aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorKOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>2009-08-18 17:11:10 -0400
committerLinus Torvalds <torvalds@linux-foundation.org>2009-08-18 19:31:13 -0400
commit0753ba01e126020bf0f8150934903b48935b697d (patch)
treefbfd7e2d0abbe724a8c5e0e17fb9af522ed2e097
parent89a4eb4b66e8f4d395e14a14d262dac4d6ca52f0 (diff)
mm: revert "oom: move oom_adj value"
The commit 2ff05b2b (oom: move oom_adj value) moveed the oom_adj value to the mm_struct. It was a very good first step for sanitize OOM. However Paul Menage reported the commit makes regression to his job scheduler. Current OOM logic can kill OOM_DISABLED process. Why? His program has the code of similar to the following. ... set_oom_adj(OOM_DISABLE); /* The job scheduler never killed by oom */ ... if (vfork() == 0) { set_oom_adj(0); /* Invoked child can be killed */ execve("foo-bar-cmd"); } .... vfork() parent and child are shared the same mm_struct. then above set_oom_adj(0) doesn't only change oom_adj for vfork() child, it's also change oom_adj for vfork() parent. Then, vfork() parent (job scheduler) lost OOM immune and it was killed. Actually, fork-setting-exec idiom is very frequently used in userland program. We must not break this assumption. Then, this patch revert commit 2ff05b2b and related commit. Reverted commit list --------------------- - commit 2ff05b2b4e (oom: move oom_adj value from task_struct to mm_struct) - commit 4d8b9135c3 (oom: avoid unnecessary mm locking and scanning for OOM_DISABLE) - commit 8123681022 (oom: only oom kill exiting tasks with attached memory) - commit 933b787b57 (mm: copy over oom_adj value at fork time) Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Paul Menage <menage@google.com> Cc: David Rientjes <rientjes@google.com> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Rik van Riel <riel@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Nick Piggin <npiggin@suse.de> Cc: Mel Gorman <mel@csn.ul.ie> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
-rw-r--r--Documentation/filesystems/proc.txt15
-rw-r--r--fs/proc/base.c19
-rw-r--r--include/linux/mm_types.h2
-rw-r--r--include/linux/sched.h1
-rw-r--r--kernel/fork.c1
-rw-r--r--mm/oom_kill.c64
6 files changed, 48 insertions, 54 deletions
diff --git a/Documentation/filesystems/proc.txt b/Documentation/filesystems/proc.txt
index fad18f9456e4..ffead13f9443 100644
--- a/Documentation/filesystems/proc.txt
+++ b/Documentation/filesystems/proc.txt
@@ -1167,13 +1167,11 @@ CHAPTER 3: PER-PROCESS PARAMETERS
11673.1 /proc/<pid>/oom_adj - Adjust the oom-killer score 11673.1 /proc/<pid>/oom_adj - Adjust the oom-killer score
1168------------------------------------------------------ 1168------------------------------------------------------
1169 1169
1170This file can be used to adjust the score used to select which processes should 1170This file can be used to adjust the score used to select which processes
1171be killed in an out-of-memory situation. The oom_adj value is a characteristic 1171should be killed in an out-of-memory situation. Giving it a high score will
1172of the task's mm, so all threads that share an mm with pid will have the same 1172increase the likelihood of this process being killed by the oom-killer. Valid
1173oom_adj value. A high value will increase the likelihood of this process being 1173values are in the range -16 to +15, plus the special value -17, which disables
1174killed by the oom-killer. Valid values are in the range -16 to +15 as 1174oom-killing altogether for this process.
1175explained below and a special value of -17, which disables oom-killing
1176altogether for threads sharing pid's mm.
