[Bug 28640] New: kernel32/thread.ok fails on ubuntu 11.10?

http://bugs.winehq.org/show_bug.cgi?id=28640

Bug #: 28640
           Summary: kernel32/thread.ok fails on ubuntu 11.10?
           Product: Wine
           Version: 1.3.30
          Platform: x86
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: kernel32
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]
    Classification: Unclassified

Don't know if this is a wine bug, an ubuntu bug, or a kernel bug yet.

kernel32/thread.ok fails reliably for me on ubuntu 11.10
(which uses linux-3.0.0, fwiw).  Logs are slightly clearer when
run on a single core, but it still fails.  Log:

dank@i7:~/wine-git/dlls/kernel32/tests$ taskset -c 0 make thread.ok
../../../tools/runtest -q -P wine -M kernel32.dll -T ../../.. -p
kernel32_test.exe.so thread.c && touch thread.ok
fixme:thread:NtQueryInformationThread Cannot get kerneltime or usertime of
other threads
thread.c:823: Test failed: SetThreadAffinityMask(thread,-1) failed to request
all processors.
fixme:thread:SetThreadIdealProcessor (0xfffffffe): stub
fixme:thread:SetThreadIdealProcessor (0xfffffffe): stub
fixme:thread:SetThreadIdealProcessor (0xfffffffe): stub
Protocol error:003b: partial read 12
wine client error:3a: read: Bad address
err:seh:raise_exception Unhandled exception code c0000005 flags 0 addr
0x690abb19
wine client error:b: write: Bad file descriptor
wine client error:27: write: Bad file descriptor

The errors vary, but there's always a bad file descriptor, and often
a crash.    Sometimes starting a wineserver in the foreground fails
because trying to get the /tmp/.wine-.../lock fails with EAGAIN, too.

I'll attach a +relay,+seh log.   Looks kinda like a null pointer read
crash near the start of the thread that runs a queued work item?

[Bug 28640] kernel32/thread.ok fails on ubuntu 11.10?

http://bugs.winehq.org/show_bug.cgi?id=28640

--- Comment #1 from Dan Kegel  2011-10-10 18:56:40 CDT ---
Created attachment 36811
  --> http://bugs.winehq.org/attachment.cgi?id=36811
+relay,+seh log of crash

[Bug 28640] kernel32/thread.ok fails on ubuntu 11.10?

http://bugs.winehq.org/show_bug.cgi?id=28640

--- Comment #2 from Dan Kegel  2011-10-10 19:07:22 CDT ---
Also filed as https://bugs.launchpad.net/ubuntu/+bug/872063

[Bug 28640] kernel32/thread.ok fails on ubuntu 11.10?

http://bugs.winehq.org/show_bug.cgi?id=28640

Austin English  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |download, testcase
                 CC|                            |[e-mail]

--- Comment #3 from Austin English  2011-10-10 19:18:27 CDT ---
My money's on gcc-4.6. I had similar problems on debian testing, downgrading
gcc fixed it.

Also using linux 3.0.0.

[Bug 28640] kernel32/thread.ok fails on ubuntu 11.10?

http://bugs.winehq.org/show_bug.cgi?id=28640

--- Comment #4 from Dan Kegel  2011-10-10 19:21:24 CDT ---
Yup.  Was using default gcc (Ubuntu/Linaro 4.6.1-9ubuntu3) 4.6.1,
recompiling just kernel32 with gcc-4.5 made the error go away.

So it's a wine bug and/or a gcc bug...

[Bug 28640] kernel32/thread.ok fails on ubuntu 11.10?

http://bugs.winehq.org/show_bug.cgi?id=28640

--- Comment #5 from Dan Kegel  2011-10-10 19:28:46 CDT ---
See also http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=641056

[Bug 28640] kernel32/thread.ok fails on ubuntu 11.10?

http://bugs.winehq.org/show_bug.cgi?id=28640

--- Comment #6 from Dan Kegel  2011-10-19 13:51:01 CDT ---
Adding -fno-omit-frame-pointer -fno-asynchronous-unwind-tables
seems to make it work even with gcc-4.6.1 here.

[Bug 28640] kernel32/thread.ok fails on ubuntu 11.10?

https://bugs.winehq.org/show_bug.cgi?id=28640

--- Comment #7 from Austin English  ---
Dan, can you please retest or mark as abandoned if you no longer can on that
setup?

[Bug 28640] kernel32/thread.ok fails on ubuntu 11.10?

https://bugs.winehq.org/show_bug.cgi?id=28640

[e-mail] changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #8 from [e-mail] ---
Those upstream bugs are fixed.

this bug should be closed as upstream/fixed

[Bug 28640] kernel32/thread.ok fails on ubuntu 11.10?

https://bugs.winehq.org/show_bug.cgi?id=28640

Austin English  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |UPSTREAM

--- Comment #9 from Austin English  ---
Not a wine bug.

[Bug 28640] kernel32/thread.ok fails on ubuntu 11.10?

https://bugs.winehq.org/show_bug.cgi?id=28640

Austin English  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #10 from Austin English  ---
Closing.