Closed
Bug 1334461
Opened 8 years ago
Closed 8 years ago
run mochitest-a11y on ubuntu 16.04 instead of 12.04
Categories
(Core :: Disability Access APIs, defect)
Core
Disability Access APIs
Tracking
()
RESOLVED
FIXED
mozilla54
Tracking | Status | |
---|---|---|
firefox54 | --- | fixed |
People
(Reporter: jmaher, Assigned: jmaher)
References
Details
Attachments
(2 files)
(deleted),
patch
|
Details | Diff | Splinter Review | |
(deleted),
patch
|
kats
:
review+
|
Details | Diff | Splinter Review |
to update our operating systems to be modern, we have been slowly converting our tests to run on 16.04! This is great as there are updated libraries, kernels, and other tools.
For a11y, this works great except for ASAN where we have leaks. In fact 13 different tests hit a leak (what appears to be the same leak). We can consider disabling the tests or fixing the leak, I would vote for fixing the leak and then reassess what remains (ideally nothing).
Assignee | ||
Comment 1•8 years ago
|
||
in accessible/tests/mochitest/actions/test_general.xul [0], we have this leak:
[task 2017-01-26T18:33:54.679829Z] 18:33:54 INFO - =================================================================
[task 2017-01-26T18:33:54.679921Z] 18:33:54 INFO - ==1125==ERROR: LeakSanitizer: detected memory leaks
[task 2017-01-26T18:33:54.679968Z] 18:33:54 INFO - Direct leak of 40 byte(s) in 1 object(s) allocated from:
[task 2017-01-26T18:33:54.680030Z] 18:33:54 INFO - #0 0x4b2654 in calloc /builds/slave/moz-toolchain/src/llvm/projects/compiler-rt/lib/asan/asan_malloc_linux.cc:66:3
[task 2017-01-26T18:33:54.680097Z] 18:33:54 INFO - #1 0x7fe144afb9ae in pa_xmalloc0 (/usr/lib/x86_64-linux-gnu/pulseaudio/libpulsecommon-8.0.so+0x149ae)
[task 2017-01-26T18:33:54.680144Z] 18:33:54 INFO - -----------------------------------------------------
[task 2017-01-26T18:33:54.680180Z] 18:33:54 INFO - Suppressions used:
[task 2017-01-26T18:33:54.680219Z] 18:33:54 INFO - count bytes template
[task 2017-01-26T18:33:54.680258Z] 18:33:54 INFO - 257 3493 libc.so
[task 2017-01-26T18:33:54.680300Z] 18:33:54 INFO - 803 25600 nsComponentManagerImpl
[task 2017-01-26T18:33:54.680344Z] 18:33:54 INFO - 53 7632 mozJSComponentLoader::LoadModule
[task 2017-01-26T18:33:54.680391Z] 18:33:54 INFO - 1 384 pixman_implementation_lookup_composite
[task 2017-01-26T18:33:54.680566Z] 18:33:54 INFO - 355 14021 libfontconfig.so
[task 2017-01-26T18:33:54.681844Z] 18:33:54 INFO - 1 32 libdl.so
[task 2017-01-26T18:33:54.682381Z] 18:33:54 INFO - 17 4348 libglib-2.0.so
[task 2017-01-26T18:33:54.682890Z] 18:33:54 INFO - 1 28 libresolv.so
[task 2017-01-26T18:33:54.683379Z] 18:33:54 INFO - -----------------------------------------------------
[task 2017-01-26T18:33:54.683882Z] 18:33:54 INFO - SUMMARY: AddressSanitizer: 40 byte(s) leaked in 1 allocation(s).
[task 2017-01-26T18:33:54.985689Z] 18:33:54 INFO - TEST-INFO | Main app process: exit 0
[task 2017-01-26T18:33:54.986641Z] 18:33:54 INFO - TEST-INFO | LeakSanitizer | To show the addresses of leaked objects add report_objects=1 to LSAN_OPTIONS
[task 2017-01-26T18:33:54.987267Z] 18:33:54 INFO - TEST-INFO | LeakSanitizer | This can be done in testing/mozbase/mozrunner/mozrunner/utils.py
[task 2017-01-26T18:33:54.987397Z] 18:33:54 ERROR - TEST-UNEXPECTED-FAIL | LeakSanitizer | leak at pa_xmalloc0
[0] https://public-artifacts.taskcluster.net/Uuw3Io36QLyJxCDFxapeiQ/0/public/logs/live_backing.log
Assignee | ||
Comment 2•8 years ago
|
||
same for:
accessible/tests/mochitest/actions/test_keys_menu.xul
- https://public-artifacts.taskcluster.net/RFieHFQuSruSUXLs6RLZSw/0/public/logs/live_backing.log
accessible/tests/mochitest/attributes/test_obj_group.xul:
- https://public-artifacts.taskcluster.net/XECTmTymQaGLfESMhhaN0w/0/public/logs/live_backing.log
accessible/tests/mochitest/events/test_contextmenu.xul:
- https://public-artifacts.taskcluster.net/dhuu13ffSbGYNIegWVZtqQ/0/public/logs/live_backing.log
accessible/tests/mochitest/events/test_focus_contextmenu.xul:
- https://public-artifacts.taskcluster.net/UzF6NRB2Sb2_P7iNyz6uCg/0/public/logs/live_backing.log
accessible/tests/mochitest/events/test_focus_general.html
- https://public-artifacts.taskcluster.net/PNWhasCsSXGflffYqEnGkQ/0/public/logs/live_backing.log
accessible/tests/mochitest/events/test_focus_general.xul:
- https://public-artifacts.taskcluster.net/eXrcoQCgR4-zphm_FaJcxQ/0/public/logs/live_backing.log
accessible/tests/mochitest/events/test_focus_listcontrols.xul
- https://public-artifacts.taskcluster.net/RgrV7zcnQnmGbX7FGaZFaQ/0/public/logs/live_backing.log
accessible/tests/mochitest/events/test_focus_menu.xul
- https://queue.taskcluster.net/v1/task/BJWNImvGSum1_u6NoTQ4dA/runs/0/artifacts/public/logs/live_backing.log
accessible/tests/mochitest/states/test_expandable.xul
- https://public-artifacts.taskcluster.net/Ll_8uLEST-6U89bLFPDi_A/0/public/logs/live_backing.log
accessible/tests/mochitest/treeupdate/test_contextmenu.xul
- https://queue.taskcluster.net/v1/task/BufcgywNSFmAAh-nyk75_w/runs/0/artifacts/public/logs/live_backing.log
accessible/tests/mochitest/treeupdate/test_menu.xul
- https://queue.taskcluster.net/v1/task/BLklut5fQmaEpW_dt6IAdA/runs/0/artifacts/public/logs/live_backing.log
accessible/tests/mochitest/treeupdate/test_menubutton.xul:
- https://public-artifacts.taskcluster.net/WuKX4dIZRjmNg_Y6c6LBZA/0/public/logs/live_backing.log
Assignee | ||
Comment 3•8 years ago
|
||
adding a patch if we choose to run on 16.04 and skip tests.