1177 1175
1178The process to be killed in an out-of-memory situation is selected among all others 1176The process to be killed in an out-of-memory situation is selected among all others
1179based on its badness score. This value equals the original memory size of the process 1177based on its badness score. This value equals the original memory size of the process
@@ -1187,9 +1185,6 @@ the parent's score if they do not share the same memory. Thus forking servers
1187are the prime candidates to be killed. Having only one 'hungry' child will make 1185are the prime candidates to be killed. Having only one 'hungry' child will make
1188parent less preferable than the child. 1186parent less preferable than the child.
1189 1187
1190/proc/<pid>/oom_adj cannot be changed for kthreads since they are immune from
1191oom-killing already.
1192
1193/proc/<pid>/oom_score shows process' current badness score. 1188/proc/<pid>/oom_score shows process' current badness score.
1194 1189
1195The following heuristics are then applied: 1190The following heuristics are then applied:
diff --git a/fs/proc/base.c b/fs/proc/base.c
index 175db258942f..6f742f6658a9 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -1003,12 +1003,7 @@ static ssize_t oom_adjust_read(struct file *file, char __user *buf,
1003 1003
1004 if (!task) 1004 if (!task)
1005 return -ESRCH; 1005 return -ESRCH;
1006 task_lock(task); 1006 oom_adjust = task->oomkilladj;
1007 if (task->mm)
1008 oom_adjust = task->mm->oom_adj;
1009 else
1010 oom_adjust = OOM_DISABLE;
1011 task_unlock(task);
1012 put_task_struct(task); 1007 put_task_struct(task);
1013 1008
1014 len = snprintf(buffer, sizeof(buffer), "%i\n", oom_adjust); 1009 len = snprintf(buffer, sizeof(buffer), "%i\n", oom_adjust);
@@ -1037,19 +1032,11 @@ static ssize_t oom_adjust_write(struct file *file, const char __user *buf,
1037 task = get_proc_task(file->f_path.dentry->d_inode); 1032 task = get_proc_task(file->f_path.dentry->d_inode);
1038 if (!task) 1033 if (!task)
1039 return -ESRCH; 1034 return -ESRCH;
1040 task_lock(task); 1035 if (oom_adjust < task->oomkilladj && !capable(CAP_SYS_RESOURCE)) {
1041 if (!task->mm) {
1042 task_unlock(task);
1043 put_task_struct(task);
1044 return -EINVAL;
1045 }
1046 if (oom_adjust < task->mm->oom_adj && !capable(CAP_SYS_RESOURCE)) {
1047 task_unlock(task);
1048 put_task_struct(task); 1036 put_task_struct(task);
1049 return -EACCES; 1037 return -EACCES;
1050 } 1038 }
1051 task->mm->oom_adj = oom_adjust; 1039 task->oomkilladj = oom_adjust;
1052 task_unlock(task);
1053 put_task_struct(task); 1040 put_task_struct(task);
1054 if (end - buffer == 0) 1041 if (end - buffer == 0)
1055 return -EIO; 1042 return -EIO;
diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index 7acc8439d9b3..0042090a4d70 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -240,8 +240,6 @@ struct mm_struct {
240 240
241 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */ 241 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
242 242
243 s8 oom_adj; /* OOM kill score adjustment (bit shift) */
244
245 cpumask_t cpu_vm_mask; 243 cpumask_t cpu_vm_mask;
246 244
247 /* Architecture-specific MM context */ 245 /* Architecture-specific MM context */
