NetBSD Problem Report #27099

Received: (qmail 27226 invoked by uid 605); 1 Oct 2004 07:10:14 -0000
Message-Id: <200410010710.i917ACas016378@lain.ziaspace.com>
Date: Fri, 1 Oct 2004 07:10:12 GMT
From: jklos@netbsd.org
Sender: gnats-bugs-owner@NetBSD.org
Reply-To: jklos@netbsd.org
To: gnats-bugs@gnats.NetBSD.org
Subject: m68k can be paniced by pkgsrc/sysutils/crashme
X-Send-Pr-Version: 3.95

>Number:         27099
>Category:       port-m68k
>Synopsis:       If one runs crashme on a 2.0 m68k machine, the system panics.
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    port-m68k-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Fri Oct 01 07:11:00 +0000 2004
>Closed-Date:    
>Last-Modified:  Sun Apr 13 00:45:00 +0000 2008
>Originator:     John Klos
>Release:        NetBSD 2.0_BETA
>Organization:
ZiaSpace Productions

>Environment:


System: NetBSD lain.ziaspace.com 2.0_BETA NetBSD 2.0_BETA (LAIN-$Revision: 1.999 $) #0: Fri Aug 27 04:36:32 UTC 2004 john@lain.ziaspace.com:/usr/src/sys/arch/macppc/compile/LAIN macppc
Architecture: m68k
Machine: amiga, mac68k
>Description:

Run pkgsrc/sysutils/crashme on a 2.0 m68k system. It crashes the system.
>How-To-Repeat:

See description.
>Fix:

Not sure yet.
>Release-Note:
>Audit-Trail:
State-Changed-From-To: open->feedback 
State-Changed-By: is 
State-Changed-When: Fri Oct 1 14:10:43 UTC 2004 
State-Changed-Why:  
You should provide more information: 
1. dmesg of the machines affected, ad/or kernel configuration files 
2. kernel traceback. Maybe produce a kernel core dump and keep it around 
for further investigation. 

Regards 
-is 

From: John Klos <john@ziaspace.com>
To: gnats-bugs@gnats.netbsd.org
Cc: is@netbsd.org, port-m68k-maintainer@netbsd.org
Subject: Re: port-m68k/27099
Date: Fri, 1 Oct 2004 17:15:06 +0000 (UTC)

 > Synopsis: If one runs crashme on a 2.0 m68k machine, the system crashes.
 >
 > State-Changed-From-To: open->feedback
 > State-Changed-By: is
 > State-Changed-When: Fri Oct 1 14:10:43 UTC 2004
 > State-Changed-Why:
 > You should proved more information:
 > 1. dmesg of the machines affected, ad/or kernel configuration files
 > 2. kernel traceback. Maybe produce a kernel core dump and keep it around
 >   for further investigation.

 Sorry. Even though I hardly have any time whatsoever right this moment, I 
 wanted to get the PR in sooner rather than later.

 The systems I've tried are two m68040 Macintosh Quadras, an m68060 Amiga 
 1200, and an m68060 Amiga 4000. The systems running 1.6 do not crash, but 
 the systems running 2.0 do.

 GENERIC kernels from 30-August for both mac68k and amiga crash, as well as 
 two custom kernels I've tried. I'm trying the latest snapshots from 
 releng, but I don't think anything's changed.

 This weekend I will try to get a monitor and keyboard onto one of the 
 systems and see if I can get into the debugger.

 John
From: Izumi Tsutsui <tsutsui@ceres.dti.ne.jp>
To: gnats-bugs@NetBSD.org
Cc: tsutsui@ceres.dti.ne.jp
Subject: Re: port-m68k/27099 (If one runs crashme on a 2.0 m68k machine, the
	 system panics.)
Date: Tue, 9 May 2006 00:16:15 +0900

 Did it cause a panic, or just hangup on your machines?
 i.e. is this the same problem with port-hp300/4265 and
 http://mail-index.netbsd.org/port-m68k/2006/03/27/0000.html
 ?
 ---
 Izumi Tsutsui

State-Changed-From-To: feedback->open
State-Changed-By: dholland@NetBSD.org
State-Changed-When: Thu, 10 Apr 2008 01:26:42 +0000
State-Changed-Why:
Feedback timeout. Anyone have an m68k to test this?


