Out of memory, OOM killer output information analysis

Keywords: Attribute xml less DBus

When out of memory occurs, the out_of_memory function chooses a process that the kernel considers guilty of allocating too much memory and kills the process.

There is a high probability of freeing up more free pages and then jumping back to retry the memory allocation operation.

Here we do not discuss out_of_memory function process selection at the expense of process strategy.

We will only discuss the meaning of the output information from the kernel when out of memory occurs.


1. OOM information

The following is a typical out of memory kernel output message:

  1. <4>[12345.342532] systemd-journal invoked oom-killer: gfp_mask=0x800d0, order=0, oom_score_adj=0  
  2. <4>[12345.351216] CPU: 1 PID: 1371 Comm: systemd-journal Tainted: G           O 3.14.31-00017-g40fab71 #1  
  3. <4>[12345.360695] Backtrace:  
  4. <4>[12345.363263] [<c0012fcc>] (dump_backtrace) from [<c00131a4>] (show_stack+0x20/0x24)  
  5. <4>[12345.371192]  r6:00000000 r5:ffffffff r4:00000000 r3:bd943631  
  6. <4>[12345.377136] [<c0013184>] (show_stack) from [<c07bbe78>] (dump_stack+0x7c/0xc8)  
  7. <4>[12345.384710] [<c07bbdfc>] (dump_stack) from [<c07ba7e4>] (dump_header.isra.14+0x74/0x188)  
  8. <4>[12345.393184]  r6:000800d0 r5:00000000 r4:e8088000 r3:00000002  
  9. <4>[12345.399126] [<c07ba770>] (dump_header.isra.14) from [<c00f8a28>] (oom_kill_process+0x230/0x3e0)  
  10. <4>[12345.408234]  r10:00000000 r8:000800d0 r7:00000000 r6:c0b89aa8 r5:000800d0 r4:e9bb79c0  
  11. <4>[12345.416462] [<c00f87f8>] (oom_kill_process) from [<c00f90c8>] (out_of_memory+0x2f4/0x354)  
  12. <4>[12345.425024]  r10:00000000 r9:00000000 r8:000800d0 r7:00000000 r6:c0b89aa8 r5:c0b89d08  
  13. <4>[12345.433249]  r4:c0b89aa8  
  14. <4>[12345.435903] [<c00f8dd4>] (out_of_memory) from [<c00fd6c8>] (__alloc_pages_nodemask+0x93c/0x988)  
  15. <4>[12345.445011]  r10:00000000 r9:c0c38fc0 r8:c0b871d8 r7:e8088000 r6:c0c39bc0 r5:00000000  
  16. <4>[12345.453234]  r4:000800d0  
  17. <4>[12345.455887] [<c00fcd8c>] (__alloc_pages_nodemask) from [<c00fd734>] (__get_free_pages+0x20/0x3c)  
  18. <4>[12345.465087]  r10:e97d36a8 r9:00000063 r8:e8089f6c r7:00000063 r6:b6f79f68 r5:e97d36a8  
  19. <4>[12345.473311]  r4:00000000  
  20. <4>[12345.475965] [<c00fd714>] (__get_free_pages) from [<c0196878>] (proc_pid_readlink+0x68/0x110)  
  21. <4>[12345.484808] [<c0196810>] (proc_pid_readlink) from [<c013dcb8>] (SyS_readlinkat+0xf0/0x104)  
  22. <4>[12345.493461]  r7:bea40520 r6:ffffff9c r5:00004000 r4:00000000  
  23. <4>[12345.499402] [<c013dbc8>] (SyS_readlinkat) from [<c000eee0>] (ret_fast_syscall+0x0/0x34)  
  24. <4>[12345.507785]  r10:00000000 r9:e8088000 r8:c000f148 r7:0000014c r6:00000063 r5:b6f79f68  
  25. <4>[12345.516011]  r4:00000064  
  26. <4>[12345.518663] Mem-info:  
  27. <4>[12345.521049] Normal per-cpu:  
  28. <4>[12345.523969] CPU    0: hi:   42, btch:   7 usd:  23  
  29. <4>[12345.528979] CPU    1: hi:   42, btch:   7 usd:  25  
  30. <4>[12345.534004] HighMem per-cpu:  
  31. <4>[12345.537013] CPU    0: hi:  186, btch:  31 usd:  27  
  32. <4>[12345.542199] CPU    1: hi:  186, btch:  31 usd:  29  
  33. <4>[12345.547247] active_anon:21860 inactive_anon:14790 isolated_anon:0  
  34. <4>[12345.547247]  active_file:41585 inactive_file:10422 isolated_file:0  
  35. <4>[12345.547247]  unevictable:0 dirty:9 writeback:205 unstable:0  
  36. <4>[12345.547247]  free:285748 slab_reclaimable:2100 slab_unreclaimable:26286  
  37. <4>[12345.547247]  mapped:26079 shmem:14857 pagetables:687 bounce:0  
  38. <4>[12345.547247]  free_cma:57779  
  39. <4>[12345.581839] Normal free:233460kB min:2488kB low:3108kB high:3732kB active_anon:17312kB   
  40. inactive_anon:10824kB active_file:128kB inactive_file:4kB unevictable:0kB isolated(anon):0kB   
  41. isolated(file):0kB present:774144kB managed:387568kB mlocked:0kB dirty:16kB writeback:76kB   
  42. mapped:3296kB shmem:10840kB slab_reclaimable:8400kB slab_unreclaimable:105144kB kernel_stack:1168kB   
  43. pagetables:2748kB unstable:0kB bounce:0kB free_cma:231116kB writeback_tmp:0kB pages_scanned:1648   
  44. all_unreclaimable? yes  
  45. <4>[12345.627014] lowmem_reserve[]: 0 10168 10168  
  46. <4>[12345.631565] HighMem free:909036kB min:512kB low:2604kB high:4696kB active_anon:70632kB   
