Message ID | 1594309987-9919-1-git-send-email-laoar.shao@gmail.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Series | [v2] mm, oom: make the calculation of oom badness more accurate | expand |
On Thu 09-07-20 11:53:07, Yafang Shao wrote: > Recently we found an issue on our production environment that when memcg > oom is triggered the oom killer doesn't chose the process with largest > resident memory but chose the first scanned process. Note that all > processes in this memcg have the same oom_score_adj, so the oom killer > should chose the process with largest resident memory. > > Bellow is part of the oom info, which is enough to analyze this issue. > [7516987.983223] memory: usage 16777216kB, limit 16777216kB, failcnt 52843037 > [7516987.983224] memory+swap: usage 16777216kB, limit 9007199254740988kB, failcnt 0 > [7516987.983225] kmem: usage 301464kB, limit 9007199254740988kB, failcnt 0 > [...] > [7516987.983293] [ pid ] uid tgid total_vm rss pgtables_bytes swapents oom_score_adj name > [7516987.983510] [ 5740] 0 5740 257 1 32768 0 -998 pause > [7516987.983574] [58804] 0 58804 4594 771 81920 0 -998 entry_point.bas > [7516987.983577] [58908] 0 58908 7089 689 98304 0 -998 cron > [7516987.983580] [58910] 0 58910 16235 5576 163840 0 -998 supervisord > [7516987.983590] [59620] 0 59620 18074 1395 188416 0 -998 sshd > [7516987.983594] [59622] 0 59622 18680 6679 188416 0 -998 python > [7516987.983598] [59624] 0 59624 1859266 5161 548864 0 -998 odin-agent > [7516987.983600] [59625] 0 59625 707223 9248 983040 0 -998 filebeat > [7516987.983604] [59627] 0 59627 416433 64239 774144 0 -998 odin-log-agent > [7516987.983607] [59631] 0 59631 180671 15012 385024 0 -998 python3 > [7516987.983612] [61396] 0 61396 791287 3189 352256 0 -998 client > [7516987.983615] [61641] 0 61641 1844642 29089 946176 0 -998 client > [7516987.983765] [ 9236] 0 9236 2642 467 53248 0 -998 php_scanner > [7516987.983911] [42898] 0 42898 15543 838 167936 0 -998 su > [7516987.983915] [42900] 1000 42900 3673 867 77824 0 -998 exec_script_vr2 > [7516987.983918] [42925] 1000 42925 36475 19033 335872 0 -998 python > [7516987.983921] [57146] 1000 57146 3673 848 73728 0 -998 exec_script_J2p > [7516987.983925] [57195] 1000 57195 186359 22958 491520 0 -998 python2 > [7516987.983928] [58376] 1000 58376 275764 14402 290816 0 -998 rosmaster > [7516987.983931] [58395] 1000 58395 155166 4449 245760 0 -998 rosout > [7516987.983935] [58406] 1000 58406 18285584 3967322 37101568 0 -998 data_sim > [7516987.984221] oom-kill:constraint=CONSTRAINT_MEMCG,nodemask=(null),cpuset=3aa16c9482ae3a6f6b78bda68a55d32c87c99b985e0f11331cddf05af6c4d753,mems_allowed=0-1,oom_memcg=/kubepods/podf1c273d3-9b36-11ea-b3df-246e9693c184,task_memcg=/kubepods/podf1c273d3-9b36-11ea-b3df-246e9693c184/1f246a3eeea8f70bf91141eeaf1805346a666e225f823906485ea0b6c37dfc3d,task=pause,pid=5740,uid=0 > [7516987.984254] Memory cgroup out of memory: Killed process 5740 (pause) total-vm:1028kB, anon-rss:4kB, file-rss:0kB, shmem-rss:0kB > [7516988.092344] oom_reaper: reaped process 5740 (pause), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB > > We can find that the first scanned process 5740 (pause) was killed, but its > rss is only one page. That is because, when we calculate the oom badness in > oom_badness(), we always ignore the negtive point and convert all of these > negtive points to 1. Now as oom_score_adj of all the processes in this > targeted memcg have the same value -998, the points of these processes are > all negtive value. As a result, the first scanned process will be killed. > > The oom_socre_adj (-998) in this memcg is set by kubelet, because it is a > a Guaranteed pod, which has higher priority to prevent from being killed by > system oom. > > To fix this issue, we should make the calculation of oom point more > accurate. We can achieve it by convert the chosen_point from 'unsigned > long' to 'long'. > > Signed-off-by: Yafang Shao <laoar.shao@gmail.com> > --- > drivers/tty/sysrq.c | 1 + > fs/proc/base.c | 7 ++++++- > include/linux/oom.h | 4 ++-- > mm/memcontrol.c | 1 + > mm/oom_kill.c | 19 ++++++++----------- > mm/page_alloc.c | 1 + > 6 files changed, 19 insertions(+), 14 deletions(-) > > diff --git a/drivers/tty/sysrq.c b/drivers/tty/sysrq.c > index 7c95afa9..e83fd46 100644 > --- a/drivers/tty/sysrq.c > +++ b/drivers/tty/sysrq.c > @@ -382,6 +382,7 @@ static void moom_callback(struct work_struct *ignored) > .memcg = NULL, > .gfp_mask = gfp_mask, > .order = -1, > + .chosen_points = LONG_MIN, It would be better to do the initialization only once when we start evaluating tasks (select_bad_process). > }; > > mutex_lock(&oom_lock); > diff --git a/fs/proc/base.c b/fs/proc/base.c > index d86c0af..bf16406 100644 > --- a/fs/proc/base.c > +++ b/fs/proc/base.c > @@ -551,8 +551,13 @@ static int proc_oom_score(struct seq_file *m, struct pid_namespace *ns, > { > unsigned long totalpages = totalram_pages() + total_swap_pages; > unsigned long points = 0; > + long badness; > > - points = oom_badness(task, totalpages) * 1000 / totalpages; > + badness = oom_badness(task, totalpages); > + if (badness != LONG_MIN) { > + /* Let's keep the range of points as [0, 2000]. */ > + points = (1000 + badness * 1000 / (long)totalpages) * 2 / 3; > + } > seq_printf(m, "%lu\n", points); This doesn't really work for OOM_SCORE_ADJ_MIN cases because they will simply print LONG_MIN rather than 0. So you want /* * Special case OOM_SCORE_ADJ_MIN for all others scale the * badness value into [0, 2000] range which we have been * exporting for a long time so userspace might depend on it */ if (badness == LONG_MIN) badness = 0; else points = (1000 + badness * 1000 / (long)totalpages) * 2 / 3 FTR. In my other email I was proposing to scale usage to the [-1000, 1000] range by points = adj + usage * 1000/ totalpages this would make the math slightly easier to follow but then I've realized that this would be much less precise so what you have is better. Btw. we used to do that in the past until a7f638f999ff4 which has changed that for this very reason. > @@ -107,7 +107,7 @@ static inline vm_fault_t check_stable_address_space(struct mm_struct *mm) > > bool __oom_reap_task_mm(struct mm_struct *mm); > > -extern unsigned long oom_badness(struct task_struct *p, > +long oom_badness(struct task_struct *p, > unsigned long totalpages); This is not really necessary. With that being addressed, you can add Acked-by: Michal Hocko <mhocko@suse.com> Thanks!