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 3ab08e4bb6b8..0f1ea4a66957 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1198,6 +1198,7 @@ struct task_struct {
1198 * a short time 1198 * a short time
1199 */ 1199 */
1200 unsigned char fpu_counter; 1200 unsigned char fpu_counter;
1201 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1201#ifdef CONFIG_BLK_DEV_IO_TRACE 1202#ifdef CONFIG_BLK_DEV_IO_TRACE
1202 unsigned int btrace_seq; 1203 unsigned int btrace_seq;
1203#endif 1204#endif
diff --git a/kernel/fork.c b/kernel/fork.c
index 021e1138556e..144326b7af50 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -426,7 +426,6 @@ static struct mm_struct * mm_init(struct mm_struct * mm, struct task_struct *p)
426 init_rwsem(&mm->mmap_sem); 426 init_rwsem(&mm->mmap_sem);
427 INIT_LIST_HEAD(&mm->mmlist); 427 INIT_LIST_HEAD(&mm->mmlist);
428 mm->flags = (current->mm) ? current->mm->flags : default_dump_filter; 428 mm->flags = (current->mm) ? current->mm->flags : default_dump_filter;
429 mm->oom_adj = (current->mm) ? current->mm->oom_adj : 0;
430 mm->core_state = NULL; 429 mm->core_state = NULL;
431 mm->nr_ptes = 0; 430 mm->nr_ptes = 0;
432 set_mm_counter(mm, file_rss, 0); 431 set_mm_counter(mm, file_rss, 0);
diff --git a/mm/oom_kill.c b/mm/oom_kill.c
index 175a67a78a99..a7b2460e922b 100644
--- a/mm/oom_kill.c
+++ b/mm/oom_kill.c
@@ -58,7 +58,6 @@ unsigned long badness(struct task_struct *p, unsigned long uptime)
58 unsigned long points, cpu_time, run_time; 58 unsigned long points, cpu_time, run_time;
59 struct mm_struct *mm; 59 struct mm_struct *mm;
60 struct task_struct *child; 60 struct task_struct *child;
61 int oom_adj;
62 61
63 task_lock(p); 62 task_lock(p);
64 mm = p->mm; 63 mm = p->mm;
@@ -66,11 +65,6 @@ unsigned long badness(struct task_struct *p, unsigned long uptime)
66 task_unlock(p); 65 task_unlock(p);
67 return 0; 66 return 0;
68 } 67 }
69 oom_adj = mm->oom_adj;
70 if (oom_adj == OOM_DISABLE) {
71 task_unlock(p);
72 return 0;
73 }
74 68
75 /* 69 /*
76 * The memory size of the process is the basis for the badness. 70 * The memory size of the process is the basis for the badness.
@@ -154,15 +148,15 @@ unsigned long badness(struct task_struct *p, unsigned long uptime)
154 points /= 8; 148 points /= 8;
155 149
156 /* 150 /*
157 * Adjust the score by oom_adj. 151 * Adjust the score by oomkilladj.
158 */ 152 */
159 if (oom_adj) { 153 if (p->oomkilladj) {
160 if (oom_adj > 0) { 154 if (p->oomkilladj > 0) {
161 if (!points) 155 if (!points)
162 points = 1; 156 points = 1;
163 points <<= oom_adj; 157 points <<= p->oomkilladj;
164 } else 158 } else
165 points >>= -(oom_adj); 159 points >>= -(p->oomkilladj);
166 } 160 }
167 161
168#ifdef DEBUG 162#ifdef DEBUG
@@ -257,8 +251,11 @@ static struct task_struct *select_bad_process(unsigned long *ppoints,
257 *ppoints = ULONG_MAX; 251 *ppoints = ULONG_MAX;
258 } 252 }
259 253
254 if (p->oomkilladj == OOM_DISABLE)
255 continue;
256
260 points = badness(p, uptime.tv_sec); 257 points = badness(p, uptime.tv_sec);
261 if (points > *ppoints) { 258 if (points > *ppoints || !chosen) {
262 chosen = p; 259 chosen = p;
263 *ppoints = points; 260 *ppoints = points;
264 } 261 }
@@ -307,7 +304,8 @@ static void dump_tasks(const struct mem_cgroup *mem)
307 } 304 }
308 printk(KERN_INFO "[%5d] %5d %5d %8lu %8lu %3d %3d %s\n", 305 printk(KERN_INFO "[%5d] %5d %5d %8lu %8lu %3d %3d %s\n",
