blob: 72f402cc9c9af77f6d6c765c166e35d0bb4764ae [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 * Simple NUMA memory policy for the Linux kernel.
3 *
4 * Copyright 2003,2004 Andi Kleen, SuSE Labs.
Christoph Lameter8bccd852005-10-29 18:16:59 -07005 * (C) Copyright 2005 Christoph Lameter, Silicon Graphics, Inc.
Linus Torvalds1da177e2005-04-16 15:20:36 -07006 * Subject to the GNU Public License, version 2.
7 *
8 * NUMA policy allows the user to give hints in which node(s) memory should
9 * be allocated.
10 *
11 * Support four policies per VMA and per process:
12 *
13 * The VMA policy has priority over the process policy for a page fault.
14 *
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
20 * is used.
Christoph Lameter8bccd852005-10-29 18:16:59 -070021 *
Linus Torvalds1da177e2005-04-16 15:20:36 -070022 * bind Only allocate memory on a specific set of nodes,
23 * no fallback.
Christoph Lameter8bccd852005-10-29 18:16:59 -070024 * 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
27 *
Linus Torvalds1da177e2005-04-16 15:20:36 -070028 * 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
32 * process policy.
Christoph Lameter8bccd852005-10-29 18:16:59 -070033 *
Linus Torvalds1da177e2005-04-16 15:20:36 -070034 * 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.
37 *
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.
42 *
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.
46 *
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.
51 *
52 * For shmfs/tmpfs/hugetlbfs shared memory the policy is shared between
53 * all users and remembered even when nobody has memory mapped.
54 */
55
56/* Notebook:
57 fix mmap readahead to honour policy and enable policy for any page cache
58 object
59 statistics for bigpages
60 global policy for page cache? currently it uses process policy. Requires
61 first item above.
62 handle mremap for shared memory (currently ignored for the policy)
63 grows down?
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.
67*/
68
69#include <linux/mempolicy.h>
70#include <linux/mm.h>
71#include <linux/highmem.h>
72#include <linux/hugetlb.h>
73#include <linux/kernel.h>
74#include <linux/sched.h>
75#include <linux/mm.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 <asm/tlbflush.h>
87#include <asm/uaccess.h>
88
89static kmem_cache_t *policy_cache;
90static kmem_cache_t *sn_cache;
91
92#define PDprintk(fmt...)
93
94/* Highest zone. An specific allocation for a zone below that is not
95 policied. */
96static int policy_zone;
97
Andi Kleend42c6992005-07-06 19:56:03 +020098struct mempolicy default_policy = {
Linus Torvalds1da177e2005-04-16 15:20:36 -070099 .refcnt = ATOMIC_INIT(1), /* never free it */
100 .policy = MPOL_DEFAULT,
101};
102
Linus Torvalds1da177e2005-04-16 15:20:36 -0700103/* Do sanity checking on a policy */
Andi Kleendfcd3c02005-10-29 18:15:48 -0700104static int mpol_check_policy(int mode, nodemask_t *nodes)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700105{
Andi Kleendfcd3c02005-10-29 18:15:48 -0700106 int empty = nodes_empty(*nodes);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700107
108 switch (mode) {
109 case MPOL_DEFAULT:
110 if (!empty)
111 return -EINVAL;
112 break;
113 case MPOL_BIND:
114 case MPOL_INTERLEAVE:
115 /* Preferred will only use the first bit, but allow
116 more for now. */
117 if (empty)
118 return -EINVAL;
119 break;
120 }
Andi Kleendfcd3c02005-10-29 18:15:48 -0700121 return nodes_subset(*nodes, node_online_map) ? 0 : -EINVAL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700122}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700123/* Generate a custom zonelist for the BIND policy. */
Andi Kleendfcd3c02005-10-29 18:15:48 -0700124static struct zonelist *bind_zonelist(nodemask_t *nodes)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700125{
126 struct zonelist *zl;
127 int num, max, nd;
128
Andi Kleendfcd3c02005-10-29 18:15:48 -0700129 max = 1 + MAX_NR_ZONES * nodes_weight(*nodes);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700130 zl = kmalloc(sizeof(void *) * max, GFP_KERNEL);
131 if (!zl)
132 return NULL;
133 num = 0;
Andi Kleendfcd3c02005-10-29 18:15:48 -0700134 for_each_node_mask(nd, *nodes) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700135 int k;
136 for (k = MAX_NR_ZONES-1; k >= 0; k--) {
137 struct zone *z = &NODE_DATA(nd)->node_zones[k];
138 if (!z->present_pages)
139 continue;
140 zl->zones[num++] = z;
141 if (k > policy_zone)
142 policy_zone = k;
143 }
144 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700145 zl->zones[num] = NULL;
146 return zl;
147}
148
149/* Create a new policy */
Andi Kleendfcd3c02005-10-29 18:15:48 -0700150static struct mempolicy *mpol_new(int mode, nodemask_t *nodes)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700151{
152 struct mempolicy *policy;
153
Andi Kleendfcd3c02005-10-29 18:15:48 -0700154 PDprintk("setting mode %d nodes[0] %lx\n", mode, nodes_addr(*nodes)[0]);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700155 if (mode == MPOL_DEFAULT)
156 return NULL;
157 policy = kmem_cache_alloc(policy_cache, GFP_KERNEL);
158 if (!policy)
159 return ERR_PTR(-ENOMEM);
160 atomic_set(&policy->refcnt, 1);
161 switch (mode) {
162 case MPOL_INTERLEAVE:
Andi Kleendfcd3c02005-10-29 18:15:48 -0700163 policy->v.nodes = *nodes;
Andi Kleen8f493d72006-01-03 00:07:28 +0100164 if (nodes_weight(*nodes) == 0) {
165 kmem_cache_free(policy_cache, policy);
166 return ERR_PTR(-EINVAL);
167 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700168 break;