From: Martin Husemann <martin@duskware.de>
To: gnats-bugs@NetBSD.org
Cc: dholland@NetBSD.org, jklos@netbsd.org
Subject: Re: port-m68k/27099 (If one runs crashme on a 2.0 m68k machine, the system panics.)
Date: Thu, 10 Apr 2008 12:39:49 +0200

 I just tried it on a 040 mac68k running a -current kernel: after ~30 minutes
 the machine hangs dead, can't break into ddb.

 Martin

From: Izumi Tsutsui <tsutsui@ceres.dti.ne.jp>
To: gnats-bugs@NetBSD.org
Cc: port-m68k-maintainer@NetBSD.org, netbsd-bugs@NetBSD.org,
        gnats-admin@NetBSD.org, dholland@NetBSD.org, jklos@NetBSD.org,
        tsutsui@ceres.dti.ne.jp
Subject: Re: port-m68k/27099 (If one runs crashme on a 2.0 m68k machine, the
	 system panics.)
Date: Thu, 10 Apr 2008 21:42:56 +0900

 I think this is dup of port-hp300/4265 (analyzed but no clue yet).
 ---
 Izumi Tsutsui

From: David Holland <dholland-bugs@netbsd.org>
To: gnats-bugs@NetBSD.org
Cc: port-m68k-maintainer@netbsd.org, gnats-admin@netbsd.org,
	netbsd-bugs@netbsd.org, jklos@netbsd.org
Subject: Re: port-m68k/27099 (If one runs crashme on a 2.0 m68k machine, the system panics.)
Date: Thu, 10 Apr 2008 15:24:34 +0000

 On Thu, Apr 10, 2008 at 10:40:01AM +0000, Martin Husemann wrote:
  > I just tried it on a 040 mac68k running a -current kernel: after ~30 minutes
  > the machine hangs dead, can't break into ddb.

 Does tsutsui's test case (the one in
 http://mail-index.netbsd.org/port-m68k/2006/03/27/0000.html) trigger
 it? If so, that's most of the work, and someone just needs to load
 trap.c up with printfs to figure out what's happening...

 (Unfortunately, I don't have a 68040, so I can't do much besides kibitz.)

 -- 
 David A. Holland
 dholland@netbsd.org

From: Izumi Tsutsui <tsutsui@ceres.dti.ne.jp>
To: dholland-bugs@NetBSD.org
Cc: gnats-bugs@NetBSD.org, port-m68k-maintainer@NetBSD.org,
        gnats-admin@NetBSD.org, netbsd-bugs@NetBSD.org, jklos@NetBSD.org,
        tsutsui@ceres.dti.ne.jp
Subject: Re: port-m68k/27099 (If one runs crashme on a 2.0 m68k machine, the
	 system panics.)
Date: Fri, 11 Apr 2008 00:38:27 +0900

 > If so, that's most of the work, and someone just needs to load
 > trap.c up with printfs to figure out what's happening...

 IIRC printf debug didn't help because kernel completely seemed
 to hang on trap. The instruction cache also looked affected
 (adding m68k_sync_icache(2) around the bomb() changed some behavor),
 but I gave up at that point.
 ---
 Izumi Tsutsui

From: Martin Husemann <martin@duskware.de>
To: Izumi Tsutsui <tsutsui@ceres.dti.ne.jp>
Cc: gnats-bugs@NetBSD.org, dholland@NetBSD.org, jklos@NetBSD.org
Subject: Re: port-m68k/27099 (If one runs crashme on a 2.0 m68k machine, the system panics.)
Date: Thu, 10 Apr 2008 18:29:11 +0200

 On Thu, Apr 10, 2008 at 09:42:56PM +0900, Izumi Tsutsui wrote:
 > I think this is dup of port-hp300/4265 (analyzed but no clue yet).

 At least your test case works fine on my machine:

 address = 0x4101400
 Illegal instruction(core dumped)

 Martin

From: Izumi Tsutsui <tsutsui@ceres.dti.ne.jp>
To: martin@duskware.de
Cc: gnats-bugs@NetBSD.org, dholland@NetBSD.org, jklos@NetBSD.org,
        tsutsui@ceres.dti.ne.jp
Subject: Re: port-m68k/27099 (If one runs crashme on a 2.0 m68k machine, the
	 system panics.)
Date: Fri, 11 Apr 2008 01:57:06 +0900

 > At least your test case works fine on my machine:

 Hmm, it still hangs on my HP9000/382.
 I heard a rumor that some of 040 had a bug on FP trap
 (and it also caused FPSP and LC040 FPU_EMULATE problems),
 but I haven't confirmed it.