On Thu, Jul 09, 2020 at 11:53:07AM -0400, Yafang Shao wrote: > Recently we found an issue on our production environment that when memcg > oom is triggered the oom killer doesn't chose the process with largest > resident memory but chose the first scanned process. Note that all > processes in this memcg have the same oom_score_adj, so the oom killer > should chose the process with largest resident memory. > > Bellow is part of the oom info, which is enough to analyze this issue. > [7516987.983223] memory: usage 16777216kB, limit 16777216kB, failcnt 52843037 > [7516987.983224] memory+swap: usage 16777216kB, limit 9007199254740988kB, failcnt 0 > [7516987.983225] kmem: usage 301464kB, limit 9007199254740988kB, failcnt 0 > [...] > [7516987.983293] [ pid ] uid tgid total_vm rss pgtables_bytes swapents oom_score_adj name > [7516987.983510] [ 5740] 0 5740 257 1 32768 0 -998 pause > [7516987.983574] [58804] 0 58804 4594 771 81920 0 -998 entry_point.bas > [7516987.983577] [58908] 0 58908 7089 689 98304 0 -998 cron > [7516987.983580] [58910] 0 58910 16235 5576 163840 0 -998 supervisord > [7516987.983590] [59620] 0 59620 18074 1395 188416 0 -998 sshd > [7516987.983594] [59622] 0 59622 18680 6679 188416 0 -998 python > [7516987.983598] [59624] 0 59624 1859266 5161 548864 0 -998 odin-agent > [7516987.983600] [59625] 0 59625 707223 9248 983040 0 -998 filebeat > [7516987.983604] [59627] 0 59627 416433 64239 774144 0 -998 odin-log-agent > [7516987.983607] [59631] 0 59631 180671 15012 385024 0 -998 python3 > [7516987.983612] [61396] 0 61396 791287 3189 352256 0 -998 client > [7516987.983615] [61641] 0 61641 1844642 29089 946176 0 -998 client > [7516987.983765] [ 9236] 0 9236 2642 467 53248 0 -998 php_scanner > [7516987.983911] [42898] 0 42898 15543 838 167936 0 -998 su > [7516987.983915] [42900] 1000 42900 3673 867 77824 0 -998 exec_script_vr2 > [7516987.983918] [42925] 1000 42925 36475 19033 335872 0 -998 python > [7516987.983921] [57146] 1000 57146 3673 848 73728 0 -998 exec_script_J2p > [7516987.983925] [57195] 1000 57195 186359 22958 491520 0 -998 python2 > [7516987.983928] [58376] 1000 58376 275764 14402 290816 0 -998 rosmaster > [7516987.983931] [58395] 1000 58395 155166 4449 245760 0 -998 rosout > [7516987.983935] [58406] 1000 58406 18285584 3967322 37101568 0 -998 data_sim > [7516987.984221] oom-kill:constraint=CONSTRAINT_MEMCG,nodemask=(null),cpuset=3aa16c9482ae3a6f6b78bda68a55d32c87c99b985e0f11331cddf05af6c4d753,mems_allowed=0-1,oom_memcg=/kubepods/podf1c273d3-9b36-11ea-b3df-246e9693c184,task_memcg=/kubepods/podf1c273d3-9b36-11ea-b3df-246e9693c184/1f246a3eeea8f70bf91141eeaf1805346a666e225f823906485ea0b6c37dfc3d,task=pause,pid=5740,uid=0 > [7516987.984254] Memory cgroup out of memory: Killed process 5740 (pause) total-vm:1028kB, anon-rss:4kB, file-rss:0kB, shmem-rss:0kB > [7516988.092344] oom_reaper: reaped process 5740 (pause), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB > > We can find that the first scanned process 5740 (pause) was killed, but its > rss is only one page. That is because, when we calculate the oom badness in > oom_badness(), we always ignore the negtive point and convert all of these > negtive points to 1. Now as oom_score_adj of all the processes in this > targeted memcg have the same value -998, the points of these processes are > all negtive value. As a result, the first scanned process will be killed. > > The oom_socre_adj (-998) in this memcg is set by kubelet, because it is a > a Guaranteed pod, which has higher priority to prevent from being killed by > system oom. > > To fix this issue, we should make the calculation of oom point more > accurate. We can achieve it by convert the chosen_point from 'unsigned > long' to 'long'. > > Signed-off-by: Yafang Shao <laoar.shao@gmail.com> Reverting this commit fixed the crash below while recovering from kenrel OOM, Reproducer: # git clone https://github.com/cailca/linux-mm # cd linux-mm; make # ./random 5 (just mmap and CoW) [ 261.362982][ T1770] Out of memory and no killable processes... [ 261.391982][ T1770] Kernel panic - not syncing: System is deadlocked on memory [ 261.428413][ T1770] CPU: 44 PID: 1770 Comm: random Not tainted 5.8.0-rc4-next-20200710 #1 [ 261.467527][ T1770] Hardware name: HP ProLiant BL660c Gen9, BIOS I38 10/17/2018 [ 261.501318][ T1770] Call Trace: [ 261.516258][ T1770] dump_stack+0x9d/0xe0 [ 261.534637][ T1770] panic+0x2a1/0x522 [ 261.551839][ T1770] ? __warn_printk+0xd6/0xd6 [ 261.572119][ T1770] ? printk+0x9a/0xc0 [ 261.589553][ T1770] ? dump_header.cold.33+0x166/0x16b [ 261.613525][ T1770] ? llist_add_batch+0x52/0x90 [ 261.635103][ T1770] out_of_memory.cold.42+0x3b/0x4f [ 261.657861][ T1770] ? oom_killer_disable+0x210/0x210 [ 261.681348][ T1770] ? mutex_trylock+0x23a/0x2a0 [ 261.703208][ T1770] __alloc_pages_slowpath.constprop.73+0x19e8/0x2420 [ 261.733036][ T1770] ? rcu_read_lock_sched_held+0xaa/0xd0 [ 261.757853][ T1770] ? warn_alloc+0x120/0x120 [ 261.777793][ T1770] ? __alloc_pages_nodemask+0x3b7/0x7d0 [ 261.802347][ T1770] ? lock_downgrade+0x720/0x720 [ 261.823934][ T1770] __alloc_pages_nodemask+0x66b/0x7d0 [ 261.847732][ T1770] ? __alloc_pages_slowpath.constprop.73+0x2420/0x2420 [ 261.878316][ T1770] ? up_read+0x1b2/0x755 [ 261.995886][ T1770] pagecache_get_page+0x15e/0x550 [ 262.017892][ T1770] ? filemap_fault+0x1321/0x1dd0 [ 262.039864][ T1770] filemap_fault+0xafb/0x1dd0 [ 262.060633][ T1770] __xfs_filemap_fault.constprop.6+0x119/0x490 [ 262.088029][ T1770] __do_fault+0xf5/0x550 [ 262.106738][ T1770] ? handle_mm_fault+0x26ec/0x2ef0 [ 262.129316][ T1770] handle_mm_fault+0x21de/0x2ef0 [ 262.151211][ T1770] ? copy_page_range+0x580/0x580 [ 262.173193][ T1770] do_user_addr_fault+0x2af/0x824 [ 262.195424][ T1770] ? trace_hardirqs_off_finish+0x1e7/0x250 [ 262.220967][ T1770] exc_page_fault+0x56/0xa0 [ 262.240805][ T1770] ? asm_exc_page_fault+0x8/0x30 [ 262.262230][ T1770] asm_exc_page_fault+0x1e/0x30 [ 262.283642][ T1770] RIP: 0033:0x7f3427adf120 [ 262.303060][ T1770] Code: Bad RIP value. [ 262.321132][ T1770] RSP: 002b:00007f34279eaeb8 EFLAGS: 00010206 [ 262.348422][ T1770] RAX: 00007f33ee9d5000 RBX: 0000000000000000 RCX: 00007f3427ae28c7 [ 262.384779][ T1770] RDX: 00007f310353e000 RSI: 00000003234abf33 RDI: 0000000000000000 [ 262.424217][ T1770] RBP: 00007f34279eaef0 R08: 00000000ffffffff R09: 0000000000000000 [ 262.463773][ T1770] R10: 0000000000000004 R11: 00007f3427adf120 R12: 00007ffc762430ce [ 262.500186][ T1770] R13: 00007ffc762430cf R14: 0000000000000000 R15: 00007f34279eafc0 [ 262.536172][ T1770] Kernel Offset: 0x31600000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff) [ 262.588968][ T1770] ---[ end Kernel panic - not syncing: System is deadlocked on memory ]--- More Console outputs prior to the crash, [ 246.610518][ T1778] random invoked oom-killer: gfp_mask=0x100dca(GFP_HIGHUSER_MOVABLE|__GFP_ZERO), order=0, oom_score_adj=0 [ 246.668114][ T1778] CPU: 16 PID: 1778 Comm: random Not tainted 5.8.0-rc4-next-20200710 #1 [ 246.706299][ T1778] Hardware name: HP ProLiant BL660c Gen9, BIOS I38 10/17/2018 [ 246.739649][ T1778] Call Trace: [ 246.754012][ T1778] dump_stack+0x9d/0xe0 [ 246.772361][ T1778] dump_header+0x104/0x830 [ 246.792033][ T1778] ? ___ratelimit+0x1c6/0x400 [ 246.812735][ T1778] oom_kill_process+0x11f/0x330 [ 246.834146][ T1778] out_of_memory+0x63f/0x1320 [ 246.856993][ T1778] ? oom_killer_disable+0x210/0x210 [ 246.883648][ T1778] ? mutex_trylock+0x23a/0x2a0 [ 246.907003][ T1778] __alloc_pages_slowpath.constprop.73+0x19e8/0x2420 [ 246.938698][ T1778] ? __lock_acquire+0xc57/0x4da0 [ 246.963601][ T1778] ? warn_alloc+0x120/0x120 [ 246.985262][ T1778] ? __alloc_pages_nodemask+0x3b7/0x7d0 [ 247.011387][ T1778] ? lock_downgrade+0x720/0x720 [ 247.034009][ T1778] __alloc_pages_nodemask+0x66b/0x7d0 [ 247.058147][ T1778] ? handle_mm_fault+0x1f3/0x2ef0 [ 247.080475][ T1778] ? __alloc_pages_slowpath.constprop.73+0x2420/0x2420 [ 247.111122][ T1778] ? mark_held_locks+0xb0/0x110 [ 247.133124][ T1778] alloc_pages_vma+0xa5/0x4c0 [ 247.156360][ T1778] handle_mm_fault+0x1bdf/0x2ef0 [ 247.179379][ T1778] ? copy_page_range+0x580/0x580 [ 247.200926][ T1778] do_user_addr_fault+0x2af/0x824 [ 247.223331][ T1778] exc_page_fault+0x56/0xa0 [ 247.243202][ T1778] ? asm_exc_page_fault+0x8/0x30 [ 247.265347][ T1778] asm_exc_page_fault+0x1e/0x30 [ 247.286788][ T1778] RIP: 0033:0x402cb9 [ 247.303935][ T1778] Code: Bad RIP value. [ 247.321938][ T1778] RSP: 002b:00007f1e2c92eec0 EFLAGS: 00010206 [ 247.349179][ T1778] RAX: 00007f1a7d0c6000 RBX: 0000000000000000 RCX: 00007f3427ae28c7 [ 247.385115][ T1778] RDX: 00007f17e4fd6000 RSI: 00000003234abf33 RDI: 0000000000000000 [ 247.420982][ T1778] RBP: 00007f1e2c92eef0 R08: 00000000ffffffff R09: 0000000000000000 [ 247.456715][ T1778] R10: 0000000000000022 R11: 0000000000000246 R12: 00007ffc762430ce [ 247.492327][ T1778] R13: 00007ffc762430cf R14: 0000000000000000 R15: 00007f1e2c92efc0 [ 247.528232][ T1778] Mem-Info: [ 247.541762][ T1778] active_anon:25261757 inactive_anon:2144151 isolated_anon:0 [ 247.541762][ T1778] active_file:138 inactive_file:81 isolated_file:21 [ 247.541762][ T1778] unevictable:0 dirty:0 writeback:0 [ 247.541762][ T1778] slab_reclaimable:26958 slab_unreclaimable:1049075 [ 247.541762][ T1778] mapped:1189 shmem:2353 pagetables:55339 bounce:0 [ 247.541762][ T1778] free:69553 free_pcp:1530 free_cma:0 [ 247.716529][ T1778] Node 0 active_anon:12887656kB inactive_anon:1955060kB active_file:0kB inactive_file:24kB unevictable:0kB isolated(anon):0kB isolated(file):0kB mapped:8kB dirty:0kB writeback:0kB shmem:504kB shmem_thp: 0kB shmem_pmdmapped: 0kB anon_thp: 13785088kB writeback_tmp:0kB kernel_stack:5024kB all_unreclaimable? no [ 247.848980][ T1778] Node 1 active_anon:29008916kB inactive_anon:3085700kB active_file:0kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB mapped:0kB dirty:0kB writeback:0kB shmem:364kB shmem_thp: 0kB shmem_pmdmapped: 0kB anon_thp: 28985344kB writeback_tmp:0kB kernel_stack:4128kB all_unreclaimable? yes [ 247.980966][ T1778] Node 2 active_anon:29063680kB inactive_anon:1757120kB active_file:212kB inactive_file:296kB unevictable:0kB isolated(anon):0kB isolated(file):8kB mapped:4696kB dirty:0kB writeback:0kB shmem:8292kB shmem_thp: 0kB shmem_pmdmapped: 0kB anon_thp: 29159424kB writeback_tmp:0kB kernel_stack:5568kB all_unreclaimable? no [ 248.117727][ T1778] Node 3 active_anon:30090240kB inactive_anon:1778724kB active_file:268kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB mapped:0kB dirty:0kB writeback:0kB shmem:252kB shmem_thp: 0kB shmem_pmdmapped: 0kB anon_thp: 30033920kB writeback_tmp:0kB kernel_stack:4480kB all_unreclaimable? yes [ 248.254628][ T1778] Node 0 DMA free:15904kB min:16kB low:28kB high:40kB reserved_highatomic:0KB active_anon:0kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB writepending:0kB present:15992kB managed:15904kB mlocked:0kB pagetables:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB [ 248.378815][ T1778] lowmem_reserve[]: 0 1810 15429 15429 [ 248.403362][ T1778] Node 0 DMA32 free:56160kB min:2180kB low:4032kB high:5884kB reserved_highatomic:0KB active_anon:1820964kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB writepending:0kB present:1948156kB managed:1880848kB mlocked:0kB pagetables:8kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB [ 248.538800][ T1778] lowmem_reserve[]: 0 0 13619 13619 [ 248.562446][ T1778] Node 0 Normal free:28424kB min:16404kB low:30348kB high:44292kB reserved_highatomic:12288KB active_anon:11066692kB inactive_anon:1955060kB active_file:0kB inactive_file:8kB unevictable:0kB writepending:0kB present:31457280kB managed:13951408kB mlocked:0kB pagetables:29868kB bounce:0kB free_pcp:828kB local_pcp:4kB free_cma:0kB [ 248.708696][ T1778] lowmem_reserve[]: 0 0 0 0 [ 248.728635][ T1778] Node 1 Normal free:56784kB min:38852kB low:71876kB high:104900kB reserved_highatomic:18432KB active_anon:29008916kB inactive_anon:3085700kB active_file:40kB inactive_file:0kB unevictable:0kB writepending:0kB present:33554432kB managed:33027152kB mlocked:0kB pagetables:63652kB bounce:0kB free_pcp:500kB local_pcp:0kB free_cma:0kB [ 248.872458][ T1778] lowmem_reserve[]: 0 0 0 0 [ 248.892503][ T1778] Node 2 Normal free:53868kB min:44996kB low:78020kB high:111044kB reserved_highatomic:10240KB active_anon:29071368kB inactive_anon:1757120kB active_file:644kB inactive_file:1132kB unevictable:0kB writepending:0kB present:33554432kB managed:33027156kB mlocked:0kB pagetables:63888kB bounce:0kB free_pcp:16036kB local_pcp:248kB free_cma:0kB [ 249.038452][ T1778] lowmem_reserve[]: 0 0 0 0 [ 249.058650][ T1778] Node 3 Normal free:41936kB min:38852kB low:71876kB high:104900kB reserved_highatomic:4096KB active_anon:30090280kB inactive_anon:1778724kB active_file:248kB inactive_file:0kB unevictable:0kB writepending:0kB present:33554432kB managed:33026648kB mlocked:0kB pagetables:63992kB bounce:0kB free_pcp:2160kB local_pcp:508kB free_cma:0kB [ 249.205090][ T1778] lowmem_reserve[]: 0 0 0 0 [ 249.226088][ T1778] Node 0 DMA: 2*4kB (U) 1*8kB (U) 1*16kB (U) 0*32kB 2*64kB (U) 1*128kB (U) 1*256kB (U) 0*512kB 1*1024kB (U) 1*2048kB (M) 3*4096kB (M) = 15904kB [ 249.292333][ T1778] Node 0 DMA32: 2*4kB (U) 3*8kB (UM) 4*16kB (UM) 2*32kB (M) 3*64kB (UM) 2*128kB (UM) 5*256kB (UM) 6*512kB (M) 6*1024kB (UM) 0*2048kB 11*4096kB (M) = 56160kB [ 249.365090][ T1778] Node 0 Normal: 117*4kB (UE) 137*8kB (UE) 469*16kB (UME) 179*32kB (UE) 58*64kB (UE) 11*128kB (UE) 0*256kB 4*512kB (M) 6*1024kB (M) 0*2048kB 0*4096kB = 28108kB [ 249.437775][ T1778] Node 1 Normal: 162*4kB (UM) 83*8kB (U) 447*16kB (UME) 52*32kB (UME) 20*64kB (UME) 4*128kB (UME) 3*256kB (ME) 48*512kB (UM) 19*1024kB (M) 0*2048kB 0*4096kB = 56720kB [ 249.513107][ T1778] Node 2 Normal: 594*4kB (UM) 347*8kB (UM) 1014*16kB (UE) 99*32kB (UME) 79*64kB (UME) 144*128kB (UM) 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 48032kB [ 249.583100][ T1778] Node 3 Normal: 225*4kB (U) 174*8kB (UM) 814*16kB (UME) 170*32kB (UE) 81*64kB (UE) 63*128kB (UE) 8*256kB (UE) 1*512kB (M) 5*1024kB (UM) 0*2048kB 0*4096kB = 41684kB [ 249.657751][ T1778] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB [ 249.704216][ T1778] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB [ 249.747344][ T1778] Node 1 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB [ 249.789951][ T1778] Node 1 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB [ 249.833177][ T1778] Node 2 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB [ 249.876892][ T1778] Node 2 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB [ 249.918701][ T1778] Node 3 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB [ 249.961959][ T1778] Node 3 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB [ 250.003640][ T1778] 830052 total pagecache pages [ 250.024344][ T1778] 827547 pages in swap cache [ 250.044627][ T1778] Swap cache stats: add 1048489, delete 221144, find 135/169 [ 250.077908][ T1778] Free swap = 0kB [ 250.094176][ T1778] Total swap = 4194300kB [ 250.112858][ T1778] 33521181 pages RAM [ 250.130761][ T1778] 0 pages HighMem/MovableOnly [ 250.152483][ T1778] 4788902 pages reserved [ 250.171254][ T1778] 0 pages hwpoisoned [ 250.189104][ T1778] Tasks state (memory values in pages): [ 250.216265][ T1778] [ pid ] uid tgid total_vm rss pgtables_bytes swapents oom_score_adj name [ 250.260075][ T1778] [ 893] 0 893 27572 1402 249856 0 0 systemd-journal [ 250.305110][ T1778] [ 928] 0 928 26661 567 229376 0 -1000 systemd-udevd [ 250.349061][ T1778] [ 1147] 0 1147 31227 173 143360 31 0 irqbalance [ 250.392400][ T1778] [ 1155] 81 1155 18672 144 167936 53 -900 dbus-daemon [ 250.435855][ T1778] [ 1157] 998 1157 508379 1349 385024 902 0 polkitd [ 250.477995][ T1778] [ 1158] 0 1158 53104 391 421888 101 0 sssd [ 250.518371][ T1778] [ 1159] 0 1159 97608 478 368640 129 0 NetworkManager [ 250.562804][ T1778] [ 1162] 994 1162 95357 173 229376 35 0 rngd [ 250.604826][ T1778] [ 1164] 995 1164 7366 48 94208 19 0 chronyd [ 250.648036][ T1778] [ 1169] 0 1169 7337 374 86016 742 0 ksmtuned [ 250.690923][ T1778] [ 1200] 0 1200 23773 130 204800 98 -1000 sshd [ 250.736317][ T1778] [ 1203] 0 1203 106493 3161 434176 1244 0 tuned [ 250.777240][ T1778] [ 1219] 0 1219 10394 100 106496 32 0 rhsmcertd [ 250.819134][ T1778] [ 1238] 