169 case MPOL_PREFERRED:
Andi Kleendfcd3c02005-10-29 18:15:48 -0700170 policy->v.preferred_node = first_node(*nodes);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700171 if (policy->v.preferred_node >= MAX_NUMNODES)
172 policy->v.preferred_node = -1;
173 break;
174 case MPOL_BIND:
175 policy->v.zonelist = bind_zonelist(nodes);
176 if (policy->v.zonelist == NULL) {
177 kmem_cache_free(policy_cache, policy);
178 return ERR_PTR(-ENOMEM);
179 }
180 break;
181 }
182 policy->policy = mode;
183 return policy;
184}
185
186/* Ensure all existing pages follow the policy. */
Nick Pigginb5810032005-10-29 18:16:12 -0700187static int check_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
Andi Kleendfcd3c02005-10-29 18:15:48 -0700188 unsigned long addr, unsigned long end, nodemask_t *nodes)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700189{
Hugh Dickins91612e02005-06-21 17:15:07 -0700190 pte_t *orig_pte;
191 pte_t *pte;
Hugh Dickins705e87c2005-10-29 18:16:27 -0700192 spinlock_t *ptl;
Hugh Dickins941150a2005-06-21 17:15:06 -0700193
Hugh Dickins705e87c2005-10-29 18:16:27 -0700194 orig_pte = pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
Hugh Dickins91612e02005-06-21 17:15:07 -0700195 do {
Linus Torvalds6aab3412005-11-28 14:34:23 -0800196 struct page *page;
Hugh Dickins91612e02005-06-21 17:15:07 -0700197 unsigned int nid;
198
199 if (!pte_present(*pte))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700200 continue;
Linus Torvalds6aab3412005-11-28 14:34:23 -0800201 page = vm_normal_page(vma, addr, *pte);
202 if (!page)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700203 continue;
Linus Torvalds6aab3412005-11-28 14:34:23 -0800204 nid = page_to_nid(page);
Andi Kleendfcd3c02005-10-29 18:15:48 -0700205 if (!node_isset(nid, *nodes))
Hugh Dickins91612e02005-06-21 17:15:07 -0700206 break;
207 } while (pte++, addr += PAGE_SIZE, addr != end);
Hugh Dickins705e87c2005-10-29 18:16:27 -0700208 pte_unmap_unlock(orig_pte, ptl);
Hugh Dickins91612e02005-06-21 17:15:07 -0700209 return addr != end;
210}
211
Nick Pigginb5810032005-10-29 18:16:12 -0700212static inline int check_pmd_range(struct vm_area_struct *vma, pud_t *pud,
Andi Kleendfcd3c02005-10-29 18:15:48 -0700213 unsigned long addr, unsigned long end, nodemask_t *nodes)
Hugh Dickins91612e02005-06-21 17:15:07 -0700214{
215 pmd_t *pmd;
216 unsigned long next;
217
218 pmd = pmd_offset(pud, addr);
219 do {
220 next = pmd_addr_end(addr, end);
221 if (pmd_none_or_clear_bad(pmd))
222 continue;
Nick Pigginb5810032005-10-29 18:16:12 -0700223 if (check_pte_range(vma, pmd, addr, next, nodes))
Hugh Dickins91612e02005-06-21 17:15:07 -0700224 return -EIO;
225 } while (pmd++, addr = next, addr != end);
226 return 0;
227}
228
Nick Pigginb5810032005-10-29 18:16:12 -0700229static inline int check_pud_range(struct vm_area_struct *vma, pgd_t *pgd,
Andi Kleendfcd3c02005-10-29 18:15:48 -0700230 unsigned long addr, unsigned long end, nodemask_t *nodes)
Hugh Dickins91612e02005-06-21 17:15:07 -0700231{
232 pud_t *pud;
233 unsigned long next;
234
235 pud = pud_offset(pgd, addr);
236 do {
237 next = pud_addr_end(addr, end);
238 if (pud_none_or_clear_bad(pud))
239 continue;
Nick Pigginb5810032005-10-29 18:16:12 -0700240 if (check_pmd_range(vma, pud, addr, next, nodes))
Hugh Dickins91612e02005-06-21 17:15:07 -0700241 return -EIO;
242 } while (pud++, addr = next, addr != end);
243 return 0;
244}
245
Nick Pigginb5810032005-10-29 18:16:12 -0700246static inline int check_pgd_range(struct vm_area_struct *vma,
Andi Kleendfcd3c02005-10-29 18:15:48 -0700247 unsigned long addr, unsigned long end, nodemask_t *nodes)
Hugh Dickins91612e02005-06-21 17:15:07 -0700248{
249 pgd_t *pgd;
250 unsigned long next;
251
Nick Pigginb5810032005-10-29 18:16:12 -0700252 pgd = pgd_offset(vma->vm_mm, addr);
Hugh Dickins91612e02005-06-21 17:15:07 -0700253 do {
254 next = pgd_addr_end(addr, end);
255 if (pgd_none_or_clear_bad(pgd))
256 continue;
Nick Pigginb5810032005-10-29 18:16:12 -0700257 if (check_pud_range(vma, pgd, addr, next, nodes))
Hugh Dickins91612e02005-06-21 17:15:07 -0700258 return -EIO;
259 } while (pgd++, addr = next, addr != end);
260 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700261}
262
263/* Step 1: check the range */
264static struct vm_area_struct *
265check_range(struct mm_struct *mm, unsigned long start, unsigned long end,
Andi Kleendfcd3c02005-10-29 18:15:48 -0700266 nodemask_t *nodes, unsigned long flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700267{
268 int err;
269 struct vm_area_struct *first, *vma, *prev;
270
271 first = find_vma(mm, start);
272 if (!first)
273 return ERR_PTR(-EFAULT);
274 prev = NULL;
275 for (vma = first; vma && vma->vm_start < end; vma = vma->vm_next) {
276 if (!vma->vm_next && vma->vm_end < end)
277 return ERR_PTR(-EFAULT);
278 if (prev && prev->vm_end < vma->vm_start)
279 return ERR_PTR(-EFAULT);
280 if ((flags & MPOL_MF_STRICT) && !is_vm_hugetlb_page(vma)) {
Andi Kleen5b952b32005-09-13 01:25:08 -0700281 unsigned long endvma = vma->vm_end;
282 if (endvma > end)
283 endvma = end;
284 if (vma->vm_start > start)
285 start = vma->vm_start;
Nick Pigginb5810032005-10-29 18:16:12 -0700286 err = check_pgd_range(vma, start, endvma, nodes);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700287 if (err) {
288 first = ERR_PTR(err);
289 break;
290 }
291 }
292 prev = vma;
293 }
294 return first;
295}
296
297/* Apply policy to a single VMA */
298static int policy_vma(struct vm_area_struct *vma, struct mempolicy *new)
299{
300 int err = 0;
301 struct mempolicy *old = vma->vm_policy;