309 p->pid, __task_cred(p)->uid, p->tgid, mm->total_vm, 306 p->pid, __task_cred(p)->uid, p->tgid, mm->total_vm,
310 get_mm_rss(mm), (int)task_cpu(p), mm->oom_adj, p->comm); 307 get_mm_rss(mm), (int)task_cpu(p), p->oomkilladj,
308 p->comm);
311 task_unlock(p); 309 task_unlock(p);
312 } while_each_thread(g, p); 310 } while_each_thread(g, p);
313} 311}
@@ -325,8 +323,11 @@ static void __oom_kill_task(struct task_struct *p, int verbose)
325 return; 323 return;
326 } 324 }
327 325
328 if (!p->mm) 326 if (!p->mm) {
327 WARN_ON(1);
328 printk(KERN_WARNING "tried to kill an mm-less task!\n");
329 return; 329 return;
330 }
330 331
331 if (verbose) 332 if (verbose)
332 printk(KERN_ERR "Killed process %d (%s)\n", 333 printk(KERN_ERR "Killed process %d (%s)\n",
@@ -348,13 +349,28 @@ static int oom_kill_task(struct task_struct *p)
348 struct mm_struct *mm; 349 struct mm_struct *mm;
349 struct task_struct *g, *q; 350 struct task_struct *g, *q;
350 351
351 task_lock(p);
352 mm = p->mm; 352 mm = p->mm;
353 if (!mm || mm->oom_adj == OOM_DISABLE) { 353
354 task_unlock(p); 354 /* WARNING: mm may not be dereferenced since we did not obtain its
355 * value from get_task_mm(p). This is OK since all we need to do is
356 * compare mm to q->mm below.
357 *
358 * Furthermore, even if mm contains a non-NULL value, p->mm may
359 * change to NULL at any time since we do not hold task_lock(p).
360 * However, this is of no concern to us.
361 */
362
363 if (mm == NULL)
355 return 1; 364 return 1;
356 } 365
357 task_unlock(p); 366 /*
367 * Don't kill the process if any threads are set to OOM_DISABLE
368 */
369 do_each_thread(g, q) {
370 if (q->mm == mm && q->oomkilladj == OOM_DISABLE)
371 return 1;
372 } while_each_thread(g, q);
373
358 __oom_kill_task(p, 1); 374 __oom_kill_task(p, 1);
359 375
360 /* 376 /*
@@ -377,11 +393,10 @@ static int oom_kill_process(struct task_struct *p, gfp_t gfp_mask, int order,
377 struct task_struct *c; 393 struct task_struct *c;
378 394
379 if (printk_ratelimit()) { 395 if (printk_ratelimit()) {
380 task_lock(current);
381 printk(KERN_WARNING "%s invoked oom-killer: " 396 printk(KERN_WARNING "%s invoked oom-killer: "
382 "gfp_mask=0x%x, order=%d, oom_adj=%d\n", 397 "gfp_mask=0x%x, order=%d, oomkilladj=%d\n",
383 current->comm, gfp_mask, order, 398 current->comm, gfp_mask, order, current->oomkilladj);
384 current->mm ? current->mm->oom_adj : OOM_DISABLE); 399 task_lock(current);
385 cpuset_print_task_mems_allowed(current); 400 cpuset_print_task_mems_allowed(current);
386 task_unlock(current); 401 task_unlock(current);
387 dump_stack(); 402 dump_stack();
@@ -394,9 +409,8 @@ static int oom_kill_process(struct task_struct *p, gfp_t gfp_mask, int order,
394 /* 409 /*
395 * If the task is already exiting, don't alarm the sysadmin or kill 410 * If the task is already exiting, don't alarm the sysadmin or kill
396 * its children or threads, just set TIF_MEMDIE so it can die quickly 411 * its children or threads, just set TIF_MEMDIE so it can die quickly
397 * if its mm is still attached.
398 */ 412 */
399 if (p->mm && (p->flags & PF_EXITING)) { 413 if (p->flags & PF_EXITING) {
400 __oom_kill_task(p, 0); 414 __oom_kill_task(p, 0);
401 return 0; 415 return 0;
402 } 416 }