Commit d4220f987cf473c65a342ca69e3eb13dea919a49
Exists in
master
and in
7 other branches
Merge branch 'hwpoison' of git://git.kernel.org/pub/scm/linux/kernel/git/ak/linux-mce-2.6
* 'hwpoison' of git://git.kernel.org/pub/scm/linux/kernel/git/ak/linux-mce-2.6: (34 commits) HWPOISON: Remove stray phrase in a comment HWPOISON: Try to allocate migration page on the same node HWPOISON: Don't do early filtering if filter is disabled HWPOISON: Add a madvise() injector for soft page offlining HWPOISON: Add soft page offline support HWPOISON: Undefine short-hand macros after use to avoid namespace conflict HWPOISON: Use new shake_page in memory_failure HWPOISON: Use correct name for MADV_HWPOISON in documentation HWPOISON: mention HWPoison in Kconfig entry HWPOISON: Use get_user_page_fast in hwpoison madvise HWPOISON: add an interface to switch off/on all the page filters HWPOISON: add memory cgroup filter memcg: add accessor to mem_cgroup.css memcg: rename and export try_get_mem_cgroup_from_page() HWPOISON: add page flags filter mm: export stable page flags HWPOISON: limit hwpoison injector to known page types HWPOISON: add fs/device filters HWPOISON: return 0 to indicate success reliably HWPOISON: make semantics of IGNORED/DELAYED clear ...
Showing 19 changed files Side-by-side Diff
- Documentation/ABI/testing/sysfs-memory-page-offline
- Documentation/vm/hwpoison.txt
- Documentation/vm/page-types.c
- MAINTAINERS
- drivers/base/memory.c
- fs/proc/page.c
- include/asm-generic/mman-common.h
- include/linux/kernel-page-flags.h
- include/linux/memcontrol.h
- include/linux/mm.h
- include/linux/page-flags.h
- mm/Kconfig
- mm/hwpoison-inject.c
- mm/internal.h
- mm/madvise.c
- mm/memcontrol.c
- mm/memory-failure.c
- mm/memory.c
- mm/page_alloc.c
Documentation/ABI/testing/sysfs-memory-page-offline
1 | +What: /sys/devices/system/memory/soft_offline_page | |
2 | +Date: Sep 2009 | |
3 | +KernelVersion: 2.6.33 | |
4 | +Contact: andi@firstfloor.org | |
5 | +Description: | |
6 | + Soft-offline the memory page containing the physical address | |
7 | + written into this file. Input is a hex number specifying the | |
8 | + physical address of the page. The kernel will then attempt | |
9 | + to soft-offline it, by moving the contents elsewhere or | |
10 | + dropping it if possible. The kernel will then be placed | |
11 | + on the bad page list and never be reused. | |
12 | + | |
13 | + The offlining is done in kernel specific granuality. | |
14 | + Normally it's the base page size of the kernel, but | |
15 | + this might change. | |
16 | + | |
17 | + The page must be still accessible, not poisoned. The | |
18 | + kernel will never kill anything for this, but rather | |
19 | + fail the offline. Return value is the size of the | |
20 | + number, or a error when the offlining failed. Reading | |
21 | + the file is not allowed. | |
22 | + | |
23 | +What: /sys/devices/system/memory/hard_offline_page | |
24 | +Date: Sep 2009 | |
25 | +KernelVersion: 2.6.33 | |
26 | +Contact: andi@firstfloor.org | |
27 | +Description: | |
28 | + Hard-offline the memory page containing the physical | |
29 | + address written into this file. Input is a hex number | |
30 | + specifying the physical address of the page. The | |
31 | + kernel will then attempt to hard-offline the page, by | |
32 | + trying to drop the page or killing any owner or | |
33 | + triggering IO errors if needed. Note this may kill | |
34 | + any processes owning the page. The kernel will avoid | |
35 | + to access this page assuming it's poisoned by the | |
36 | + hardware. | |
37 | + | |
38 | + The offlining is done in kernel specific granuality. | |
39 | + Normally it's the base page size of the kernel, but | |
40 | + this might change. | |
41 | + | |
42 | + Return value is the size of the number, or a error when | |
43 | + the offlining failed. | |
44 | + Reading the file is not allowed. |
Documentation/vm/hwpoison.txt
... | ... | @@ -92,16 +92,62 @@ |
92 | 92 | |
93 | 93 | Testing: |
94 | 94 | |
95 | -madvise(MADV_POISON, ....) | |
95 | +madvise(MADV_HWPOISON, ....) | |
96 | 96 | (as root) |
97 | 97 | Poison a page in the process for testing |
98 | 98 | |
99 | 99 | |
100 | 100 | hwpoison-inject module through debugfs |
101 | - /sys/debug/hwpoison/corrupt-pfn | |
102 | 101 | |
103 | -Inject hwpoison fault at PFN echoed into this file | |
102 | +/sys/debug/hwpoison/ | |
104 | 103 | |
104 | +corrupt-pfn | |
105 | + | |
106 | +Inject hwpoison fault at PFN echoed into this file. This does | |
107 | +some early filtering to avoid corrupted unintended pages in test suites. | |
108 | + | |
109 | +unpoison-pfn | |
110 | + | |
111 | +Software-unpoison page at PFN echoed into this file. This | |
112 | +way a page can be reused again. | |
113 | +This only works for Linux injected failures, not for real | |
114 | +memory failures. | |
115 | + | |
116 | +Note these injection interfaces are not stable and might change between | |
117 | +kernel versions | |
118 | + | |
119 | +corrupt-filter-dev-major | |
120 | +corrupt-filter-dev-minor | |
121 | + | |
122 | +Only handle memory failures to pages associated with the file system defined | |
123 | +by block device major/minor. -1U is the wildcard value. | |
124 | +This should be only used for testing with artificial injection. | |
125 | + | |
126 | +corrupt-filter-memcg | |
127 | + | |
128 | +Limit injection to pages owned by memgroup. Specified by inode number | |
129 | +of the memcg. | |
130 | + | |
131 | +Example: | |
132 | + mkdir /cgroup/hwpoison | |
133 | + | |
134 | + usemem -m 100 -s 1000 & | |
135 | + echo `jobs -p` > /cgroup/hwpoison/tasks | |
136 | + | |
137 | + memcg_ino=$(ls -id /cgroup/hwpoison | cut -f1 -d' ') | |
138 | + echo $memcg_ino > /debug/hwpoison/corrupt-filter-memcg | |
139 | + | |
140 | + page-types -p `pidof init` --hwpoison # shall do nothing | |
141 | + page-types -p `pidof usemem` --hwpoison # poison its pages | |
142 | + | |
143 | +corrupt-filter-flags-mask | |
144 | +corrupt-filter-flags-value | |
145 | + | |
146 | +When specified, only poison pages if ((page_flags & mask) == value). | |
147 | +This allows stress testing of many kinds of pages. The page_flags | |
148 | +are the same as in /proc/kpageflags. The flag bits are defined in | |
149 | +include/linux/kernel-page-flags.h and documented in | |
150 | +Documentation/vm/pagemap.txt | |
105 | 151 | |
106 | 152 | Architecture specific MCE injector |
107 | 153 |
Documentation/vm/page-types.c
1 | 1 | /* |
2 | 2 | * page-types: Tool for querying page flags |
3 | 3 | * |
4 | + * This program is free software; you can redistribute it and/or modify it | |
5 | + * under the terms of the GNU General Public License as published by the Free | |
6 | + * Software Foundation; version 2. | |
7 | + * | |
8 | + * This program is distributed in the hope that it will be useful, but WITHOUT | |
9 | + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or | |
10 | + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for | |
11 | + * more details. | |
12 | + * | |
13 | + * You should find a copy of v2 of the GNU General Public License somewhere on | |
14 | + * your Linux system; if not, write to the Free Software Foundation, Inc., 59 | |
15 | + * Temple Place, Suite 330, Boston, MA 02111-1307 USA. | |
16 | + * | |
4 | 17 | * Copyright (C) 2009 Intel corporation |
5 | 18 | * |