302
303 PDprintk("vma %lx-%lx/%lx vm_ops %p vm_file %p set_policy %p\n",
304 vma->vm_start, vma->vm_end, vma->vm_pgoff,
305 vma->vm_ops, vma->vm_file,
306 vma->vm_ops ? vma->vm_ops->set_policy : NULL);
307
308 if (vma->vm_ops && vma->vm_ops->set_policy)
309 err = vma->vm_ops->set_policy(vma, new);
310 if (!err) {
311 mpol_get(new);
312 vma->vm_policy = new;
313 mpol_free(old);
314 }
315 return err;
316}
317
318/* Step 2: apply policy to a range and do splits. */
319static int mbind_range(struct vm_area_struct *vma, unsigned long start,
320 unsigned long end, struct mempolicy *new)
321{
322 struct vm_area_struct *next;
323 int err;
324
325 err = 0;
326 for (; vma && vma->vm_start < end; vma = next) {
327 next = vma->vm_next;
328 if (vma->vm_start < start)
329 err = split_vma(vma->vm_mm, vma, start, 1);
330 if (!err && vma->vm_end > end)
331 err = split_vma(vma->vm_mm, vma, end, 0);
332 if (!err)
333 err = policy_vma(vma, new);
334 if (err)
335 break;
336 }
337 return err;
338}
339
Christoph Lameter8bccd852005-10-29 18:16:59 -0700340static int contextualize_policy(int mode, nodemask_t *nodes)
341{
342 if (!nodes)
343 return 0;
344
345 /* Update current mems_allowed */
346 cpuset_update_current_mems_allowed();
347 /* Ignore nodes not set in current->mems_allowed */
348 cpuset_restrict_to_mems_allowed(nodes->bits);
349 return mpol_check_policy(mode, nodes);
350}
351
352long do_mbind(unsigned long start, unsigned long len,
353 unsigned long mode, nodemask_t *nmask, unsigned long flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700354{
355 struct vm_area_struct *vma;
356 struct mm_struct *mm = current->mm;
357 struct mempolicy *new;
358 unsigned long end;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700359 int err;
360
361 if ((flags & ~(unsigned long)(MPOL_MF_STRICT)) || mode > MPOL_MAX)
362 return -EINVAL;
363 if (start & ~PAGE_MASK)
364 return -EINVAL;
365 if (mode == MPOL_DEFAULT)
366 flags &= ~MPOL_MF_STRICT;
367 len = (len + PAGE_SIZE - 1) & PAGE_MASK;
368 end = start + len;
369 if (end < start)
370 return -EINVAL;
371 if (end == start)
372 return 0;
Christoph Lameter5fcbb232005-10-29 18:17:00 -0700373 if (mpol_check_policy(mode, nmask))
Christoph Lameter8bccd852005-10-29 18:16:59 -0700374 return -EINVAL;
375 new = mpol_new(mode, nmask);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700376 if (IS_ERR(new))
377 return PTR_ERR(new);
378
379 PDprintk("mbind %lx-%lx mode:%ld nodes:%lx\n",start,start+len,
Andi Kleendfcd3c02005-10-29 18:15:48 -0700380 mode,nodes_addr(nodes)[0]);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700381
382 down_write(&mm->mmap_sem);
Christoph Lameter8bccd852005-10-29 18:16:59 -0700383 vma = check_range(mm, start, end, nmask, flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700384 err = PTR_ERR(vma);
385 if (!IS_ERR(vma))
386 err = mbind_range(vma, start, end, new);
387 up_write(&mm->mmap_sem);
388 mpol_free(new);
389 return err;
390}
391
392/* Set the process memory policy */
Christoph Lameter8bccd852005-10-29 18:16:59 -0700393long do_set_mempolicy(int mode, nodemask_t *nodes)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700394{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700395 struct mempolicy *new;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700396
Christoph Lameter8bccd852005-10-29 18:16:59 -0700397 if (contextualize_policy(mode, nodes))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700398 return -EINVAL;
Christoph Lameter8bccd852005-10-29 18:16:59 -0700399 new = mpol_new(mode, nodes);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700400 if (IS_ERR(new))
401 return PTR_ERR(new);
402 mpol_free(current->mempolicy);
403 current->mempolicy = new;
404 if (new && new->policy == MPOL_INTERLEAVE)
Andi Kleendfcd3c02005-10-29 18:15:48 -0700405 current->il_next = first_node(new->v.nodes);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700406 return 0;
407}
408
409/* Fill a zone bitmap for a policy */
Andi Kleendfcd3c02005-10-29 18:15:48 -0700410static void get_zonemask(struct mempolicy *p, nodemask_t *nodes)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700411{
412 int i;
413
Andi Kleendfcd3c02005-10-29 18:15:48 -0700414 nodes_clear(*nodes);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700415 switch (p->policy) {
416 case MPOL_BIND:
417 for (i = 0; p->v.zonelist->zones[i]; i++)
Christoph Lameter8bccd852005-10-29 18:16:59 -0700418 node_set(p->v.zonelist->zones[i]->zone_pgdat->node_id,
419 *nodes);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700420 break;
421 case MPOL_DEFAULT:
422 break;
423 case MPOL_INTERLEAVE:
Andi Kleendfcd3c02005-10-29 18:15:48 -0700424 *nodes = p->v.nodes;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700425 break;
426 case MPOL_PREFERRED:
427 /* or use current node instead of online map? */
428 if (p->v.preferred_node < 0)
Andi Kleendfcd3c02005-10-29 18:15:48 -0700429 *nodes = node_online_map;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700430 else
Andi Kleendfcd3c02005-10-29 18:15:48 -0700431 node_set(p->v.preferred_node, *nodes);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700432 break;
433 default:
434 BUG();
435 }
436}
437
438static int lookup_node(struct mm_struct *mm, unsigned long addr)
439{
440 struct page *p;
441 int err;
442
443 err = get_user_pages(current, mm, addr & PAGE_MASK, 1, 0, 0, &p, NULL);
444 if (err >= 0) {
445 err = page_to_nid(p);
446 put_page(p);
447 }
448 return err;
449}
450
Linus Torvalds1da177e2005-04-16 15:20:36 -0700451/* Retrieve NUMA policy */
