->lock);
1524 mpol_free(new2->policy);
1525 kmem_cache_free(sn_cache, new2);
1530 void mpol_shared_policy_init(struct shared_policy *info, int policy,
1531 nodemask_t *policy_nodes)
1533 info->root = RB_ROOT;
1534 spin_lock_init(&info->lock);
1536 if (policy != MPOL_DEFAULT) {
1537 struct mempolicy *newpol;
1539 /* Falls back to MPOL_DEFAULT on any error */
1540 newpol = mpol_new(policy, policy_nodes);
1541 if (!IS_ERR(newpol)) {
1542 /* Create pseudo-vma that contains just the policy */
1543 struct vm_area_struct pvma;
1545 memset(&pvma, 0, sizeof(struct vm_area_struct));
1546 /* Policy covers entire file */
1547 pvma.vm_end = TASK_SIZE;
1548 mpol_set_shared_policy(info, &pvma, newpol);
1554 int mpol_set_shared_policy(struct shared_policy *info,
1555 struct vm_area_struct *vma, struct mempolicy *npol)
1558 struct sp_node *new = NULL;
1559 unsigned long sz = vma_pages(vma);
1561 PDprintk("set_shared_policy %lx sz %lu %d %lx\n",
1563 sz, npol? npol->policy : -1,
1564 npol ? nodes_addr(npol->v.nodes)[0] : -1);
1567 new = sp_alloc(vma->vm_pgoff, vma->vm_pgoff + sz, npol);
1571 err = shared_policy_replace(info, vma->vm_pgoff, vma->vm_pgoff+sz, new);
1573 kmem_cache_free(sn_cache, new);
1577 /* Free a backing policy store on inode delete. */
1578 void mpol_free_shared_policy(struct shared_policy *p)
1581 struct rb_node *next;
1583 if (!p->root.rb_node)
1585 spin_lock(&p->lock);
1586 next = rb_first(&p->root);
1588 n = rb_entry(next, struct sp_node, nd);
1589 next = rb_next(&n->nd);
1590 rb_erase(&n->nd, &p->root);
1591 mpol_free(n->policy);
1592 kmem_cache_free(sn_cache, n);
1594 spin_unlock(&p->lock);
1597 /* assumes fs == KERNEL_DS */
1598 void __init numa_policy_init(void)
1600 policy_cache = kmem_cache_create("numa_policy",
1601 sizeof(struct mempolicy),
1602 0, SLAB_PANIC, NULL, NULL);
1604 sn_cache = kmem_cache_create("shared_policy_node",
1605 sizeof(struct sp_node),
1606 0, SLAB_PANIC, NULL, NULL);
1608 /* Set interleaving policy for system init. This way not all
1609 the data structures allocated at system boot end up in node zero. */
1611 if (do_set_mempolicy(MPOL_INTERLEAVE, &node_online_map))
1612 printk("numa_policy_init: interleaving failed\n");
1615 /* Reset policy of current process to default */
1616 void numa_default_policy(void)
1618 do_set_mempolicy(MPOL_DEFAULT, NULL);
1621 /* Migrate a policy to a different set of nodes */
1622 void mpol_rebind_policy(struct mempolicy *pol, const nodemask_t *newmask)
1624 nodemask_t *mpolmask;
1629 mpolmask = &pol->cpuset_mems_allowed;
1630 if (nodes_equal(*mpolmask, *newmask))
1633 switch (pol->policy) {
1636 case MPOL_INTERLEAVE:
1637 nodes_remap(tmp, pol->v.nodes, *mpolmask, *newmask);
1639 *mpolmask = *newmask;
1640 current->il_next = node_remap(current->il_next,
1641 *mpolmask, *newmask);
1643 case MPOL_PREFERRED:
1644 pol->v.preferred_node = node_remap(pol->v.preferred_node,
1645 *mpolmask, *newmask);
1646 *mpolmask = *newmask;
1651 struct zonelist *zonelist;
1654 for (z = pol->v.zonelist->zones; *z; z++)
1655 node_set((*z)->zone_pgdat->node_id, nodes);
1656 nodes_remap(tmp, nodes, *mpolmask, *newmask);
1659 zonelist = bind_zonelist(&nodes);
1661 /* If no mem, then zonelist is NULL and we keep old zonelist.
1662 * If that old zonelist has no remaining mems_allowed nodes,
1663 * then zonelist_policy() will "FALL THROUGH" to MPOL_DEFAULT.
1667 /* Good - got mem - substitute new zonelist */
1668 kfree(pol->v.zonelist);
1669 pol->v.zonelist = zonelist;
1671 *mpolmask = *newmask;
1681 * Wrapper for mpol_rebind_policy() that just requires task
1682 * pointer, and updates task mempolicy.
1685 void mpol_rebind_task(struct task_struct *tsk, const nodemask_t *new)
1687 mpol_rebind_policy(tsk->mempolicy, new);
1691 * Rebind each vma in mm to new nodemask.
1693 * Call holding a reference to mm. Takes mm->mmap_sem during call.