  47. inactive_anon:48336kB active_file:166212kB inactive_file:41684kB unevictable:0kB isolated(anon):0kB   
  48. isolated(file):0kB present:1301504kB managed:1301504kB mlocked:0kB dirty:20kB writeback:744kB   
  49. mapped:101020kB shmem:48588kB slab_reclaimable:0kB slab_unreclaimable:0kB kernel_stack:0kB   
  50. pagetables:0kB unstable:0kB bounce:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0   
  51. all_unreclaimable? no  
  52. <4>[12345.675614] lowmem_reserve[]: 0 0 0  
  53. <4>[12345.679437] Normal: 1165*4kB (MRC) 1122*8kB (RC) 1119*16kB (RC) 1118*32kB (C) 1068*64kB (RC)   
  54. 748*128kB (C) 0*256kB 0*512kB 0*1024kB 1*2048kB (R) 0*4096kB 0*8192kB = 233460kB  
  55. <4>[12345.695797] HighMem: 99*4kB (M) 1148*8kB (UM) 1314*16kB (UM) 880*32kB (UM) 327*64kB (M)   
  56. 87*128kB (M) 34*256kB (M) 38*512kB (M) 12*1024kB (M) 10*2048kB (M) 3*4096kB (M) 91*8192kB (UMR) = 909516kB  
  57. <4>[12345.714293] 66770 total pagecache pages  
  58. <4>[12345.718309] 0 pages in swap cache  
  59. <4>[12345.724832] Swap cache stats: add 0, delete 0, find 0/0  
  60. <4>[12345.730308] Free swap  = 0kB  
  61. <4>[12345.733412] Total swap = 0kB  
  62. <4>[12345.747245] 520192 pages of RAM  
  63. <4>[12345.750577] 286253 free pages  
  64. <4>[12345.753778] 97924 reserved pages  
  65. <4>[12345.757258] 28061 slab pages  
  66. <4>[12345.760574] 115601 pages shared  
  67. <4>[12345.764283] 0 pages swap cached  
  68. <6>[12345.767572] [ pid ]   uid  tgid total_vm      rss nr_ptes swapents oom_score_adj name  
  69. <6>[12345.775906] [ 1366]     0  1366      459      125       3        0             0 sh  
  70. <6>[12345.785861] [ 1367]     0  1367      665      235       4        0             0 propertyd  
  71. <6>[12345.794802] [ 1368]     0  1368    26553     8835      58        0             0 seed  
  72. <6>[12345.803296] [ 1371]     0  1371     1648      772       5        0             0 systemd-journal  
  73. <6>[12345.812792] [ 1375]     0  1375      750      300       4        0         -1000 systemd-udevd  
  74. <6>[12345.822449] [ 2416]  1040  2416     3852      510       7        0             0 secd  
  75. <6>[12345.831341] [ 2419]     0  2419     6678      923       9        0             0 storagemanagerd  
  76. <6>[12345.840944] [ 2420]     0  2420     1267      497       5        0             0 connmand  
  77. <6>[12345.849566] [ 2422]     0  2422     4484      687       8        0             0 uuid  
  78. <6>[12345.857843] [ 2424]     0  2424     1161      358       5        0             0 connman-vpnd  
  79. <6>[12345.867271] [ 2427]  1000  2427     1593      461       6        0             0 logboxd  
  80. <6>[12345.875846] [ 2432]     0  2432     9483     1718      15        0             0 cmns  
  81. <6>[12345.884104] [ 2451]    81  2451     1355      474       4        0          -900 dbus-daemon  
  82. <6>[12345.893018] [ 2532]     0  2532    11794      246      10        0             0 adbd  
  83. <6>[12345.901304] [ 2535]     0  2535     1502      347       5        0             0 wpa_supplicant  
  84. <6>[12345.910473] [ 2536]     0  2536    12820      866      12        0             0 udisksd  
  85. <6>[12345.919119] [ 2537]     0  2537     1898      527       6        0             0 tyid  
  86. <6>[12345.927361] [ 2540]     0  2540    10076     2157      16        0             0 datamanagerd  
  87. <6>[12345.936349] [ 2554]     0  2554     5983      574       7        0             0 connectivityser  
  88. <6>[12345.945635] [ 2558]     0  2558    10604     5388      21        0             0 weston  
  89. <6>[12345.964101] [ 2589]     0  2589    14597     1917      17        0             0 pagemanagerd  
  90. <6>[12345.973272] [ 2590]     0  2590     3832      515       7        0             0 amt  
  91. <6>[12345.981730] [ 2593]     0  2593     6176     1343      12        0             0 weston-desktop-  
  92. <6>[12345.991046] [ 2599]     0  2599     7185      761      12        0             0 scim-launcher  
  93. <6>[12346.098925] [ 5580]     0  5580      458      116       3        0             0 sh  
  94. <6>[12346.107065] [ 5581]     0  5581      492      175       3        0             0 gzip  
  95. <3>[12346.115335] Out of memory: Kill process 5575 thread_x score 481 or sacrifice child  
  96. <3>[12346.124212] Killed process 5575 thread_x total-vm:106212kB, anon-rss:18036kB, file-rss:2704kB  