6 | 19 | * Authors: Wu Fengguang <fengguang.wu@intel.com> |
7 | - * | |
8 | - * Released under the General Public License (GPL). | |
9 | 20 | */ |
10 | 21 | |
11 | 22 | #define _LARGEFILE64_SOURCE |
MAINTAINERS
... | ... | @@ -2377,6 +2377,15 @@ |
2377 | 2377 | S: Maintained |
2378 | 2378 | F: drivers/hwmon/hdaps.c |
2379 | 2379 | |
2380 | +HWPOISON MEMORY FAILURE HANDLING | |
2381 | +M: Andi Kleen <andi@firstfloor.org> | |
2382 | +L: linux-mm@kvack.org | |
2383 | +L: linux-kernel@vger.kernel.org | |
2384 | +T: git git://git.kernel.org/pub/scm/linux/kernel/git/ak/linux-mce-2.6.git hwpoison | |
2385 | +S: Maintained | |
2386 | +F: mm/memory-failure.c | |
2387 | +F: mm/hwpoison-inject.c | |
2388 | + | |
2380 | 2389 | HYPERVISOR VIRTUAL CONSOLE DRIVER |
2381 | 2390 | L: linuxppc-dev@ozlabs.org |
2382 | 2391 | S: Odd Fixes |
drivers/base/memory.c
... | ... | @@ -341,7 +341,65 @@ |
341 | 341 | } |
342 | 342 | #endif |
343 | 343 | |
344 | +#ifdef CONFIG_MEMORY_FAILURE | |
344 | 345 | /* |
346 | + * Support for offlining pages of memory | |
347 | + */ | |
348 | + | |
349 | +/* Soft offline a page */ | |
350 | +static ssize_t | |
351 | +store_soft_offline_page(struct class *class, const char *buf, size_t count) | |
352 | +{ | |
353 | + int ret; | |
354 | + u64 pfn; | |
355 | + if (!capable(CAP_SYS_ADMIN)) | |
356 | + return -EPERM; | |
357 | + if (strict_strtoull(buf, 0, &pfn) < 0) | |
358 | + return -EINVAL; | |
359 | + pfn >>= PAGE_SHIFT; | |
360 | + if (!pfn_valid(pfn)) | |
361 | + return -ENXIO; | |
362 | + ret = soft_offline_page(pfn_to_page(pfn), 0); | |
363 | + return ret == 0 ? count : ret; | |
364 | +} | |
365 | + | |
366 | +/* Forcibly offline a page, including killing processes. */ | |
367 | +static ssize_t | |
368 | +store_hard_offline_page(struct class *class, const char *buf, size_t count) | |
369 | +{ | |
370 | + int ret; | |
371 | + u64 pfn; | |
372 | + if (!capable(CAP_SYS_ADMIN)) | |
373 | + return -EPERM; | |
374 | + if (strict_strtoull(buf, 0, &pfn) < 0) | |
375 | + return -EINVAL; | |
376 | + pfn >>= PAGE_SHIFT; | |
377 | + ret = __memory_failure(pfn, 0, 0); | |
378 | + return ret ? ret : count; | |
379 | +} | |
380 | + | |
381 | +static CLASS_ATTR(soft_offline_page, 0644, NULL, store_soft_offline_page); | |
382 | +static CLASS_ATTR(hard_offline_page, 0644, NULL, store_hard_offline_page); | |
383 | + | |
384 | +static __init int memory_fail_init(void) | |
385 | +{ | |
386 | + int err; | |
387 | + | |
388 | + err = sysfs_create_file(&memory_sysdev_class.kset.kobj, | |
389 | + &class_attr_soft_offline_page.attr); | |
390 | + if (!err) | |
391 | + err = sysfs_create_file(&memory_sysdev_class.kset.kobj, | |
392 | + &class_attr_hard_offline_page.attr); | |
393 | + return err; | |
394 | +} | |
395 | +#else | |
396 | +static inline int memory_fail_init(void) | |
397 | +{ | |
398 | + return 0; | |
399 | +} | |
400 | +#endif | |
401 | + | |
402 | +/* | |
345 | 403 | * Note that phys_device is optional. It is here to allow for |
346 | 404 | * differentiation between which *physical* devices each |
347 | 405 | * section belongs to... |
... | ... | @@ -471,6 +529,9 @@ |
471 | 529 | } |
472 | 530 | |
473 | 531 | err = memory_probe_init(); |
532 | + if (!ret) | |
533 | + ret = err; | |
534 | + err = memory_fail_init(); | |
474 | 535 | if (!ret) |
475 | 536 | ret = err; |
476 | 537 | err = block_size_init(); |
fs/proc/page.c
... | ... | @@ -8,6 +8,7 @@ |
8 | 8 | #include <linux/proc_fs.h> |
9 | 9 | #include <linux/seq_file.h> |
10 | 10 | #include <linux/hugetlb.h> |
11 | +#include <linux/kernel-page-flags.h> | |
11 | 12 | #include <asm/uaccess.h> |
12 | 13 | #include "internal.h" |
13 | 14 | |
14 | 15 | |
... | ... | @@ -71,52 +72,12 @@ |
71 | 72 | * physical page flags. |
72 | 73 | */ |
73 | 74 | |
74 | -/* These macros are used to decouple internal flags from exported ones */ | |
75 | - | |
76 | -#define KPF_LOCKED 0 | |
77 | -#define KPF_ERROR 1 | |
78 | -#define KPF_REFERENCED 2 | |
79 | -#define KPF_UPTODATE 3 | |
80 | -#define KPF_DIRTY 4 | |
81 | -#define KPF_LRU 5 | |
82 | -#define KPF_ACTIVE 6 | |
83 | -#define KPF_SLAB 7 | |
84 | -#define KPF_WRITEBACK 8 | |
85 | -#define KPF_RECLAIM 9 | |
86 | -#define KPF_BUDDY 10 | |
87 | - | |
88 | -/* 11-20: new additions in 2.6.31 */ | |
89 | -#define KPF_MMAP 11 | |
90 | -#define KPF_ANON 12 | |
91 | -#define KPF_SWAPCACHE 13 | |
92 | -#define KPF_SWAPBACKED 14 | |
93 | -#define KPF_COMPOUND_HEAD 15 | |
94 | -#define KPF_COMPOUND_TAIL 16 | |
95 | -#define KPF_HUGE 17 | |
96 | -#define KPF_UNEVICTABLE 18 | |
97 | -#define KPF_HWPOISON 19 | |
98 | -#define KPF_NOPAGE 20 | |
99 | - | |
100 | -#define KPF_KSM 21 | |
101 | - | |
102 | -/* kernel hacking assistances | |
103 | - * WARNING: subject to change, never rely on them! | |
104 | - */ | |
105 | -#define KPF_RESERVED 32 | |
106 | -#define KPF_MLOCKED 33 | |
107 | -#define KPF_MAPPEDTODISK 34 | |
108 | -#define KPF_PRIVATE 35 | |
109 | -#define KPF_PRIVATE_2 36 | |
110 | -#define KPF_OWNER_PRIVATE 37 | |
111 | -#define KPF_ARCH 38 | |
112 | -#define KPF_UNCACHED 39 | |
113 | - | |
114 | 75 | static inline u64 kpf_copy_bit(u64 kflags, int ubit, int kbit) |
115 | 76 | { |
116 | 77 | return ((kflags >> kbit) & 1) << ubit; |
117 | 78 | } |
118 | 79 | |
119 | -static u64 get_uflags(struct page *page) | |
80 | +u64 stable_page_flags(struct page *page) | |
120 | 81 | { |
121 | 82 | u64 k; |
122 | 83 | u64 u; |
... | ... | @@ -219,7 +180,7 @@ |
219 | 180 | else |
220 | 181 | ppage = NULL; |
221 | 182 | |
222 | - if (put_user(get_uflags(ppage), out)) { | |
183 | + if (put_user(stable_page_flags(ppage), out)) { | |
223 | 184 | ret = -EFAULT; |
224 | 185 | break; |
225 | 186 | } |
include/asm-generic/mman-common.h
... | ... | @@ -40,6 +40,7 @@ |
40 | 40 | #define MADV_DONTFORK 10 /* don't inherit across fork */ |
41 | 41 | #define MADV_DOFORK 11 /* do inherit across fork */ |
42 | 42 | #define MADV_HWPOISON 100 /* poison a page for testing */ |
43 | +#define MADV_SOFT_OFFLINE 101 /* soft offline page for testing */ | |
43 | 44 | |
44 | 45 | #define MADV_MERGEABLE 12 /* KSM may merge identical pages */ |
45 | 46 | #define MADV_UNMERGEABLE 13 /* KSM may not merge identical pages */ |
include/linux/kernel-page-flags.h
1 | +#ifndef LINUX_KERNEL_PAGE_FLAGS_H | |
2 | +#define LINUX_KERNEL_PAGE_FLAGS_H | |
3 | + | |
4 | +/* | |
5 | + * Stable page flag bits exported to user space | |
6 | + */ | |
7 | + | |
8 | +#define KPF_LOCKED 0 | |
9 | +#define KPF_ERROR 1 | |
10 | +#define KPF_REFERENCED 2 | |
11 | +#define KPF_UPTODATE 3 | |
12 | +#define KPF_DIRTY 4 | |
13 | +#define KPF_LRU 5 | |
14 | +#define KPF_ACTIVE 6 | |
15 | +#define KPF_SLAB 7 | |
16 | +#define KPF_WRITEBACK 8 | |
17 | +#define KPF_RECLAIM 9 | |
18 | +#define KPF_BUDDY 10 | |
19 | + | |
20 | +/* 11-20: new additions in 2.6.31 */ | |
21 | +#define KPF_MMAP 11 | |
22 | +#define KPF_ANON 12 | |
23 | +#define KPF_SWAPCACHE 13 | |
24 | +#define KPF_SWAPBACKED 14 | |
25 | +#define KPF_COMPOUND_HEAD 15 | |
26 | +#define KPF_COMPOUND_TAIL 16 | |
27 | +#define KPF_HUGE 17 | |
28 | +#define KPF_UNEVICTABLE 18 | |
29 | +#define KPF_HWPOISON 19 | |
30 | +#define KPF_NOPAGE 20 | |
31 | + | |
32 | +#define KPF_KSM 21 | |
33 | + | |
34 | +/* kernel hacking assistances | |
35 | + * WARNING: subject to change, never rely on them! | |
36 | + */ | |
