Bug 89646

Summary: [BSW ppgtt]igt/gem_close_race/gem-close-race OOM killer
Product: DRI Reporter: lu hua <huax.lu>
Component: DRM/IntelAssignee: Intel GFX Bugs mailing list <intel-gfx-bugs>
Status: CLOSED DUPLICATE QA Contact: Intel GFX Bugs mailing list <intel-gfx-bugs>
Severity: normal    
Priority: highest CC: intel-gfx-bugs, michel.thierry
Version: unspecified   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
dmesg none

Description lu hua 2015-03-18 08:06:42 UTC 
Created attachment 114430 [details]
dmesg

==System Environment==
--------------------------
Regression: not sure, caused by ppgtt

no-working platforms: BSW

==kernel==
--------------------------
drm-intel-nightly/3753ea527e539d3358a6368e77cb29ecb2e71c42
commit 3753ea527e539d3358a6368e77cb29ecb2e71c42
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Tue Mar 17 22:36:04 2015 +0100

    drm-intel-nightly: 2015y-03m-17d-21h-35m-20s UTC integration manifest

==Bug detailed description==
-----------------------------
It causes OOM killer on BSW. add i915.enable_ppgtt=0, it works well.

output:
IGT-Version: 1.10-g505645c (x86_64) (Linux: 4.0.0-rc4_drm-intel-nightly_3753ea_20150318+ x86_64)
Killed

[   52.076005] Call Trace:
[   52.076020]  [<ffffffff817942e1>] ? dump_stack+0x40/0x50
[   52.076028]  [<ffffffff8179161e>] ? dump_header.isra.11+0x6b/0x196
[   52.076071]  [<ffffffffa009e3f0>] ? i915_gem_shrinker_oom+0x82/0x1bb [i915]
[   52.076080]  [<ffffffff810d37ca>] ? oom_kill_process+0xb5/0x374
[   52.076089]  [<ffffffff81041b21>] ? has_ns_capability_noaudit+0xd/0x14
[   52.076095]  [<ffffffff810d3f16>] ? __out_of_memory+0x440/0x462
[   52.076101]  [<ffffffff810d4063>] ? out_of_memory+0x52/0x67
[   52.076110]  [<ffffffff810d7d2c>] ? __alloc_pages_nodemask+0x66e/0x6fc
[   52.076118]  [<ffffffff811046d3>] ? alloc_pages_current+0xad/0xca
[   52.076124]  [<ffffffff810d2970>] ? filemap_fault+0x21f/0x37e
[   52.076134]  [<ffffffff810ee38e>] ? do_set_pte+0x8b/0x98
[   52.076139]  [<ffffffff810ec68f>] ? __do_fault+0x3d/0x80
[   52.076145]  [<ffffffff810efb0c>] ? handle_mm_fault+0x358/0xc4d
[   52.076153]  [<ffffffff8105edd9>] ? set_next_entity+0x32/0x55
[   52.076159]  [<ffffffff81060a13>] ? pick_next_task_fair+0xe5/0x3dc
[   52.076169]  [<ffffffff8103293a>] ? __do_page_fault+0x21c/0x3d8
[   52.076176]  [<ffffffff8110fa75>] ? vfs_read+0xd2/0x110
[   52.076184]  [<ffffffff8179b5e2>] ? page_fault+0x22/0x30
[   52.076188] Mem-Info:
[   52.076193] Node 0 DMA per-cpu:
[   52.076198] CPU    0: hi:    0, btch:   1 usd:   0
[   52.076201] CPU    1: hi:    0, btch:   1 usd:   0
[   52.076205] CPU    2: hi:    0, btch:   1 usd:   0
[   52.076209] CPU    3: hi:    0, btch:   1 usd:   0
[   52.076212] Node 0 DMA32 per-cpu:
[   52.076216] CPU    0: hi:  186, btch:  31 usd:   0
[   52.076219] CPU    1: hi:  186, btch:  31 usd:   0
[   52.076224] CPU    2: hi:  186, btch:  31 usd:   0
[   52.076228] CPU    3: hi:  186, btch:  31 usd:  30
[   52.076231] Node 0 Normal per-cpu:
[   52.076235] CPU    0: hi:  186, btch:  31 usd:  31
[   52.076238] CPU    1: hi:  186, btch:  31 usd:  30
[   52.076243] CPU    2: hi:  186, btch:  31 usd:   0
[   52.076246] CPU    3: hi:  186, btch:  31 usd:  30


