[Xitami] xitami aborts

Ewen McNeill xitami@lists.xitami.org
Mon, 03 Sep 2001 10:22:22 +1200


In message <EX-20010828-144824-0004@exchange.fiserv-missive1.fiserv.net>, "Bowde
n, Russell" writes:
>>>[Xitami "just disappears" after a while]
>I installed 2.4d9 and still have the same problem.  The stdout file contains on
>ly the normal startup messages.  There's nothing in the stderr file except the 
>message I echo there when starting Xitami.  The Console and xitami logs do not 
>show anything abnormal.  

Hmm.  Normally when Xitami crashes out the error handler tries to print
out some information about which module it was in, and so on, which is
occassionally useful.

>Is there any information at all that I could get from these cores that might 
>point to the problem?  Since I haven't seen any other reports of this 
>problem, maybe it's peculiar to UnixWare.

The most useful thing to get out of core dump files -- assuming you've
left the symbols attached to the executable -- is a trace of the
functions on the stack at the time that it crashed.  That tells us both
a possible area where it is having problems, and also over the course of
a few crashes whether it's the same thing or a "random" thing causing
the problem.  ("random" problems are much harder to find, being
generally memory overwrite issues.)

>From memory under Unixware you'll need to use dbx to get a stack trace,
unless you've installed (and used) the GNU compiler suite, in which case
gdb will probably work.  (The relevant gdb command is "bt"; I don't
remember off hand what the command is in dbx.)

Ewen

-- 
Ewen McNeill, Technical Consultant, iMatix Corporation  www.imatix.com