37 | +#define KPF_RESERVED 32 | |
38 | +#define KPF_MLOCKED 33 | |
39 | +#define KPF_MAPPEDTODISK 34 | |
40 | +#define KPF_PRIVATE 35 | |
41 | +#define KPF_PRIVATE_2 36 | |
42 | +#define KPF_OWNER_PRIVATE 37 | |
43 | +#define KPF_ARCH 38 | |
44 | +#define KPF_UNCACHED 39 | |
45 | + | |
46 | +#endif /* LINUX_KERNEL_PAGE_FLAGS_H */ |
include/linux/memcontrol.h
... | ... | @@ -73,6 +73,7 @@ |
73 | 73 | extern void mem_cgroup_out_of_memory(struct mem_cgroup *mem, gfp_t gfp_mask); |
74 | 74 | int task_in_mem_cgroup(struct task_struct *task, const struct mem_cgroup *mem); |
75 | 75 | |
76 | +extern struct mem_cgroup *try_get_mem_cgroup_from_page(struct page *page); | |
76 | 77 | extern struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p); |
77 | 78 | |
78 | 79 | static inline |
... | ... | @@ -85,6 +86,8 @@ |
85 | 86 | return cgroup == mem; |
86 | 87 | } |
87 | 88 | |
89 | +extern struct cgroup_subsys_state *mem_cgroup_css(struct mem_cgroup *mem); | |
90 | + | |
88 | 91 | extern int |
89 | 92 | mem_cgroup_prepare_migration(struct page *page, struct mem_cgroup **ptr); |
90 | 93 | extern void mem_cgroup_end_migration(struct mem_cgroup *mem, |
... | ... | @@ -202,6 +205,11 @@ |
202 | 205 | { |
203 | 206 | } |
204 | 207 | |
208 | +static inline struct mem_cgroup *try_get_mem_cgroup_from_page(struct page *page) | |
209 | +{ | |
210 | + return NULL; | |
211 | +} | |
212 | + | |
205 | 213 | static inline int mm_match_cgroup(struct mm_struct *mm, struct mem_cgroup *mem) |
206 | 214 | { |
207 | 215 | return 1; |
... | ... | @@ -211,6 +219,11 @@ |
211 | 219 | const struct mem_cgroup *mem) |
212 | 220 | { |
213 | 221 | return 1; |
222 | +} | |
223 | + | |
224 | +static inline struct cgroup_subsys_state *mem_cgroup_css(struct mem_cgroup *mem) | |
225 | +{ | |
226 | + return NULL; | |
214 | 227 | } |
215 | 228 | |
216 | 229 | static inline int |
include/linux/mm.h
... | ... | @@ -1331,11 +1331,17 @@ |
1331 | 1331 | size_t size); |
1332 | 1332 | extern void refund_locked_memory(struct mm_struct *mm, size_t size); |
1333 | 1333 | |
1334 | +enum mf_flags { | |
1335 | + MF_COUNT_INCREASED = 1 << 0, | |
1336 | +}; | |
1334 | 1337 | extern void memory_failure(unsigned long pfn, int trapno); |
1335 | -extern int __memory_failure(unsigned long pfn, int trapno, int ref); | |
1338 | +extern int __memory_failure(unsigned long pfn, int trapno, int flags); | |
1339 | +extern int unpoison_memory(unsigned long pfn); | |
1336 | 1340 | extern int sysctl_memory_failure_early_kill; |
1337 | 1341 | extern int sysctl_memory_failure_recovery; |
1342 | +extern void shake_page(struct page *p, int access); | |
1338 | 1343 | extern atomic_long_t mce_bad_pages; |
1344 | +extern int soft_offline_page(struct page *page, int flags); | |
1339 | 1345 | |
1340 | 1346 | #endif /* __KERNEL__ */ |
1341 | 1347 | #endif /* _LINUX_MM_H */ |
include/linux/page-flags.h
... | ... | @@ -275,12 +275,14 @@ |
275 | 275 | |
276 | 276 | #ifdef CONFIG_MEMORY_FAILURE |
277 | 277 | PAGEFLAG(HWPoison, hwpoison) |
278 | -TESTSETFLAG(HWPoison, hwpoison) | |
278 | +TESTSCFLAG(HWPoison, hwpoison) | |
279 | 279 | #define __PG_HWPOISON (1UL << PG_hwpoison) |
280 | 280 | #else |
281 | 281 | PAGEFLAG_FALSE(HWPoison) |
282 | 282 | #define __PG_HWPOISON 0 |
283 | 283 | #endif |
284 | + | |
285 | +u64 stable_page_flags(struct page *page); | |
284 | 286 | |
285 | 287 | static inline int PageUptodate(struct page *page) |
286 | 288 | { |
mm/Kconfig
... | ... | @@ -251,8 +251,9 @@ |
251 | 251 | special hardware support and typically ECC memory. |
252 | 252 | |
253 | 253 | config HWPOISON_INJECT |
254 | - tristate "Poison pages injector" | |
254 | + tristate "HWPoison pages injector" | |
255 | 255 | depends on MEMORY_FAILURE && DEBUG_KERNEL |
256 | + select PROC_PAGE_MONITOR | |
256 | 257 | |
257 | 258 | config NOMMU_INITIAL_TRIM_EXCESS |
258 | 259 | int "Turn on mmap() excess space trimming before booting" |
mm/hwpoison-inject.c
... | ... | @@ -3,18 +3,68 @@ |
3 | 3 | #include <linux/debugfs.h> |
4 | 4 | #include <linux/kernel.h> |
5 | 5 | #include <linux/mm.h> |
6 | +#include <linux/swap.h> | |
7 | +#include <linux/pagemap.h> | |
8 | +#include "internal.h" | |
6 | 9 | |
7 | -static struct dentry *hwpoison_dir, *corrupt_pfn; | |
10 | +static struct dentry *hwpoison_dir; | |
8 | 11 | |
9 | 12 | static int hwpoison_inject(void *data, u64 val) |
10 | 13 | { |
14 | + unsigned long pfn = val; | |
15 | + struct page *p; | |
16 | + int err; | |
17 | + | |
11 | 18 | if (!capable(CAP_SYS_ADMIN)) |
12 | 19 | return -EPERM; |
13 | - printk(KERN_INFO "Injecting memory failure at pfn %Lx\n", val); | |
14 | - return __memory_failure(val, 18, 0); | |
20 | + | |
21 | + if (!hwpoison_filter_enable) | |
22 | + goto inject; | |
23 | + if (!pfn_valid(pfn)) | |
24 | + return -ENXIO; | |
25 | + | |
26 | + p = pfn_to_page(pfn); | |
27 | + /* | |
28 | + * This implies unable to support free buddy pages. | |
29 | + */ | |
30 | + if (!get_page_unless_zero(p)) | |
31 | + return 0; | |
32 | + | |
33 | + if (!PageLRU(p)) | |
34 | + shake_page(p, 0); | |
35 | + /* | |
36 | + * This implies unable to support non-LRU pages. | |
37 | + */ | |
38 | + if (!PageLRU(p)) | |
39 | + return 0; | |
40 | + | |
41 | + /* | |
42 | + * do a racy check with elevated page count, to make sure PG_hwpoison | |
43 | + * will only be set for the targeted owner (or on a free page). | |
44 | + * We temporarily take page lock for try_get_mem_cgroup_from_page(). | |
45 | + * __memory_failure() will redo the check reliably inside page lock. | |
46 | + */ | |
47 | + lock_page(p); | |
48 | + err = hwpoison_filter(p); | |
49 | + unlock_page(p); | |
50 | + if (err) | |
51 | + return 0; | |
52 | + | |
53 | +inject: | |
54 | + printk(KERN_INFO "Injecting memory failure at pfn %lx\n", pfn); | |
55 | + return __memory_failure(pfn, 18, MF_COUNT_INCREASED); | |
15 | 56 | } |
16 | 57 | |
58 | +static int hwpoison_unpoison(void *data, u64 val) | |
59 | +{ | |
60 | + if (!capable(CAP_SYS_ADMIN)) | |
61 | + return -EPERM; | |
62 | + | |
63 | + return unpoison_memory(val); | |
64 | +} | |
65 | + | |
17 | 66 | DEFINE_SIMPLE_ATTRIBUTE(hwpoison_fops, NULL, hwpoison_inject, "%lli\n"); |
67 | +DEFINE_SIMPLE_ATTRIBUTE(unpoison_fops, NULL, hwpoison_unpoison, "%lli\n"); | |
18 | 68 | |
19 | 69 | static void pfn_inject_exit(void) |
20 | 70 | { |
21 | 71 | |
22 | 72 | |
23 | 73 | |
... | ... | @@ -24,16 +74,63 @@ |
24 | 74 | |
25 | 75 | static int pfn_inject_init(void) |
26 | 76 | { |
77 | + struct dentry *dentry; | |
78 | + | |
27 | 79 | hwpoison_dir = debugfs_create_dir("hwpoison", NULL); |
28 | 80 | if (hwpoison_dir == NULL) |
29 | 81 | return -ENOMEM; |
30 | - corrupt_pfn = debugfs_create_file("corrupt-pfn", 0600, hwpoison_dir, | |
82 | + | |
83 | + /* | |
84 | + * Note that the below poison/unpoison interfaces do not involve | |
85 | + * hardware status change, hence do not require hardware support. | |
86 | + * They are mainly for testing hwpoison in software level. | |
87 | + */ | |
88 | + dentry = debugfs_create_file("corrupt-pfn", 0600, hwpoison_dir, | |
31 | 89 | NULL, &hwpoison_fops); |
32 | - if (corrupt_pfn == NULL) { | |
33 | - pfn_inject_exit(); | |
34 | - return -ENOMEM; | |
35 | - } | |
90 | + if (!dentry) | |
91 | + goto fail; | |
92 | + | |
93 | + dentry = debugfs_create_file("unpoison-pfn", 0600, hwpoison_dir, | |
94 | + NULL, &unpoison_fops); | |
