Re: gcore and core dumps...

[Alfred M. Szmidt]

   > marco@c366:/tmp$ ./crash 
   > Segmentation fault (core dumped)
   > marco@c366:/tmp$ showtrans /servers/crash
   > /hurd/symlink crash-suspend
   > marco@c366:/tmp$ showtrans /servers/crash-suspend 
   > /hurd/crash --suspend

   Strange.  
   Well, attach gdb to the crash server and see what it thinks it's doing.

It dumps core because c->proc is MACH_PORT_NULL.  And then when we
call dead_crasher(), we will dump core, since c->proc is
MACH_PORT_NULL.  No idea why proc_task2proc gives back user_proc ==
MACH_PORT_NULL though, will look at that when I install debuggable
libraries.

,----
| Breakpoint 2, S_crash_dump_task (port=32, reply_port=32, reply_type=32, 
task=0, core_file=33, 
|     signo=11, sigcode=0, sigerror=2, exc=1, code=2, subcode=0, ctty_id=34)
|     at ../../hurd/trans/crash.c:156
| [...snip...]
| (gdb) n
| 211                   c->proc = user_proc;
| (gdb) 
| 223                   err = MIG_NO_REPLY;
| (gdb) p c->proc
| $6 = 0
| [...snip...]
| Breakpoint 1, dead_crasher (ptr=0x804e8c0) at ../../hurd/trans/crash.c:384
| 384       if (c->proc != MACH_PORT_NULL)
| (gdb) 
| 395           error_t err = dump_core (c->task, c->core_file, c->core_limit,
| (gdb) p c->proc
| $7 = 0
| (gdb) 
`----

Cheers.