https://bugs.winehq.org/show_bug.cgi?id=55283
Bug ID: 55283 Summary: amstream:amstream systematically crashes and times out on gitlab-debian-32 Product: Wine Version: unspecified Hardware: x86-64 OS: Linux Status: NEW Severity: major Priority: P2 Component: quartz Assignee: wine-bugs@winehq.org Reporter: fgouget@codeweavers.com Distribution: ---
amstream:amstream systematically crashes and times out on gitlab-debian-32:
Unhandled exception: assertion failed in wow64 32-bit code (0xf7f19559). Backtrace: =>0 0xf7f9a559 __kernel_vsyscall+0x9() in [vdso].so (0xf7f34ff4) 1 0xf7da21d7 in libc.so.6 (+0x8a1d7) (0xf7f34ff4) 2 0xf7d510d1 in libc.so.6 (+0x390d1) (0xf7f34ff4) 3 0xf7d3a26a in libc.so.6 (+0x2226a) (0xf7f34ff4) 4 0x71a4de79 in libllvm-15.so.1 (+0x875e79) (0x0022e910) 5 0x00000040 (0x00000040) 0xf7f9a559 __kernel_vsyscall+0x9 in [vdso].so: pop %ebp
See https://test.winehq.org/data/patterns.html#amstream:amstream
The failures started on 2023-07-10 which corresponds to the GitLab CI switch to Debian 12. What makes this very bad is that the failure is systematic so every single merge request has been getting a GitLab CI error since then.
So it is pretty urgent to either: 1. Fix the test if something is wrong with it. 2. Fix the gitlab-debian-32 test environment if it is wrong. 3. Skip amstream:amstream if neither of the above is doable in a reasonable timeframe.
https://bugs.winehq.org/show_bug.cgi?id=55283
François Gouget fgouget@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Keywords| |regression, source, | |testcase Regression SHA1| |5799e89f15b5584f186dcc2fd24 | |b98e0a7bd6c10
https://bugs.winehq.org/show_bug.cgi?id=55283
François Gouget fgouget@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |z.figura12@gmail.com
--- Comment #1 from François Gouget fgouget@codeweavers.com --- I submitted an MR disabling this test in case there is no quick fix: https://gitlab.winehq.org/wine/wine/-/merge_requests/3361
https://bugs.winehq.org/show_bug.cgi?id=55283
Alexandre Julliard julliard@winehq.org changed:
What |Removed |Added ---------------------------------------------------------------------------- Resolution|--- |DUPLICATE Status|NEW |RESOLVED
--- Comment #2 from Alexandre Julliard julliard@winehq.org --- That's another symptom of bug 52213.
*** This bug has been marked as a duplicate of bug 52213 ***
https://bugs.winehq.org/show_bug.cgi?id=55283
Bernhard Übelacker bernhardu@mailbox.org changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |bernhardu@mailbox.org
--- Comment #3 from Bernhard Übelacker bernhardu@mailbox.org --- Created attachment 74897 --> https://bugs.winehq.org/attachment.cgi?id=74897 details for backtrace translation
Just if it is still of any interest. I think the above backtrace can be manually translated with the help of the dbgsym packages to following functions:
1 0xf7da21d7 in libc.so.6 (+0x8a1d7) (0xf7f34ff4) 0xb7c8a1d0 <__pthread_kill_implementation+272>: call *%gs:0x10 0xb7c8a1d7
2 0xf7d510d1 in libc.so.6 (+0x390d1) (0xf7f34ff4) 0xb7c390cc <__GI_raise+28>: call 0xb7c8a230 <__GI___pthread_kill> 0xb7c390d1
3 0xf7d3a26a in libc.so.6 (+0x2226a) (0xf7f34ff4) 0xb7c22265 <__GI_abort+232>: call 0xb7c390b0 <__GI_raise> 0xb7c2226a
4 0x71a4de79 in libllvm-15.so.1 (+0x875e79) (0x0022e910) 0xb0875e74 <_ZN4llvm18report_fatal_errorERKNS_5TwineEb+468>: call 0xb0776400 abort@plt 0xb0875e79 <llvm::report_fatal_error(llvm::Twine const&, bool)+468>
https://bugs.winehq.org/show_bug.cgi?id=55283
Zeb Figura z.figura12@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Resolution|DUPLICATE |FIXED
--- Comment #4 from Zeb Figura z.figura12@gmail.com --- (In reply to Alexandre Julliard from comment #2)
That's another symptom of bug 52213.
