more info (was:Re: glibc 2.3 broken

[Marcus Brinkmann]

On Sun, Oct 13, 2002 at 06:06:14PM +0200, Marcus Brinkmann wrote:
> (to Jeff) You are right, glibc 2.3 is quite broken for us.
> 
> I don't even come to a login prompt.  postinst segfaults, runsystem
> segfaults, and the system soon crashes.

ok, I found out a bit more.  I could reproduce this in my build tree with
the following:

marcus@ajax:~/libc/cvs/debian/glibc-2.3/i386-gnu/obj$ 
LD_LIBRARY_PATH=.:hurd:mach:nis:login:crypt:dlfcn:elf:hesiod:locale:malloc:math:nss:resolv:rt
 elf/ld.so /bin/sh
sh-2.05b$ 

After that, everything is still normal.  Then entering any command (or
anything at all) and pressing return makes sh go wild:

sh-2.05b$ hello<return>

marcus@ajax:~/libc/cvs/debian/glibc-2.3/i386-gnu/obj$ ps aux 224
USER       PID %CPU %MEM    SZ   RSS TT STAT   START     TIME COMMAND
marcus     224 86.4  0.2  147M 1.81M p0 R     9:05PM  0:06.60 /bin/sh

It seemed to me that portinfo only reported one additional port.

After trying this a couple of time I got a "panic: zalloc: zone ipc ports 
exhausted".

Here is a backtrace from the glibc part of the sh process while it has a lot
of CPU usage (the other parts where from sh without debugging symbols and
obviously bogus):

(gdb) bt full
#0  0x010ecd79 in ptmalloc_lock_all () at
../sysdeps/mach/i386/machine-lock.h:52
        ar_ptr = (struct malloc_state *) 0x0
#1  0x01123b69 in __fork () at ../sysdeps/mach/hurd/fork.c:101
        env = {{__jmpbuf = {18602200, 18837456, 0, 16873976, 16873472, 
17971976},
    __mask_was_saved = 0, __saved_mask = 0}}
        pid = 93340
        i = 1034514604
        err = 91052
        ss = (struct hurd_sigstate * volatile) 0x11f8808

Hope this helps,
Marcus

-- 
`Rhubarb is no Egyptian god.' GNU      http://www.gnu.org    marcus@gnu.org
Marcus Brinkmann              The Hurd http://www.gnu.org/software/hurd/
Marcus.Brinkmann@ruhr-uni-bochum.de
http://www.marcus-brinkmann.de/