New issue
Advanced search Search tips

Issue 753365 link

Starred by 1 user

Issue metadata

Status: WontFix
Owner: ----
Closed: Aug 2017
Components:
EstimatedDays: ----
NextAction: ----
OS: Linux
Pri: 2
Type: ----



Sign in to add a comment

webkit_layout_tests flakily failing MSAN during System dependencies check

Project Member Reported by joh...@chromium.org, Aug 8 2017

Issue description

webkit_layout_tests flakily failing MSAN during System dependencies check

Builders failed on: 
- WebKit Linux Trusty MSAN: 
  https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Trusty%20MSAN

It fails around one in 6 builds.

None of the individual layout tests fail; instead one of the shards sometimes fails to start up - it shows as e.g. "shard #9 (failed)".

Analysing the failed shard shows the following error:


06:01:04.929 12129 Checking system dependencies ...
06:01:04.930 12129 DISPLAY = :100
06:01:06.246 12129 "/b/s/w/ir/out/Release/content_shell --check-layout-test-sys-deps" took 1.32s
06:01:06.246 12129 System dependencies check failed.
06:01:06.246 12129 To override, invoke with --nocheck-sys-deps
06:01:06.247 12129 
06:01:06.247 12129 Xlib:  extension "RANDR" missing on display ":100".

DevTools listening on ws://127.0.0.1:59500/devtools/browser/5e8e91e7-14ad-4cd7-bad3-a434fed0e292
Uninitialized bytes in __interceptor_strlen at offset 20 inside [0x71f000001c00, 3280)
==11034:11034==WARNING: MemorySanitizer: use-of-uninitialized-value
    #0 0x7f30496eb4c8  (/usr/lib/x86_64-linux-gnu/mesa/libGL.so.1+0x194c8)
    #1 0x7f30496f0599  (/usr/lib/x86_64-linux-gnu/mesa/libGL.so.1+0x1e599)
    #2 0x7f30496ecf3a  (/usr/lib/x86_64-linux-gnu/mesa/libGL.so.1+0x1af3a)
    #3 0x7f30496ed0ee  (/usr/lib/x86_64-linux-gnu/mesa/libGL.so.1+0x1b0ee)
    #4 0xa174ff7  (/b/s/w/ir/out/Release/content_shell+0xa174ff7)
    #5 0xa176862  (/b/s/w/ir/out/Release/content_shell+0xa176862)
    #6 0xa16c7f5  (/b/s/w/ir/out/Release/content_shell+0xa16c7f5)
    #7 0xa3a88ca  (/b/s/w/ir/out/Release/content_shell+0xa3a88ca)
    #8 0xa3a3555  (/b/s/w/ir/out/Release/content_shell+0xa3a3555)
    #9 0xa3a2db2  (/b/s/w/ir/out/Release/content_shell+0xa3a2db2)
    #10 0xa446a07  (/b/s/w/ir/out/Release/content_shell+0xa446a07)
    #11 0x1119ad85  (/b/s/w/ir/out/Release/content_shell+0x1119ad85)
    #12 0x48502f0  (/b/s/w/ir/out/Release/content_shell+0x48502f0)
    #13 0x4853066  (/b/s/w/ir/out/Release/content_shell+0x4853066)
    #14 0xc02c4ba  (/b/s/w/ir/out/Release/content_shell+0xc02c4ba)
    #15 0x163fffe  (/b/s/w/ir/out/Release/content_shell+0x163fffe)
    #16 0x4a90c7  (/b/s/w/ir/out/Release/content_shell+0x4a90c7)
    #17 0x7f3059649f44  (/lib/x86_64-linux-gnu/libc.so.6+0x21f44)
    #18 0x43cffa  (/b/s/w/ir/out/Release/content_shell+0x43cffa)

  Uninitialized value was created by a heap allocation
    #0 0x4602ad  (/b/s/w/ir/out/Release/content_shell+0x4602ad)
    #1 0x7f30496f0ae3  (/usr/lib/x86_64-linux-gnu/mesa/libGL.so.1+0x1eae3)

SUMMARY: MemorySanitizer: use-of-uninitialized-value (/usr/lib/x86_64-linux-gnu/mesa/libGL.so.1+0x194c8) 
Exiting
 
It's been happening since at least June 20th:

https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Trusty%20MSAN/builds/1419

https://chromium-swarm.appspot.com/user/task/36dd4a54f32cdb10

Some other builds where it happened (there are many more!): 2262, 2185, 2159, 2149, 1493, 1446
> (/usr/lib/x86_64-linux-gnu/mesa/libGL.so.1+0x194c8)

^ this should have been loaded from prebuilt instrumented library directory in chromium build tree. Is the builder using use_prebuilt_instrumented_libraries=1?

See http://dev.chromium.org/developers/testing/memorysanitizer, note the "OpenGL" paragraph(s).

Components: Blink>Infra Infra
Labels: Pri-2
Status: Untriaged (was: Available)
Taking a couple of guesses at the right component for this.
Components: -Infra
Status: WontFix (was: Untriaged)
The last failing build (that also has the error in the bug description) is https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Trusty%20MSAN/builds/2626

It is followed by 17 green builds in a row, so I assume the problem is fixed?

Sign in to add a comment