0 1238 54766 421 438272 216 0 sssd_be [ 250.861075][ T1778] [ 1246] 0 1246 55309 406 462848 0 0 sssd_nss [ 250.903953][ T1778] [ 1255] 0 1255 24020 219 208896 29 0 systemd-logind [ 250.947724][ T1778] [ 1263] 0 1263 3918 25 77824 8 0 agetty [ 250.987890][ T1778] [ 1272] 0 1272 23433 337 225280 17 0 systemd [ 251.029612][ T1778] [ 1285] 0 1285 37329 427 299008 259 0 (sd-pam) [ 251.072439][ T1778] [ 1517] 0 1517 54398 193 184320 46 0 rsyslogd [ 251.116327][ T1778] [ 1527] 0 1527 9024 196 102400 25 0 crond [ 251.158985][ T1778] [ 1531] 0 1531 63396 180 118784 24 -1000 restraintd [ 251.204944][ T1778] [ 1541] 0 1541 6312 71 81920 17 -1000 10_bash_login [ 251.254455][ T1778] [ 1568] 0 1568 2425 6 61440 49 0 make [ 251.296110][ T1778] [ 1572] 0 1572 6312 92 81920 10 0 runtest.sh [ 251.340690][ T1778] [ 1658] 0 1658 38227 293 299008 19 0 sshd [ 251.382404][ T1778] [ 1674] 0 1674 38227 264 294912 51 0 sshd [ 251.424319][ T1778] [ 1686] 0 1686 6346 45 94208 105 0 bash [ 251.466076][ T1778] [ 1768] 0 1768 1639 0 61440 28 0 random [ 251.508756][ T1778] [ 1769] 0 1769 32924969 26568800 221474816 1042033 0 random [ 251.551728][ T1778] [ 1793] 0 1793 1821 25 61440 0 0 sleep [ 251.593685][ T1778] [ 1798] 0 1798 1821 15 57344 0 0 sleep [ 251.635656][ T1778] oom-kill:constraint=CONSTRAINT_NONE,nodemask=(null),cpuset=/,mems_allowed=0-3,global_oom,task_memcg=/system.slice/dbus.service,task=dbus-daemon,pid=1155,uid=81 [ 251.716666][ T1778] Out of memory: Killed process 1155 (dbus-daemon) total-vm:74688kB, anon-rss:572kB, file-rss:4kB, shmem-rss:0kB, UID:81 pgtables:164kB oom_score_adj:-900 [ 251.795110][ T353] oom_reaper: reaped process 1155 (dbus-daemon), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB [ 251.877615][ T1793] sleep (1793) used greatest stack depth: 18696 bytes left [ 256.237437][ T1770] random invoked oom-killer: gfp_mask=0x100cca(GFP_HIGHUSER_MOVABLE), order=0, oom_score_adj=0 [ 256.289783][ T1770] CPU: 44 PID: 1770 Comm: random Not tainted 5.8.0-rc4-next-20200710 #1 [ 256.329061][ T1770] Hardware name: HP ProLiant BL660c Gen9, BIOS I38 10/17/2018 [ 256.363958][ T1770] Call Trace: [ 256.378672][ T1770] dump_stack+0x9d/0xe0 [ 256.397817][ T1770] dump_header+0x104/0x830 [ 256.418135][ T1770] ? oom_evaluate_task+0x2fb/0x550 [ 256.441580][ T1770] out_of_memory+0xf17/0x1320 [ 256.463209][ T1770] ? oom_killer_disable+0x210/0x210 [ 256.486890][ T1770] ? mutex_trylock+0x23a/0x2a0 [ 256.509086][ T1770] __alloc_pages_slowpath.constprop.73+0x19e8/0x2420 [ 256.539952][ T1770] ? rcu_read_lock_sched_held+0xaa/0xd0 [ 256.565885][ T1770] ? warn_alloc+0x120/0x120 [ 256.586540][ T1770] ? __alloc_pages_nodemask+0x3b7/0x7d0 [ 256.611648][ T1770] ? lock_downgrade+0x720/0x720 [ 256.633613][ T1770] __alloc_pages_nodemask+0x66b/0x7d0 [ 256.658355][ T1770] ? __alloc_pages_slowpath.constprop.73+0x2420/0x2420 [ 256.690335][ T1770] ? up_read+0x1b2/0x755 [ 256.709484][ T1770] pagecache_get_page+0x15e/0x550 [ 256.732772][ T1770] ? filemap_fault+0x1321/0x1dd0 [ 256.755246][ T1770] filemap_fault+0xafb/0x1dd0 [ 256.778313][ T1770] __xfs_filemap_fault.constprop.6+0x119/0x490 [ 256.808802][ T1770] __do_fault+0xf5/0x550 [ 256.828058][ T1770] ? handle_mm_fault+0x26ec/0x2ef0 [ 256.851543][ T1770] handle_mm_fault+0x21de/0x2ef0 [ 256.874265][ T1770] ? copy_page_range+0x580/0x580 [ 256.897346][ T1770] do_user_addr_fault+0x2af/0x824 [ 256.919942][ T1770] ? trace_hardirqs_off_finish+0x1e7/0x250 [ 256.947004][ T1770] exc_page_fault+0x56/0xa0 [ 256.967920][ T1770] ? asm_exc_page_fault+0x8/0x30 [ 256.991703][ T1770] asm_exc_page_fault+0x1e/0x30 [ 257.013863][ T1770] RIP: 0033:0x7f3427adf120 [ 257.034132][ T1770] Code: Bad RIP value. [ 257.053045][ T1770] RSP: 002b:00007f34279eaeb8 EFLAGS: 00010206 [ 257.081173][ T1770] RAX: 00007f33ee9d5000 RBX: 0000000000000000 RCX: 00007f3427ae28c7 [ 257.118606][ T1770] RDX: 00007f310353e000 RSI: 00000003234abf33 RDI: 0000000000000000 [ 257.155693][ T1770] RBP: 00007f34279eaef0 R08: 00000000ffffffff R09: 0000000000000000 [ 257.192771][ T1770] R10: 0000000000000004 R11: 00007f3427adf120 R12: 00007ffc762430ce [ 257.229803][ T1770] R13: 00007ffc762430cf R14: 0000000000000000 R15: 00007f34279eafc0 [ 257.267345][ T1770] Mem-Info: [ 257.282676][ T1770] active_anon:25449780 inactive_anon:1965016 isolated_anon:0 [ 257.282676][ T1770] active_file:65 inactive_file:0 isolated_file:0 [ 257.282676][ T1770] unevictable:0 dirty:0 writeback:0 [ 257.282676][ T1770] slab_reclaimable:27023 slab_unreclaimable:1048444 [ 257.282676][ T1770] mapped:392 shmem:2334 pagetables:55543 bounce:0 [ 257.282676][ T1770] free:62065 free_pcp:554 free_cma:0 [ 257.461086][ T1770] Node 0 active_anon:12935916kB inactive_anon:1909368kB active_file:36kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB mapped:12kB dirty:0kB writeback:0kB shmem:504kB shmem_thp: 0kB shmem_pmdmapped: 0kB anon_thp: 13744128kB writeback_tmp:0kB kernel_stack:5056kB all_unreclaimable? yes [ 257.596604][ T1770] Node 1 active_anon:29673204kB inactive_anon:2420336kB active_file:0kB inactive_file:16kB unevictable:0kB isolated(anon):0kB isolated(file):0kB mapped:0kB dirty:0kB writeback:0kB shmem:352kB shmem_thp: 0kB shmem_pmdmapped: 0kB anon_thp: 28454912kB writeback_tmp:0kB kernel_stack:4128kB all_unreclaimable? yes [ 257.729603][ T1770] Node 2 active_anon:29105592kB inactive_anon:1743448kB active_file:80kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB mapped:1664kB dirty:0kB writeback:0kB shmem:8244kB shmem_thp: 0kB shmem_pmdmapped: 0kB anon_thp: 28469248kB writeback_tmp:0kB kernel_stack:5568kB all_unreclaimable? yes [ 257.869589][ T1770] Node 3 active_anon:30084408kB inactive_anon:1786912kB active_file:172kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB mapped:0kB dirty:0kB writeback:0kB shmem:236kB shmem_thp: 0kB shmem_pmdmapped: 0kB anon_thp: 29853696kB writeback_tmp:0kB kernel_stack:4448kB all_unreclaimable? yes [ 258.002724][ T1770] Node 0 DMA free:15904kB min:16kB low:28kB high:40kB reserved_highatomic:0KB active_anon:0kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB writepending:0kB present:15992kB managed:15904kB mlocked:0kB pagetables:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB [ 258.125796][ T1770] lowmem_reserve[]: 0 1810 15429 15429 [ 258.150081][ T1770] Node 0 DMA32 free:56160kB min:2180kB low:4032kB high:5884kB reserved_highatomic:0KB active_anon:1820964kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB writepending:0kB present:1948156kB managed:1880848kB mlocked:0kB pagetables:8kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB [ 258.282380][ T1770] lowmem_reserve[]: 0 0 13619 13619 [ 258.306345][ T1770] Node 0 Normal free:27708kB min:16404kB low:30348kB high:44292kB reserved_highatomic:12288KB active_anon:11114952kB inactive_anon:1909368kB active_file:0kB inactive_file:0kB unevictable:0kB writepending:0kB present:31457280kB managed:13951408kB mlocked:0kB pagetables:29840kB bounce:0kB free_pcp:1100kB local_pcp:0kB free_cma:0kB [ 258.451411][ T1770] lowmem_reserve[]: 0 0 0 0 [ 258.471323][ T1770] Node 1 Normal free:56716kB min:38852kB low:71876kB high:104900kB reserved_highatomic:18432KB active_anon:29673208kB inactive_anon:2420336kB active_file:0kB inactive_file:0kB unevictable:0kB writepending:0kB present:33554432kB managed:33027152kB mlocked:0kB pagetables:63984kB bounce:0kB free_pcp:724kB local_pcp:0kB free_cma:0kB [ 258.611278][ T1770] lowmem_reserve[]: 0 0 0 0 [ 258.631152][ T1770] Node 2 Normal free:48984kB min:38852kB low:71876kB high:104900kB reserved_highatomic:10240KB active_anon:29105592kB inactive_anon:1743448kB active_file:108kB inactive_file:0kB unevictable:0kB writepending:0kB present:33554432kB managed:33027156kB mlocked:0kB pagetables:64248kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB [ 258.771172][ T1770] lowmem_reserve[]: 0 0 0 0 [ 258.791140][ T1770] Node 3 Normal free:42624kB min:38852kB low:71876kB high:104900kB reserved_highatomic:4096KB active_anon:30084408kB inactive_anon:1786912kB active_file:220kB inactive_file:0kB unevictable:0kB writepending:0kB present:33554432kB managed:33026648kB mlocked:0kB pagetables:64092kB bounce:0kB