Path: utzoo!attcan!uunet!cs.utexas.edu!romp!auschs!d75!dundee.austin.ibm.com!au0005
From: au0005@dundee.austin.ibm.com
Newsgroups: comp.sys.ibm.pc.rt
Subject: Re: Need help with kernel message from AIX 2.2.1 !!!
Keywords: AIX, RT, impending doom
Message-ID: <3283@d75.UUCP>
Date: 2 Aug 90 17:01:11 GMT
References: <212@gort.cs.utexas.edu>
Sender: news@d75.UUCP
Reply-To: au0005@dundee.austin.ibm.com ()
Organization: IBM Austin, TX
Lines: 49


In article <212@gort.cs.utexas.edu>, jason@cs.utexas.edu (Jason Martin
Levitt) writes:
> From: jason@cs.utexas.edu (Jason Martin Levitt)
> Subject: Need help with kernel message from AIX 2.2.1 !!!
> Date: 16 Jul 90 09:44:15 GMT
> 
> 
>     I keep getting this message from the AIX 2.2.1 kernel:
> 
>   Core failure: 255
> 
>     No one responded to my posting a couple of weeks ago about
> this message. I'm really worried that my machine may fail abruptly
> since I seem to be getting the message more frequently now. 
>     Can someone at IBM *please* grep through the AIX 2.2.1 kernel
> source code and find this string? 
> 
>         --Jason     jason@cs.utexas.edu   (512) 326-9102

Hello Jason,

Yes, this message does come from core.c. The value printed comes from
u.u_error, after a failure to dump a core file for some specific reason.

All the calls that core.c makes should return a proper error code,
like your 28, which is ENOSPC.

However, one of the first things that core.c does is grab a bunch of
space from the stack, so that he can parse a filename in there for the
core file.

He does this using the copyout() function, which only returns -1 on
failure. Since the type of u.u_error == char ( which is unsigned ),
this shows as a Core Failure: 255.

Basically, if you get a 255 for an error code, you have a problem with
the stack OR this filename. i.e, no stack space available, the filename
causes the stack to grow beyond limits etc. 

Hope that this helps.

Regards, 

Peter May.


#include <standard.disclaimer>
/*  My Comments are my own: I do not represent IBM. */