95 | + if (!dentry) | |
96 | + goto fail; | |
97 | + | |
98 | + dentry = debugfs_create_u32("corrupt-filter-enable", 0600, | |
99 | + hwpoison_dir, &hwpoison_filter_enable); | |
100 | + if (!dentry) | |
101 | + goto fail; | |
102 | + | |
103 | + dentry = debugfs_create_u32("corrupt-filter-dev-major", 0600, | |
104 | + hwpoison_dir, &hwpoison_filter_dev_major); | |
105 | + if (!dentry) | |
106 | + goto fail; | |
107 | + | |
108 | + dentry = debugfs_create_u32("corrupt-filter-dev-minor", 0600, | |
109 | + hwpoison_dir, &hwpoison_filter_dev_minor); | |
110 | + if (!dentry) | |
111 | + goto fail; | |
112 | + | |
113 | + dentry = debugfs_create_u64("corrupt-filter-flags-mask", 0600, | |
114 | + hwpoison_dir, &hwpoison_filter_flags_mask); | |
115 | + if (!dentry) | |
116 | + goto fail; | |
117 | + | |
118 | + dentry = debugfs_create_u64("corrupt-filter-flags-value", 0600, | |
119 | + hwpoison_dir, &hwpoison_filter_flags_value); | |
120 | + if (!dentry) | |
121 | + goto fail; | |
122 | + | |
123 | +#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP | |
124 | + dentry = debugfs_create_u64("corrupt-filter-memcg", 0600, | |
125 | + hwpoison_dir, &hwpoison_filter_memcg); | |
126 | + if (!dentry) | |
127 | + goto fail; | |
128 | +#endif | |
129 | + | |
36 | 130 | return 0; |
131 | +fail: | |
132 | + pfn_inject_exit(); | |
133 | + return -ENOMEM; | |
37 | 134 | } |
38 | 135 | |
39 | 136 | module_init(pfn_inject_init); |
mm/internal.h
... | ... | @@ -50,6 +50,9 @@ |
50 | 50 | */ |
51 | 51 | extern void __free_pages_bootmem(struct page *page, unsigned int order); |
52 | 52 | extern void prep_compound_page(struct page *page, unsigned long order); |
53 | +#ifdef CONFIG_MEMORY_FAILURE | |
54 | +extern bool is_free_buddy_page(struct page *page); | |
55 | +#endif | |
53 | 56 | |
54 | 57 | |
55 | 58 | /* |
... | ... | @@ -247,4 +250,13 @@ |
247 | 250 | #define ZONE_RECLAIM_SOME 0 |
248 | 251 | #define ZONE_RECLAIM_SUCCESS 1 |
249 | 252 | #endif |
253 | + | |
254 | +extern int hwpoison_filter(struct page *p); | |
255 | + | |
256 | +extern u32 hwpoison_filter_dev_major; | |
257 | +extern u32 hwpoison_filter_dev_minor; | |
258 | +extern u64 hwpoison_filter_flags_mask; | |
259 | +extern u64 hwpoison_filter_flags_value; | |
260 | +extern u64 hwpoison_filter_memcg; | |
261 | +extern u32 hwpoison_filter_enable; |
mm/madvise.c
... | ... | @@ -9,6 +9,7 @@ |
9 | 9 | #include <linux/pagemap.h> |
10 | 10 | #include <linux/syscalls.h> |
11 | 11 | #include <linux/mempolicy.h> |
12 | +#include <linux/page-isolation.h> | |
12 | 13 | #include <linux/hugetlb.h> |
13 | 14 | #include <linux/sched.h> |
14 | 15 | #include <linux/ksm.h> |
... | ... | @@ -222,7 +223,7 @@ |
222 | 223 | /* |
223 | 224 | * Error injection support for memory error handling. |
224 | 225 | */ |
225 | -static int madvise_hwpoison(unsigned long start, unsigned long end) | |
226 | +static int madvise_hwpoison(int bhv, unsigned long start, unsigned long end) | |
226 | 227 | { |
227 | 228 | int ret = 0; |
228 | 229 | |
229 | 230 | |
230 | 231 | |
... | ... | @@ -230,15 +231,21 @@ |
230 | 231 | return -EPERM; |
231 | 232 | for (; start < end; start += PAGE_SIZE) { |
232 | 233 | struct page *p; |
233 | - int ret = get_user_pages(current, current->mm, start, 1, | |
234 | - 0, 0, &p, NULL); | |
234 | + int ret = get_user_pages_fast(start, 1, 0, &p); | |
235 | 235 | if (ret != 1) |
236 | 236 | return ret; |
237 | + if (bhv == MADV_SOFT_OFFLINE) { | |
238 | + printk(KERN_INFO "Soft offlining page %lx at %lx\n", | |
239 | + page_to_pfn(p), start); | |
240 | + ret = soft_offline_page(p, MF_COUNT_INCREASED); | |
241 | + if (ret) | |
242 | + break; | |
243 | + continue; | |
244 | + } | |
237 | 245 | printk(KERN_INFO "Injecting memory failure for page %lx at %lx\n", |
238 | 246 | page_to_pfn(p), start); |
239 | 247 | /* Ignore return value for now */ |
240 | - __memory_failure(page_to_pfn(p), 0, 1); | |
241 | - put_page(p); | |
248 | + __memory_failure(page_to_pfn(p), 0, MF_COUNT_INCREASED); | |
242 | 249 | } |
243 | 250 | return ret; |
244 | 251 | } |
... | ... | @@ -335,8 +342,8 @@ |
335 | 342 | size_t len; |
336 | 343 | |
337 | 344 | #ifdef CONFIG_MEMORY_FAILURE |
338 | - if (behavior == MADV_HWPOISON) | |
339 | - return madvise_hwpoison(start, start+len_in); | |
345 | + if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE) | |
346 | + return madvise_hwpoison(behavior, start, start+len_in); | |
340 | 347 | #endif |
341 | 348 | if (!madvise_behavior_valid(behavior)) |
342 | 349 | return error; |
mm/memcontrol.c
... | ... | @@ -283,6 +283,11 @@ |
283 | 283 | return &mem->info.nodeinfo[nid]->zoneinfo[zid]; |
284 | 284 | } |
285 | 285 | |
286 | +struct cgroup_subsys_state *mem_cgroup_css(struct mem_cgroup *mem) | |
287 | +{ | |
288 | + return &mem->css; | |
289 | +} | |
290 | + | |
286 | 291 | static struct mem_cgroup_per_zone * |
287 | 292 | page_cgroup_zoneinfo(struct page_cgroup *pc) |
288 | 293 | { |
289 | 294 | |
290 | 295 | |
291 | 296 | |
... | ... | @@ -1536,25 +1541,22 @@ |
1536 | 1541 | return container_of(css, struct mem_cgroup, css); |
1537 | 1542 | } |
1538 | 1543 | |
1539 | -static struct mem_cgroup *try_get_mem_cgroup_from_swapcache(struct page *page) | |
1544 | +struct mem_cgroup *try_get_mem_cgroup_from_page(struct page *page) | |
1540 | 1545 | { |
1541 | - struct mem_cgroup *mem; | |
1546 | + struct mem_cgroup *mem = NULL; | |
1542 | 1547 | struct page_cgroup *pc; |
1543 | 1548 | unsigned short id; |
1544 | 1549 | swp_entry_t ent; |
1545 | 1550 | |
1546 | 1551 | VM_BUG_ON(!PageLocked(page)); |
1547 | 1552 | |
1548 | - if (!PageSwapCache(page)) | |
1549 | - return NULL; | |
1550 | - | |
1551 | 1553 | pc = lookup_page_cgroup(page); |
1552 | 1554 | lock_page_cgroup(pc); |
1553 | 1555 | if (PageCgroupUsed(pc)) { |
1554 | 1556 | mem = pc->mem_cgroup; |
1555 | 1557 | if (mem && !css_tryget(&mem->css)) |
1556 | 1558 | mem = NULL; |
1557 | - } else { | |
1559 | + } else if (PageSwapCache(page)) { | |
1558 | 1560 | ent.val = page_private(page); |
1559 | 1561 | id = lookup_swap_cgroup(ent); |
1560 | 1562 | rcu_read_lock(); |
... | ... | @@ -1874,7 +1876,7 @@ |
1874 | 1876 | */ |
1875 | 1877 | if (!PageSwapCache(page)) |
1876 | 1878 | goto charge_cur_mm; |
1877 | - mem = try_get_mem_cgroup_from_swapcache(page); | |
1879 | + mem = try_get_mem_cgroup_from_page(page); | |
1878 | 1880 | if (!mem) |
1879 | 1881 | goto charge_cur_mm; |
1880 | 1882 | *ptr = mem; |
mm/memory-failure.c
... | ... | @@ -34,12 +34,16 @@ |
34 | 34 | #include <linux/kernel.h> |
35 | 35 | #include <linux/mm.h> |
36 | 36 | #include <linux/page-flags.h> |
37 | +#include <linux/kernel-page-flags.h> | |
37 | 38 | #include <linux/sched.h> |
38 | 39 | #include <linux/ksm.h> |
39 | 40 | #include <linux/rmap.h> |
40 | 41 | #include <linux/pagemap.h> |
41 | 42 | #include <linux/swap.h> |
42 | 43 | #include <linux/backing-dev.h> |
44 | +#include <linux/migrate.h> | |
45 | +#include <linux/page-isolation.h> | |
46 | +#include <linux/suspend.h> | |
43 | 47 | #include "internal.h" |
44 | 48 | |
45 | 49 | int sysctl_memory_failure_early_kill __read_mostly = 0; |
46 | 50 | |
... | ... | @@ -48,7 +52,121 @@ |
48 | 52 | |
49 | 53 | atomic_long_t mce_bad_pages __read_mostly = ATOMIC_LONG_INIT(0); |
50 | 54 | |
55 | +u32 hwpoison_filter_enable = 0; | |
56 | +u32 hwpoison_filter_dev_major = ~0U; | |