Christoph Lameter8bccd852005-10-29 18:16:59 -0700452long do_get_mempolicy(int *policy, nodemask_t *nmask,
453 unsigned long addr, unsigned long flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700454{
Christoph Lameter8bccd852005-10-29 18:16:59 -0700455 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700456 struct mm_struct *mm = current->mm;
457 struct vm_area_struct *vma = NULL;
458 struct mempolicy *pol = current->mempolicy;
459
Paul Jackson68860ec2005-10-30 15:02:36 -0800460 cpuset_update_current_mems_allowed();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700461 if (flags & ~(unsigned long)(MPOL_F_NODE|MPOL_F_ADDR))
462 return -EINVAL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700463 if (flags & MPOL_F_ADDR) {
464 down_read(&mm->mmap_sem);
465 vma = find_vma_intersection(mm, addr, addr+1);
466 if (!vma) {
467 up_read(&mm->mmap_sem);
468 return -EFAULT;
469 }
470 if (vma->vm_ops && vma->vm_ops->get_policy)
471 pol = vma->vm_ops->get_policy(vma, addr);
472 else
473 pol = vma->vm_policy;
474 } else if (addr)
475 return -EINVAL;
476
477 if (!pol)
478 pol = &default_policy;
479
480 if (flags & MPOL_F_NODE) {
481 if (flags & MPOL_F_ADDR) {
482 err = lookup_node(mm, addr);
483 if (err < 0)
484 goto out;
Christoph Lameter8bccd852005-10-29 18:16:59 -0700485 *policy = err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700486 } else if (pol == current->mempolicy &&
487 pol->policy == MPOL_INTERLEAVE) {
Christoph Lameter8bccd852005-10-29 18:16:59 -0700488 *policy = current->il_next;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700489 } else {
490 err = -EINVAL;
491 goto out;
492 }
493 } else
Christoph Lameter8bccd852005-10-29 18:16:59 -0700494 *policy = pol->policy;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700495
496 if (vma) {
497 up_read(&current->mm->mmap_sem);
498 vma = NULL;
499 }
500
Linus Torvalds1da177e2005-04-16 15:20:36 -0700501 err = 0;
Christoph Lameter8bccd852005-10-29 18:16:59 -0700502 if (nmask)
503 get_zonemask(pol, nmask);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700504
505 out:
506 if (vma)
507 up_read(&current->mm->mmap_sem);
508 return err;
509}
510
Christoph Lameter8bccd852005-10-29 18:16:59 -0700511/*
512 * User space interface with variable sized bitmaps for nodelists.
513 */
514
515/* Copy a node mask from user space. */
516static int get_nodes(nodemask_t *nodes, unsigned long __user *nmask,
517 unsigned long maxnode)
518{
519 unsigned long k;
520 unsigned long nlongs;
521 unsigned long endmask;
522
523 --maxnode;
524 nodes_clear(*nodes);
525 if (maxnode == 0 || !nmask)
526 return 0;
527
528 nlongs = BITS_TO_LONGS(maxnode);
529 if ((maxnode % BITS_PER_LONG) == 0)
530 endmask = ~0UL;
531 else
532 endmask = (1UL << (maxnode % BITS_PER_LONG)) - 1;
533
534 /* When the user specified more nodes than supported just check
535 if the non supported part is all zero. */
536 if (nlongs > BITS_TO_LONGS(MAX_NUMNODES)) {
537 if (nlongs > PAGE_SIZE/sizeof(long))
538 return -EINVAL;
539 for (k = BITS_TO_LONGS(MAX_NUMNODES); k < nlongs; k++) {
540 unsigned long t;
541 if (get_user(t, nmask + k))
542 return -EFAULT;
543 if (k == nlongs - 1) {
544 if (t & endmask)
545 return -EINVAL;
546 } else if (t)
547 return -EINVAL;
548 }
549 nlongs = BITS_TO_LONGS(MAX_NUMNODES);
550 endmask = ~0UL;
551 }
552
553 if (copy_from_user(nodes_addr(*nodes), nmask, nlongs*sizeof(unsigned long)))
554 return -EFAULT;
555 nodes_addr(*nodes)[nlongs-1] &= endmask;
556 return 0;
557}
558
559/* Copy a kernel node mask to user space */
560static int copy_nodes_to_user(unsigned long __user *mask, unsigned long maxnode,
561 nodemask_t *nodes)
562{
563 unsigned long copy = ALIGN(maxnode-1, 64) / 8;
564 const int nbytes = BITS_TO_LONGS(MAX_NUMNODES) * sizeof(long);
565
566 if (copy > nbytes) {
567 if (copy > PAGE_SIZE)
568 return -EINVAL;
569 if (clear_user((char __user *)mask + nbytes, copy - nbytes))
570 return -EFAULT;
571 copy = nbytes;
572 }
573 return copy_to_user(mask, nodes_addr(*nodes), copy) ? -EFAULT : 0;
574}
575
576asmlinkage long sys_mbind(unsigned long start, unsigned long len,
577 unsigned long mode,
578 unsigned long __user *nmask, unsigned long maxnode,
579 unsigned flags)
580{
581 nodemask_t nodes;
582 int err;
583
584 err = get_nodes(&nodes, nmask, maxnode);
585 if (err)
586 return err;
587 return do_mbind(start, len, mode, &nodes, flags);
588}
589
590/* Set the process memory policy */
591asmlinkage long sys_set_mempolicy(int mode, unsigned long __user *nmask,
592 unsigned long maxnode)
593{
594 int err;
595 nodemask_t nodes;
596
597 if (mode < 0 || mode > MPOL_MAX)
598 return -EINVAL;
599 err = get_nodes(&nodes, nmask, maxnode);
600 if (err)
601 return err;
602 return do_set_mempolicy(mode, &nodes);
603}
604
605/* Retrieve NUMA policy */
606asmlinkage long sys_get_mempolicy(int __user *policy,
607 unsigned long __user *nmask,
608 unsigned long maxnode,
609 unsigned long addr, unsigned long flags)
610{
611 int err, pval;
612 nodemask_t nodes;
613
614 if (nmask != NULL && maxnode < MAX_NUMNODES)
615 return -EINVAL;
616
617 err = do_get_mempolicy(&pval, &nodes, addr, flags);
618
619 if (err)
620 return err;
621
622 if (policy && put_user(pval, policy))
623 return -EFAULT;
624
625 if (nmask)
626 err = copy_nodes_to_user(nmask, maxnode, &nodes);
627
628 return err;
629}
630
Linus Torvalds1da177e2005-04-16 15:20:36 -0700631#ifdef CONFIG_COMPAT
632
633asmlinkage long compat_sys_get_mempolicy(int __user *policy,
634 compat_ulong_t __user *nmask,
