https://bugs.winehq.org/show_bug.cgi?id=54347
Bug ID: 54347 Summary: winealsa's MIDI support sometimes causes a dmloader:loader stack overflow in Wine Product: Wine Version: unspecified Hardware: x86-64 OS: Linux Status: NEW Severity: normal Priority: P2 Component: winealsa.drv Assignee: wine-bugs@winehq.org Reporter: fgouget@codeweavers.com Distribution: ---
winealsa's MIDI support sometimes causes a dmloader:loader stack overflow in Wine:
01e0:Call KERNEL32.SetThreadDescription(fffffffe,01e1c034 L"winealsa_midi_notify") ret=01e1128e ... 01e0:err:virtual:virtual_setup_exception stack overflow 900 bytes addr 0x7bc56a8d stack 0x1e90c7c (0x1e90000-0x1e91000-0x2090000)
On test.winehq.org this probably corresponds to the c0000005 crashes though to be sure one would need to have access to the Wine traces and +relay or something to identify the overflowing thread:
0530:loader: 104 tests executed (0 marked as todo, 0 as flaky, 0 failures), 0 skipped. dmloader:loader:0530 done (-1073741819) in 0s 88B Test failed: crash (c0000005)
See https://test.winehq.org/data/patterns.html#dmloader:loader
Looking at the corresponding TestBot jobs one can see two instances per run but there are not enough traces to match the thread to winealsa's midi thread:
05a4:err:virtual:virtual_setup_exception stack overflow 900 bytes addr 0x7bc56a8d stack 0x4dc0c7c (0x4dc0000-0x4dc1000-0x4fc0000) 05a4:err:virtual:virtual_setup_exception nested exception on signal stack addr (nil) stack 0x7ff6ecac ... 0210:err:virtual:virtual_setup_exception stack overflow 900 bytes addr 0x7bc56a8d stack 0x2fd0c7c (0x2fd0000-0x2fd1000-0x31d0000) 0210:err:virtual:virtual_setup_exception nested exception on signal stack addr (nil) stack 0x7ff9ecac
It's also hard to match those to specific tests because the 'Running' lines are out of sync but the second instance looks like it could be from dmloader:loader while the former would be from an earlier test, possibly amstream:amstream.
https://bugs.winehq.org/show_bug.cgi?id=54347
François Gouget fgouget@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |hdavies@codeweavers.com Keywords| |source, testcase
https://bugs.winehq.org/show_bug.cgi?id=54347
François Gouget fgouget@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Summary|winealsa's MIDI support |winealsa's MIDI support |sometimes causes a |sometimes causes |dmloader:loader stack |amstream:amstream, |overflow in Wine |dmloader:loader and | |winmm:mci stack overflows | |in Wine
--- Comment #1 from François Gouget fgouget@codeweavers.com --- This issue can also happen in amstream:amstream and winmm:mci which makes sense since they also cause winealsa.drv to be loaded. winmm:midi obviously also causes winealsa's midi support to be initialized but it seems to only get the critical section issue (see bug 54346).
https://bugs.winehq.org/show_bug.cgi?id=54347
François Gouget fgouget@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Summary|winealsa's MIDI support |winealsa's MIDI support |sometimes causes |sometimes causes |amstream:amstream, |amstream:amstream, |dmloader:loader and |dmloader:loader, |winmm:mci stack overflows |quartz:filtermapper, |in Wine |qcap:capturegraph and | |winmm:mci stack overflows | |in Wine
--- Comment #2 from François Gouget fgouget@codeweavers.com --- I have confirmed this to also impact quartz:filtermapper and qcap:capturegraph.