NetBSD Problem Report #31494

From www@netbsd.org  Thu Oct  6 11:32:20 2005
Return-Path: <www@netbsd.org>
Received: by narn.netbsd.org (Postfix, from userid 31301)
	id E3DAF63B850; Thu,  6 Oct 2005 11:32:20 +0000 (UTC)
Message-Id: <20051006113220.E3DAF63B850@narn.netbsd.org>
Date: Thu,  6 Oct 2005 11:32:20 +0000 (UTC)
From: ericg@conceptid.ca
Reply-To: ericg@conceptid.ca
To: gnats-bugs@netbsd.org
Subject: HP NetServer LPr SMP Kernel hang at esiop or siop probing
X-Send-Pr-Version: www-1.0

>Number:         31494
>Category:       port-i386
>Synopsis:       HP NetServer LPr SMP Kernel hang at esiop or siop probing
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    port-i386-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Oct 06 11:33:00 +0000 2005
>Last-Modified:  Thu Nov 24 13:38:35 +0000 2005
>Originator:     Eric
>Release:        3.0Beta
>Organization:
Conceptid.ca
>Environment:
I can't uname -a, the machine is frozen.  HP NetServer LPr dual P3-800, on-board SymLogic 53C895 and another HP NETRAID-1Si RAID PCI adapter with 2 x 18G SCSI hot-swap disk.
>Description:
For 3.0_BETA:
Kernel Generic works.
Kernel SMP freeze at esiop probing
Kernel SMP ACPI reboot very fast in the booting process.
Kernel SMP without esiop support, try to use siop and freeze at probing in the boot sequence.

For 2.0 and 2.0.2 Kernel, same thing.

Firmware is the newest for the pci card.
It crash at the on-board Scsi.  I didn't find where to disable it in bios.  I don't need the on-board scsi... All bioses are the newest.  I try another HP NetServer, same thing.

>How-To-Repeat:
see description.
>Fix:
no.

>Release-Note:

>Audit-Trail:
From: Manuel Bouyer <bouyer@antioche.eu.org>
To: gnats-bugs@NetBSD.org
Cc: port-i386-maintainer@NetBSD.org, gnats-admin@NetBSD.org,
	netbsd-bugs@NetBSD.org
Subject: Re: port-i386/31494: HP NetServer LPr SMP Kernel freeze at esiop or siop probing
Date: Thu, 6 Oct 2005 21:04:24 +0200

 On Thu, Oct 06, 2005 at 11:33:01AM +0000, ericg@conceptid.ca wrote:
 > I can't uname -a, the machine is frozen.  HP NetServer LPr dual P3-800, on-board SymLogic 53C895 and another HP NETRAID-1Si RAID PCI adapter with 2 x 18G SCSI hot-swap disk.
 > >Description:
 > For 3.0_BETA:
 > Kernel Generic works.
 > Kernel SMP freeze at esiop probing

 This is usually an indication that there is interrupt routing problems.

 > Kernel SMP ACPI reboot very fast in the booting process.
 > Kernel SMP without esiop support, try to use siop and freeze at probing in the boot sequence.

 You can disable siop too :)

 -- 
 Manuel Bouyer <bouyer@antioche.eu.org>
      NetBSD: 26 ans d'experience feront toujours la difference
 --

From: Eric Gendron <ericg@conceptid.ca>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-i386/31494
Date: Tue, 11 Oct 2005 18:12:49 -0400

 I try FreeBSD 5.0Beta and compile SMP kernel, that's work without any
 other modifications.

 So it's probably not an interrupt routing problem...


From: Eric Gendron <ericg@conceptid.ca>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-i386/31494
Date: Wed, 12 Oct 2005 05:29:37 -0400

 I've compiled netbsd withoud esiop and siop with MP support from 3.0beta
 CVS.  Seems to work since 15 minutes.

 But I think it shouldn't crash with GENERIC.MP anyway!

 It's ok for me cause I really don't care about the on-board scsi
 interface at all.

 The interrupt shouldn't be the same, I verified in BIOS and at startup.
 I think it's about one of the options in specific GENERIC.MP file?