[   52.076417] [ pid ]   uid  tgid total_vm      rss nr_ptes nr_pmds swapents oom_score_adj name
[   52.076427] [ 2428]     0  2428     1113        0       7       2       25             0 sh
[   52.076432] [ 2458]     0  2458     4435        1      13       2      629             0 initctl
[   52.076438] [ 2460]     0  2460     8255        1      18       2      120             0 mountall
[   52.076443] [ 2552]     0  2552     4936        1      13       2      139             0 upstart-udev-br
[   52.076450] [ 2563]     0  2563    12436        1      28       3      222         -1000 systemd-udevd
[   52.076455] [ 3309]     0  3309     5857        1      17       3       68             0 rpcbind
[   52.076460] [ 3371]     0  3371     7444        0      20       3       77             0 rpc.idmapd
[   52.076466] [ 3374]   102  3374     9889        1      24       3      172             0 dbus-daemon
[   52.076471] [ 3377]   117  3377     5388        1      16       3      115             0 rpc.statd
[   52.076476] [ 3467]     0  3467    82550        0      64       3      282             0 ModemManager
[   52.076481] [ 3483]     0  3483    10864        1      26       3       86             0 systemd-logind
[   52.076487] [ 3507]     0  3507    89204        0      71       4      468             0 NetworkManager
[   52.076492] [ 3541]     0  3541    73632        1      45       3      227             0 polkitd
[   52.076497] [ 3561]   101  3561    65535        0      31       4      219             0 rsyslogd
[   52.076502] [ 3596]   111  3596     8090        0      21       3       80             0 avahi-daemon
[   52.076507] [ 3599]     0  3599     2560        1       9       2      573             0 dhclient
[   52.076513] [ 3608]     0  3608     5006        1      13       3       40             0 getty
[   52.076518] [ 3613]     0  3613     5006        1      13       3       40             0 getty
[   52.076523] [ 3620]     0  3620     5006        1      14       3       39             0 getty
[   52.076528] [ 3621]     0  3621     5006        1      13       3       37             0 getty
[   52.076533] [ 3624]     0  3624     5006        1      13       3       40             0 getty
[   52.076540] [ 3646]     0  3646    15343        1      34       2      169         -1000 sshd
[   52.076545] [ 3653]     0  3653     4799        1      14       3       57             0 irqbalance
[   52.076550] [ 3654]     0  3654     5916        1      17       3       62             0 cron
[   52.076555] [ 3660]     0  3660     1094        1       8       3       44             0 acpid
[   52.076561] [ 3661]   111  3661     8058        0      20       3       64             0 avahi-daemon
[   52.076566] [ 3682]   109  3682    90848        1      78       3      362             0 whoopsie
[   52.076571] [ 3683]   106  3683     9288        0      21       3       82             0 kerneloops
[   52.076576] [ 3735]     0  3735    19215        1      39       2      279             0 cupsd
[   52.076581] [ 3762]     0  3762     1113        1       5       2       44             0 ondemand
[   52.076586] [ 3769]     0  3769     1088        0       7       3       23             0 sleep
[   52.076592] [ 3777]     0  3777     5006        1      13       3       40             0 getty
[   52.076597] [ 3809]     0  3809    18840        0      41       3      229             0 cups-browsed
[   52.076602] [ 3892] 65534  3892     8808        1      23       3       64             0 dnsmasq
[   52.076607] [ 3938]     0  3938     3960        1      13       2      190             0 upstart-file-br
[   52.076613] [ 3983]     0  3983     4093        1      11       2      324             0 upstart-socket-
[   52.076619] [ 4233]     0  4233    27447        1      55       2      252             0 sshd
[   52.076624] [ 4295]     0  4295     6813        1      17       3      622             0 bash
[   52.076630] [ 4307]     0  4307    18524        0      40       4      126          1000 gem_close_race
[   52.076635] Out of memory: Kill process 4307 (gem_close_race) score 999 or sacrifice child
[   52.076779] Killed process 4307 (gem_close_race) total-vm:74096kB, anon-rss:0kB, file-rss:0kB

==Reproduce steps==
---------------------------- 
1. ./gem_close_race --run-subtest gem-close-race
Comment 1 Michel Thierry 2015-03-23 13:09:59 UTC 
"deferred allocation / dynamic page allocation" (in review) is needed for this test to pass in its current form. 

gem-close-race opens too many connections and a 'fully allocated' ppgtt is created for each ctx.

The same rationale applies to igt/gem_ctx_thrash.

*** This bug has been marked as a duplicate of bug 87477 ***
Comment 2 Jari Tahvanainen 2017-07-03 14:00:48 UTC 
Closing verified+duplicate as duplicate of closed+fixed.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.