free_pcp:356kB local_pcp:40kB free_cma:0kB [ 258.939974][ T1770] lowmem_reserve[]: 0 0 0 0 [ 258.959555][ T1770] Node 0 DMA: 2*4kB (U) 1*8kB (U) 1*16kB (U) 0*32kB 2*64kB (U) 1*128kB (U) 1*256kB (U) 0*512kB 1*1024kB (U) 1*2048kB (M) 3*4096kB (M) = 15904kB [ 259.026918][ T1770] Node 0 DMA32: 2*4kB (U) 3*8kB (UM) 4*16kB (UM) 2*32kB (M) 3*64kB (UM) 2*128kB (UM) 5*256kB (UM) 6*512kB (M) 6*1024kB (UM) 0*2048kB 11*4096kB (M) = 56160kB [ 259.099706][ T1770] Node 0 Normal: 69*4kB (UME) 142*8kB (UME) 472*16kB (UME) 203*32kB (UME) 68*64kB (UME) 14*128kB (UME) 0*256kB 0*512kB 6*1024kB (M) 0*2048kB 0*4096kB = 27748kB [ 259.175171][ T1770] Node 1 Normal: 105*4kB (UME) 77*8kB (UME) 438*16kB (UME) 28*32kB (UME) 12*64kB (UME) 3*128kB (UE) 2*256kB (E) 20*512kB (UM) 35*1024kB (M) 0*2048kB 0*4096kB = 56684kB [ 259.252959][ T1770] Node 2 Normal: 526*4kB (UM) 406*8kB (UM) 1107*16kB (UME) 156*32kB (UME) 159*64kB (UME) 82*128kB (UM) 1*256kB (M) 0*512kB 0*1024kB 0*2048kB 0*4096kB = 48984kB [ 259.326006][ T1770] Node 3 Normal: 282*4kB (UM) 199*8kB (UM) 828*16kB (UME) 167*32kB (UME) 83*64kB (UME) 75*128kB (UME) 9*256kB (UME) 0*512kB 4*1024kB (UM) 0*2048kB 0*4096kB = 42624kB [ 259.406007][ T1770] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB [ 259.452447][ T1770] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB [ 259.496823][ T1770] Node 1 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB [ 259.539971][ T1770] Node 1 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB [ 259.581942][ T1770] Node 2 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB [ 259.625200][ T1770] Node 2 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB [ 259.666879][ T1770] Node 3 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB [ 259.709798][ T1770] Node 3 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB [ 259.751068][ T1770] 742508 total pagecache pages [ 259.772116][ T1770] 740154 pages in swap cache [ 259.792405][ T1770] Swap cache stats: add 1048884, delete 308889, find 186/397 [ 259.825548][ T1770] Free swap = 0kB [ 259.843259][ T1770] Total swap = 4194300kB [ 259.863815][ T1770] 33521181 pages RAM [ 259.882683][ T1770] 0 pages HighMem/MovableOnly [ 259.904128][ T1770] 4788902 pages reserved [ 259.922912][ T1770] 0 pages hwpoisoned [ 259.940059][ T1770] Tasks state (memory values in pages): [ 259.964779][ T1770] [ pid ] uid tgid total_vm rss pgtables_bytes swapents oom_score_adj name [ 260.007423][ T1770] [ 893] 0 893 27572 680 249856 0 0 systemd-journal [ 260.052192][ T1770] [ 928] 0 928 26661 555 229376 12 -1000 systemd-udevd [ 260.096152][ T1770] [ 1147] 0 1147 31227 169 143360 35 0 irqbalance [ 260.138753][ T1770] [ 1157] 998 1157 508379 1412 385024 867 0 polkitd [ 260.181125][ T1770] [ 1158] 0 1158 53104 384 421888 108 0 sssd [ 260.222429][ T1770] [ 1159] 0 1159 97608 531 368640 106 0 NetworkManager [ 260.267509][ T1770] [ 1162] 994 1162 95357 172 229376 36 0 rngd [ 260.308103][ T1770] [ 1164] 995 1164 7366 45 94208 22 0 chronyd [ 260.350334][ T1770] [ 1169] 0 1169 7337 371 86016 745 0 ksmtuned [ 260.395957][ T1770] [ 1200] 0 1200 23773 128 204800 100 -1000 sshd [ 260.436408][ T1770] [ 1203] 0 1203 106493 3196 434176 1214 0 tuned [ 260.476988][ T1770] [ 1219] 0 1219 10394 99 106496 33 0 rhsmcertd [ 260.519120][ T1770] [ 1238] 0 1238 54766 420 438272 217 0 sssd_be [ 260.560519][ T1770] [ 1246] 0 1246 55309 406 462848 0 0 sssd_nss [ 260.602315][ T1770] [ 1255] 0 1255 24020 226 208896 23 0 systemd-logind [ 260.646396][ T1770] [ 1263] 0 1263 3918 23 77824 10 0 agetty [ 260.687288][ T1770] [ 1272] 0 1272 23433 342 225280 16 0 systemd [ 260.728686][ T1770] [ 1285] 0 1285 37329 420 299008 266 0 (sd-pam) [ 260.770420][ T1770] [ 1517] 0 1517 54398 166 184320 46 0 rsyslogd [ 260.812273][ T1770] [ 1527] 0 1527 9024 190 102400 31 0 crond [ 260.853747][ T1770] [ 1531] 0 1531 63396 179 118784 25 -1000 restraintd [ 260.901574][ T1770] [ 1541] 0 1541 6312 71 81920 17 -1000 10_bash_login [ 260.945269][ T1770] [ 1568] 0 1568 2425 6 61440 49 0 make [ 260.985486][ T1770] [ 1572] 0 1572 6312 89 81920 12 0 runtest.sh [ 261.028161][ T1770] [ 1658] 0 1658 38227 291 299008 21 0 sshd [ 261.068328][ T1770] [ 1674] 0 1674 38227 262 294912 53 0 sshd [ 261.108488][ T1770] [ 1686] 0 1686 6346 45 94208 105 0 bash [ 261.148602][ T1770] [ 1768] 0 1768 1639 0 61440 28 0 random [ 261.189874][ T1770] [ 1769] 0 1769 32924969 26661064 222212096 1042153 0 random [ 261.233064][ T1770] [ 1798] 0 1798 1821 15 57344 0 0 sleep [ 261.274004][ T1770] [ 1799] 0 1799 15922 63 143360 0 0 systemd-cgroups [ 261.320765][ T1770] [ 1801] 0 1801 1090 23 53248 0 0 sleep [ 261.362982][ T1770] Out of memory and no killable processes... [ 261.391982][ T1770] Kernel panic - not syncing: System is deadlocked on memory [ 261.428413][ T1770] CPU: 44 PID: 1770 Comm: random Not tainted 5.8.0-rc4-next-20200710 #1 [ 261.467527][ T1770] Hardware name: HP ProLiant BL660c Gen9, BIOS I38 10/17/2018 [ 261.501318][ T1770] Call Trace: [ 261.516258][ T1770] dump_stack+0x9d/0xe0 [ 261.534637][ T1770] panic+0x2a1/0x522 [ 261.551839][ T1770] ? __warn_printk+0xd6/0xd6 [ 261.572119][ T1770] ? printk+0x9a/0xc0 [ 261.589553][ T1770] ? dump_header.cold.33+0x166/0x16b [ 261.613525][ T1770] ? llist_add_batch+0x52/0x90 [ 261.635103][ T1770] out_of_memory.cold.42+0x3b/0x4f [ 261.657861][ T1770] ? oom_killer_disable+0x210/0x210 [ 261.681348][ T1770] ? mutex_trylock+0x23a/0x2a0 [ 261.703208][ T1770] __alloc_pages_slowpath.constprop.73+0x19e8/0x2420 [ 261.733036][ T1770] ? rcu_read_lock_sched_held+0xaa/0xd0 [ 261.757853][ T1770] ? warn_alloc+0x120/0x120 [ 261.777793][ T1770] ? __alloc_pages_nodemask+0x3b7/0x7d0 [ 261.802347][ T1770] ? lock_downgrade+0x720/0x720 [ 261.823934][ T1770] __alloc_pages_nodemask+0x66b/0x7d0 [ 261.847732][ T1770] ? __alloc_pages_slowpath.constprop.73+0x2420/0x2420 [ 261.878316][ T1770] ? up_read+0x1b2/0x755 [ 261.995886][ T1770] pagecache_get_page+0x15e/0x550 [ 262.017892][ T1770] ? filemap_fault+0x1321/0x1dd0 [ 262.039864][ T1770] filemap_fault+0xafb/0x1dd0 [ 262.060633][ T1770] __xfs_filemap_fault.constprop.6+0x119/0x490 [ 262.088029][ T1770] __do_fault+0xf5/0x550 [ 262.106738][ T1770] ? handle_mm_fault+0x26ec/0x2ef0 [ 262.129316][ T1770] handle_mm_fault+0x21de/0x2ef0 [ 262.151211][ T1770] ? copy_page_range+0x580/0x580 [ 262.173193][ T1770] do_user_addr_fault+0x2af/0x824 [ 262.195424][ T1770] ? trace_hardirqs_off_finish+0x1e7/0x250 [ 262.220967][ T1770] exc_page_fault+0x56/0xa0 [ 262.240805][ T1770] ? asm_exc_page_fault+0x8/0x30 [ 262.262230][ T1770] asm_exc_page_fault+0x1e/0x30 [ 262.283642][ T1770] RIP: 0033:0x7f3427adf120 [ 262.303060][ T1770] Code: Bad RIP value. [ 262.321132][ T1770] RSP: 002b:00007f34279eaeb8 EFLAGS: 00010206 [ 262.348422][ T1770] RAX: 00007f33ee9d5000 RBX: 0000000000000000 RCX: 00007f3427ae28c7 [ 262.384779][ T1770] RDX: 00007f310353e000 RSI: 00000003234abf33 RDI: 0000000000000000 [ 262.424217][ T1770] RBP: 00007f34279eaef0 R08: 00000000ffffffff R09: 0000000000000000 [ 262.463773][ T1770] R10: 0000000000000004 R11: 00007f3427adf120 R12: 00007ffc762430ce [ 262.500186][ T1770] R13: 00007ffc762430cf R14: 0000000000000000 R15: 00007f34279eafc0 [ 262.536172][ T1770] Kernel Offset: 0x31600000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff) [ 262.588968][ T1770] ---[ end Kernel panic - not syncing: System is deadlocked on memory ]--- > --- > drivers/tty/sysrq.c | 1 + > fs/proc/base.c | 7 ++++++- > include/linux/oom.h | 4 ++-- > mm/memcontrol.c | 1 + > mm/oom_kill.c | 19 ++++++++----------- > mm/page_alloc.c | 1 + > 6 files changed, 19 insertions(+), 14 deletions(-) > > diff --git a/drivers/tty/sysrq.c b/drivers/tty/sysrq.c > index 7c95afa9..e83fd46 100644 > --- a/drivers/tty/sysrq.c > +++ b/drivers/tty/sysrq.c > @@ -382,6 +382,7 @@ static void moom_callback(struct work_struct *ignored) > .memcg = NULL, > .gfp_mask = gfp_mask, > .order = -1, > + .chosen_points = LONG_MIN, > }; > > mutex_lock(&oom_lock); > diff --git a/fs/proc/base.c b/fs/proc/base.c > index d86c0af..bf16406 100644 > --- a/fs/proc/base.c > +++ b/fs/proc/base.c > @@ -551,8 +551,13 @@ static int proc_oom_score(struct seq_file *m, struct