https://bugs.winehq.org/show_bug.cgi?id=44484
Bug ID: 44484 Summary: Wine segfaults after running Matlab's compiler (mcc) Product: Wine Version: unspecified Hardware: x86-64 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: -unknown Assignee: wine-bugs@winehq.org Reporter: julius.schwartzenberg@gmail.com Distribution: ---
With current Git (ff4a02fe3cf2eb096508ac7d553986c3e4830372) when running the compiler from Matlab 2016b 64-bit version, everything works fine, but Wine always segfaults in the end. Even when just letting it list its arguments, there will be a segfault in the end.
The application itself works fine. Somehow when it quits, a segfault is triggered.
I tested it on RHEL6 and RHEL7, on both systems I see this behavior.
This message is printed a lot: 002d:fixme:msvcp:_Mtx_init_in_situ unknown flags ignored: 2
After that this message is printed once: 0030:fixme:win:UnregisterDeviceNotification (handle=0xcafecafe), STUB!
Then there is the segfault.
https://bugs.winehq.org/show_bug.cgi?id=44484
--- Comment #1 from Julius Schwartzenberg julius.schwartzenberg@gmail.com --- I tried setting msvcp120 and msvcr120 to native (as they are included), but that did not make any difference.
https://bugs.winehq.org/show_bug.cgi?id=44484
--- Comment #2 from Dmitry Timoshkov dmitry@baikal.ru --- (In reply to Julius Schwartzenberg from comment #1)
I tried setting msvcp120 and msvcr120 to native (as they are included), but that did not make any difference.
You either need to install them with winetricks or manually delete fake Wine manifests. There are multiple bugs resolved as WONTFIX for this broken behaviour.
https://bugs.winehq.org/show_bug.cgi?id=44484
--- Comment #3 from Julius Schwartzenberg julius.schwartzenberg@gmail.com --- Are you sure? I think Wine picked up the native versions after I did that as the FIXME messages from the Wine versions disappears.
What I meant was that Wine still segfaults after running mcc when these are set to native.
I'm just not completely sure what I could do to get more information about why Wine segfaults. I'm doing a build now with CFLAGS='-g' and hope that gdb will show some more information then.
https://bugs.winehq.org/show_bug.cgi?id=44484
--- Comment #4 from Dmitry Timoshkov dmitry@baikal.ru --- (In reply to Julius Schwartzenberg from comment #3)
Are you sure? I think Wine picked up the native versions after I did that as the FIXME messages from the Wine versions disappears.
See https://bugs.winehq.org/show_bug.cgi?id=33604 and https://bugs.winehq.org/show_bug.cgi?id=28225
To make sure that native versions are picked up create a log with +module and search there for a dll name you are interested in.
https://bugs.winehq.org/show_bug.cgi?id=44484
--- Comment #5 from Julius Schwartzenberg julius.schwartzenberg@gmail.com --- I just tried running a -g build, but I still get this with gdb: Program received signal SIGSEGV, Segmentation fault. 0x000000001aabedd5 in ?? () (gdb) bt #0 0x000000001aabedd5 in ?? () #1 0x000000000000001f in ?? () #2 0x00000000c956bf79 in ?? () #3 0x00000000085fba08 in ?? () #4 0x000000001aa7e1cc in ?? () #5 0x0000000008ecd000 in ?? () #6 0x00000000000000b6 in ?? () #7 0x00000000085fba08 in ?? () #8 0x000000001aa663f0 in ?? () #9 0x000000001aa663f0 in ?? () #10 0x0000000000000000 in ?? ()
I checked the native module loading and I see messages like this: 0031:trace:module:load_dll Found L"Z:\home\juschwar\matlab\R2016b\bin\win64\MSVCR120.dll" for L"MSVCR120.dll" at 0x240000, count=-1 0031:trace:module:load_dll Found L"Z:\home\juschwar\matlab\R2016b\bin\win64\MSVCP120.dll" for L"MSVCP120.dll" at 0x180000000, count=-1
So I think it's picking up the native DLLs without problems.
https://bugs.winehq.org/show_bug.cgi?id=44484
--- Comment #6 from Nikolay Sivov bunglehead@gmail.com --- Msvcr120/msvcp120 don't use sxs manifests, I think msvcr90/msvcp90 were the last to use that. For newer modules only regular override settings should matter.
https://bugs.winehq.org/show_bug.cgi?id=44484
Fabian Maurer dark.shadow4@web.de changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |dark.shadow4@web.de
--- Comment #7 from Fabian Maurer dark.shadow4@web.de --- Is there a demo version that can be used for testing?
https://bugs.winehq.org/show_bug.cgi?id=44484
--- Comment #8 from Julius Schwartzenberg julius.schwartzenberg@gmail.com --- Unfortunately not. The trial version of Matlab does not have the compiler included.