|
From: Dan K. <da...@ke...> - 2008-11-12 07:02:26
|
With valgrind from svn, I'm getting this error fairly frequently when running any of Wine's 3D unit tests. It seems to be triggered by a bug in the platform's nvidia opengl driver. Would you like more info? ==17594== Conditional jump or move depends on uninitialised value(s) ==17594== at 0x6A66915: _nv000098gl (in /usr/lib/libGLcore.so.177.80) --17594-- VALGRIND INTERNAL ERROR: Valgrind received a signal 11 (SIGSEGV) - exiting --17594-- si_code=2; Faulting address: 0xBEEF0017; sp: 0x62b5cd7c valgrind: the 'impossible' happened: Killed by fatal signal ==17594== at 0x3803A9B8: vgPlain_get_StackTrace_wrk (m_stacktrace.c:167) ==17594== by 0x3803AB0F: vgPlain_get_StackTrace (m_stacktrace.c:485) ==17594== by 0x38025008: record_ExeContext_wrk (m_execontext.c:316) ==17594== by 0x38001C6B: die_and_free_mem (mc_malloc_wrappers.c:300) ==17594== by 0x38002847: vgMemCheck_free (mc_malloc_wrappers.c:328) ==17594== by 0x3804F052: vgPlain_scheduler (scheduler.c:1303) ==17594== by 0x380639F8: run_a_thread_NORETURN (syswrap-linux.c:89) sched status: running_tid=1 Thread 1: status = VgTs_Runnable |
|
From: Julian S. <js...@ac...> - 2008-11-12 08:11:39
|
On Wednesday 12 November 2008, Dan Kegel wrote: > With valgrind from svn, I'm getting this error fairly frequently > when running any of Wine's 3D unit tests. It seems to > be triggered by a bug in the platform's nvidia opengl driver. > Would you like more info? Yes, all the parameters to get_StackTrace_wrk that are not marked /*OUT*/. J |
|
From: Dan K. <da...@ke...> - 2009-03-22 01:45:28
|
On Wed, Nov 12, 2008 at 12:59 AM, Julian Seward <js...@ac...> wrote: > On Wednesday 12 November 2008, Dan Kegel wrote: >> With valgrind from svn, I'm getting this error fairly frequently >> when running any of Wine's 3D unit tests. It seems to >> be triggered by a bug in the platform's nvidia opengl driver. >> Would you like more info? > > Yes, all the parameters to get_StackTrace_wrk that are not > marked /*OUT*/. I'm getting this reliably now running chromium's base_unittests.exe under current valgrind trunk. To repeat: wget http://kegel.com/wine/base_unittests.tgz tar -xzvf base_unittests.tgz wine notepad & // to avoid having to valgrind service startup valgrind -v --trace-children=yes wine chrome/Debug/base_unittests.exe I turned on the debug flag in m_stacktrace.c, as that seems to print out what you were after. That generated a huge amount of output; here's just the last bit: ... n_ips=12 fp_min=0x7f21eb34 fp_max_orig=0xfe8f6ffc, fp_max=0xfe8f6ffc ip=0x4cb70f2 fp=0x7f21eb7c ipsF[1]=0x06f65cdb ipsF[2]=0x09ea9c4d ipsF[3]=0x09ea9f27 ipsF[4]=0x09ed15c8 ipsF[5]=0x09d009e0 ipsF[6]=0x09d01435 ipsF[7]=0x09d01ebc ipsF[8]=0x09cf0284 ipsF[9]=0x083ba3ca ipsF[10]=0x083ba7f6 ipsF[11]=0x083bac2a n_ips=12 fp_min=0x7f21eaf4 fp_max_orig=0xfe8f6ffc, fp_max=0xfe8f6ffc ip=0x4cb903e fp=0x7f21eb2c ipsF[1]=0x04cb91bf ipsF[2]=0x06f672ce ipsF[3]=0x0732de98 --21633-- VALGRIND INTERNAL ERROR: Valgrind