Closed Bug 1317605 Opened 8 years ago Closed 7 years ago

Intermittent xpcom/tests/unit/test_nsIProcess.js | xpcshell return code: 0

Categories

(Core :: XPCOM, defect)

Product:
Core
Component:
XPCOM
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED INCOMPLETE

People

(Reporter: intermittent-bug-filer, Unassigned)

References

Details

(Keywords: bulk-close-intermittents, intermittent-failure) 

This log seems typical in recent failures:

02:50:07     INFO -  TEST-START | xpcom/tests/unit/test_nsIProcess.js
02:50:08  WARNING -  TEST-UNEXPECTED-FAIL | xpcom/tests/unit/test_nsIProcess.js | xpcshell return code: 0
02:50:08     INFO -  TEST-INFO took 612ms
02:50:08     INFO -  >>>>>>>
02:50:08     INFO -  PROCESS | 28514 | [28514] WARNING: Couldn't get the user appdata directory. Crash events may not be produced.: file /builds/slave/autoland-lx-d-0000000000000000/build/src/toolkit/crashreporter/nsExceptionHandler.cpp, line 2861
02:50:08     INFO -  (xpcshell/head.js) | test MAIN run_test pending (1)
02:50:08     INFO -  TEST-PASS | xpcom/tests/unit/test_nsIProcess.js | test_kill - [test_kill : 23] true == true
02:50:08     INFO -  TEST-PASS | xpcom/tests/unit/test_nsIProcess.js | test_kill - [test_kill : 33] true == true
02:50:08     INFO -  NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIProcess.kill]
02:50:08     INFO -  test_kill@/builds/slave/test/build/tests/xpcshell/tests/xpcom/tests/unit/test_nsIProcess.js:35:3
02:50:08     INFO -  run_test@/builds/slave/test/build/tests/xpcshell/tests/xpcom/tests/unit/test_nsIProcess.js:178:3
02:50:08     INFO -  _execute_test@/builds/slave/test/build/tests/xpcshell/head.js:535:7
02:50:08     INFO -  @-e:1:1
See Also: → 1325609
Nathan - Any ideas? Why might process.kill() be intermittently failing at https://dxr.mozilla.org/mozilla-central/rev/b1c31c4a0a678194931779e0f13fba7b508eb109/xpcom/tests/unit/test_nsIProcess.js#35, on Linux? In addition to this bug, there are additional failures in bug 1325609 (bug 1317155) which seem to be the same failure, reported differently.
Flags: needinfo?(nfroyd)
The code for process.kill() lives at:

http://dxr.mozilla.org/mozilla-central/source/xpcom/threads/nsProcessCommon.cpp#599

which says for Linux, there are a couple of different failure modes:

- mThread doesn't exist
- mProcess doesn't exist
- we couldn't kill mProcess (mainly the process pid doesn't actually exist)

I'm not sure which case we're hitting; I could put together a patch that would return different error codes for each case to help narrow it down.  But at first glance, none of those cases should be happening.

...unless the test process we're using:

http://dxr.mozilla.org/mozilla-central/source/xpcom/tests/TestBlockingProcess.cpp#4

is somehow seeing something on its stdin, intermittently, and returning before we get a chance to interrupt it?
Flags: needinfo?(nfroyd)
(In reply to Nathan Froyd [:froydnj] from comment #11)
> I'm not sure which case we're hitting; I could put together a patch that
> would return different error codes for each case to help narrow it down. 

Thanks Nathan. Don't worry about it - I'll narrow it down.
I tried to reproduce this on try with additional logging, but had no luck. 

I notice that this bug only happens on linux32/debug **on buildbot** - never on taskcluster. We have just recently stopped running linux32/debug on buildbot, so I expect these failures to die off here on their own. Bug 1325609 - which might be the taskcluster version of this bug - remains problematic.
Status: NEW → RESOLVED
Closed: 7 years ago
Keywords: bulk-close-intermittents
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.