https://bugs.winehq.org/show_bug.cgi?id=55153
Bug ID: 55153 Summary: comdlg32:filedlg sometimes crashes on Windows 7 Product: Wine Version: unspecified Hardware: x86-64 OS: Windows Status: NEW Severity: normal Priority: P2 Component: comdlg32 Assignee: wine-bugs@winehq.org Reporter: fgouget@codeweavers.com
comdlg32:filedlg sometimes crashes on Windows 7:
comdlg32:filedlg start dlls/comdlg32/tests/filedlg.c 0ef8:filedlg: unhandled exception c0000005 at 766523AD comdlg32:filedlg:0ef8 done (-1073741819) in 5s 56B
See https://test.winehq.org/data/patterns.html#comdlg32:filedlg
Where c0000005 == STATUS_ACCESS_VIOLATION
The first crash happened on 2023-02-28 and there has been a bit over 1.5 per month since.
https://bugs.winehq.org/show_bug.cgi?id=55153
François Gouget fgouget@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Keywords| |source, testcase
https://bugs.winehq.org/show_bug.cgi?id=55153
Fabian Maurer dark.shadow4@web.de changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |dark.shadow4@web.de
--- Comment #1 from Fabian Maurer dark.shadow4@web.de --- I ran hundreds of tests on my Win7 VM, but I can't replicate. Is there a way to get the exact same version and configuration of windows as in the VM?
https://bugs.winehq.org/show_bug.cgi?id=55153
--- Comment #2 from François Gouget fgouget@codeweavers.com --- There is a description of the VM configuration on the TestBot when you click on "Windows 7 Ultimate / vm1":
Installed Internet Explorer 11 + all updates up to 2020-11-25. Inserted the MemTest86 CD. [CPU:2*IvyBridge-IBRS RAM:2GB disk:scsi-unmap eth:e1000-metered snd:ich9 GPU:vga display:spice virtio:0.1.185 testagentd:1.8]
I tried reproducing it on w7u (using the 2023-06-22 build since it failed) but I cannot reproduce it either. I think it's just one of these heisenfailures which is going to make it very hard to figure out :-(
For now we know it happens in one of the functions below since we don't have the "some interactive resizable dialog tests" skip message in the crash case: test_DialogCancel(); test_create_view_window2(); test_create_view_template(); test_arrange(); test_resize(); test_ok(); test_getfolderpath(); test_mru();
https://bugs.winehq.org/show_bug.cgi?id=55153
--- Comment #3 from Fabian Maurer dark.shadow4@web.de --- Reminds me, why don't we get the usual "this is the last test seen before the exception" message here?
https://bugs.winehq.org/show_bug.cgi?id=55153
--- Comment #4 from François Gouget fgouget@codeweavers.com --- We don't always get those.
There are some cases where it's because the crash happens in a thread that never called ok() or trace() and thus did not set up the TLS data.
But the format of the "unhandled exception" message varies depending on the platform (for reasons unknown onmy end) so I suspect that there are some cases where our exception filter is not being called at all; in which case one would not get the "last seen test" message either.
https://bugs.winehq.org/show_bug.cgi?id=55153
--- Comment #5 from Fabian Maurer dark.shadow4@web.de --- Would it be possible to add simple logging and then run a few batches of tests on the testbot until the error occurs? Since that is the only thing to reproduce the crash... Although it would probably put quite a bit of load on the testbot.