57 | +u32 hwpoison_filter_dev_minor = ~0U; | |
58 | +u64 hwpoison_filter_flags_mask; | |
59 | +u64 hwpoison_filter_flags_value; | |
60 | +EXPORT_SYMBOL_GPL(hwpoison_filter_enable); | |
61 | +EXPORT_SYMBOL_GPL(hwpoison_filter_dev_major); | |
62 | +EXPORT_SYMBOL_GPL(hwpoison_filter_dev_minor); | |
63 | +EXPORT_SYMBOL_GPL(hwpoison_filter_flags_mask); | |
64 | +EXPORT_SYMBOL_GPL(hwpoison_filter_flags_value); | |
65 | + | |
66 | +static int hwpoison_filter_dev(struct page *p) | |
67 | +{ | |
68 | + struct address_space *mapping; | |
69 | + dev_t dev; | |
70 | + | |
71 | + if (hwpoison_filter_dev_major == ~0U && | |
72 | + hwpoison_filter_dev_minor == ~0U) | |
73 | + return 0; | |
74 | + | |
75 | + /* | |
76 | + * page_mapping() does not accept slab page | |
77 | + */ | |
78 | + if (PageSlab(p)) | |
79 | + return -EINVAL; | |
80 | + | |
81 | + mapping = page_mapping(p); | |
82 | + if (mapping == NULL || mapping->host == NULL) | |
83 | + return -EINVAL; | |
84 | + | |
85 | + dev = mapping->host->i_sb->s_dev; | |
86 | + if (hwpoison_filter_dev_major != ~0U && | |
87 | + hwpoison_filter_dev_major != MAJOR(dev)) | |
88 | + return -EINVAL; | |
89 | + if (hwpoison_filter_dev_minor != ~0U && | |
90 | + hwpoison_filter_dev_minor != MINOR(dev)) | |
91 | + return -EINVAL; | |
92 | + | |
93 | + return 0; | |
94 | +} | |
95 | + | |
96 | +static int hwpoison_filter_flags(struct page *p) | |
97 | +{ | |
98 | + if (!hwpoison_filter_flags_mask) | |
99 | + return 0; | |
100 | + | |
101 | + if ((stable_page_flags(p) & hwpoison_filter_flags_mask) == | |
102 | + hwpoison_filter_flags_value) | |
103 | + return 0; | |
104 | + else | |
105 | + return -EINVAL; | |
106 | +} | |
107 | + | |
51 | 108 | /* |
109 | + * This allows stress tests to limit test scope to a collection of tasks | |
110 | + * by putting them under some memcg. This prevents killing unrelated/important | |
111 | + * processes such as /sbin/init. Note that the target task may share clean | |
112 | + * pages with init (eg. libc text), which is harmless. If the target task | |
113 | + * share _dirty_ pages with another task B, the test scheme must make sure B | |
114 | + * is also included in the memcg. At last, due to race conditions this filter | |
115 | + * can only guarantee that the page either belongs to the memcg tasks, or is | |
116 | + * a freed page. | |
117 | + */ | |
118 | +#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP | |
119 | +u64 hwpoison_filter_memcg; | |
120 | +EXPORT_SYMBOL_GPL(hwpoison_filter_memcg); | |
121 | +static int hwpoison_filter_task(struct page *p) | |
122 | +{ | |
123 | + struct mem_cgroup *mem; | |
124 | + struct cgroup_subsys_state *css; | |
125 | + unsigned long ino; | |
126 | + | |
127 | + if (!hwpoison_filter_memcg) | |
128 | + return 0; | |
129 | + | |
130 | + mem = try_get_mem_cgroup_from_page(p); | |
131 | + if (!mem) | |
132 | + return -EINVAL; | |
133 | + | |
134 | + css = mem_cgroup_css(mem); | |
135 | + /* root_mem_cgroup has NULL dentries */ | |
136 | + if (!css->cgroup->dentry) | |
137 | + return -EINVAL; | |
138 | + | |
139 | + ino = css->cgroup->dentry->d_inode->i_ino; | |
140 | + css_put(css); | |
141 | + | |
142 | + if (ino != hwpoison_filter_memcg) | |
143 | + return -EINVAL; | |
144 | + | |
145 | + return 0; | |
146 | +} | |
147 | +#else | |
148 | +static int hwpoison_filter_task(struct page *p) { return 0; } | |
149 | +#endif | |
150 | + | |
151 | +int hwpoison_filter(struct page *p) | |
152 | +{ | |
153 | + if (!hwpoison_filter_enable) | |
154 | + return 0; | |
155 | + | |
156 | + if (hwpoison_filter_dev(p)) | |
157 | + return -EINVAL; | |
158 | + | |
159 | + if (hwpoison_filter_flags(p)) | |
160 | + return -EINVAL; | |
161 | + | |
162 | + if (hwpoison_filter_task(p)) | |
163 | + return -EINVAL; | |
164 | + | |
165 | + return 0; | |
166 | +} | |
167 | +EXPORT_SYMBOL_GPL(hwpoison_filter); | |
168 | + | |
169 | +/* | |
52 | 170 | * Send all the processes who have the page mapped an ``action optional'' |
53 | 171 | * signal. |
54 | 172 | */ |
... | ... | @@ -83,6 +201,36 @@ |
83 | 201 | } |
84 | 202 | |
85 | 203 | /* |
204 | + * When a unknown page type is encountered drain as many buffers as possible | |
205 | + * in the hope to turn the page into a LRU or free page, which we can handle. | |
206 | + */ | |
207 | +void shake_page(struct page *p, int access) | |
208 | +{ | |
209 | + if (!PageSlab(p)) { | |
210 | + lru_add_drain_all(); | |
211 | + if (PageLRU(p)) | |
212 | + return; | |
213 | + drain_all_pages(); | |
214 | + if (PageLRU(p) || is_free_buddy_page(p)) | |
215 | + return; | |
216 | + } | |
217 | + | |
218 | + /* | |
219 | + * Only all shrink_slab here (which would also | |
220 | + * shrink other caches) if access is not potentially fatal. | |
221 | + */ | |
222 | + if (access) { | |
223 | + int nr; | |
224 | + do { | |
225 | + nr = shrink_slab(1000, GFP_KERNEL, 1000); | |
226 | + if (page_count(p) == 0) | |
227 | + break; | |
228 | + } while (nr > 10); | |
229 | + } | |
230 | +} | |
231 | +EXPORT_SYMBOL_GPL(shake_page); | |
232 | + | |
233 | +/* | |
86 | 234 | * Kill all processes that have a poisoned page mapped and then isolate |
87 | 235 | * the page. |
88 | 236 | * |
... | ... | @@ -177,7 +325,6 @@ |
177 | 325 | * In case something went wrong with munmapping |
178 | 326 | * make sure the process doesn't catch the |
179 | 327 | * signal and then access the memory. Just kill it. |
180 | - * the signal handlers | |
181 | 328 | */ |
182 | 329 | if (fail || tk->addr_valid == 0) { |
183 | 330 | printk(KERN_ERR |
184 | 331 | |
185 | 332 | |
186 | 333 | |
187 | 334 | |
188 | 335 | |
189 | 336 | |
190 | 337 | |
191 | 338 | |
... | ... | @@ -314,33 +461,49 @@ |
314 | 461 | */ |
315 | 462 | |
316 | 463 | enum outcome { |
317 | - FAILED, /* Error handling failed */ | |
464 | + IGNORED, /* Error: cannot be handled */ | |
465 | + FAILED, /* Error: handling failed */ | |
318 | 466 | DELAYED, /* Will be handled later */ |
319 | - IGNORED, /* Error safely ignored */ | |
320 | 467 | RECOVERED, /* Successfully recovered */ |
321 | 468 | }; |
322 | 469 | |
323 | 470 | static const char *action_name[] = { |
471 | + [IGNORED] = "Ignored", | |
324 | 472 | [FAILED] = "Failed", |
325 | 473 | [DELAYED] = "Delayed", |
326 | - [IGNORED] = "Ignored", | |
327 | 474 | [RECOVERED] = "Recovered", |
328 | 475 | }; |
329 | 476 | |
330 | 477 | /* |
331 | - * Error hit kernel page. | |
332 | - * Do nothing, try to be lucky and not touch this instead. For a few cases we | |
333 | - * could be more sophisticated. | |
478 | + * XXX: It is possible that a page is isolated from LRU cache, | |
479 | + * and then kept in swap cache or failed to remove from page cache. | |
480 | + * The page count will stop it from being freed by unpoison. | |
481 | + * Stress tests should be aware of this memory leak problem. | |
334 | 482 | */ |
335 | -static int me_kernel(struct page *p, unsigned long pfn) | |
483 | +static int delete_from_lru_cache(struct page *p) | |
336 | 484 | { |
337 | - return DELAYED; | |
485 | + if (!isolate_lru_page(p)) { | |
486 | + /* | |
487 | + * Clear sensible page flags, so that the buddy system won't | |
488 | + * complain when the page is unpoison-and-freed. | |
489 | + */ | |
490 | + ClearPageActive(p); | |
491 | + ClearPageUnevictable(p); | |
492 | + /* | |
493 | + * drop the page count elevated by isolate_lru_page() | |
