NetBSD Problem Report #46228

From www@NetBSD.org  Mon Mar 19 12:42:28 2012
Return-Path: <www@NetBSD.org>
Received: from mail.netbsd.org (mail.netbsd.org [149.20.53.66])
	by www.NetBSD.org (Postfix) with ESMTP id BD00A63E104
	for <gnats-bugs@gnats.NetBSD.org>; Mon, 19 Mar 2012 12:42:27 +0000 (UTC)
Message-Id: <20120319124226.C7F1263D96B@www.NetBSD.org>
Date: Mon, 19 Mar 2012 12:42:26 +0000 (UTC)
From: aniou@smutek.pl
Reply-To: aniou@smutek.pl
To: gnats-bugs@NetBSD.org
Subject: bootloader can't load kernel on Dell 2850
X-Send-Pr-Version: www-1.0

>Number:         46228
>Category:       bin
>Synopsis:       bootloader can't load kernel on Dell 2850
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    bin-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Mar 19 12:45:00 +0000 2012
>Last-Modified:  Tue Mar 27 08:35:01 +0000 2012
>Originator:     Piotr Meyer
>Release:        6.0_BETA
>Organization:
>Environment:
NetBSD  6.0_BETA NetBSD 6.0_BETA (GENERIC) amd64
>Description:
Bootloader on freshly installed system (Dell 2850) isn't able to see or load kernel. Screenshots from remote management are available at:

http://aniou.smutek.pl/netbsd_boot_failp1.jpg and
http://aniou.smutek.pl/netbsd_boot_failp2.jpg

System was installed in very standard way (boot loader at mbr,
defaults for filesystem - modulo / without -o log). 

I was not able to start either by original /boot or copied from
i386 version, BUT both bootloaders from install CD and /usr/mdec/boot
extracted from old 5.99.25 release (i386) works fine.


controller and drive info:


# dmesg|grep -E '(amr|ld)'
amr has not been converted to device_t
amr0 at pci2 dev 14 function 0: AMI RAID <PERC 4e/Di>
amr0: interrupting at ioapic0 pin 18
amr0: firmware 5B2D, BIOS H435, 256MB RAM
ld0 at amr0 unit 0: RAID 5, optimal
ld0: 135 GB, 17684 cyl, 255 head, 63 sec, 512 bytes/sect x 284098560 sectors
ld0: GPT GUID: ebf97c8b-48fa-11e1-b694-001143e38f3a
dk0 at ld0: ebf9df89-48fa-11e1-b694-001143e38f3a
dk1 at ld0: 4425145b-48fb-11e1-b694-001143e38f3a
dk2 at ld0: 4901fb92-48fb-11e1-b694-001143e38f3a
dk3 at ld0: 4d3194cc-48fb-11e1-b694-001143e38f3a
dk4 at ld0: 5012c357-48fb-11e1-b694-001143e38f3a
boot device: ld0
root on ld0a dumps on ld0b

>How-To-Repeat:
Try install on another Dell/PERC/amd64?
>Fix:
Copy of old /boot from 5.99.25 works for me, but it may doesn't fit as general solution for this problem ;)

>Audit-Trail:
From: Manuel Bouyer <bouyer@antioche.eu.org>
To: gnats-bugs@NetBSD.org
Cc: gnats-admin@NetBSD.org, netbsd-bugs@NetBSD.org
Subject: Re: bin/46228: bootloader can't load kernel on Dell 2850
Date: Mon, 19 Mar 2012 13:54:27 +0100

 On Mon, Mar 19, 2012 at 12:45:00PM +0000, aniou@smutek.pl wrote:
 > >Description:
 > Bootloader on freshly installed system (Dell 2850) isn't able to see or load kernel. Screenshots from remote management are available at:
 > 
 > http://aniou.smutek.pl/netbsd_boot_failp1.jpg and
 > http://aniou.smutek.pl/netbsd_boot_failp2.jpg

 what does 'ls' at the boot prompt say for hd0a ?

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

From: Piotr Meyer <aniou@smutek.pl>
To: gnats-bugs@NetBSD.org
Cc: gnats-admin@netbsd.org, netbsd-bugs@netbsd.org
Subject: Re: bin/46228: bootloader can't load kernel on Dell 2850
Date: Mon, 19 Mar 2012 14:11:00 +0100

 On Mon, Mar 19, 2012 at 12:55:03PM +0000, Manuel Bouyer wrote:

 >  > http://aniou.smutek.pl/netbsd_boot_failp1.jpg and
 >  > http://aniou.smutek.pl/netbsd_boot_failp2.jpg
 >  
 >  what does 'ls' at the boot prompt say for hd0a ?

