Bug 663657 collects test failure reports for Android test runs that hang: If the browser or the test hangs, no output is generated and eventually the harness kills the test and cleans up. We recently determined that the logs reported in that bug were not accurate, because part of the log was buffered when the process was killed. After fixing the log buffering problem, logs reported in that bug have been more consistent. Several of the recent (accurate) logs show that test_unsafeBidiChars.xhtml hangs intermittently: KWierso https://tbpl.mozilla.org/php/getParsedLog.php?id=33000618&tree=Fx-Team Android 2.2 Armv6 Tegra fx-team opt test mochitest-8 on 2014-01-14 16:39:11 revision: 74335bb5706a slave: tegra-163 145856 INFO TEST-END | /tests/toolkit/xre/test/test_fpuhandler.html | finished in 253ms 145857 INFO TEST-START | /tests/uriloader/exthandler/tests/mochitest/test_badMimeType.html 145858 INFO TEST-PASS | /tests/uriloader/exthandler/tests/mochitest/test_badMimeType.html | badMimeType has changed, making this test invalid 145859 INFO TEST-PASS | /tests/uriloader/exthandler/tests/mochitest/test_badMimeType.html | The test shouldn't trigger a crash 145860 INFO TEST-INFO | MEMORY STAT vsize after test: 595505152 145861 INFO TEST-INFO | MEMORY STAT residentFast after test: 194707456 145862 INFO TEST-INFO | MEMORY STAT heapAllocated after test: 54718208 145863 INFO TEST-END | /tests/uriloader/exthandler/tests/mochitest/test_badMimeType.html | finished in 189ms 145864 INFO TEST-START | /tests/uriloader/exthandler/tests/mochitest/test_unsafeBidiChars.xhtml command timed out: 2400 seconds without output, attempting to kill process killed by signal 9 program finished with exit code -1 elapsedTime=2786.596891 Also: https://tbpl.mozilla.org/php/getParsedLog.php?id=32980456&tree=Mozilla-Central#error0 https://tbpl.mozilla.org/php/getParsedLog.php?id=32974610&tree=Mozilla-Inbound#error0 This intermittent hang was previously reported in bug 787911, but that was closed when it could not be reproduced. There is also one report of a crash in this test in bug 900256.

So, there isn't really any actionable data here. Nothing in the logs, no stack, etc. Can we at least get a stack trace of what was happening when the process was killed? Also, are you sure that there are no more log buffering working against us?

I am 99% sure that there are no more log buffering problems. I fully agree that there isn't much to go on here. I will see what I can do to get better logs (it may take a while though).

Reviewing the reports in bug 663657 since 2014-01-14 that appear to be hangs during test_unsafeBidiChars, it seems to me that all were run on one of: tegra-369 tegra-353 tegra-163 :edmorley -- Check my work?

(In reply to Geoff Brown [:gbrown] from comment #3) > Reviewing the reports in bug 663657 since 2014-01-14 that appear to be hangs > during test_unsafeBidiChars, it seems to me that all were run on one of: > tegra-369 > tegra-353 > tegra-163 > > :edmorley -- Check my work? (Sorry was away :-)) The failure occurs only on those three plus one more, tegra-249: https://tbpl.mozilla.org/php/getParsedLog.php?id=33453291&tree=Mozilla-Inbound https://tbpl.mozilla.org/php/getParsedLog.php?id=33609429&tree=Mozilla-Inbound https://tbpl.mozilla.org/php/getParsedLog.php?id=33596542&tree=Mozilla-Central https://tbpl.mozilla.org/php/getParsedLog.php?id=33474485&tree=B2g-Inbound Strange! (But not unheard of!) :-)