2 OOM Information Analysis

2.1

  1. <4>[12345.342532] systemd-journal invoked oom-killer: gfp_mask=0x800d0, order=0, oom_score_adj=0  
Systemd-journal: The current process is systemd-journal, which triggers oom-killer when requesting page assignment

gfp_mask=0x800d0: is the GFP flag of alloc_page. For the current scene, it represents u GFP_RECLAIMABLE | u GFP_HIGH | u GFP_IO | u GFP_FS.

order=0: This means that the order of alloc_page is 0, which means that only 1 ^ 0 = 1 page is requested.

oom_score_adj=0: Indicates the probability that the process will be killed. The value of oom_score_adj is 0(never kill)~1000(always kill)


  1. <4>[12345.351216] CPU: 1 PID: 1371 Comm: systemd-journal Tainted: G           O 3.14.31-00017-g40fab71 #1  
  2. <4>[12345.360695] Backtrace:  
  3. <4>[12345.363263] [<c0012fcc>] (dump_backtrace) from [<c00131a4>] (show_stack+0x20/0x24)  
  4. <4>[12345.371192]  r6:00000000 r5:ffffffff r4:00000000 r3:bd943631  
  5. <4>[12345.377136] [<c0013184>] (show_stack) from [<c07bbe78>] (dump_stack+0x7c/0xc8)  
  6. <4>[12345.384710] [<c07bbdfc>] (dump_stack) from [<c07ba7e4>] (dump_header.isra.14+0x74/0x188)  
  7. <4>[12345.393184]  r6:000800d0 r5:00000000 r4:e8088000 r3:00000002  
  8. <4>[12345.399126] [<c07ba770>] (dump_header.isra.14) from [<c00f8a28>] (oom_kill_process+0x230/0x3e0)  
  9. <4>[12345.408234]  r10:00000000 r8:000800d0 r7:00000000 r6:c0b89aa8 r5:000800d0 r4:e9bb79c0  
  10. <4>[12345.416462] [<c00f87f8>] (oom_kill_process) from [<c00f90c8>] (out_of_memory+0x2f4/0x354)  
  11. <4>[12345.425024]  r10:00000000 r9:00000000 r8:000800d0 r7:00000000 r6:c0b89aa8 r5:c0b89d08  
  12. <4>[12345.433249]  r4:c0b89aa8  
  13. <4>[12345.435903] [<c00f8dd4>] (out_of_memory) from [<c00fd6c8>] (__alloc_pages_nodemask+0x93c/0x988)  
  14. <4>[12345.445011]  r10:00000000 r9:c0c38fc0 r8:c0b871d8 r7:e8088000 r6:c0c39bc0 r5:00000000  
  15. <4>[12345.453234]  r4:000800d0  
  16. <4>[12345.455887] [<c00fcd8c>] (__alloc_pages_nodemask) from [<c00fd734>] (__get_free_pages+0x20/0x3c)  
  17. <4>[12345.465087]  r10:e97d36a8 r9:00000063 r8:e8089f6c r7:00000063 r6:b6f79f68 r5:e97d36a8  
  18. <4>[12345.473311]  r4:00000000  
  19. <4>[12345.475965] [<c00fd714>] (__get_free_pages) from [<c0196878>] (proc_pid_readlink+0x68/0x110)  
  20. <4>[12345.484808] [<c0196810>] (proc_pid_readlink) from [<c013dcb8>] (SyS_readlinkat+0xf0/0x104)  
  21. <4>[12345.493461]  r7:bea40520 r6:ffffff9c r5:00004000 r4:00000000  
  22. <4>[12345.499402] [<c013dbc8>] (SyS_readlinkat) from [<c000eee0>] (ret_fast_syscall+0x0/0x34)  
  23. <4>[12345.507785]  r10:00000000 r9:e8088000 r8:c000f148 r7:0000014c r6:00000063 r5:b6f79f68  
  24. <4>[12345.516011]  r4:00000064  

The output of dump_header - > dump_stack triggers the call function stack of OOM, starting with ret_fast_syscall.

From this output, it can be inferred that a paging operation triggered by system D-JOURNAL calling readlink system call resulted in OOM.


2.2