1696 void mpol_rebind_mm(struct mm_struct *mm, nodemask_t *new)
1698 struct vm_area_struct *vma;
1700 down_write(&mm->mmap_sem);
1701 for (vma = mm->mmap; vma; vma = vma->vm_next)
1702 mpol_rebind_policy(vma->vm_policy, new);
1703 up_write(&mm->mmap_sem);
1707 * Display pages allocated per node and memory policy via /proc.
1710 static const char *policy_types[] = { "default", "prefer", "bind",
1714 * Convert a mempolicy into a string.
1715 * Returns the number of characters in buffer (if positive)
1716 * or an error (negative)
1718 static inline int mpol_to_str(char *buffer, int maxlen, struct mempolicy *pol)
1723 int mode = pol ? pol->policy : MPOL_DEFAULT;
1730 case MPOL_PREFERRED:
1732 node_set(pol->v.preferred_node, nodes);
1736 get_zonemask(pol, &nodes);
1739 case MPOL_INTERLEAVE:
1740 nodes = pol->v.nodes;
1748 l = strlen(policy_types[mode]);
1749 if (buffer + maxlen < p + l + 1)
1752 strcpy(p, policy_types[mode]);
1755 if (!nodes_empty(nodes)) {
1756 if (buffer + maxlen < p + 2)
1759 p += nodelist_scnprintf(p, buffer + maxlen - p, nodes);
1765 unsigned long pages;
1767 unsigned long active;
1768 unsigned long writeback;
1769 unsigned long mapcount_max;
1770 unsigned long dirty;
1771 unsigned long swapcache;
1772 unsigned long node[MAX_NUMNODES];
1775 static void gather_stats(struct page *page, void *private, int pte_dirty)
1777 struct numa_maps *md = private;
1778 int count = page_mapcount(page);
1781 if (pte_dirty || PageDirty(page))
1784 if (PageSwapCache(page))
1787 if (PageActive(page))
1790 if (PageWriteback(page))
1796 if (count > md->mapcount_max)
1797 md->mapcount_max = count;
1799 md->node[page_to_nid(page)]++;
1802 #ifdef CONFIG_HUGETLB_PAGE
1803 static void check_huge_range(struct vm_area_struct *vma,
1804 unsigned long start, unsigned long end,
1805 struct numa_maps *md)
1810 for (addr = start; addr < end; addr += HPAGE_SIZE) {
1811 pte_t *ptep = huge_pte_offset(vma->vm_mm, addr & HPAGE_MASK);
1821 page = pte_page(pte);
1825 gather_stats(page, md, pte_dirty(*ptep));
1829 static inline void check_huge_range(struct vm_area_struct *vma,
1830 unsigned long start, unsigned long end,
1831 struct numa_maps *md)
1836 int show_numa_map(struct seq_file *m, void *v)
1838 struct task_struct *task = m->private;
1839 struct vm_area_struct *vma = v;
1840 struct numa_maps *md;
1841 struct file *file = vma->vm_file;
1842 struct mm_struct *mm = vma->vm_mm;
1849 md = kzalloc(sizeof(struct numa_maps), GFP_KERNEL);
1853 mpol_to_str(buffer, sizeof(buffer),
1854 get_vma_policy(task, vma, vma->vm_start));
1856 seq_printf(m, "%08lx %s", vma->vm_start, buffer);
1859 seq_printf(m, " file=");
1860 seq_path(m, file->f_vfsmnt, file->f_dentry, "\n\t= ");
1861 } else if (vma->vm_start <= mm->brk && vma->vm_end >= mm->start_brk) {
1862 seq_printf(m, " heap");
1863 } else if (vma->vm_start <= mm->start_stack &&
1864 vma->vm_end >= mm->start_stack) {
1865 seq_printf(m, " stack");
1868 if (is_vm_hugetlb_page(vma)) {
1869 check_huge_range(vma, vma->vm_start, vma->vm_end, md);
1870 seq_printf(m, " huge");
1872 check_pgd_range(vma, vma->vm_start, vma->vm_end,
1873 &node_online_map, MPOL_MF_STATS, md);
1880 seq_printf(m," anon=%lu",md->anon);
1883 seq_printf(m," dirty=%lu",md->dirty);
1885 if (md->pages != md->anon && md->pages != md->dirty)
1886 seq_printf(m, " mapped=%lu", md->pages);
1888 if (md->mapcount_max > 1)
1889 seq_printf(m, " mapmax=%lu", md->mapcount_max);
1892 seq_printf(m," swapcache=%lu", md->swapcache);
1894 if (md->active < md->pages && !is_vm_hugetlb_page(vma))
1895 seq_printf(m," active=%lu", md->active);
1898 seq_printf(m," writeback=%lu", md->writeback);
1900 for_each_online_node(n)
1902 seq_printf(m, " N%d=%lu", n, md->node[n]);
1907 if (m->count < m->size)
1908 m->version = (vma != get_gate_vma(task)) ? vma->vm_start : 0;