494 | + */ | |
495 | + page_cache_release(p); | |
496 | + return 0; | |
497 | + } | |
498 | + return -EIO; | |
338 | 499 | } |
339 | 500 | |
340 | 501 | /* |
341 | - * Already poisoned page. | |
502 | + * Error hit kernel page. | |
503 | + * Do nothing, try to be lucky and not touch this instead. For a few cases we | |
504 | + * could be more sophisticated. | |
342 | 505 | */ |
343 | -static int me_ignore(struct page *p, unsigned long pfn) | |
506 | +static int me_kernel(struct page *p, unsigned long pfn) | |
344 | 507 | { |
345 | 508 | return IGNORED; |
346 | 509 | } |
... | ... | @@ -355,14 +518,6 @@ |
355 | 518 | } |
356 | 519 | |
357 | 520 | /* |
358 | - * Free memory | |
359 | - */ | |
360 | -static int me_free(struct page *p, unsigned long pfn) | |
361 | -{ | |
362 | - return DELAYED; | |
363 | -} | |
364 | - | |
365 | -/* | |
366 | 521 | * Clean (or cleaned) page cache page. |
367 | 522 | */ |
368 | 523 | static int me_pagecache_clean(struct page *p, unsigned long pfn) |
... | ... | @@ -371,6 +526,8 @@ |
371 | 526 | int ret = FAILED; |
372 | 527 | struct address_space *mapping; |
373 | 528 | |
529 | + delete_from_lru_cache(p); | |
530 | + | |
374 | 531 | /* |
375 | 532 | * For anonymous pages we're done the only reference left |
376 | 533 | * should be the one m_f() holds. |
377 | 534 | |
... | ... | @@ -500,14 +657,20 @@ |
500 | 657 | /* Trigger EIO in shmem: */ |
501 | 658 | ClearPageUptodate(p); |
502 | 659 | |
503 | - return DELAYED; | |
660 | + if (!delete_from_lru_cache(p)) | |
661 | + return DELAYED; | |
662 | + else | |
663 | + return FAILED; | |
504 | 664 | } |
505 | 665 | |
506 | 666 | static int me_swapcache_clean(struct page *p, unsigned long pfn) |
507 | 667 | { |
508 | 668 | delete_from_swap_cache(p); |
509 | 669 | |
510 | - return RECOVERED; | |
670 | + if (!delete_from_lru_cache(p)) | |
671 | + return RECOVERED; | |
672 | + else | |
673 | + return FAILED; | |
511 | 674 | } |
512 | 675 | |
513 | 676 | /* |
... | ... | @@ -550,7 +713,6 @@ |
550 | 713 | #define tail (1UL << PG_tail) |
551 | 714 | #define compound (1UL << PG_compound) |
552 | 715 | #define slab (1UL << PG_slab) |
553 | -#define buddy (1UL << PG_buddy) | |
554 | 716 | #define reserved (1UL << PG_reserved) |
555 | 717 | |
556 | 718 | static struct page_state { |
... | ... | @@ -559,8 +721,11 @@ |
559 | 721 | char *msg; |
560 | 722 | int (*action)(struct page *p, unsigned long pfn); |
561 | 723 | } error_states[] = { |
562 | - { reserved, reserved, "reserved kernel", me_ignore }, | |
563 | - { buddy, buddy, "free kernel", me_free }, | |
724 | + { reserved, reserved, "reserved kernel", me_kernel }, | |
725 | + /* | |
726 | + * free pages are specially detected outside this table: | |
727 | + * PG_buddy pages only make a small fraction of all free pages. | |
728 | + */ | |
564 | 729 | |
565 | 730 | /* |
566 | 731 | * Could in theory check if slab page is free or if we can drop |
... | ... | @@ -587,7 +752,6 @@ |
587 | 752 | |
588 | 753 | { lru|dirty, lru|dirty, "LRU", me_pagecache_dirty }, |
589 | 754 | { lru|dirty, lru, "clean LRU", me_pagecache_clean }, |
590 | - { swapbacked, swapbacked, "anonymous", me_pagecache_clean }, | |
591 | 755 | |
592 | 756 | /* |
593 | 757 | * Catchall entry: must be at end. |
594 | 758 | |
595 | 759 | |
596 | 760 | |
... | ... | @@ -595,20 +759,31 @@ |
595 | 759 | { 0, 0, "unknown page state", me_unknown }, |
596 | 760 | }; |
597 | 761 | |
762 | +#undef dirty | |
763 | +#undef sc | |
764 | +#undef unevict | |
765 | +#undef mlock | |
766 | +#undef writeback | |
767 | +#undef lru | |
768 | +#undef swapbacked | |
769 | +#undef head | |
770 | +#undef tail | |
771 | +#undef compound | |
772 | +#undef slab | |
773 | +#undef reserved | |
774 | + | |
598 | 775 | static void action_result(unsigned long pfn, char *msg, int result) |
599 | 776 | { |
600 | - struct page *page = NULL; | |
601 | - if (pfn_valid(pfn)) | |
602 | - page = pfn_to_page(pfn); | |
777 | + struct page *page = pfn_to_page(pfn); | |
603 | 778 | |
604 | 779 | printk(KERN_ERR "MCE %#lx: %s%s page recovery: %s\n", |
605 | 780 | pfn, |
606 | - page && PageDirty(page) ? "dirty " : "", | |
781 | + PageDirty(page) ? "dirty " : "", | |
607 | 782 | msg, action_name[result]); |
608 | 783 | } |
609 | 784 | |
610 | 785 | static int page_action(struct page_state *ps, struct page *p, |
611 | - unsigned long pfn, int ref) | |
786 | + unsigned long pfn) | |
612 | 787 | { |
613 | 788 | int result; |
614 | 789 | int count; |
615 | 790 | |
616 | 791 | |
... | ... | @@ -616,18 +791,22 @@ |
616 | 791 | result = ps->action(p, pfn); |
617 | 792 | action_result(pfn, ps->msg, result); |
618 | 793 | |
619 | - count = page_count(p) - 1 - ref; | |
620 | - if (count != 0) | |
794 | + count = page_count(p) - 1; | |
795 | + if (ps->action == me_swapcache_dirty && result == DELAYED) | |
796 | + count--; | |
797 | + if (count != 0) { | |
621 | 798 | printk(KERN_ERR |
622 | 799 | "MCE %#lx: %s page still referenced by %d users\n", |
623 | 800 | pfn, ps->msg, count); |
801 | + result = FAILED; | |
802 | + } | |
624 | 803 | |
625 | 804 | /* Could do more checks here if page looks ok */ |
626 | 805 | /* |
627 | 806 | * Could adjust zone counters here to correct for the missing page. |
628 | 807 | */ |
629 | 808 | |
630 | - return result == RECOVERED ? 0 : -EBUSY; | |
809 | + return (result == RECOVERED || result == DELAYED) ? 0 : -EBUSY; | |
631 | 810 | } |
632 | 811 | |
633 | 812 | #define N_UNMAP_TRIES 5 |
... | ... | @@ -636,7 +815,7 @@ |
636 | 815 | * Do all that is necessary to remove user space mappings. Unmap |
637 | 816 | * the pages and send SIGBUS to the processes if the data was dirty. |
638 | 817 | */ |
639 | -static void hwpoison_user_mappings(struct page *p, unsigned long pfn, | |
818 | +static int hwpoison_user_mappings(struct page *p, unsigned long pfn, | |
640 | 819 | int trapno) |
641 | 820 | { |
642 | 821 | enum ttu_flags ttu = TTU_UNMAP | TTU_IGNORE_MLOCK | TTU_IGNORE_ACCESS; |
643 | 822 | |
644 | 823 | |
... | ... | @@ -646,16 +825,19 @@ |
646 | 825 | int i; |
647 | 826 | int kill = 1; |
648 | 827 | |
649 | - if (PageReserved(p) || PageCompound(p) || PageSlab(p) || PageKsm(p)) | |
650 | - return; | |
828 | + if (PageReserved(p) || PageSlab(p)) | |
829 | + return SWAP_SUCCESS; | |
651 | 830 | |
652 | 831 | /* |
653 | 832 | * This check implies we don't kill processes if their pages |
654 | 833 | * are in the swap cache early. Those are always late kills. |
655 | 834 | */ |
656 | 835 | if (!page_mapped(p)) |
657 | - return; | |
836 | + return SWAP_SUCCESS; | |
658 | 837 | |
838 | + if (PageCompound(p) || PageKsm(p)) | |
839 | + return SWAP_FAIL; | |
840 | + | |
659 | 841 | if (PageSwapCache(p)) { |
660 | 842 | printk(KERN_ERR |
661 | 843 | "MCE %#lx: keeping poisoned page in swap cache\n", pfn); |
... | ... | @@ -665,6 +847,8 @@ |
665 | 847 | /* |
666 | 848 | * Propagate the dirty bit from PTEs to struct page first, because we |
667 | 849 | * need this to decide if we should kill or just drop the page. |
850 | + * XXX: the dirty test could be racy: set_page_dirty() may not always | |
851 | + * be called inside page lock (it's recommended but not enforced). | |
668 | 852 | */ |
669 | 853 | mapping = page_mapping(p); |
670 | 854 | if (!PageDirty(p) && mapping && mapping_cap_writeback_dirty(mapping)) { |
671 | 855 | |
672 | 856 | |