*** This bug has been marked as a duplicate of bug 52213 ***
I don't think this is 52213? 52213 is about the case where a thread crashes as a result of being terminated. Here there's an assertion failure and the test never completes. 52213 may be exacerbating it in some way, but fundamentally I figure that this bug is about the assertion failure.
https://bugs.winehq.org/show_bug.cgi?id=55283
Zeb Figura z.figura12@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Resolution|FIXED |DUPLICATE
--- Comment #5 from Zeb Figura z.figura12@gmail.com --- I did not mean to change this to FIXED; not sure how that happened...
*** This bug has been marked as a duplicate of bug 52213 ***
https://bugs.winehq.org/show_bug.cgi?id=55283
--- Comment #6 from Zeb Figura z.figura12@gmail.com --- From https://gitlab.winehq.org/AurimasF/wine/-/jobs/23374 I also see this:
LLVM ERROR: Unable to allocate section memory! wine: Assertion failed at address F7F3E559 (thread 04e0), starting debugger...
We're running out of memory in amstream tests, really? That's surprising. It probably doesn't help that GStreamer plugins (from initial scan, I guess) are taking up about 300 MB, but even then, where the hell is the rest of the address space going?
https://bugs.winehq.org/show_bug.cgi?id=55283
--- Comment #7 from Zeb Figura z.figura12@gmail.com --- Also, the fact that this *only* affects amstream is baffling. If it was a GStreamer problem, why are other multimedia modules not affected? If it was a D3D problem, why aren't quartz, ddraw or d2d1 affected? I don't think amstream is a the heaviest user of quartz, and it surely isn't the heaviest user of d3d. Do we just have one really bad leak in the tests?
https://bugs.winehq.org/show_bug.cgi?id=55283
--- Comment #8 from Bernhard Übelacker bernhardu@mailbox.org --- I tried to search for the llvm message to which debian returned just one hit: https://codesearch.debian.net/search?q=Unable+to+allocate+section+memory+pac... https://sources.debian.org/src/llvm-toolchain-15/1:15.0.7-6/llvm/lib/Executi...
An internet search returns a few hits related to resource limits. e.g. https://issues.apache.org/jira/browse/IMPALA-5605 Which values might be in place for gitlab CI?
https://bugs.winehq.org/show_bug.cgi?id=55283
Alistair Leslie-Hughes leslie_alistair@hotmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #9 from Alistair Leslie-Hughes leslie_alistair@hotmail.com --- Close duplicate.
https://bugs.winehq.org/show_bug.cgi?id=55283
Zeb Figura z.figura12@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Resolution|DUPLICATE |--- Status|CLOSED |REOPENED
--- Comment #10 from Zeb Figura z.figura12@gmail.com --- Unresolving duplicate; I've fixed some leaks and it doesn't seem to be crashing the CI anymore, at least from a local test.
https://bugs.winehq.org/show_bug.cgi?id=55283
--- Comment #11 from Zeb Figura z.figura12@gmail.com --- *** Bug 54095 has been marked as a duplicate of this bug. ***
https://bugs.winehq.org/show_bug.cgi?id=55283
--- Comment #12 from Alexandre Julliard julliard@winehq.org --- Thanks! It's still crashing on my box so that was indeed a different issue, sorry about the confusion.
https://bugs.winehq.org/show_bug.cgi?id=55283
Zeb Figura z.figura12@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution|--- |FIXED Fixed by SHA1| |70dc4a6a1d9e96a2d94f8fa99e7 | |a89640325af96
--- Comment #13 from Zeb Figura z.figura12@gmail.com --- gitlab seems to be fixed after https://source.winehq.org/git/wine.git/commitdiff/70dc4a6a1d9e96a2d94f8fa99e7a89640325af96. I suspect there's still a deeper problem here—simply leaking two d3d devices shouldn't be enough to push us over the edge—but at least the crash is gone.
https://bugs.winehq.org/show_bug.cgi?id=55283
Alexandre Julliard julliard@winehq.org changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #14 from Alexandre Julliard julliard@winehq.org --- Closing bugs fixed in 8.14.