pid_namespace *ns, > { > unsigned long totalpages = totalram_pages() + total_swap_pages; > unsigned long points = 0; > + long badness; > > - points = oom_badness(task, totalpages) * 1000 / totalpages; > + badness = oom_badness(task, totalpages); > + if (badness != LONG_MIN) { > + /* Let's keep the range of points as [0, 2000]. */ > + points = (1000 + badness * 1000 / (long)totalpages) * 2 / 3; > + } > seq_printf(m, "%lu\n", points); > > return 0; > diff --git a/include/linux/oom.h b/include/linux/oom.h > index c696c26..f022f58 100644 > --- a/include/linux/oom.h > +++ b/include/linux/oom.h > @@ -48,7 +48,7 @@ struct oom_control { > /* Used by oom implementation, do not set */ > unsigned long totalpages; > struct task_struct *chosen; > - unsigned long chosen_points; > + long chosen_points; > > /* Used to print the constraint info. */ > enum oom_constraint constraint; > @@ -107,7 +107,7 @@ static inline vm_fault_t check_stable_address_space(struct mm_struct *mm) > > bool __oom_reap_task_mm(struct mm_struct *mm); > > -extern unsigned long oom_badness(struct task_struct *p, > +long oom_badness(struct task_struct *p, > unsigned long totalpages); > > extern bool out_of_memory(struct oom_control *oc); > diff --git a/mm/memcontrol.c b/mm/memcontrol.c > index 1962232..df73b30 100644 > --- a/mm/memcontrol.c > +++ b/mm/memcontrol.c > @@ -1559,6 +1559,7 @@ static bool mem_cgroup_out_of_memory(struct mem_cgroup *memcg, gfp_t gfp_mask, > .memcg = memcg, > .gfp_mask = gfp_mask, > .order = order, > + .chosen_points = LONG_MIN, > }; > bool ret; > > diff --git a/mm/oom_kill.c b/mm/oom_kill.c > index 6e94962..2dd5a90 100644 > --- a/mm/oom_kill.c > +++ b/mm/oom_kill.c > @@ -196,17 +196,17 @@ static bool is_dump_unreclaim_slabs(void) > * predictable as possible. The goal is to return the highest value for the > * task consuming the most memory to avoid subsequent oom failures. > */ > -unsigned long oom_badness(struct task_struct *p, unsigned long totalpages) > +long oom_badness(struct task_struct *p, unsigned long totalpages) > { > long points; > long adj; > > if (oom_unkillable_task(p)) > - return 0; > + return LONG_MIN; > > p = find_lock_task_mm(p); > if (!p) > - return 0; > + return LONG_MIN; > > /* > * Do not even consider tasks which are explicitly marked oom > @@ -218,7 +218,7 @@ unsigned long oom_badness(struct task_struct *p, unsigned long totalpages) > test_bit(MMF_OOM_SKIP, &p->mm->flags) || > in_vfork(p)) { > task_unlock(p); > - return 0; > + return LONG_MIN; > } > > /* > @@ -233,11 +233,7 @@ unsigned long oom_badness(struct task_struct *p, unsigned long totalpages) > adj *= totalpages / 1000; > points += adj; > > - /* > - * Never return 0 for an eligible task regardless of the root bonus and > - * oom_score_adj (oom_score_adj can't be OOM_SCORE_ADJ_MIN here). > - */ > - return points > 0 ? points : 1; > + return points; > } > > static const char * const oom_constraint_text[] = { > @@ -336,12 +332,12 @@ static int oom_evaluate_task(struct task_struct *task, void *arg) > * killed first if it triggers an oom, then select it. > */ > if (oom_task_origin(task)) { > - points = ULONG_MAX; > + points = LONG_MAX; > goto select; > } > > points = oom_badness(task, oc->totalpages); > - if (!points || points < oc->chosen_points) > + if (points == LONG_MIN || points < oc->chosen_points) > goto next; > > select: > @@ -1128,6 +1124,7 @@ void pagefault_out_of_memory(void) > .memcg = NULL, > .gfp_mask = 0, > .order = 0, > + .chosen_points = LONG_MIN, > }; > > if (mem_cgroup_oom_synchronize(true)) > diff --git a/mm/page_alloc.c b/mm/page_alloc.c > index e028b87c..8eec9d65 100644 > --- a/mm/page_alloc.c > +++ b/mm/page_alloc.c > @@ -3896,6 +3896,7 @@ void warn_alloc(gfp_t gfp_mask, nodemask_t *nodemask, const char *fmt, ...) > .memcg = NULL, > .gfp_mask = gfp_mask, > .order = order, > + .chosen_points = LONG_MIN, > }; > struct page *page; > > -- > 1.8.3.1 > >
On Fri 10-07-20 08:42:53, Qian Cai wrote: > On Thu, Jul 09, 2020 at 11:53:07AM -0400, Yafang Shao wrote: > > Recently we found an issue on our production environment that when memcg > > oom is triggered the oom killer doesn't chose the process with largest > > resident memory but chose the first scanned process. Note that all > > processes in this memcg have the same oom_score_adj, so the oom killer > > should chose the process with largest resident memory. > > > > Bellow is part of the oom info, which is enough to analyze this issue. > > [7516987.983223] memory: usage 16777216kB, limit 16777216kB, failcnt 52843037 > > [7516987.983224] memory+swap: usage 16777216kB, limit 9007199254740988kB, failcnt 0 > > [7516987.983225] kmem: usage 301464kB, limit 9007199254740988kB, failcnt 0 > > [...] > > [7516987.983293] [ pid ] uid tgid total_vm rss pgtables_bytes swapents oom_score_adj name > > [7516987.983510] [ 5740] 0 5740 257 1 32768 0 -998 pause > > [7516987.983574] [58804] 0 58804 4594 771 81920 0 -998 entry_point.bas > > [7516987.983577] [58908] 0 58908 7089 689 98304 0 -998 cron > > [7516987.983580] [58910] 0 58910 16235 5576 163840 0 -998 supervisord > > [7516987.983590] [59620] 0 59620 18074 1395 188416 0 -998 sshd > > [7516987.983594] [59622] 0 59622 18680 6679 188416 0 -998 python > > [7516987.983598] [59624] 0 59624 1859266 5161 548864 0 -998 odin-agent > > [7516987.983600] [59625] 0 59625 707223 9248 983040 0 -998 filebeat > > [7516987.983604] [59627] 0 59627 416433 64239 774144 0 -998 odin-log-agent > > [7516987.983607] [59631] 0 59631 180671 15012 385024 0 -998 python3 > > [7516987.983612] [61396] 0 61396 791287 3189 352256 0 -998 client > > [7516987.983615] [61641] 0 61641 1844642 29089 946176 0 -998 client > > [7516987.983765] [ 9236] 0 9236 2642 467 53248 0 -998 php_scanner > > [7516987.983911] [42898] 0 42898 15543 838 167936 0 -998 su > > [7516987.983915] [42900] 1000 42900 3673 867 77824 0 -998 exec_script_vr2 > > [7516987.983918] [42925] 1000 42925 36475 19033 335872 0 -998 python > > [7516987.983921] [57146] 1000 57146 3673 848 73728 0 -998 exec_script_J2p > > [7516987.983925] [57195] 1000 57195 186359 22958 491520 0 -998 python2 > > [7516987.983928] [58376] 1000 58376 275764 14402 290816 0 -998 rosmaster > > [7516987.983931] [58395] 1000 58395 155166 4449 245760 0 -998 rosout > > [7516987.983935] [58406] 1000 58406 18285584 3967322 37101568 0 -998 data_sim > > [7516987.984221] oom-kill:constraint=CONSTRAINT_MEMCG,nodemask=(null),cpuset=3aa16c9482ae3a6f6b78bda68a55d32c87c99b985e0f11331cddf05af6c4d753,mems_allowed=0-1,oom_memcg=/kubepods/podf1c273d3-9b36-11ea-b3df-246e9693c184,task_memcg=/kubepods/podf1c273d3-9b36-11ea-b3df-246e9693c184/1f246a3eeea8f70bf91141eeaf1805346a666e225f823906485ea0b6c37dfc3d,task=pause,pid=5740,uid=0 > > [7516987.984254] Memory cgroup out of memory: Killed process 5740 (pause) total-vm:1028kB, anon-rss:4kB, file-rss:0kB, shmem-rss:0kB > > [7516988.092344] oom_reaper: reaped process 5740 (pause), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB > > > > We can find that the first scanned process 5740 (pause) was killed, but its > > rss is only one page. That is because, when we calculate the oom badness in > > oom_badness(), we always ignore the negtive point and convert all of these > > negtive points to 1. Now as oom_score_adj of all the processes in this > > targeted memcg have the same value -998, the points of these processes are > > all negtive value. As a result, the first scanned process will be killed. > > > > The oom_socre_adj (-998) in this memcg is set by kubelet, because it is a > > a Guaranteed pod, which has higher priority to prevent from being killed by > > system oom. > > > > To fix this issue, we should make the calculation of oom point more > > accurate. We can achieve it by convert the chosen_point from 'unsigned > > long' to 'long'. > > > > Signed-off-by: Yafang Shao <laoar.shao@gmail.com> > > Reverting this commit fixed the crash below while recovering from kenrel OOM, I suspect that the previous version of the patch has been tested (in Linux next). Does this version exhibit the same problem? I will have a closer look. Is the full dmesg available somewhere?