... | ... | @@ -716,11 +900,12 @@ |
716 | 900 | */ |
717 | 901 | kill_procs_ao(&tokill, !!PageDirty(p), trapno, |
718 | 902 | ret != SWAP_SUCCESS, pfn); |
903 | + | |
904 | + return ret; | |
719 | 905 | } |
720 | 906 | |
721 | -int __memory_failure(unsigned long pfn, int trapno, int ref) | |
907 | +int __memory_failure(unsigned long pfn, int trapno, int flags) | |
722 | 908 | { |
723 | - unsigned long lru_flag; | |
724 | 909 | struct page_state *ps; |
725 | 910 | struct page *p; |
726 | 911 | int res; |
727 | 912 | |
... | ... | @@ -729,13 +914,15 @@ |
729 | 914 | panic("Memory failure from trap %d on page %lx", trapno, pfn); |
730 | 915 | |
731 | 916 | if (!pfn_valid(pfn)) { |
732 | - action_result(pfn, "memory outside kernel control", IGNORED); | |
733 | - return -EIO; | |
917 | + printk(KERN_ERR | |
918 | + "MCE %#lx: memory outside kernel control\n", | |
919 | + pfn); | |
920 | + return -ENXIO; | |
734 | 921 | } |
735 | 922 | |
736 | 923 | p = pfn_to_page(pfn); |
737 | 924 | if (TestSetPageHWPoison(p)) { |
738 | - action_result(pfn, "already hardware poisoned", IGNORED); | |
925 | + printk(KERN_ERR "MCE %#lx: already hardware poisoned\n", pfn); | |
739 | 926 | return 0; |
740 | 927 | } |
741 | 928 | |
... | ... | @@ -752,9 +939,15 @@ |
752 | 939 | * In fact it's dangerous to directly bump up page count from 0, |
753 | 940 | * that may make page_freeze_refs()/page_unfreeze_refs() mismatch. |
754 | 941 | */ |
755 | - if (!get_page_unless_zero(compound_head(p))) { | |
756 | - action_result(pfn, "free or high order kernel", IGNORED); | |
757 | - return PageBuddy(compound_head(p)) ? 0 : -EBUSY; | |
942 | + if (!(flags & MF_COUNT_INCREASED) && | |
943 | + !get_page_unless_zero(compound_head(p))) { | |
944 | + if (is_free_buddy_page(p)) { | |
945 | + action_result(pfn, "free buddy", DELAYED); | |
946 | + return 0; | |
947 | + } else { | |
948 | + action_result(pfn, "high order kernel", IGNORED); | |
949 | + return -EBUSY; | |
950 | + } | |
758 | 951 | } |
759 | 952 | |
760 | 953 | /* |
761 | 954 | |
... | ... | @@ -766,14 +959,19 @@ |
766 | 959 | * walked by the page reclaim code, however that's not a big loss. |
767 | 960 | */ |
768 | 961 | if (!PageLRU(p)) |
769 | - lru_add_drain_all(); | |
770 | - lru_flag = p->flags & lru; | |
771 | - if (isolate_lru_page(p)) { | |
962 | + shake_page(p, 0); | |
963 | + if (!PageLRU(p)) { | |
964 | + /* | |
965 | + * shake_page could have turned it free. | |
966 | + */ | |
967 | + if (is_free_buddy_page(p)) { | |
968 | + action_result(pfn, "free buddy, 2nd try", DELAYED); | |
969 | + return 0; | |
970 | + } | |
772 | 971 | action_result(pfn, "non LRU", IGNORED); |
773 | 972 | put_page(p); |
774 | 973 | return -EBUSY; |
775 | 974 | } |
776 | - page_cache_release(p); | |
777 | 975 | |
778 | 976 | /* |
779 | 977 | * Lock the page and wait for writeback to finish. |
780 | 978 | |
781 | 979 | |
782 | 980 | |
783 | 981 | |
784 | 982 | |
... | ... | @@ -781,26 +979,48 @@ |
781 | 979 | * and in many cases impossible, so we just avoid it here. |
782 | 980 | */ |
783 | 981 | lock_page_nosync(p); |
982 | + | |
983 | + /* | |
984 | + * unpoison always clear PG_hwpoison inside page lock | |
985 | + */ | |
986 | + if (!PageHWPoison(p)) { | |
987 | + printk(KERN_ERR "MCE %#lx: just unpoisoned\n", pfn); | |
988 | + res = 0; | |
989 | + goto out; | |
990 | + } | |
991 | + if (hwpoison_filter(p)) { | |
992 | + if (TestClearPageHWPoison(p)) | |
993 | + atomic_long_dec(&mce_bad_pages); | |
994 | + unlock_page(p); | |
995 | + put_page(p); | |
996 | + return 0; | |
997 | + } | |
998 | + | |
784 | 999 | wait_on_page_writeback(p); |
785 | 1000 | |
786 | 1001 | /* |
787 | 1002 | * Now take care of user space mappings. |
1003 | + * Abort on fail: __remove_from_page_cache() assumes unmapped page. | |
788 | 1004 | */ |
789 | - hwpoison_user_mappings(p, pfn, trapno); | |
1005 | + if (hwpoison_user_mappings(p, pfn, trapno) != SWAP_SUCCESS) { | |
1006 | + printk(KERN_ERR "MCE %#lx: cannot unmap page, give up\n", pfn); | |
1007 | + res = -EBUSY; | |
1008 | + goto out; | |
1009 | + } | |
790 | 1010 | |
791 | 1011 | /* |
792 | 1012 | * Torn down by someone else? |
793 | 1013 | */ |
794 | - if ((lru_flag & lru) && !PageSwapCache(p) && p->mapping == NULL) { | |
1014 | + if (PageLRU(p) && !PageSwapCache(p) && p->mapping == NULL) { | |
795 | 1015 | action_result(pfn, "already truncated LRU", IGNORED); |
796 | - res = 0; | |
1016 | + res = -EBUSY; | |
797 | 1017 | goto out; |
798 | 1018 | } |
799 | 1019 | |
800 | 1020 | res = -EBUSY; |
801 | 1021 | for (ps = error_states;; ps++) { |
802 | - if (((p->flags | lru_flag)& ps->mask) == ps->res) { | |
803 | - res = page_action(ps, p, pfn, ref); | |
1022 | + if ((p->flags & ps->mask) == ps->res) { | |
1023 | + res = page_action(ps, p, pfn); | |
804 | 1024 | break; |
805 | 1025 | } |
806 | 1026 | } |
... | ... | @@ -830,5 +1050,237 @@ |
830 | 1050 | void memory_failure(unsigned long pfn, int trapno) |
831 | 1051 | { |
832 | 1052 | __memory_failure(pfn, trapno, 0); |
1053 | +} | |
1054 | + | |
1055 | +/** | |
1056 | + * unpoison_memory - Unpoison a previously poisoned page | |
1057 | + * @pfn: Page number of the to be unpoisoned page | |
1058 | + * | |
1059 | + * Software-unpoison a page that has been poisoned by | |
1060 | + * memory_failure() earlier. | |
1061 | + * | |
1062 | + * This is only done on the software-level, so it only works | |
1063 | + * for linux injected failures, not real hardware failures | |
1064 | + * | |
1065 | + * Returns 0 for success, otherwise -errno. | |
1066 | + */ | |
1067 | +int unpoison_memory(unsigned long pfn) | |
1068 | +{ | |
1069 | + struct page *page; | |
1070 | + struct page *p; | |
1071 | + int freeit = 0; | |
1072 | + | |
1073 | + if (!pfn_valid(pfn)) | |
1074 | + return -ENXIO; | |
1075 | + | |
1076 | + p = pfn_to_page(pfn); | |
1077 | + page = compound_head(p); | |
1078 | + | |
1079 | + if (!PageHWPoison(p)) { | |
1080 | + pr_debug("MCE: Page was already unpoisoned %#lx\n", pfn); | |
1081 | + return 0; | |
1082 | + } | |
1083 | + | |
1084 | + if (!get_page_unless_zero(page)) { | |
1085 | + if (TestClearPageHWPoison(p)) | |
1086 | + atomic_long_dec(&mce_bad_pages); | |
1087 | + pr_debug("MCE: Software-unpoisoned free page %#lx\n", pfn); | |
1088 | + return 0; | |
1089 | + } | |
1090 | + | |
1091 | + lock_page_nosync(page); | |
1092 | + /* | |
1093 | + * This test is racy because PG_hwpoison is set outside of page lock. | |
1094 | + * That's acceptable because that won't trigger kernel panic. Instead, | |
1095 | + * the PG_hwpoison page will be caught and isolated on the entrance to | |
1096 | + * the free buddy page pool. | |
1097 | + */ | |
1098 | + if (TestClearPageHWPoison(p)) { | |
1099 | + pr_debug("MCE: Software-unpoisoned page %#lx\n", pfn); | |
1100 | + atomic_long_dec(&mce_bad_pages); | |
1101 | + freeit = 1; | |
1102 | + } | |
1103 | + unlock_page(page); | |
1104 | + | |
1105 | + put_page(page); | |
1106 | + if (freeit) | |
1107 | + put_page(page); | |
1108 | + | |
1109 | + return 0; | |
1110 | +} | |
1111 | +EXPORT_SYMBOL(unpoison_memory); | |
1112 | + | |
1113 | +static struct page *new_page(struct page *p, unsigned long private, int **x) | |
1114 | +{ | |
1115 | + int nid = page_to_nid(p); | |
1116 | + return alloc_pages_exact_node(nid, GFP_HIGHUSER_MOVABLE, 0); | |
1117 | +} | |
1118 | + | |