635 compat_ulong_t maxnode,
636 compat_ulong_t addr, compat_ulong_t flags)
637{
638 long err;
639 unsigned long __user *nm = NULL;
640 unsigned long nr_bits, alloc_size;
641 DECLARE_BITMAP(bm, MAX_NUMNODES);
642
643 nr_bits = min_t(unsigned long, maxnode-1, MAX_NUMNODES);
644 alloc_size = ALIGN(nr_bits, BITS_PER_LONG) / 8;
645
646 if (nmask)
647 nm = compat_alloc_user_space(alloc_size);
648
649 err = sys_get_mempolicy(policy, nm, nr_bits+1, addr, flags);
650
651 if (!err && nmask) {
652 err = copy_from_user(bm, nm, alloc_size);
653 /* ensure entire bitmap is zeroed */
654 err |= clear_user(nmask, ALIGN(maxnode-1, 8) / 8);
655 err |= compat_put_bitmap(nmask, bm, nr_bits);
656 }
657
658 return err;
659}
660
661asmlinkage long compat_sys_set_mempolicy(int mode, compat_ulong_t __user *nmask,
662 compat_ulong_t maxnode)
663{
664 long err = 0;
665 unsigned long __user *nm = NULL;
666 unsigned long nr_bits, alloc_size;
667 DECLARE_BITMAP(bm, MAX_NUMNODES);
668
669 nr_bits = min_t(unsigned long, maxnode-1, MAX_NUMNODES);
670 alloc_size = ALIGN(nr_bits, BITS_PER_LONG) / 8;
671
672 if (nmask) {
673 err = compat_get_bitmap(bm, nmask, nr_bits);
674 nm = compat_alloc_user_space(alloc_size);
675 err |= copy_to_user(nm, bm, alloc_size);
676 }
677
678 if (err)
679 return -EFAULT;
680
681 return sys_set_mempolicy(mode, nm, nr_bits+1);
682}
683
684asmlinkage long compat_sys_mbind(compat_ulong_t start, compat_ulong_t len,
685 compat_ulong_t mode, compat_ulong_t __user *nmask,
686 compat_ulong_t maxnode, compat_ulong_t flags)
687{
688 long err = 0;
689 unsigned long __user *nm = NULL;
690 unsigned long nr_bits, alloc_size;
Andi Kleendfcd3c02005-10-29 18:15:48 -0700691 nodemask_t bm;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700692
693 nr_bits = min_t(unsigned long, maxnode-1, MAX_NUMNODES);
694 alloc_size = ALIGN(nr_bits, BITS_PER_LONG) / 8;
695
696 if (nmask) {
Andi Kleendfcd3c02005-10-29 18:15:48 -0700697 err = compat_get_bitmap(nodes_addr(bm), nmask, nr_bits);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700698 nm = compat_alloc_user_space(alloc_size);
Andi Kleendfcd3c02005-10-29 18:15:48 -0700699 err |= copy_to_user(nm, nodes_addr(bm), alloc_size);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700700 }
701
702 if (err)
703 return -EFAULT;
704
705 return sys_mbind(start, len, mode, nm, nr_bits+1, flags);
706}
707
708#endif
709
710/* Return effective policy for a VMA */
Christoph Lameter6e21c8f2005-09-03 15:54:45 -0700711struct mempolicy *
712get_vma_policy(struct task_struct *task, struct vm_area_struct *vma, unsigned long addr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700713{
Christoph Lameter6e21c8f2005-09-03 15:54:45 -0700714 struct mempolicy *pol = task->mempolicy;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700715
716 if (vma) {
717 if (vma->vm_ops && vma->vm_ops->get_policy)
Christoph Lameter8bccd852005-10-29 18:16:59 -0700718 pol = vma->vm_ops->get_policy(vma, addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700719 else if (vma->vm_policy &&
720 vma->vm_policy->policy != MPOL_DEFAULT)
721 pol = vma->vm_policy;
722 }
723 if (!pol)
724 pol = &default_policy;
725 return pol;
726}
727
728/* Return a zonelist representing a mempolicy */
Al Virodd0fc662005-10-07 07:46:04 +0100729static struct zonelist *zonelist_policy(gfp_t gfp, struct mempolicy *policy)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700730{
731 int nd;
732
733 switch (policy->policy) {
734 case MPOL_PREFERRED:
735 nd = policy->v.preferred_node;
736 if (nd < 0)
737 nd = numa_node_id();
738 break;
739 case MPOL_BIND:
740 /* Lower zones don't get a policy applied */
741 /* Careful: current->mems_allowed might have moved */
Al Viroaf4ca452005-10-21 02:55:38 -0400742 if (gfp_zone(gfp) >= policy_zone)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700743 if (cpuset_zonelist_valid_mems_allowed(policy->v.zonelist))
744 return policy->v.zonelist;
745 /*FALL THROUGH*/
746 case MPOL_INTERLEAVE: /* should not happen */
747 case MPOL_DEFAULT:
748 nd = numa_node_id();
749 break;
750 default:
751 nd = 0;
752 BUG();
753 }
Al Viroaf4ca452005-10-21 02:55:38 -0400754 return NODE_DATA(nd)->node_zonelists + gfp_zone(gfp);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700755}
756
757/* Do dynamic interleaving for a process */
758static unsigned interleave_nodes(struct mempolicy *policy)
759{
760 unsigned nid, next;
761 struct task_struct *me = current;
762
763 nid = me->il_next;
Andi Kleendfcd3c02005-10-29 18:15:48 -0700764 next = next_node(nid, policy->v.nodes);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700765 if (next >= MAX_NUMNODES)
Andi Kleendfcd3c02005-10-29 18:15:48 -0700766 next = first_node(policy->v.nodes);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700767 me->il_next = next;
768 return nid;
769}
770
771/* Do static interleaving for a VMA with known offset. */
772static unsigned offset_il_node(struct mempolicy *pol,
773 struct vm_area_struct *vma, unsigned long off)
774{
Andi Kleendfcd3c02005-10-29 18:15:48 -0700775 unsigned nnodes = nodes_weight(pol->v.nodes);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700776 unsigned target = (unsigned)off % nnodes;
777 int c;
778 int nid = -1;
779
780 c = 0;
781 do {
Andi Kleendfcd3c02005-10-29 18:15:48 -0700782 nid = next_node(nid, pol->v.nodes);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700783 c++;
784 } while (c <= target);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700785 return nid;
786}
787
788/* Allocate a page in interleaved policy.