  1. <4>[12345.518663] Mem-info:  
  2. <4>[12345.521049] Normal per-cpu:  
  3. <4>[12345.523969] CPU    0: hi:   42, btch:   7 usd:  23  
  4. <4>[12345.528979] CPU    1: hi:   42, btch:   7 usd:  25  
  5. <4>[12345.534004] HighMem per-cpu:  
  6. <4>[12345.537013] CPU    0: hi:  186, btch:  31 usd:  27  
  7. <4>[12345.542199] CPU    1: hi:  186, btch:  31 usd:  29  
  8. <4>[12345.547247] active_anon:21860 inactive_anon:14790 isolated_anon:0  
  9. <4>[12345.547247]  active_file:41585 inactive_file:10422 isolated_file:0  
  10. <4>[12345.547247]  unevictable:0 dirty:9 writeback:205 unstable:0  
  11. <4>[12345.547247]  free:285748 slab_reclaimable:2100 slab_unreclaimable:26286  
  12. <4>[12345.547247]  mapped:26079 shmem:14857 pagetables:687 bounce:0  
  13. <4>[12345.547247]  free_cma:57779  
  14. <4>[12345.581839] Normal free:233460kB min:2488kB low:3108kB high:3732kB active_anon:17312kB   
  15. inactive_anon:10824kB active_file:128kB inactive_file:4kB unevictable:0kB isolated(anon):0kB   
  16. isolated(file):0kB present:774144kB managed:387568kB mlocked:0kB dirty:16kB writeback:76kB   
  17. mapped:3296kB shmem:10840kB slab_reclaimable:8400kB slab_unreclaimable:105144kB kernel_stack:1168kB   
  18. pagetables:2748kB unstable:0kB bounce:0kB free_cma:231116kB writeback_tmp:0kB pages_scanned:1648   
  19. all_unreclaimable? yes  
  20. <4>[12345.627014] lowmem_reserve[]: 0 10168 10168  
  21. <4>[12345.631565] HighMem free:909036kB min:512kB low:2604kB high:4696kB active_anon:70632kB   
  22. inactive_anon:48336kB active_file:166212kB inactive_file:41684kB unevictable:0kB isolated(anon):0kB   
  23. isolated(file):0kB present:1301504kB managed:1301504kB mlocked:0kB dirty:20kB writeback:744kB   
  24. mapped:101020kB shmem:48588kB slab_reclaimable:0kB slab_unreclaimable:0kB kernel_stack:0kB   
  25. pagetables:0kB unstable:0kB bounce:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no  
  26. <4>[12345.675614] lowmem_reserve[]: 0 0 0  
  27. <4>[12345.679437] Normal: 1165*4kB (MRC) 1122*8kB (RC) 1119*16kB (RC) 1118*32kB (C) 1068*64kB (RC)   
  28. 748*128kB (C) 0*256kB 0*512kB 0*1024kB 1*2048kB (R) 0*4096kB 0*8192kB = 233460kB  
  29. <4>[12345.695797] HighMem: 99*4kB (M) 1148*8kB (UM) 1314*16kB (UM) 880*32kB (UM) 327*64kB (M)   
  30. 87*128kB (M) 34*256kB (M) 38*512kB (M) 12*1024kB (M) 10*2048kB (M) 3*4096kB (M) 91*8192kB (UMR) = 909516kB  
  31. <4>[12345.714293] 66770 total pagecache pages  
  32. <4>[12345.718309] 0 pages in swap cache  
  33. <4>[12345.724832] Swap cache stats: add 0, delete 0, find 0/0  
  34. <4>[12345.730308] Free swap  = 0kB  
  35. <4>[12345.733412] Total swap = 0kB  
  36. <4>[12345.747245] 520192 pages of RAM  
  37. <4>[12345.750577] 286253 free pages  
  38. <4>[12345.753778] 97924 reserved pages  
  39. <4>[12345.757258] 28061 slab pages  
  40. <4>[12345.760574] 115601 pages shared  
  41. <4>[12345.764283] 0 pages swap cached  
Dump_header - > show_mem outputs the current system memory information.

