Segfaults in RtlTryEnterCriticalSection ?

[Sylvain Petreolle]

What information could be usefull to debug a segfault in
RtlTryEnterCriticalSection ?
Since March beginning I have segfaults into programs that used to run
before.

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 8192 (LWP 8925)]
0x400ba904 in RtlTryEnterCriticalSection (crit=0x40240570) at
../../include/winbase.h:1918
1918        __asm__ __volatile__( ".byte 0x64\n\tmovl 0x24,%0" : "=r"
(ret) );
(gdb) bt
#0  0x400ba904 in RtlTryEnterCriticalSection (crit=0x40240570) at
../../include/winbase.h:1918
#1  0x4009355c in TryEnterCriticalSection (crit=0x40240570) at
../../memory/heap.c:814
#2  0x400b15f3 in __pthread_mutex_trylock (mutex=0x4212b180) at
../../scheduler/pthread.c:320
#3  0x42073999 in calloc () from /lib/i686/libc.so.6
#4  0x410c69e7 in s8790 () from /usr/X11R6/lib/modules/dri/fglrx_dri.so

(gdb) list -
1908    {
1909        DWORD ret;
1910        __asm__ __volatile__( ".byte 0x64\n\tmovl 0x20,%0" : "=r"
(ret) );
1911        return ret;
1912    }
1913
1914    extern inline DWORD WINAPI GetCurrentThreadId(void);
1915    extern inline DWORD WINAPI GetCurrentThreadId(void)
1916    {
1917        DWORD ret;
gdb) disass $pc-32 $pc+32
Dump of assembler code from 0x400ba8e4 to 0x400ba924:
0x400ba8e4 <RtlTryEnterCriticalSection+32>:     or     $0x74,%al
0x400ba8e6 <RtlTryEnterCriticalSection+34>:     or    
%ecx,0x24348bc8(%ecx)
0x400ba8ec <RtlTryEnterCriticalSection+40>:     leave
0x400ba8ed <RtlTryEnterCriticalSection+41>:     ret    $0x4
0x400ba8f0 <RtlTryEnterCriticalSection+44>:     mov    $0x1,%eax
0x400ba8f5 <RtlTryEnterCriticalSection+49>:     lock xadd %eax,(%edx)
0x400ba8f9 <RtlTryEnterCriticalSection+53>:     incl   0x8(%esi)
0x400ba8fc <RtlTryEnterCriticalSection+56>:     mov    $0x1,%ecx
0x400ba901 <RtlTryEnterCriticalSection+61>:     jmp    0x400ba8e7
<RtlTryEnterCriticalSection+35>
0x400ba903 <RtlTryEnterCriticalSection+63>:     nop
0x400ba904 <RtlTryEnterCriticalSection+64>:     mov    %fs:0x24,%eax
0x400ba90a <RtlTryEnterCriticalSection+70>:     mov    %eax,0xc(%esi)
0x400ba90d <RtlTryEnterCriticalSection+73>:     movl   $0x1,0x8(%esi)
0x400ba914 <RtlTryEnterCriticalSection+80>:     jmp    0x400ba8fc
<RtlTryEnterCriticalSection+56>
0x400ba916 <RtlTryEnterCriticalSection+82>:     mov    %esi,%esi
0x400ba918 <RtlLeaveCriticalSection>:   push   %ebp
0x400ba919 <RtlLeaveCriticalSection+1>: mov    %esp,%ebp
0x400ba91b <RtlLeaveCriticalSection+3>: mov    0x8(%ebp),%ecx
0x400ba91e <RtlLeaveCriticalSection+6>: push   %ebx
0x400ba91f <RtlLeaveCriticalSection+7>: mov    0x8(%ecx),%eax
0x400ba922 <RtlLeaveCriticalSection+10>:        dec    %eax
0x400ba923 <RtlLeaveCriticalSection+11>:        call   0x400ba928
<RtlLeaveCriticalSection+16>
End of assembler dump.

=====
Sylvain Petreolle
[e-mail] 
Fight Spam ! EuroCauce: http://www.euro.cauce.org/en/index.html
ICQ #170597259

"Don't think you are. Know you are." Morpheus, in "Matrix".

___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en français !
Yahoo! Mail : http://fr.mail.yahoo.com

[Eric Pouech]

Sylvain Petreolle wrote:
> What information could be usefull to debug a segfault in
> RtlTryEnterCriticalSection ?
> Since March beginning I have segfaults into programs that used to run
> before.

could you check whether %fs is still a valid value ? (ie the one the 
thread got upon startup)
A+

[Sylvain Petreolle]

After doing tests, it seems fs is always set as
fs             0xffff   65535
when it crashes. of course its a wrong/unitialized value ?

--- Eric Pouech  a écrit : > Sylvain Petreolle
wrote:
> > What information could be usefull to debug a segfault in
> > RtlTryEnterCriticalSection ?
> > Since March beginning I have segfaults into programs that used to
>
> run
> > before.
>
> could you check whether %fs is still a valid value ? (ie the one the 
> thread got upon startup)
> A+
>
> -- 
> Eric Pouech

=====
Sylvain Petreolle
[e-mail] 
Fight Spam ! EuroCauce: http://www.euro.cauce.org/en/index.html
ICQ #170597259

"Don't think you are. Know you are." Morpheus, in "Matrix".

___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en français !
Yahoo! Mail : http://fr.mail.yahoo.com

[Ove Kaaven]

fre, 2003-03-14 kl. 16:37 skrev Sylvain Petreolle:
> After doing tests, it seems fs is always set as
> fs             0xffff   65535
> when it crashes. of course its a wrong/unitialized value ?

No, it's an LDT entry used by the ATI FireGL drivers (see the backtrace
in your original post). TransGaming has already tried to notify ATI
about the problem and get them to not do this in their next driver
release.