From owner-freebsd-hackers  Mon Oct  7 07:10:18 1996
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id HAA28768
          for hackers-outgoing; Mon, 7 Oct 1996 07:10:18 -0700 (PDT)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id HAA28763
          for <hackers@freebsd.org>; Mon, 7 Oct 1996 07:10:13 -0700 (PDT)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.7.6/8.6.9) id AAA31883; Tue, 8 Oct 1996 00:03:55 +1000
Date: Tue, 8 Oct 1996 00:03:55 +1000
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199610071403.AAA31883@godzilla.zeta.org.au>
To: garyj@frt.dec.com, hackers@freebsd.org
Subject: Re: Another question: kernel crash dumps 8)
Sender: owner-hackers@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

>> >integrate the fix into the existing gdb 
>> >
>> >> (Hint: ``frame frame->tf_ebp frame->tf_eip'')
>
>in this case I think Amancio's right. Should be fairly trivial to
>incorporate the changes. The kvm stuff in the new gdb isn't that
>different from what's in the old version.

Erm, I thought we were talking about tracing trap frames in gdb.
You can sort of do it by hand using the above hint.  The above usage of
the frame statement depends on an improvement to gdb written by Joerg.
I hope it is in the new gdb port.

gdb never supported tracing trap frames directly, and the above method
loses some information.  For a complete trace it would need to know
about magic addresses inside the kernel.  ddb knows about "calltrap",
"Xresume", and "_Xsyscall".  These may change whenever someone
improves the trap handling, so they are hard to support.  In fact,
"calltrap" has already gone away in my version of exception.s (but
not in db_trace.c :-().

Bruce