2.2.1

<4>[12345.521049] Normal per-cpu:
<4>[12345.523969] CPU    0: hi:   42, btch:   7 usd:  23
<4>[12345.528979] CPU    1: hi:   42, btch:   7 usd:  25
<4>[12345.534004] HighMem per-cpu:
<4>[12345.537013] CPU    0: hi:  186, btch:  31 usd:  27
<4>[12345.542199] CPU    1: hi:  186, btch:  31 usd:  29
Each memory management area defines a "per CPU" page frame cache. All "per CPU" caches contain pre-allocated page frames that are used to satisfy a single page memory request from a local CPU.

CPU    0: hi:   42, btch:   7 usd:  23

Represents CPU 0,

hi: 42 represents the upper limit. If this number is exceeded, the batch page frames are released into the buddy system.

btch: 7 represents the size of a page block when adding or deleting a page frame to a cache

Usd: Number of Page Boxes in 23 Page Box Cache

2.2.2

  1. <4>[12345.547247] active_anon:21860 inactive_anon:14790 isolated_anon:0  
  2. <4>[12345.547247]  active_file:41585 inactive_file:10422 isolated_file:0  
  3. <4>[12345.547247]  unevictable:0 dirty:9 writeback:205 unstable:0  
  4. <4>[12345.547247]  free:285748 slab_reclaimable:2100 slab_unreclaimable:26286  
  5. <4>[12345.547247]  mapped:26079 shmem:14857 pagetables:687 bounce:0  
  6. <4>[12345.547247]  free_cma:57779  

active_anon: Anonymous mapping of activities, "activities" refers to recent visits, and "anonymity" refers to page mappings that are not related to any data source.

inactive_anon: Inactive Anonymous Mapping

isolated_anon: DON'T KNOW

active_file: Active file mapping, page mapping and disk file association

inactive_file: Inactive file mapping

isolated_file: DON'T KNOW

unevictable:

Dirty: dirty pages that represent the content of pages and the original content on fast devices.

Writteback: The current page is in a write-back state

unstable:

free: free pages

Slab_relaimable: Recyclable pages in slab cache

Slab_unreclaimable: Pages that cannot be recycled in slab cache

mapped: BH_MAPPED, which means that this page is used as a buffer map for fast devices. Note that this map is different from anon and file mappings.

shmem: Pages for shared memory mapping

pagetable: Pages occupied by page tables, that is, the number of pages occupied by PTE PTD

bounce:

Free_cma: The free page of the continuous memory allocator.