1119 | +/* | |
1120 | + * Safely get reference count of an arbitrary page. | |
1121 | + * Returns 0 for a free page, -EIO for a zero refcount page | |
1122 | + * that is not free, and 1 for any other page type. | |
1123 | + * For 1 the page is returned with increased page count, otherwise not. | |
1124 | + */ | |
1125 | +static int get_any_page(struct page *p, unsigned long pfn, int flags) | |
1126 | +{ | |
1127 | + int ret; | |
1128 | + | |
1129 | + if (flags & MF_COUNT_INCREASED) | |
1130 | + return 1; | |
1131 | + | |
1132 | + /* | |
1133 | + * The lock_system_sleep prevents a race with memory hotplug, | |
1134 | + * because the isolation assumes there's only a single user. | |
1135 | + * This is a big hammer, a better would be nicer. | |
1136 | + */ | |
1137 | + lock_system_sleep(); | |
1138 | + | |
1139 | + /* | |
1140 | + * Isolate the page, so that it doesn't get reallocated if it | |
1141 | + * was free. | |
1142 | + */ | |
1143 | + set_migratetype_isolate(p); | |
1144 | + if (!get_page_unless_zero(compound_head(p))) { | |
1145 | + if (is_free_buddy_page(p)) { | |
1146 | + pr_debug("get_any_page: %#lx free buddy page\n", pfn); | |
1147 | + /* Set hwpoison bit while page is still isolated */ | |
1148 | + SetPageHWPoison(p); | |
1149 | + ret = 0; | |
1150 | + } else { | |
1151 | + pr_debug("get_any_page: %#lx: unknown zero refcount page type %lx\n", | |
1152 | + pfn, p->flags); | |
1153 | + ret = -EIO; | |
1154 | + } | |
1155 | + } else { | |
1156 | + /* Not a free page */ | |
1157 | + ret = 1; | |
1158 | + } | |
1159 | + unset_migratetype_isolate(p); | |
1160 | + unlock_system_sleep(); | |
1161 | + return ret; | |
1162 | +} | |
1163 | + | |
1164 | +/** | |
1165 | + * soft_offline_page - Soft offline a page. | |
1166 | + * @page: page to offline | |
1167 | + * @flags: flags. Same as memory_failure(). | |
1168 | + * | |
1169 | + * Returns 0 on success, otherwise negated errno. | |
1170 | + * | |
1171 | + * Soft offline a page, by migration or invalidation, | |
1172 | + * without killing anything. This is for the case when | |
1173 | + * a page is not corrupted yet (so it's still valid to access), | |
1174 | + * but has had a number of corrected errors and is better taken | |
1175 | + * out. | |
1176 | + * | |
1177 | + * The actual policy on when to do that is maintained by | |
1178 | + * user space. | |
1179 | + * | |
1180 | + * This should never impact any application or cause data loss, | |
1181 | + * however it might take some time. | |
1182 | + * | |
1183 | + * This is not a 100% solution for all memory, but tries to be | |
1184 | + * ``good enough'' for the majority of memory. | |
1185 | + */ | |
1186 | +int soft_offline_page(struct page *page, int flags) | |
1187 | +{ | |
1188 | + int ret; | |
1189 | + unsigned long pfn = page_to_pfn(page); | |
1190 | + | |
1191 | + ret = get_any_page(page, pfn, flags); | |
1192 | + if (ret < 0) | |
1193 | + return ret; | |
1194 | + if (ret == 0) | |
1195 | + goto done; | |
1196 | + | |
1197 | + /* | |
1198 | + * Page cache page we can handle? | |
1199 | + */ | |
1200 | + if (!PageLRU(page)) { | |
1201 | + /* | |
1202 | + * Try to free it. | |
1203 | + */ | |
1204 | + put_page(page); | |
1205 | + shake_page(page, 1); | |
1206 | + | |
1207 | + /* | |
1208 | + * Did it turn free? | |
1209 | + */ | |
1210 | + ret = get_any_page(page, pfn, 0); | |
1211 | + if (ret < 0) | |
1212 | + return ret; | |
1213 | + if (ret == 0) | |
1214 | + goto done; | |
1215 | + } | |
1216 | + if (!PageLRU(page)) { | |
1217 | + pr_debug("soft_offline: %#lx: unknown non LRU page type %lx\n", | |
1218 | + pfn, page->flags); | |
1219 | + return -EIO; | |
1220 | + } | |
1221 | + | |
1222 | + lock_page(page); | |
1223 | + wait_on_page_writeback(page); | |
1224 | + | |
1225 | + /* | |
1226 | + * Synchronized using the page lock with memory_failure() | |
1227 | + */ | |
1228 | + if (PageHWPoison(page)) { | |
1229 | + unlock_page(page); | |
1230 | + put_page(page); | |
1231 | + pr_debug("soft offline: %#lx page already poisoned\n", pfn); | |
1232 | + return -EBUSY; | |
1233 | + } | |
1234 | + | |
1235 | + /* | |
1236 | + * Try to invalidate first. This should work for | |
1237 | + * non dirty unmapped page cache pages. | |
1238 | + */ | |
1239 | + ret = invalidate_inode_page(page); | |
1240 | + unlock_page(page); | |
1241 | + | |
1242 | + /* | |
1243 | + * Drop count because page migration doesn't like raised | |
1244 | + * counts. The page could get re-allocated, but if it becomes | |
1245 | + * LRU the isolation will just fail. | |
1246 | + * RED-PEN would be better to keep it isolated here, but we | |
1247 | + * would need to fix isolation locking first. | |
1248 | + */ | |
1249 | + put_page(page); | |
1250 | + if (ret == 1) { | |
1251 | + ret = 0; | |
1252 | + pr_debug("soft_offline: %#lx: invalidated\n", pfn); | |
1253 | + goto done; | |
1254 | + } | |
1255 | + | |
1256 | + /* | |
1257 | + * Simple invalidation didn't work. | |
1258 | + * Try to migrate to a new page instead. migrate.c | |
1259 | + * handles a large number of cases for us. | |
1260 | + */ | |
1261 | + ret = isolate_lru_page(page); | |
1262 | + if (!ret) { | |
1263 | + LIST_HEAD(pagelist); | |
1264 | + | |
1265 | + list_add(&page->lru, &pagelist); | |
1266 | + ret = migrate_pages(&pagelist, new_page, MPOL_MF_MOVE_ALL, 0); | |
1267 | + if (ret) { | |
1268 | + pr_debug("soft offline: %#lx: migration failed %d, type %lx\n", | |
1269 | + pfn, ret, page->flags); | |
1270 | + if (ret > 0) | |
1271 | + ret = -EIO; | |
1272 | + } | |
1273 | + } else { | |
1274 | + pr_debug("soft offline: %#lx: isolation failed: %d, page count %d, type %lx\n", | |
1275 | + pfn, ret, page_count(page), page->flags); | |
1276 | + } | |
1277 | + if (ret) | |
1278 | + return ret; | |
1279 | + | |
1280 | +done: | |
1281 | + atomic_long_add(1, &mce_bad_pages); | |
1282 | + SetPageHWPoison(page); | |
1283 | + /* keep elevated page count for bad page */ | |
1284 | + return ret; | |
833 | 1285 | } |
mm/memory.c
... | ... | @@ -2555,6 +2555,10 @@ |
2555 | 2555 | ret = VM_FAULT_MAJOR; |
2556 | 2556 | count_vm_event(PGMAJFAULT); |
2557 | 2557 | } else if (PageHWPoison(page)) { |
2558 | + /* | |
2559 | + * hwpoisoned dirty swapcache pages are kept for killing | |
2560 | + * owner processes (which may be unknown at hwpoison time) | |
2561 | + */ | |
2558 | 2562 | ret = VM_FAULT_HWPOISON; |
2559 | 2563 | delayacct_clear_flag(DELAYACCT_PF_SWAPIN); |
2560 | 2564 | goto out_release; |
mm/page_alloc.c
... | ... | @@ -5091,4 +5091,25 @@ |
5091 | 5091 | spin_unlock_irqrestore(&zone->lock, flags); |
5092 | 5092 | } |
5093 | 5093 | #endif |
5094 | + | |
5095 | +#ifdef CONFIG_MEMORY_FAILURE | |
5096 | +bool is_free_buddy_page(struct page *page) | |
5097 | +{ | |
5098 | + struct zone *zone = page_zone(page); | |
5099 | + unsigned long pfn = page_to_pfn(page); | |
5100 | + unsigned long flags; | |
5101 | + int order; | |
5102 | + | |
5103 | + spin_lock_irqsave(&zone->lock, flags); | |
5104 | + for (order = 0; order < MAX_ORDER; order++) { | |
5105 | + struct page *page_head = page - (pfn & ((1 << order) - 1)); | |
5106 | + | |
5107 | + if (PageBuddy(page_head) && page_order(page_head) >= order) | |
5108 | + break; | |
5109 | + } | |
5110 | + spin_unlock_irqrestore(&zone->lock, flags); | |
5111 | + | |
5112 | + return order < MAX_ORDER; | |
5113 | +} | |
5114 | +#endif |