Subject: Re: i386 MP panic: amap_pp_adjref: negative reference count
To: Sami Kantoluoto <>
From: Andrew Brown <>
List: current-users
Date: 12/16/2002 09:50:29
>> what does the traceback in ddb look like?  do you get a crash dump?
>Traceback attached, and no crash dump.

how bizarre.  nothing like the code path that led to my panic, except
at the end where it panics.

can you make a crash dump, or does the machine have trouble doing that?

>> can you build a kernel with -g so that gdb can do full port-mortem
>> debugging?
>Anything to do in ddb that can help? (I can do that immediately because I
>haven't rebooted the machine yet)

"ps/a" in ddb will show stuff like this (this is from an alpha, so the
pointers are wider):

   db> ps/a
    PID     COMMAND      STRUCT PROC *            UAREA *     VMSPACE/VM_MAP
    296       getty 0xfffffc000298f1d0 0xfffffe0016b1c000 0xfffffc0002998120
    270        cron 0xfffffc000362dac0 0xfffffe0016e56000 0xfffffc0002998fc0
    265       inetd 0xfffffc000ad72ee8 0xfffffe0016e6a000 0xfffffc0002999680

from your ps output:

   db{0}> ps
    PID        PPID       PGRP        UID S   FLAGS          COMMAND    WAIT
   >22798     22758      11496          0 7 0x180006             idlc
    22758     22757      11496          0 3 0x84086             idlc  select
    22757     22686      11496          0 3 0x84086             tcsh   pause

process 22798 is the running process.  the traceback shows it's in the
middle of execve().  if you can get the vm_map pointer for that
process and do a "show map/f nnnnn" on it, that would be great.

i fear, however, that the process has already had its vmspace
eliminated, and it will be empty.  let's see, though.

|-----< "CODE WARRIOR" >-----|             * "ah!  i see you have the internet (Andrew Brown)                that goes *ping*!"       * "information is power -- share the wealth."