Hi!

First, let me state that I am just trying a shot into the dark :-)

On Mon, Jul 10, 2000 at 06:20:20PM +0200, Kallweit, Heiner wrote:
> > From: "Kallweit, Heiner" <[EMAIL PROTECTED]>
> > 
> > Heiner.Kallweit> we use OpenSSL 0.9.5a together with Libwww in a CGI
> > Heiner.Kallweit> on a HP-UX 11.0 platform.
> > Heiner.Kallweit> About every 100th request I get a coredump. Strange
> > Heiner.Kallweit> enough, the coredump happens in different functions
> > Heiner.Kallweit> in libcrypto.sl. When I do a backtrace with gdb I can
> > Heiner.Kallweit> see that the crash always happens at the beginning of
> > Heiner.Kallweit> the handshake. But the origin of the backtrace is at
> > Heiner.Kallweit> slightly different places in the handshake_func.
> > 
> > The origin perhaps, but do all the crashes end in CRYPTO_thread_id()?
> > Could you tell us what functions it ends up happening in (the one in
> > frame #0)?
> No, it crashes in different functions, but always inside the handshake_func.
> I include another example:
> 
> #0  0xc0205e34 in __errno () from /usr/lib/libc.2
> #1  0xc1dc62f4 in sock_read (b=0x4002bac8, out=0x40031070
> "\026\003\001\003\345\002", 
>     outl=5) at bss_sock.c:191

By checking out bss_sock.c the "clear_socket_error()" macro is being used,
which effectively does a (see include/openssl/e_os.h) "errno=0".

On HP-UX (10.20) errno is an "extern int errno", as defined in <errno.h>.

The question now is, why a function "__errno()" is triggered at that point??

What is the reason for the coredump? Is it a segmentation fault or a
bus error? "file core" will tell you.

Core dumps with different backtraces might be caused by a corrupted
malloc structure (overwriting at the begin or the end of an malloc()ed
array). Have you ever tried to run you program with Efence enabled?
The point of failure can be very far from the actual bug in the program.

Best regards,
        Lutz
-- 
Lutz Jaenicke                             [EMAIL PROTECTED]
BTU Cottbus               http://www.aet.TU-Cottbus.DE/personen/jaenicke/
Lehrstuhl Allgemeine Elektrotechnik                  Tel. +49 355 69-4129
Universitaetsplatz 3-4, D-03044 Cottbus              Fax. +49 355 69-4153
______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
User Support Mailing List                    [EMAIL PROTECTED]
Automated List Manager                           [EMAIL PROTECTED]

Reply via email to