Mail Archives: cygwin-developers/2002/01/09/16:03:36
On Wed, Jan 09, 2002 at 09:58:04PM +0100, Corinna Vinschen wrote:
> On Wed, Jan 09, 2002 at 09:30:20PM +0100, Corinna Vinschen wrote:
> > On Wed, Jan 09, 2002 at 03:00:52PM -0500, Norman Vine wrote:
> > > (gdb) bt full
> > > #0 strncpy (dst0=0xa01d712 "", src0=0x0, count=109)
> > > at /src/cygwin/src/newlib/libc/string/strncpy.c:116
> > > dst0 = 0xa01d712 ""
> > > count = 109
> > > dst = 0xa01d712 ""
> > > src = 0x0
> > > aligned_dst = (long int *) 0xa01d712
> > > aligned_src = (long int *) 0x22f924
> > > #1 0x61033758 in cygwin_getsockname (fd=7, addr=0xa01d710,
> > > namelen=0x22f924)
> >
> > Is postgreSQL using socketpairs, perhaps?
>
> Nope, forget that. I think I've found it. Could you please try it
> again with the latest from CVS? I've just checked in a patch.
To be more enlighting (at least a bit):
I haven't had the postgreSQL sources handy so I first poked a bit
around. Then I downloaded these sources and found that the affected
postgreSQL function StreamConnection() accepts a connection. This
was the hint I missed. My change to return a correct path for AF_LOCAL
sockets in a call to getsockname() requires calling a fhandler_socket
method set_sun_path(). I called it in cygwin_bind() but I forgot to
call it for the newly created socket in a call to cygwin_accept.
I hope that fixes it.
Corinna
--
Corinna Vinschen Please, send mails regarding Cygwin to
Cygwin Developer mailto:cygwin AT cygwin DOT com
Red Hat, Inc.
- Raw text -