received a signal 11 (SIGSEGV) - exiting --21633-- si_code=2; Faulting address: 0xFBAD8005; sp: 0x4798d3c valgrind: the 'impossible' happened: Killed by fatal signal n_ips=100 fp_min=0x4798d3c fp_max_orig=0x4798ff0, fp_max=0x4798ffc ip=0x3803b138 fp=0x4798d64 ipsF[1]=0x3803b2cf ipsF[2]=0x38025178 ipsF[3]=0x38002e7b ipsF[4]=0x380033d5 ipsF[5]=0x3804f982 ipsF[6]=0x3806c378 ipsF[7]=0xffffffff ==21633== at 0x3803B138: vgPlain_get_StackTrace_wrk (m_stacktrace.c:162) ==21633== by 0x3803B2CF: vgPlain_get_StackTrace (m_stacktrace.c:483) ==21633== by 0x38025178: record_ExeContext_wrk (m_execontext.c:316) ==21633== by 0x38002E7B: vgMemCheck_new_block (mc_malloc_wrappers.c:214) ==21633== by 0x380033D5: vgMemCheck_malloc (mc_malloc_wrappers.c:235) ==21633== by 0x3804F982: vgPlain_scheduler (scheduler.c:1303) ==21633== by 0x3806C378: run_a_thread_NORETURN (syswrap-linux.c:89) I've put the log with and without debug=1, compressed with rzip, up at http://kegel.com/wine/log2.rz and http://kegel.com/wine/log3.rz. - Dan |
|
From: Dan K. <da...@ke...> - 2009-03-22 01:47:46
|
On Sat, Mar 21, 2009 at 6:38 PM, Dan Kegel <da...@ke...> wrote:
> On Wed, Nov 12, 2008 at 12:59 AM, Julian Seward <js...@ac...> wrote:
>> On Wednesday 12 November 2008, Dan Kegel wrote:
>>> With valgrind from svn, I'm getting this error fairly frequently
>>> when running any of Wine's 3D unit tests. It seems to
>>> be triggered by a bug in the platform's nvidia opengl driver.
>>> Would you like more info?
>>
>> Yes, all the parameters to get_StackTrace_wrk that are not
>> marked /*OUT*/.
>
> I'm getting this reliably now running chromium's base_unittests.exe
> under current valgrind trunk.
And, bonus deal, I'm still getting it on the 3d tests. To repeat that scenario,
$ cd wine-git/dlls/d3d9/tests
$ valgrind --trace-children=yes ~/wine32/wine d3d9_test.exe.so volume.c
And here's the tail end of the long output for that case:
...
n_ips=12 fp_min=0x7f2046c0 fp_max_orig=0xfe980ffc, fp_max=0xfe980ffc
ip=0x4cb903e fp=0x7f2046f8
ipsF[1]=0x096e8112
ipsF[2]=0x096e8313
ipsF[3]=0x096ebc8a
ipsF[4]=0x096c60ad
--21819-- VALGRIND INTERNAL ERROR: Valgrind received a signal 11
(SIGSEGV) - exiting
--21819-- si_code=2; Faulting address: 0xC1D0004C; sp: 0x4798d3c
valgrind: the 'impossible' happened:
Killed by fatal signal
n_ips=100 fp_min=0x4798d3c fp_max_orig=0x4798ff0, fp_max=0x4798ffc
ip=0x3803b138 fp=0x4798d64
ipsF[1]=0x3803b2cf
ipsF[2]=0x38025178
ipsF[3]=0x38002e7b
ipsF[4]=0x380033d5
ipsF[5]=0x3804f982
ipsF[6]=0x3806c378
ipsF[7]=0xffffffff
==21819== at 0x3803B138: vgPlain_get_StackTrace_wrk (m_stacktrace.c:162)
==21819== by 0x3803B2CF: vgPlain_get_StackTrace (m_stacktrace.c:483)
==21819== by 0x38025178: record_ExeContext_wrk (m_execontext.c:316)
==21819== by 0x38002E7B: vgMemCheck_new_block (mc_malloc_wrappers.c:214)
==21819== by 0x380033D5: vgMemCheck_malloc (mc_malloc_wrappers.c:235)
==21819== by 0x3804F982: vgPlain_scheduler (scheduler.c:1303)
==21819== by 0x3806C378: run_a_thread_NORETURN (syswrap-linux.c:89)
|