 Could you also try to add 
 	m68k_sync_icache(bomb, 4096);
 just before the (*bomb)() call?
 (with <sys/typesh>, <m68k/sync_icache.h> and -lm68k)
 ---
 Izumi Tsutsui

From: Martin Husemann <martin@duskware.de>
To: Izumi Tsutsui <tsutsui@ceres.dti.ne.jp>
Cc: gnats-bugs@NetBSD.org, dholland@NetBSD.org, jklos@NetBSD.org
Subject: Re: port-m68k/27099 (If one runs crashme on a 2.0 m68k machine, the system panics.)
Date: Thu, 10 Apr 2008 20:02:02 +0200

 On Fri, Apr 11, 2008 at 01:57:06AM +0900, Izumi Tsutsui wrote:
 > Could you also try to add 
 > 	m68k_sync_icache(bomb, 4096);
 > just before the (*bomb)() call?
 > (with <sys/typesh>, <m68k/sync_icache.h> and -lm68k)

 Same result, still properly dumps core.

 Martin

From: "Michael L. Hitch" <mhitch@lightning.msu.montana.edu>
To: gnats-bugs@NetBSD.org
Cc: port-m68k-maintainer@netbsd.org, netbsd-bugs@netbsd.org, 
    gnats-admin@netbsd.org, dholland@NetBSD.org, jklos@netbsd.org
Subject: Re: port-m68k/27099 (If one runs crashme on a 2.0 m68k machine, the
 system panics.)
Date: Thu, 10 Apr 2008 20:38:01 -0600 (MDT)

 On Thu, 10 Apr 2008, dholland@NetBSD.org wrote:

 > Synopsis: If one runs crashme on a 2.0 m68k machine, the system panics.
 >
 > State-Changed-From-To: feedback->open
 > State-Changed-By: dholland@NetBSD.org
 > State-Changed-When: Thu, 10 Apr 2008 01:26:42 +0000
 > State-Changed-Why:
 > Feedback timeout. Anyone have an m68k to test this?

    I just tried this on an Amiga 4000 with a 68060 running 4.0, and after 
 try 32, my system paniced with a memory fault.  It appears to be at 
 _060_isp_cas_restart+0x1586 on a pflush a2@ instruction with a2 containing 
 0x2b70.  That offset seems rather large, so I suspect _060_isp_cas_restart 
 isn't involved.  A binary integer kernel package follows that, and it's 
 somewhere within one of the 060 support routines.  A stack trace results 
 in another fault.  Gdb probably can't figure out how to track back through
 that code.

    A second attempt resulted in a kernel hang with the machine locked up.

 --
 Michael L. Hitch			mhitch@montana.edu
 Computer Consultant
 Information Technology Center
 Montana State University	Bozeman, MT	USA

From: David Holland <dholland-bugs@netbsd.org>
To: Izumi Tsutsui <tsutsui@ceres.dti.ne.jp>
Cc: dholland-bugs@NetBSD.org, gnats-bugs@NetBSD.org,
	port-m68k-maintainer@NetBSD.org, gnats-admin@NetBSD.org,
	netbsd-bugs@NetBSD.org, jklos@NetBSD.org
Subject: Re: port-m68k/27099 (If one runs crashme on a 2.0 m68k machine, the system panics.)
Date: Sun, 13 Apr 2008 00:40:57 +0000

 On Fri, Apr 11, 2008 at 12:38:27AM +0900, Izumi Tsutsui wrote:
  > > If so, that's most of the work, and someone just needs to load
  > > trap.c up with printfs to figure out what's happening...
  > 
  > IIRC printf debug didn't help because kernel completely seemed
  > to hang on trap. The instruction cache also looked affected
  > (adding m68k_sync_icache(2) around the bomb() changed some behavor),
  > but I gave up at that point.

 That just makes it more fun... :-)  first order of business I guess
 would be to figure out if it makes it into the kernel at all, by
 adding something in locore. (Dunno what; on i386 one tends to do this
 by scribbling on the screen.)

 Well, maybe the first order of business is to dig around for known
 processor errata.

 -- 
 David A. Holland
 dholland@netbsd.org

>Unformatted:

NetBSD Home
NetBSD PR Database Search

(Contact us) $NetBSD: query-full-pr,v 1.39 2013/11/01 18:47:49 spz Exp $
$NetBSD: gnats_config.sh,v 1.8 2006/05/07 09:23:38 tsutsui Exp $
Copyright © 1994-2007 The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.