Assignee | ||
Comment 4•8 years ago
|
||
:kats, you had interest in running on 16.04, possibly looking at this leak mentioned here would save us some work in other suites.
Flags: needinfo?(bugmail)
Comment 5•8 years ago
|
||
Yup I can try to debug this. I'm doing a llvm build now so that I can run ASAN locally. Leaving needinfo on me for now.
That being said, wouldn't it be better to add the leaked block to the list of ignored leaks rather than to skip the tests entirely on ASAN builds?
Comment 6•8 years ago
|
||
... in fact, the leaks file already lists existing leaks in libpulsecommon [1]. I suspect it's just the version number of the library that has changed in going to 16.04, so the leak is showing up as "new" again.
[1] http://searchfox.org/mozilla-central/rev/8fa84ca6444e2e01fb405e078f6d2c8da0e55723/build/sanitizers/lsan_suppressions.txt#69
Assignee | ||
Comment 7•8 years ago
|
||
oh, interesting, I wasn't aware of lsan_suppressions.txt :)
I just pushed to try to test if updating this fixes the leak:
https://treeherder.mozilla.org/#/jobs?repo=try&revision=b00fc6e6e0b0e2fee72d9b4110d081cb312e312c
Comment 8•8 years ago
|
||
I just checked a recent asan tc-M(a11y) mozilla-inbound log [1] which shows this:
[task 2017-01-27T14:35:05.831468Z] 14:35:05 INFO - -----------------------------------------------------
[task 2017-01-27T14:35:05.831775Z] 14:35:05 INFO - Suppressions used:
[task 2017-01-27T14:35:05.832011Z] 14:35:05 INFO - count bytes template
.. [snip] ..
[task 2017-01-27T14:35:05.837231Z] 14:35:05 INFO - 1 40 libpulsecommon-1.1.so
.. [snip] ..
So the single 40-byte leak is in fact a pre-existing problem even on Ubuntu 12.04, and is getting ignored because of that leak suppression entry. I think it should be fine to just update the suppression file to ignore it in both libpulsecommon-1.1.so and libpulsecommon-8.0.so, and then remove the -1.1.so entry after all the jobs have been migrated to 16.04. Does that sound reasonable?
[1] https://public-artifacts.taskcluster.net/bFUW00FkTuOwPIhkpjQpgg/0/public/logs/live_backing.log
Flags: needinfo?(bugmail)
Assignee | ||
Comment 9•8 years ago
|
||
yes, let me confirm on the try push and that would be an ideal route forward- thanks for digging this up and helping move this bug forward :)
Assignee | ||
Comment 10•8 years ago
|
||
Comment 11•8 years ago
|
||
Comment on attachment 8831154 [details] [diff] [review]
update libpulsecommon in lsan_suppressions.txt and run a11y on 16.04
Review of attachment 8831154 [details] [diff] [review]:
-----------------------------------------------------------------
::: build/sanitizers/lsan_suppressions.txt
@@ +66,5 @@
> leak:libp11-kit.so
> leak:libpixman-1.so
> leak:libpulse.so
> leak:libpulsecommon-1.1.so
> +leak:libpulsecommon-8.0.so
Maybe add a comment here saying that libpulsecommon-1.1 is on Ubuntu 12.04 and -8.0 is on 16.04?
Attachment #8831154 -
Flags: review?(bugmail) → review+
Comment 12•8 years ago
|
||
Pushed by jmaher@mozilla.com:
https://hg.mozilla.org/integration/mozilla-inbound/rev/6a065534a5a2
run mochitest-a11y on ubuntu 16.04 instead of 12.04. r=kats
Comment 13•8 years ago
|
||
bugherder |
Status: ASSIGNED → RESOLVED
Closed: 8 years ago
status-firefox54:
--- → fixed
Resolution: --- → FIXED
Target Milestone: --- → mozilla54
You need to log in
before you can comment on or make changes to this bug.
Description
•