On Fri 10-07-20 14:58:54, Michal Hocko wrote:
[...]
> I will have a closer look. Is the full dmesg available somewhere?
Ups, I have missed this:
diff --git a/mm/oom_kill.c b/mm/oom_kill.c
index 2dd5a90f2f81..7f01835862f4 100644
--- a/mm/oom_kill.c
+++ b/mm/oom_kill.c
@@ -306,7 +306,7 @@ static enum oom_constraint constrained_alloc(struct oom_control *oc)
static int oom_evaluate_task(struct task_struct *task, void *arg)
{
struct oom_control *oc = arg;
- unsigned long points;
+ long points;
if (oom_unkillable_task(task))
goto next;
Does it help?
On Fri, Jul 10, 2020 at 9:07 PM Michal Hocko <mhocko@kernel.org> wrote: > > On Fri 10-07-20 14:58:54, Michal Hocko wrote: > [...] > > I will have a closer look. Is the full dmesg available somewhere? > > Ups, I have missed this: > diff --git a/mm/oom_kill.c b/mm/oom_kill.c > index 2dd5a90f2f81..7f01835862f4 100644 > --- a/mm/oom_kill.c > +++ b/mm/oom_kill.c > @@ -306,7 +306,7 @@ static enum oom_constraint constrained_alloc(struct oom_control *oc) > static int oom_evaluate_task(struct task_struct *task, void *arg) > { > struct oom_control *oc = arg; > - unsigned long points; > + long points; > > if (oom_unkillable_task(task)) > goto next; > > Does it help? > -- Thanks Michal. This should be the fix. I did it in our in-house kernel, but when I posted it to upstream I missed it. That's my fault.
On Fri, Jul 10, 2020 at 8:10 PM Michal Hocko <mhocko@kernel.org> wrote: > > On Thu 09-07-20 11:53:07, Yafang Shao wrote: > > Recently we found an issue on our production environment that when memcg > > oom is triggered the oom killer doesn't chose the process with largest > > resident memory but chose the first scanned process. Note that all > > processes in this memcg have the same oom_score_adj, so the oom killer > > should chose the process with largest resident memory. > > > > Bellow is part of the oom info, which is enough to analyze this issue. > > [7516987.983223] memory: usage 16777216kB, limit 16777216kB, failcnt 52843037 > > [7516987.983224] memory+swap: usage 16777216kB, limit 9007199254740988kB, failcnt 0 > > [7516987.983225] kmem: usage 301464kB, limit 9007199254740988kB, failcnt 0 > > [...] > > [7516987.983293] [ pid ] uid tgid total_vm rss pgtables_bytes swapents oom_score_adj name > > [7516987.983510] [ 5740] 0 5740 257 1 32768 0 -998 pause > > [7516987.983574] [58804] 0 58804 4594 771 81920 0 -998 entry_point.bas > > [7516987.983577] [58908] 0 58908 7089 689 98304 0 -998 cron > > [7516987.983580] [58910] 0 58910 16235 5576 163840 0 -998 supervisord > > [7516987.983590] [59620] 0 59620 18074 1395 188416 0 -998 sshd > > [7516987.983594] [59622] 0 59622 18680 6679 188416 0 -998 python > > [7516987.983598] [59624] 0 59624 1859266 5161 548864 0 -998 odin-agent > > [7516987.983600] [59625] 0 59625 707223 9248 983040 0 -998 filebeat > > [7516987.983604] [59627] 0 59627 416433 64239 774144 0 -998 odin-log-agent > > [7516987.983607] [59631] 0 59631 180671 15012 385024 0 -998 python3 > > [7516987.983612] [61396] 0 61396 791287 3189 352256 0 -998 client > > [7516987.983615] [61641] 0 61641 1844642 29089 946176 0 -998 client > > [7516987.983765] [ 9236] 0 9236 2642 467 53248 0 -998 php_scanner > > [7516987.983911] [42898] 0 42898 15543 838 167936 0 -998 su > > [7516987.983915] [42900] 1000 42900 3673 867 77824 0 -998 exec_script_vr2 > > [7516987.983918] [42925] 1000 42925 36475 19033 335872 0 -998 python > > [7516987.983921] [57146] 1000 57146 3673 848 73728 0 -998 exec_script_J2p > > [7516987.983925] [57195] 1000 57195 186359 22958 491520 0 -998 python2 > > [7516987.983928] [58376] 1000 58376 275764 14402 290816 0 -998 rosmaster > > [7516987.983931] [58395] 1000 58395 155166 4449 245760 0 -998 rosout > > [7516987.983935] [58406] 1000 58406 18285584 3967322 37101568 0 -998 data_sim > > [7516987.984221] oom-kill:constraint=CONSTRAINT_MEMCG,nodemask=(null),cpuset=3aa16c9482ae3a6f6b78bda68a55d32c87c99b985e0f11331cddf05af6c4d753,mems_allowed=0-1,oom_memcg=/kubepods/podf1c273d3-9b36-11ea-b3df-246e9693c184,task_memcg=/kubepods/podf1c273d3-9b36-11ea-b3df-246e9693c184/1f246a3eeea8f70bf91141eeaf1805346a666e225f823906485ea0b6c37dfc3d,task=pause,pid=5740,uid=0 > > [7516987.984254] Memory cgroup out of memory: Killed process 5740 (pause) total-vm:1028kB, anon-rss:4kB, file-rss:0kB, shmem-rss:0kB > > [7516988.092344] oom_reaper: reaped process 5740 (pause), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB > > > > We can find that the first scanned process 5740 (pause) was killed, but its > > rss is only one page. That is because, when we calculate the oom badness in > > oom_badness(), we always ignore the negtive point and convert all of these > > negtive points to 1. Now as oom_score_adj of all the processes in this > > targeted memcg have the same value -998, the points of these processes are > > all negtive value. As a result, the first scanned process will be killed. > > > > The oom_socre_adj (-998) in this memcg is set by kubelet, because it is a > > a Guaranteed pod, which has higher priority to prevent from being killed by > > system oom. > > > > To fix this issue, we should make the calculation of oom point more > > accurate. We can achieve it by convert the chosen_point from 'unsigned > > long' to 'long'. > > > > Signed-off-by: Yafang Shao <laoar.shao@gmail.com> > > --- > > drivers/tty/sysrq.c | 1 + > > fs/proc/base.c | 7 ++++++- > > include/linux/oom.h | 4 ++-- > > mm/memcontrol.c | 1 + > > mm/oom_kill.c | 19 ++++++++----------- > > mm/page_alloc.c | 1 + > > 6 files changed, 19 insertions(+), 14 deletions(-) > > > > diff --git a/drivers/tty/sysrq.c b/drivers/tty/sysrq.c > > index 7c95afa9..e83fd46 100644 > > --- a/drivers/tty/sysrq.c > > +++ b/drivers/tty/sysrq.c > > @@ -382,6 +382,7 @@ static void moom_callback(struct work_struct *ignored) > > .memcg = NULL, > > .gfp_mask = gfp_mask, > > .order = -1, > > + .chosen_points = LONG_MIN, > > It would be better to do the initialization only once when we start > evaluating tasks (select_bad_process). > I used to initialize it in constrained_alloc() in the previous version, but I found that is not proper, so I change the initialization in the definitions of each oom_control. select_bad_process() should be a better choice. I will update it. > > }; > > > > mutex_lock(&oom_lock); > > diff --git a/fs/proc/base.c b/fs/proc/base.c > > index d86c0af..bf16406 100644 > > --- a/fs/proc/base.c > > +++ b/fs/proc/base.c > > @@ -551,8 +551,13 @@ static int proc_oom_score(struct seq_file *m, struct pid_namespace *ns, > > { > > unsigned long totalpages = totalram_pages() + total_swap_pages; > > unsigned long points = 0; > > + long badness; > > > > - points = oom_badness(task, totalpages) * 1000 / totalpages; > > + badness = oom_badness(task, totalpages); > > + if (badness != LONG_MIN) { > > + /* Let's keep the range of points as [0, 2000]. */ > > + points = (1000 + badness * 1000 / (long)totalpages) * 2 / 3; > > + } > > seq_printf(m, "%lu\n", points); > > This doesn't really work for OOM_SCORE_ADJ_MIN cases because they > will simply print LONG_MIN rather than 0. > The point has be initlialize to 0: unsigned long points = 0; So for OOM_SCORE_ADJ_MIN cases, it will print 0, seq_printf(m, "%lu\n", points); But.. > So you want > /* > * Special case OOM_SCORE_ADJ_MIN for all others scale the > * badness value into [0, 2000] range which we have been > * exporting for a long time so userspace might depend on it > */ the comment is useful, I will update it with your comment. Thanks. > if (badness == LONG_MIN) > badness = 0; > else > points = (1000 + badness * 1000 / (long)totalpages) * 2 / 3 > > FTR. In my other email I was proposing to scale usage to the [-1000, 1000] > range by > points = adj + usage * 1000/ totalpages > > this would make the math slightly easier to follow but then I've > realized that this would be much less precise so what you have is > better. Btw. we used to do that in the past until a7f638f999ff4 > which has changed that for this very reason. > > > @@ -107,7 +107,7 @@ static inline vm_fault_t check_stable_address_space(struct mm_struct *mm) > > > > bool __oom_reap_task_mm(struct mm_struct *mm); > > > > -extern unsigned long oom_badness(struct task_struct *p, > > +long oom_badness(struct task_struct *p, > > unsigned long totalpages); > > This is not really necessary. > > With that being addressed, you can add > Acked-by: Michal Hocko <mhocko@suse.com> > Thanks for the review.