From: Manuel Bouyer <bouyer@antioche.eu.org>
To: gnats-bugs@NetBSD.org
Cc: port-i386-maintainer@NetBSD.org, gnats-admin@NetBSD.org,
	netbsd-bugs@NetBSD.org
Subject: Re: port-i386/31494
Date: Wed, 12 Oct 2005 20:29:55 +0200

 On Tue, Oct 11, 2005 at 10:20:02PM +0000, Eric Gendron wrote:
 > The following reply was made to PR port-i386/31494; it has been noted by GNATS.
 > 
 > From: Eric Gendron <ericg@conceptid.ca>
 > To: gnats-bugs@NetBSD.org
 > Cc: 
 > Subject: Re: port-i386/31494
 > Date: Tue, 11 Oct 2005 18:12:49 -0400
 > 
 >  I try FreeBSD 5.0Beta and compile SMP kernel, that's work without any
 >  other modifications.
 >  
 >  So it's probably not an interrupt routing problem...

 It's not relevant. The kernel does the interrupt routing, so this would
 just prove that FreeBSD can do proper interrupt routing on this system,
 while NetBSD doesn't. This may be fixed by some interrupt-specific
 option in the kernel config.

 -- 
 Manuel Bouyer <bouyer@antioche.eu.org>
      NetBSD: 26 ans d'experience feront toujours la difference
 --

From: Manuel Bouyer <bouyer@antioche.eu.org>
To: gnats-bugs@NetBSD.org
Cc: port-i386-maintainer@NetBSD.org, gnats-admin@NetBSD.org,
	netbsd-bugs@NetBSD.org
Subject: Re: port-i386/31494
Date: Wed, 12 Oct 2005 20:32:49 +0200

 On Wed, Oct 12, 2005 at 09:30:02AM +0000, Eric Gendron wrote:
 > The following reply was made to PR port-i386/31494; it has been noted by GNATS.
 > 
 > From: Eric Gendron <ericg@conceptid.ca>
 > To: gnats-bugs@NetBSD.org
 > Cc: 
 > Subject: Re: port-i386/31494
 > Date: Wed, 12 Oct 2005 05:29:37 -0400
 > 
 >  I've compiled netbsd withoud esiop and siop with MP support from 3.0beta
 >  CVS.  Seems to work since 15 minutes.
 >  
 >  But I think it shouldn't crash with GENERIC.MP anyway!

 I looked at adding a timeout to the (s)siop reset code. But it adds
 too much code (and possible bugs) for a workaround for something that
 has to be fixed anyway.

 >  
 >  It's ok for me cause I really don't care about the on-board scsi
 >  interface at all.
 >  
 >  The interrupt shouldn't be the same, I verified in BIOS and at startup.
 >  I think it's about one of the options in specific GENERIC.MP file?

 Yes. either PCIBIOS or ACPI. Did you try the GENERIC.MPACPI kernel config ?

 -- 
 Manuel Bouyer <bouyer@antioche.eu.org>
      NetBSD: 26 ans d'experience feront toujours la difference
 --

From: Eric Gendron <ericg@conceptid.ca>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-i386/31494
Date: Thu, 13 Oct 2005 00:40:07 -0400

 Like I said at firts post... of course I tried GENERIC.MPACPI:

 "Kernel SMP ACPI reboot very fast in the booting process."

 means crashing by rebooting... very soon in the boot process.

 I also tried changing the pci card to another pci slot... and check to
 be sure there is no sharing IRQ with on-board SCSI.

 I also insist, this time out arise only when MP kernel is used.  Generic
 single cpu kernel works well.

From: Manuel Bouyer <bouyer@antioche.eu.org>
To: gnats-bugs@NetBSD.org
Cc: port-i386-maintainer@NetBSD.org, gnats-admin@NetBSD.org,
	netbsd-bugs@NetBSD.org
Subject: Re: port-i386/31494
Date: Thu, 13 Oct 2005 23:02:57 +0200

 On Thu, Oct 13, 2005 at 05:35:01AM +0000, Eric Gendron wrote:
 >  
 >  I also insist, this time out arise only when MP kernel is used.  Generic
 >  single cpu kernel works well.

 Probablty because the MP kernel uses MPBIOS/ioapic. You could try adding
 the PCIBIOS and PCI_INTR_FIXUP options to GENERIC.MP.

 -- 
 Manuel Bouyer <bouyer@antioche.eu.org>
      NetBSD: 26 ans d'experience feront toujours la difference
 --

>Unformatted:

Home	PR Database Search