I am puzzled as well, looking at the backtrace. It mentions something
about waiting for a critical section, so that indeed points to the
Mutex. Wonder if the crash gets fixed after the Mutex is removed?
Teus.
On Fri, 2009-01-23 at 21:07 -0800, Dan Dennison wrote:
I'm trying to debug a crash inside libxml2 on Windows that happens
inside of config_xml_vector_string_get().
I notice this comment in the code:
// A static mutex was introduced to try to solve the problem of
rogue
// resource creation, but it appeared that it didn't solve the
problem.
// Yet the mutex was left, because the configurations are
accessed in
// a thread, so it is always good to have a mutex in place.
Immediately after that a mutex called
config_xml_vector_string_get_mutex gets initialized.
Shortly thereafter:
inputbuffer = xmlParserInputBufferCreateMem(contents, strlen
(contents), XML_CHAR_ENCODING_NONE);
gets called, and BE segfaults with the attached stack frame.
Thoughts? This only happens when I push either OK or Cancel on a
project note.
dan :)
segfault and backtrace:
Program received signal SIGSEGV, Segmentation fault.
0x7c91b1fa in ntdll!RtlpWaitForCriticalSection ()
from C:\WINDOWS\system32\ntdll.dll
(gdb) bt
#0 0x7c91b1fa in ntdll!RtlpWaitForCriticalSection ()
from C:\WINDOWS\system32\ntdll.dll
#1 0x7c901046 in ntdll!RtlEnumerateGenericTableLikeADirectory ()
from C:\WINDOWS\system32\ntdll.dll
#2 0x000e4ad8 in ?? ()
#3 0x1004ab98 in xmlGetGlobalState () from C:\beta-debug\bin
\libxml2.dll
#4 0x100e4ad8 in ?? ()
#5 0x00000000 in ?? ()