On Aug 3 17:39, Corinna Vinschen wrote:
> On Aug 3 17:17, Novaelec wrote:
> > Hello Corinna,
> >
> > I think the problem is in cygserver because the bug is related to semaphores
> > when I use gdb. [...]
>
> Sorry to say that, but to me it looks like a bug in postmaster. Looking
> into the stacktrace printed when the SEGV occurs, I'm getting this:
>
> $ addr2line -e /bin/cygwin1.dll 610D6127 610028FE 610173A9 61017A6E 61017D1F
> 6109E76C 610942FF
> /home/corinna/src/cygwin/vanilla/newlib/libc/machine/i386/memset.S:45
> /home/corinna/src/cygwin/vanilla/winsup/cygwin/cygheap.cc:331
> /home/corinna/src/cygwin/vanilla/winsup/cygwin/dtable.cc:375
> /home/corinna/src/cygwin/vanilla/winsup/cygwin/dtable.cc:486
> /home/corinna/src/cygwin/vanilla/winsup/cygwin/dtable.cc:532
> /home/corinna/src/cygwin/vanilla/winsup/cygwin/syscalls.cc:127
>
> This is the stacktrace of a dup() call on the tty slave descriptor. It
> looks like a heap corruption to me. It has certainly nothing to do with
> semaphores.
FYI, I tracked the problem down to a point that I can savely say,
it's neither a bug in postgresql, nor in cygserver. It has nothing
to do with semaphores. As noted above, it's a heap problem, but
it's entirely inside Cygwin. Unfortunately I have no fix so far.
For the curious, it's reproducible with
int fd;
for (;;)
fd = dup (0);
Corinna
--
Corinna Vinschen Please, send mails regarding Cygwin to
Cygwin Project Co-Leader mailto:cygwin@cygwin.com
Red Hat, Inc.
--
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Problem reports: http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ: http://cygwin.com/faq/
