From your stack trace, I first noticed msvcrt.dll!free and that led me
to believe it was a problem of that kind. malloc() and free() do
different memory checks in debug (msvcrtXXd.dll) and release
(msvcrtXX.dll) versions. An attempt to mix both libraries in the same
executable might cause an access violation.
After a closer look in your stack trace, I noticed that your
program stop is caused by a SIGTRAP (my original hypothesis would be
confirmed by a SIGSEGV). This means one checking inside free() has
failed. Could be a double free() or a pointer that hasn't been returned
directly by malloc() or the actual pointer might have been overwritten
due to an overflow.
I believe the next step is to place a breakpoint at
E:\libssh-0.5.1\src\dh.c:179 (/break dh.c:179/) and then step (/n/) the
execution.
Regards,
Paulo
On 11/08/11 10:24, voker57 wrote:
On Wed, Aug 10, 2011 at 07:42:12PM -0300, Paulo Panhoto wrote:
This seems to be a case where debug and release builds are mixed. More
likely Release DLL with Debug executable.
Can you please elaborate on that? The only executable I'm using is test program
I compile, OpenSSL is 1.0.0d pre-compiled version from
http://www.slproweb.com/products/Win32OpenSSL.html, libssh is built with debug.