diff --git a/drivers/tty/sysrq.c b/drivers/tty/sysrq.c index 7c95afa9..e83fd46 100644 --- a/drivers/tty/sysrq.c +++ b/drivers/tty/sysrq.c @@ -382,6 +382,7 @@ static void moom_callback(struct work_struct *ignored) .memcg = NULL, .gfp_mask = gfp_mask, .order = -1, + .chosen_points = LONG_MIN, }; mutex_lock(&oom_lock); diff --git a/fs/proc/base.c b/fs/proc/base.c index d86c0af..bf16406 100644 --- a/fs/proc/base.c +++ b/fs/proc/base.c @@ -551,8 +551,13 @@ static int proc_oom_score(struct seq_file *m, struct pid_namespace *ns, { unsigned long totalpages = totalram_pages() + total_swap_pages; unsigned long points = 0; + long badness; - points = oom_badness(task, totalpages) * 1000 / totalpages; + badness = oom_badness(task, totalpages); + if (badness != LONG_MIN) { + /* Let's keep the range of points as [0, 2000]. */ + points = (1000 + badness * 1000 / (long)totalpages) * 2 / 3; + } seq_printf(m, "%lu\n", points); return 0; diff --git a/include/linux/oom.h b/include/linux/oom.h index c696c26..f022f58 100644 --- a/include/linux/oom.h +++ b/include/linux/oom.h @@ -48,7 +48,7 @@ struct oom_control { /* Used by oom implementation, do not set */ unsigned long totalpages; struct task_struct *chosen; - unsigned long chosen_points; + long chosen_points; /* Used to print the constraint info. */ enum oom_constraint constraint; @@ -107,7 +107,7 @@ static inline vm_fault_t check_stable_address_space(struct mm_struct *mm) bool __oom_reap_task_mm(struct mm_struct *mm); -extern unsigned long oom_badness(struct task_struct *p, +long oom_badness(struct task_struct *p, unsigned long totalpages); extern bool out_of_memory(struct oom_control *oc); diff --git a/mm/memcontrol.c b/mm/memcontrol.c index 1962232..df73b30 100644 --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -1559,6 +1559,7 @@ static bool mem_cgroup_out_of_memory(struct mem_cgroup *memcg, gfp_t gfp_mask, .memcg = memcg, .gfp_mask = gfp_mask, .order = order, + .chosen_points = LONG_MIN, }; bool ret; diff --git a/mm/oom_kill.c b/mm/oom_kill.c index 6e94962..2dd5a90 100644 --- a/mm/oom_kill.c +++ b/mm/oom_kill.c @@ -196,17 +196,17 @@ static bool is_dump_unreclaim_slabs(void) * predictable as possible. The goal is to return the highest value for the * task consuming the most memory to avoid subsequent oom failures. */ -unsigned long oom_badness(struct task_struct *p, unsigned long totalpages) +long oom_badness(struct task_struct *p, unsigned long totalpages) { long points; long adj; if (oom_unkillable_task(p)) - return 0; + return LONG_MIN; p = find_lock_task_mm(p); if (!p) - return 0; + return LONG_MIN; /* * Do not even consider tasks which are explicitly marked oom @@ -218,7 +218,7 @@ unsigned long oom_badness(struct task_struct *p, unsigned long totalpages) test_bit(MMF_OOM_SKIP, &p->mm->flags) || in_vfork(p)) { task_unlock(p); - return 0; + return LONG_MIN; } /* @@ -233,11 +233,7 @@ unsigned long oom_badness(struct task_struct *p, unsigned long totalpages) adj *= totalpages / 1000; points += adj; - /* - * Never return 0 for an eligible task regardless of the root bonus and - * oom_score_adj (oom_score_adj can't be OOM_SCORE_ADJ_MIN here). - */ - return points > 0 ? points : 1; + return points; } static const char * const oom_constraint_text[] = { @@ -336,12 +332,12 @@ static int oom_evaluate_task(struct task_struct *task, void *arg) * killed first if it triggers an oom, then select it. */ if (oom_task_origin(task)) { - points = ULONG_MAX; + points = LONG_MAX; goto select; } points = oom_badness(task, oc->totalpages); - if (!points || points < oc->chosen_points) + if (points == LONG_MIN || points < oc->chosen_points) goto next; select: @@ -1128,6 +1124,7 @@ void pagefault_out_of_memory(void) .memcg = NULL, .gfp_mask = 0, .order = 0, + .chosen_points = LONG_MIN, }; if (mem_cgroup_oom_synchronize(true)) diff --git a/mm/page_alloc.c b/mm/page_alloc.c index e028b87c..8eec9d65 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -3896,6 +3896,7 @@ void warn_alloc(gfp_t gfp_mask, nodemask_t *nodemask, const char *fmt, ...) .memcg = NULL, .gfp_mask = gfp_mask, .order = order, + .chosen_points = LONG_MIN, }; struct page *page;
Recently we found an issue on our production environment that when memcg oom is triggered the oom killer doesn't chose the process with largest resident memory but chose the first scanned process. Note that all processes in this memcg have the same oom_score_adj, so the oom killer should chose the process with largest resident memory. Bellow is part of the oom info, which is enough to analyze this issue. [7516987.983223] memory: usage 16777216kB, limit 16777216kB, failcnt 52843037 [7516987.983224] memory+swap: usage 16777216kB, limit 9007199254740988kB, failcnt 0 [7516987.983225] kmem: usage 301464kB, limit 9007199254740988kB, failcnt 0 [...] [7516987.983293] [ pid ] uid tgid total_vm rss pgtables_bytes swapents oom_score_adj name [7516987.983510] [ 5740] 0 5740 257 1 32768 0 -998 pause [7516987.983574] [58804] 0 58804 4594 771 81920 0 -998 entry_point.bas [7516987.983577] [58908] 0 58908 7089 689 98304 0 -998 cron [7516987.983580] [58910] 0 58910 16235 5576 163840 0 -998 supervisord [7516987.983590] [59620] 0 59620 18074 1395 188416 0 -998 sshd [7516987.983594] [59622] 0 59622 18680 6679 188416 0 -998 python [7516987.983598] [59624] 0 59624 1859266 5161 548864 0 -998 odin-agent [7516987.983600] [59625] 0 59625 707223 9248 983040 0 -998 filebeat [7516987.983604] [59627] 0 59627 416433 64239 774144 0 -998 odin-log-agent [7516987.983607] [59631] 0 59631 180671 15012 385024 0 -998 python3 [7516987.983612] [61396] 0 61396 791287 3189 352256 0 -998 client [7516987.983615] [61641] 0 61641 1844642 29089 946176 0 -998 client [7516987.983765] [ 9236] 0 9236 2642 467 53248 0 -998 php_scanner [7516987.983911] [42898] 0 42898 15543 838 167936 0 -998 su [7516987.983915] [42900] 1000 42900 3673 867 77824 0 -998 exec_script_vr2 [7516987.983918] [42925] 1000 42925 36475 19033 335872 0 -998 python [7516987.983921] [57146] 1000 57146 3673 848 73728 0 -998 exec_script_J2p [7516987.983925] [57195] 1000 57195 186359 22958 491520 0 -998 python2 [7516987.983928] [58376] 1000 58376 275764 14402 290816 0 -998 rosmaster [7516987.983931] [58395] 1000 58395 155166 4449 245760 0 -998 rosout [7516987.983935] [58406] 1000 58406 18285584 3967322 37101568 0 -998 data_sim [7516987.984221] oom-kill:constraint=CONSTRAINT_MEMCG,nodemask=(null),cpuset=3aa16c9482ae3a6f6b78bda68a55d32c87c99b985e0f11331cddf05af6c4d753,mems_allowed=0-1,oom_memcg=/kubepods/podf1c273d3-9b36-11ea-b3df-246e9693c184,task_memcg=/kubepods/podf1c273d3-9b36-11ea-b3df-246e9693c184/1f246a3eeea8f70bf91141eeaf1805346a666e225f823906485ea0b6c37dfc3d,task=pause,pid=5740,uid=0 [7516987.984254] Memory cgroup out of memory: Killed process 5740 (pause) total-vm:1028kB, anon-rss:4kB, file-rss:0kB, shmem-rss:0kB [7516988.092344] oom_reaper: reaped process 5740 (pause), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB We can find that the first scanned process 5740 (pause) was killed, but its rss is only one page. That is because, when we calculate the oom badness in oom_badness(), we always ignore the negtive point and convert all of these negtive points to 1. Now as oom_score_adj of all the processes in this targeted memcg have the same value -998, the points of these processes are all negtive value. As a result, the first scanned process will be killed. The oom_socre_adj (-998) in this memcg is set by kubelet, because it is a a Guaranteed pod, which has higher priority to prevent from being killed by system oom. To fix this issue, we should make the calculation of oom point more accurate. We can achieve it by convert the chosen_point from 'unsigned long' to 'long'. Signed-off-by: Yafang Shao <laoar.shao@gmail.com> --- drivers/tty/sysrq.c | 1 + fs/proc/base.c | 7 ++++++- include/linux/oom.h | 4 ++-- mm/memcontrol.c | 1 + mm/oom_kill.c | 19 ++++++++----------- mm/page_alloc.c | 1 + 6 files changed, 19 insertions(+), 14 deletions(-)