789 Own path because it needs to do special accounting. */
Andi Kleen662f3a02005-10-29 18:15:49 -0700790static struct page *alloc_page_interleave(gfp_t gfp, unsigned order,
791 unsigned nid)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700792{
793 struct zonelist *zl;
794 struct page *page;
795
Al Viroaf4ca452005-10-21 02:55:38 -0400796 zl = NODE_DATA(nid)->node_zonelists + gfp_zone(gfp);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700797 page = __alloc_pages(gfp, order, zl);
798 if (page && page_zone(page) == zl->zones[0]) {
Christoph Lametere7c8d5c2005-06-21 17:14:47 -0700799 zone_pcp(zl->zones[0],get_cpu())->interleave_hit++;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700800 put_cpu();
801 }
802 return page;
803}
804
805/**
806 * alloc_page_vma - Allocate a page for a VMA.
807 *
808 * @gfp:
809 * %GFP_USER user allocation.
810 * %GFP_KERNEL kernel allocations,
811 * %GFP_HIGHMEM highmem/user allocations,
812 * %GFP_FS allocation should not call back into a file system.
813 * %GFP_ATOMIC don't sleep.
814 *
815 * @vma: Pointer to VMA or NULL if not available.
816 * @addr: Virtual Address of the allocation. Must be inside the VMA.
817 *
818 * This function allocates a page from the kernel page pool and applies
819 * a NUMA policy associated with the VMA or the current process.
820 * When VMA is not NULL caller must hold down_read on the mmap_sem of the
821 * mm_struct of the VMA to prevent it from going away. Should be used for
822 * all allocations for pages that will be mapped into
823 * user space. Returns NULL when no page can be allocated.
824 *
825 * Should be called with the mm_sem of the vma hold.
826 */
827struct page *
Al Virodd0fc662005-10-07 07:46:04 +0100828alloc_page_vma(gfp_t gfp, struct vm_area_struct *vma, unsigned long addr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700829{
Christoph Lameter6e21c8f2005-09-03 15:54:45 -0700830 struct mempolicy *pol = get_vma_policy(current, vma, addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700831
832 cpuset_update_current_mems_allowed();
833
834 if (unlikely(pol->policy == MPOL_INTERLEAVE)) {
835 unsigned nid;
836 if (vma) {
837 unsigned long off;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700838 off = vma->vm_pgoff;
839 off += (addr - vma->vm_start) >> PAGE_SHIFT;
840 nid = offset_il_node(pol, vma, off);
841 } else {
842 /* fall back to process interleaving */
843 nid = interleave_nodes(pol);
844 }
845 return alloc_page_interleave(gfp, 0, nid);
846 }
847 return __alloc_pages(gfp, 0, zonelist_policy(gfp, pol));
848}
849
850/**
851 * alloc_pages_current - Allocate pages.
852 *
853 * @gfp:
854 * %GFP_USER user allocation,
855 * %GFP_KERNEL kernel allocation,
856 * %GFP_HIGHMEM highmem allocation,
857 * %GFP_FS don't call back into a file system.
858 * %GFP_ATOMIC don't sleep.
859 * @order: Power of two of allocation size in pages. 0 is a single page.
860 *
861 * Allocate a page from the kernel page pool. When not in
862 * interrupt context and apply the current process NUMA policy.
863 * Returns NULL when no page can be allocated.
864 *
865 * Don't call cpuset_update_current_mems_allowed() unless
866 * 1) it's ok to take cpuset_sem (can WAIT), and
867 * 2) allocating for current task (not interrupt).
868 */
Al Virodd0fc662005-10-07 07:46:04 +0100869struct page *alloc_pages_current(gfp_t gfp, unsigned order)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700870{
871 struct mempolicy *pol = current->mempolicy;
872
873 if ((gfp & __GFP_WAIT) && !in_interrupt())
874 cpuset_update_current_mems_allowed();
875 if (!pol || in_interrupt())
876 pol = &default_policy;
877 if (pol->policy == MPOL_INTERLEAVE)
878 return alloc_page_interleave(gfp, order, interleave_nodes(pol));
879 return __alloc_pages(gfp, order, zonelist_policy(gfp, pol));
880}
881EXPORT_SYMBOL(alloc_pages_current);
882
883/* Slow path of a mempolicy copy */
884struct mempolicy *__mpol_copy(struct mempolicy *old)
885{
886 struct mempolicy *new = kmem_cache_alloc(policy_cache, GFP_KERNEL);
887
888 if (!new)
889 return ERR_PTR(-ENOMEM);
890 *new = *old;
891 atomic_set(&new->refcnt, 1);
892 if (new->policy == MPOL_BIND) {
893 int sz = ksize(old->v.zonelist);
894 new->v.zonelist = kmalloc(sz, SLAB_KERNEL);
895 if (!new->v.zonelist) {
896 kmem_cache_free(policy_cache, new);
897 return ERR_PTR(-ENOMEM);
898 }
899 memcpy(new->v.zonelist, old->v.zonelist, sz);
900 }
901 return new;
902}
903
904/* Slow path of a mempolicy comparison */
905int __mpol_equal(struct mempolicy *a, struct mempolicy *b)
906{
907 if (!a || !b)
908 return 0;
909 if (a->policy != b->policy)
910 return 0;
911 switch (a->policy) {
912 case MPOL_DEFAULT:
913 return 1;
914 case MPOL_INTERLEAVE:
Andi Kleendfcd3c02005-10-29 18:15:48 -0700915 return nodes_equal(a->v.nodes, b->v.nodes);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700916 case MPOL_PREFERRED:
917 return a->v.preferred_node == b->v.preferred_node;
918 case MPOL_BIND: {
919 int i;
920 for (i = 0; a->v.zonelist->zones[i]; i++)
921 if (a->v.zonelist->zones[i] != b->v.zonelist->zones[i])
922 return 0;
923 return b->v.zonelist->zones[i] == NULL;
924 }
925 default:
926 BUG();
927 return 0;
928 }
929}
930
931/* Slow path of a mpol destructor. */
932void __mpol_free(struct mempolicy *p)
933{
934 if (!atomic_dec_and_test(&p->refcnt))
935 return;
936 if (p->policy == MPOL_BIND)
937 kfree(p->v.zonelist);
938 p->policy = MPOL_DEFAULT;
939 kmem_cache_free(policy_cache, p);
940}
941
942/*
943 * Hugetlb policy. Same as above, just works with node numbers instead of
944 * zonelists.
