On Fri, 2001-12-07 at 10:34, Justin Erenkrantz wrote:
> On Fri, Dec 07, 2001 at 10:30:29AM -0800, Ian Holsman wrote:
> > hey Justin
> > I mailed something about this a couple of days ago.
> > The patch stopped some of the core dumps, but aaron & I
> > found other points.
> > I never patched it as I wasn't confident that I wasn't breaking other
> > things along the way.
>
> Yeah, I saw that. I just care that we track it down before 2.0.31.
> I think the best way to do it is add it to STATUS. =)
>
> As a note, I don't always get a coredump though. It'll just hang
> and not respond. I wonder if two different things are going on.
>
> Anyway, the coredumps I do get are like:
>
> #0 0xdfb4566d in apr_sockaddr_port_get (port=0x8047af6, sockaddr=0x0)
> at ../unix/sa_common.c:132
> 132 *port = ntohs(sockaddr->sa.sin.sin_port);
> (gdb) bt
> #0 0xdfb4566d in apr_sockaddr_port_get (port=0x8047af6, sockaddr=0x0)
> at ../unix/sa_common.c:132
> #1 0x80ca5eb in alloc_listener (process=0x8116f5c, addr=0x80f8303 "::",
> port=80) at listen.c:240
> #2 0x80caa16 in ap_set_listener (cmd=0x8047c2c, dummy=0x0,
> ips=0x816847c "80")
> at listen.c:363
> #3 0x80bd9db in invoke_cmd (cmd=0x810ede4, parms=0x8047c2c,
> mconfig=0x0,
> args=0x8160b06 "") at config.c:660
> #4 0x80be92d in ap_walk_config_sub (current=0x8160ae4, parms=0x8047c2c,
> section_vector=0x815d8d4) at config.c:1006
> #5 0x80be9dd in ap_walk_config (current=0x8160ae4, parms=0x8047c2c,
> section_vector=0x815d8d4) at config.c:1043
> #6 0x80bf67c in ap_process_config_tree (s=0x811c534,
> conftree=0x81602f4,
> p=0x8118efc, ptemp=0x815b214) at config.c:1457
> #7 0x80c255c in main (argc=1, argv=0x8047d04) at main.c:429
>
yeah.. that was the first one.
the patch was to not call set_listener when we have a NULL socket.
it stopped this problem, and some of the restarts, but we found another
blocker
I can see if I can reproduce
> Is this what you see? -- justin
