crash dump generated via @CRASH

[Chris Torek]

In article <319 at brl-smoke.ARPA> narten at purdue.edu (Thomas Narten) writes:
>Can anyone tell me how to get adb to perform a (meaningful) traceback
>out of a crash dump [...]?  It appears that all @CRASH does is
>dump -1 into the PC and PSL and then resume UNIX, [...]

That is correct.  This also crashes VMS, which is why DEC supplies
the command.  (Actually, CRASH.COM also prints out all the stack
pointers.)

>The normal traceback using *(scb-4)$c [...] doesn't do the trick.

It works for me:

	% cd /usr/crash
	% adb -k vm{unix,core}.221
	sbr 6868 slr 5d70
	p0br 9600 p0lr 4c6 p1br	aa00 p1lr ffe4
	panic: Segmentation fault
	*(scb-4)$c
	_boot()	from _boot+fb
	_boot(0,0) from	_panic+3a
	_panic(80052930) from _trap+ac
	_trap()	from _Xtransflt+1d
	_Xtransflt(808d8d0c,0) from _sonewconn+93
	_sonewconn(808d5e0c) from 80007b1c
	_tcp_input(808d8e00,8006ac00) from 80005f8d
	[...]

(Incidentally, this is an old crash dump, and I even know what the
bug is, but do not yet have a clean fix.  But these are rare:  two
or three times a year on a heavily loaded machine.)
-- 
In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 1415)
UUCP:	seismo!umcp-cs!chris
CSNet:	chris at umcp-cs		ARPA:	chris at mimsy.umd.edu