945 */
946
947/* Find first node suitable for an allocation */
948int mpol_first_node(struct vm_area_struct *vma, unsigned long addr)
949{
Christoph Lameter6e21c8f2005-09-03 15:54:45 -0700950 struct mempolicy *pol = get_vma_policy(current, vma, addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700951
952 switch (pol->policy) {
953 case MPOL_DEFAULT:
954 return numa_node_id();
955 case MPOL_BIND:
956 return pol->v.zonelist->zones[0]->zone_pgdat->node_id;
957 case MPOL_INTERLEAVE:
958 return interleave_nodes(pol);
959 case MPOL_PREFERRED:
960 return pol->v.preferred_node >= 0 ?
961 pol->v.preferred_node : numa_node_id();
962 }
963 BUG();
964 return 0;
965}
966
967/* Find secondary valid nodes for an allocation */
968int mpol_node_valid(int nid, struct vm_area_struct *vma, unsigned long addr)
969{
Christoph Lameter6e21c8f2005-09-03 15:54:45 -0700970 struct mempolicy *pol = get_vma_policy(current, vma, addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700971
972 switch (pol->policy) {
973 case MPOL_PREFERRED:
974 case MPOL_DEFAULT:
975 case MPOL_INTERLEAVE:
976 return 1;
977 case MPOL_BIND: {
978 struct zone **z;
979 for (z = pol->v.zonelist->zones; *z; z++)
980 if ((*z)->zone_pgdat->node_id == nid)
981 return 1;
982 return 0;
983 }
984 default:
985 BUG();
986 return 0;
987 }
988}
989
990/*
991 * Shared memory backing store policy support.
992 *
993 * Remember policies even when nobody has shared memory mapped.
994 * The policies are kept in Red-Black tree linked from the inode.
995 * They are protected by the sp->lock spinlock, which should be held
996 * for any accesses to the tree.
997 */
998
999/* lookup first element intersecting start-end */
1000/* Caller holds sp->lock */
1001static struct sp_node *
1002sp_lookup(struct shared_policy *sp, unsigned long start, unsigned long end)
1003{
1004 struct rb_node *n = sp->root.rb_node;
1005
1006 while (n) {
1007 struct sp_node *p = rb_entry(n, struct sp_node, nd);
1008
1009 if (start >= p->end)
1010 n = n->rb_right;
1011 else if (end <= p->start)
1012 n = n->rb_left;
1013 else
1014 break;
1015 }
1016 if (!n)
1017 return NULL;
1018 for (;;) {
1019 struct sp_node *w = NULL;
1020 struct rb_node *prev = rb_prev(n);
1021 if (!prev)
1022 break;
1023 w = rb_entry(prev, struct sp_node, nd);
1024 if (w->end <= start)
1025 break;
1026 n = prev;
1027 }
1028 return rb_entry(n, struct sp_node, nd);
1029}
1030
1031/* Insert a new shared policy into the list. */
1032/* Caller holds sp->lock */
1033static void sp_insert(struct shared_policy *sp, struct sp_node *new)
1034{
1035 struct rb_node **p = &sp->root.rb_node;
1036 struct rb_node *parent = NULL;
1037 struct sp_node *nd;
1038
1039 while (*p) {
1040 parent = *p;
1041 nd = rb_entry(parent, struct sp_node, nd);
1042 if (new->start < nd->start)
1043 p = &(*p)->rb_left;
1044 else if (new->end > nd->end)
1045 p = &(*p)->rb_right;
1046 else
1047 BUG();
1048 }
1049 rb_link_node(&new->nd, parent, p);
1050 rb_insert_color(&new->nd, &sp->root);
1051 PDprintk("inserting %lx-%lx: %d\n", new->start, new->end,
1052 new->policy ? new->policy->policy : 0);
1053}
1054
1055/* Find shared policy intersecting idx */
1056struct mempolicy *
1057mpol_shared_policy_lookup(struct shared_policy *sp, unsigned long idx)
1058{
1059 struct mempolicy *pol = NULL;
1060 struct sp_node *sn;
1061
1062 if (!sp->root.rb_node)
1063 return NULL;
1064 spin_lock(&sp->lock);
1065 sn = sp_lookup(sp, idx, idx+1);
1066 if (sn) {
1067 mpol_get(sn->policy);
1068 pol = sn->policy;
1069 }
1070 spin_unlock(&sp->lock);
1071 return pol;
1072}
1073
1074static void sp_delete(struct shared_policy *sp, struct sp_node *n)
1075{
1076 PDprintk("deleting %lx-l%x\n", n->start, n->end);
1077 rb_erase(&n->nd, &sp->root);
1078 mpol_free(n->policy);
1079 kmem_cache_free(sn_cache, n);
1080}
1081
1082struct sp_node *
1083sp_alloc(unsigned long start, unsigned long end, struct mempolicy *pol)
1084{
1085 struct sp_node *n = kmem_cache_alloc(sn_cache, GFP_KERNEL);
1086
1087 if (!n)
1088 return NULL;
1089 n->start = start;
1090 n->end = end;
1091 mpol_get(pol);
1092 n->policy = pol;
1093 return n;
1094}
1095
1096/* Replace a policy range. */
1097static int shared_policy_replace(struct shared_policy *sp, unsigned long start,
1098 unsigned long end, struct sp_node *new)
1099{
1100 struct sp_node *n, *new2 = NULL;
1101
1102restart:
1103 spin_lock(&sp->lock);
1104 n = sp_lookup(sp, start, end);
1105 /* Take care of old policies in the same range. */
1106 while (n && n->start < end) {
1107 struct rb_node *next = rb_next(&n->nd);
1108 if (n->start >= start) {
1109 if (n->end <= end)
1110 sp_delete(sp, n);
1111 else
1112 n->start = end;
1113 } else {
1114 /* Old policy spanning whole new range. */
1115 if (n->end > end) {
1116 if (!new2) {