2.2.3

  1. <4>[12345.581839] Normal free:233460kB min:2488kB low:3108kB high:3732kB active_anon:17312kB inactive_anon:10824kB   
  2. active_file:128kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:774144kB   
  3. managed:387568kB mlocked:0kB dirty:16kB writeback:76kB mapped:3296kB shmem:10840kB slab_reclaimable:8400kB   
  4. slab_unreclaimable:105144kB kernel_stack:1168kB pagetables:2748kB unstable:0kB bounce:0kB free_cma:231116kB   
  5. writeback_tmp:0kB pages_scanned:1648 all_unreclaimable? yes  
  6. <4>[12345.627014] lowmem_reserve[]: 0 10168 10168  
  7. <4>[12345.631565] HighMem free:909036kB min:512kB low:2604kB high:4696kB active_anon:70632kB inactive_anon:48336kB   
  8. active_file:166212kB inactive_file:41684kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:1301504kB   
  9. managed:1301504kB mlocked:0kB dirty:20kB writeback:744kB mapped:101020kB shmem:48588kB slab_reclaimable:0kB   
  10. slab_unreclaimable:0kB kernel_stack:0kB pagetables:0kB unstable:0kB bounce:0kB free_cma:0kB writeback_tmp:0kB   
  11. pages_scanned:0 all_unreclaimable? no  
  12. <4>[12345.675614] lowmem_reserve[]: 0 0 0  

Normal free: Free space in Normal zone

min, low, high are several watermarks for normal zone to perform page replacement

lowmem_reserve: Represents the number of allocatable pages reserved for other zones by this zone

present: Represents the physical memory size of a zone

Managed: The size of present ation memory managed by buddy system, managed = preset - reserved

The other values can be referred to in Section 2.2.2. They have similar meanings except that the values represent Normal zone s.

Note 1. Several items are unique to Normal, such as kernel_stack, pagetables, free_cma, slab_reclaimable, slab_unreclaimable, because the pages of normal zone are directly mapped and are used by the kernel.

For highmem, it is mainly used for anonymous mapping, file mapping, mapped, and shared memory.


2.2.4

  1. <4>[12345.679437] Normal: 1165*4kB (MRC) 1122*8kB (RC) 1119*16kB (RC) 1118*32kB (C) 1068*64kB (RC)   
  2. 748*128kB (C) 0*256kB 0*512kB 0*1024kB 1*2048kB (R) 0*4096kB 0*8192kB = 233460kB  
  3. <4>[12345.695797] HighMem: 99*4kB (M) 1148*8kB (UM) 1314*16kB (UM) 880*32kB (UM) 327*64kB (M)   
  4. 87*128kB (M) 34*256kB (M) 38*512kB (M) 12*1024kB (M) 10*2048kB (M) 3*4096kB (M) 91*8192kB (UMR) = 909516kB  
buddy system information, order range 0-11

M denotes moveable

R denotes Reserve

C means CMA.

U stands for unmovable

E denotes reclaimable

1. Only (C) means that this freelist can only be allocated to allocations with ALLOC_CMA flags.

2. Highmem has no C flag because continuous memory allocation occurs only in Normal zone

3. MRC, which means that this freelist has CMA memory, Reserve memory and Movable memory


3 Who triggered OOM

Several factors influence the occurrence of OOM

1. The size of the order allocated and the way the system treats the order

2. Which zone does the allocation occur?

3. Zone's Watermark Size

4. Degree of memory fragmentation

5. It is said that the constant allocation of address space will also lead to the occurrence of OOM (not yet verified).


For the OOM information above, we can see that there is a lot of free space in the system: 233460KB, but OOM still happens.

The first order allocated is 0, so fragmentation is irrelevant. gfp_mask=0x800d0 indicates that the allocation occurs in the Normal partition and the type is Reclaimable. Reclaimable also means that memory cannot be allocated from CMA.

Since it is not the order that is too large that causes the allocation failure, it is the free space less than the memory watermarking min that causes OOM killer.

  1. static bool __zone_watermark_ok(struct zone *z, unsigned int order,  
  2.             unsigned long mark, int classzone_idx, int alloc_flags,  
  3.             long free_pages)  
  4. {   
  5.      ...  
  6.   
  7. #ifdef CONFIG_CMA  
  8.     /* If allocation can't use CMA areas don't use free CMA pages */  
  9.     if (!(alloc_flags & ALLOC_CMA))  
  10.         free_cma = zone_page_state(z, NR_FREE_CMA_PAGES);  
  11. #endif  
  12.   
  13.     if (free_pages - free_cma <= min + z->lowmem_reserve[classzone_idx])  
  14.         return false;  
  15.     ...  
  16. }   
The allocation type is Reclaimable, which causes free space to be subtracted from CMA free space by 233460KB - 223116kB = 2352KB, less than min watermarking. The system starts OOM killer.

Posted by thegreatone2176 on Sun, 16 Dec 2018 16:21:05 -0800

Hot Keywords

©2024 Programmer Group