http://bugs.winehq.org/show_bug.cgi?id=32354
Bug #: 32354 Summary: testbot: A crashing test is not detected Product: Wine-Testbot Version: unspecified Platform: x86 OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: unknown AssignedTo: wine-bugs@winehq.org ReportedBy: wine.dev@web.de Classification: Unclassified
A 64bit test crashed on all 64bit testbot machines, but the summary has "0" as "Number of failures".
Test run for the broken patch: http://testbot.winehq.org/JobDetails.pl?Key=22963&log_301=1#k301
Later test: http://testbot.winehq.org/JobDetails.pl?Key=23106&log_302=1#k302
(I send a fix for the broken code in wine in some minutes)
http://bugs.winehq.org/show_bug.cgi?id=32354
Detlef Riekenberg wine.dev@web.de changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |fgouget@codeweavers.com
http://bugs.winehq.org/show_bug.cgi?id=32354
--- Comment #1 from François Gouget fgouget@codeweavers.com 2012-12-17 16:26:58 CST --- I don't see 'Test failed:' anywhere for these tests. So the '0 failures' report seems plausible. That may mean the test itself may not be doing all that it needs to detect errors.
http://bugs.winehq.org/show_bug.cgi?id=32354
--- Comment #2 from Detlef Riekenberg wine.dev@web.de 2012-12-21 07:59:40 CST --- (In reply to comment #1)
I don't see 'Test failed:' anywhere for these tests.
That's the issue. The crash is not detected by the script. winetest.exe detected the crash (See the last line): advapi32:security done (3221226356)
So the '0 failures' report seems plausible.
That's the result from the first summary line (from the slave process): security: 18 tests executed (0 marked as todo, 0 failures), 0 skipped.
The script remember the "0 failures" here and failed to detect, that the summary line (from the test.exe) before the last line is missing
A winetest.exe output for a different crash (no slave process): comdlg32:filedlg done (-1073741819)
http://test.winehq.org/data/f30c18f3d245f40a0aca8708b30431f9949ed6ee/win7_fg...
Simple asumption: Start a slave process with a test (to get a summary line) and i expect, that this crash on your machine will be undetected.
http://bugs.winehq.org/show_bug.cgi?id=32354
François Gouget fgouget@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|wine-bugs@winehq.org |fgouget@codeweavers.com
http://bugs.winehq.org/show_bug.cgi?id=32354
--- Comment #3 from François Gouget fgouget@codeweavers.com 2013-04-12 10:06:08 CDT --- Created attachment 44156 --> http://bugs.winehq.org/attachment.cgi?id=44156 Attempt to reproduce the issue
http://bugs.winehq.org/show_bug.cgi?id=32354
François Gouget fgouget@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Attachment #44156|0 |1 is obsolete| |
--- Comment #4 from François Gouget fgouget@codeweavers.com 2013-04-12 10:37:35 CDT --- Created attachment 44157 --> http://bugs.winehq.org/attachment.cgi?id=44157 Patch to reproduce this bug
I tried to reproduce this bug by causing a simple crash in the parent but this did not work because the folloing line gets added to the log:
security: unhandled exception c0000005 at 00428C4B
The original misbehaving jobs are gone but fortunately I found the commit that fixed the original crash: commit 53bd69491bc6b71bd59e0442b84dbb6e3aee4944. So I created this patch which can be used to reproduce this issue by running it on a 64-bit Windows version.
https://bugs.winehq.org/show_bug.cgi?id=32354
--- Comment #5 from François Gouget fgouget@codeweavers.com --- The attachment no longer lets one reproduce the bug, probably because changes elsewhere in the test fixed some memory corruption issue so this LocaFree() call no longer causes the process to exit silently.
Fortunately reproducing this bug is as simple as just calling exit(0) anywhere after a subprocess has been run. I added a test for this in the TestBot test suite: https://github.com/fgouget/wine/tree/wtbsuite
https://bugs.winehq.org/show_bug.cgi?id=32354
François Gouget fgouget@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |ASSIGNED
--- Comment #6 from François Gouget fgouget@codeweavers.com --- A patch has been submitted to fix this: https://www.winehq.org/pipermail/wine-patches/2017-January/156836.html
https://bugs.winehq.org/show_bug.cgi?id=32354
François Gouget fgouget@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Resolution|--- |FIXED Status|ASSIGNED |RESOLVED
--- Comment #7 from François Gouget fgouget@codeweavers.com --- In the end this got fixed differently. The test result lines now contain the pid which makes it possible to distinguish the subprocess result line from that of the main process.
Here is the relevant TestBot patch:
commit 75f60f8068e640281ff4e624ebafb11bc604767a Author: Francois Gouget fgouget@codeweavers.com Date: Sat Feb 18 11:09:18 2017 +0100
testbot/WineRunTask: Better check the task report.
This makes the WineRunTask analysis match the expected results on the report test. In particular it now leverages the pid traces to detect when a test has no test summary line for its main process.
Signed-off-by: Francois Gouget fgouget@codeweavers.com Signed-off-by: Alexandre Julliard julliard@winehq.org
There are other patches for stricter scanning of the test results. The test.winehq.org site also has a similar set of patches.
https://bugs.winehq.org/show_bug.cgi?id=32354
Austin English austinenglish@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #8 from Austin English austinenglish@gmail.com --- Closing.
-
wine-bugs＠winehq.org