1117 spin_unlock(&sp->lock);
1118 new2 = sp_alloc(end, n->end, n->policy);
1119 if (!new2)
1120 return -ENOMEM;
1121 goto restart;
1122 }
1123 n->end = start;
1124 sp_insert(sp, new2);
1125 new2 = NULL;
1126 break;
1127 } else
1128 n->end = start;
1129 }
1130 if (!next)
1131 break;
1132 n = rb_entry(next, struct sp_node, nd);
1133 }
1134 if (new)
1135 sp_insert(sp, new);
1136 spin_unlock(&sp->lock);
1137 if (new2) {
1138 mpol_free(new2->policy);
1139 kmem_cache_free(sn_cache, new2);
1140 }
1141 return 0;
1142}
1143
1144int mpol_set_shared_policy(struct shared_policy *info,
1145 struct vm_area_struct *vma, struct mempolicy *npol)
1146{
1147 int err;
1148 struct sp_node *new = NULL;
1149 unsigned long sz = vma_pages(vma);
1150
1151 PDprintk("set_shared_policy %lx sz %lu %d %lx\n",
1152 vma->vm_pgoff,
1153 sz, npol? npol->policy : -1,
Andi Kleendfcd3c02005-10-29 18:15:48 -07001154 npol ? nodes_addr(npol->v.nodes)[0] : -1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001155
1156 if (npol) {
1157 new = sp_alloc(vma->vm_pgoff, vma->vm_pgoff + sz, npol);
1158 if (!new)
1159 return -ENOMEM;
1160 }
1161 err = shared_policy_replace(info, vma->vm_pgoff, vma->vm_pgoff+sz, new);
1162 if (err && new)
1163 kmem_cache_free(sn_cache, new);
1164 return err;
1165}
1166
1167/* Free a backing policy store on inode delete. */
1168void mpol_free_shared_policy(struct shared_policy *p)
1169{
1170 struct sp_node *n;
1171 struct rb_node *next;
1172
1173 if (!p->root.rb_node)
1174 return;
1175 spin_lock(&p->lock);
1176 next = rb_first(&p->root);
1177 while (next) {
1178 n = rb_entry(next, struct sp_node, nd);
1179 next = rb_next(&n->nd);
Andi Kleen90c50292005-07-27 11:43:50 -07001180 rb_erase(&n->nd, &p->root);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001181 mpol_free(n->policy);
1182 kmem_cache_free(sn_cache, n);
1183 }
1184 spin_unlock(&p->lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001185}
1186
1187/* assumes fs == KERNEL_DS */
1188void __init numa_policy_init(void)
1189{
1190 policy_cache = kmem_cache_create("numa_policy",
1191 sizeof(struct mempolicy),
1192 0, SLAB_PANIC, NULL, NULL);
1193
1194 sn_cache = kmem_cache_create("shared_policy_node",
1195 sizeof(struct sp_node),
1196 0, SLAB_PANIC, NULL, NULL);
1197
1198 /* Set interleaving policy for system init. This way not all
1199 the data structures allocated at system boot end up in node zero. */
1200
Christoph Lameter8bccd852005-10-29 18:16:59 -07001201 if (do_set_mempolicy(MPOL_INTERLEAVE, &node_online_map))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001202 printk("numa_policy_init: interleaving failed\n");
1203}
1204
Christoph Lameter8bccd852005-10-29 18:16:59 -07001205/* Reset policy of current process to default */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001206void numa_default_policy(void)
1207{
Christoph Lameter8bccd852005-10-29 18:16:59 -07001208 do_set_mempolicy(MPOL_DEFAULT, NULL);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001209}
Paul Jackson68860ec2005-10-30 15:02:36 -08001210
1211/* Migrate a policy to a different set of nodes */
1212static void rebind_policy(struct mempolicy *pol, const nodemask_t *old,
1213 const nodemask_t *new)
1214{
1215 nodemask_t tmp;
1216
1217 if (!pol)
1218 return;
1219
1220 switch (pol->policy) {
1221 case MPOL_DEFAULT:
1222 break;
1223 case MPOL_INTERLEAVE:
1224 nodes_remap(tmp, pol->v.nodes, *old, *new);
1225 pol->v.nodes = tmp;
1226 current->il_next = node_remap(current->il_next, *old, *new);
1227 break;
1228 case MPOL_PREFERRED:
1229 pol->v.preferred_node = node_remap(pol->v.preferred_node,
1230 *old, *new);
1231 break;
1232 case MPOL_BIND: {
1233 nodemask_t nodes;
1234 struct zone **z;
1235 struct zonelist *zonelist;
1236
1237 nodes_clear(nodes);
1238 for (z = pol->v.zonelist->zones; *z; z++)
1239 node_set((*z)->zone_pgdat->node_id, nodes);
1240 nodes_remap(tmp, nodes, *old, *new);
1241 nodes = tmp;
1242
1243 zonelist = bind_zonelist(&nodes);
1244
1245 /* If no mem, then zonelist is NULL and we keep old zonelist.
1246 * If that old zonelist has no remaining mems_allowed nodes,
1247 * then zonelist_policy() will "FALL THROUGH" to MPOL_DEFAULT.
1248 */
1249
1250 if (zonelist) {
1251 /* Good - got mem - substitute new zonelist */
1252 kfree(pol->v.zonelist);
1253 pol->v.zonelist = zonelist;
1254 }
1255 break;
1256 }
1257 default:
1258 BUG();
1259 break;
1260 }
1261}
1262
1263/*
1264 * Someone moved this task to different nodes. Fixup mempolicies.
1265 *
1266 * TODO - fixup current->mm->vma and shmfs/tmpfs/hugetlbfs policies as well,
1267 * once we have a cpuset mechanism to mark which cpuset subtree is migrating.
1268 */
1269void numa_policy_rebind(const nodemask_t *old, const nodemask_t *new)
1270{
1271 rebind_policy(current->mempolicy, old, new);
1272}