NetBSD Problem Report #4332

Received: (qmail 28684 invoked from network); 24 Oct 1997 12:39:53 -0000
Message-Id: <199710241239.OAA12962@macone.cs.uni-bonn.de>
Date: Fri, 24 Oct 1997 14:39:20 +0200 (CEST)
From: ignatios@macone.cs.uni-bonn.de
Reply-To: is@netbsd.org
To: gnats-bugs@gnats.netbsd.org
Subject: PR 3123 test program makes 68060 panic if compiled with old gcc
X-Send-Pr-Version: 3.95

>Number:         4332
>Category:       port-amiga
>Synopsis:       PR3123 test program makes 68060 panic if compiled with old gcc
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    is
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Fri Oct 24 05:50:02 +0000 1997
>Closed-Date:    
>Last-Modified:  Thu Jan 17 20:23:57 +0000 2002
>Originator:     Ignatios Souvatzis
>Release:        19971020
>Organization:
	University of Bonn, CS department, Chair V
>Environment:

NetBSD/Amiga on 68060
>Description:
If you run the test program from PR 3123, compiled with the faulty gcc, 
it will not get a SIGBUS, but panic the machine.

The FSLW reported by the panic says "invalid root level pointer", which is 
very fishy, as the virtual address in error is in the user stack area (and in
its first page, which should be there as soon as the process has started to
run).

>How-To-Repeat:
	cc -o trampoline trampoline.c
	./trampoline
>Fix:
	Workaround: fix gcc to produce correct code.
	Of course, the trap handler needs to be fixed.
>Release-Note:
>Audit-Trail:
Responsible-Changed-From-To: gnats-admin->is 
Responsible-Changed-By: is 
Responsible-Changed-When: Fri Oct 24 06:26:39 PDT 1997 
Responsible-Changed-Why:  
I'm looking into this.  
>Unformatted:

Home	PR Database Search