 'ls: No such file or directory' 

 (but kernel exists and is bootable by older versions of /boot)

 -- 
 Piotr 'aniou' Meyer

From: Piotr Meyer <aniou@smutek.pl>
To: gnats-bugs@NetBSD.org
Cc: gnats-admin@netbsd.org, netbsd-bugs@netbsd.org, bouyer@antioche.eu.org
Subject: Re: bin/46228: bootloader can't load kernel on Dell 2850
Date: Mon, 26 Mar 2012 22:02:00 +0200

 On Mon, Mar 19, 2012 at 02:11:00PM +0100, Piotr Meyer wrote:
 > On Mon, Mar 19, 2012 at 12:55:03PM +0000, Manuel Bouyer wrote:
 >  
 > >  > http://aniou.smutek.pl/netbsd_boot_failp1.jpg and
 > >  > http://aniou.smutek.pl/netbsd_boot_failp2.jpg
 > >  
 > >  what does 'ls' at the boot prompt say for hd0a ?
 > 
 > 'ls: No such file or directory' 
 > 
 > (but kernel exists and is bootable by older versions of /boot)

 I found that /usr/mdec/boot built without GPT support boots fine.

 After enabling CPPFLAGS+= -DNO_GPT in sys/arch/i386/stand/lib/Makefile
 re-compiled /usr/mdec/boot works - compiled again, but with above 
 definition commented out - fails exactly as in my first mail. 

 Maybe something left on disk? I had FreeBSD 9.x previously installed 
 on this host.

 -- 
 Piotr 'aniou' Meyer

From: Manuel Bouyer <bouyer@antioche.eu.org>
To: Piotr Meyer <aniou@smutek.pl>
Cc: gnats-bugs@NetBSD.org, gnats-admin@NetBSD.org, netbsd-bugs@NetBSD.org
Subject: Re: bin/46228: bootloader can't load kernel on Dell 2850
Date: Mon, 26 Mar 2012 22:07:23 +0200

 On Mon, Mar 26, 2012 at 10:02:00PM +0200, Piotr Meyer wrote:
 > I found that /usr/mdec/boot built without GPT support boots fine.
 > 
 > After enabling CPPFLAGS+= -DNO_GPT in sys/arch/i386/stand/lib/Makefile
 > re-compiled /usr/mdec/boot works - compiled again, but with above 
 > definition commented out - fails exactly as in my first mail. 
 > 
 > Maybe something left on disk? I had FreeBSD 9.x previously installed 
 > on this host.

 yes, that could be the problem.
 Maybe try gpt destroy -r ?

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

From: Piotr Meyer <aniou@smutek.pl>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: bin/46228: bootloader can't load kernel on Dell 2850
Date: Tue, 27 Mar 2012 10:32:41 +0200

 On Mon, Mar 26, 2012 at 08:10:05PM +0000, Manuel Bouyer wrote:

 >  > I found that /usr/mdec/boot built without GPT support boots fine.
 >  > 
 >  > After enabling CPPFLAGS+= -DNO_GPT in sys/arch/i386/stand/lib/Makefile
 >  > re-compiled /usr/mdec/boot works - compiled again, but with above 
 >  > definition commented out - fails exactly as in my first mail. 
 >  > 
 >  > Maybe something left on disk? I had FreeBSD 9.x previously installed 
 >  > on this host.
 >  
 >  yes, that could be the problem.
 >  Maybe try gpt destroy -r ?

 Still no luck - gpt refuse any form of cooperation, all operations
 (remove, migrate, show) gave me simply:

  'gpt: error: bogus map'

 Finally I found some suggestions: http://markmail.org/message/efyk7bbzvssidvl5
 - looks like I got mix of GPT and MBR partitions:

 # gpt -vvv show ld0
 gpt: rld0d: mediasize=145458462720; sectorsize=512; blocks=284098560
 gpt: rld0d: MBR at sector 0
 gpt: rld0d: MBR part: type=169, start=63, size=284098497
 gpt: rld0d: Pri GPT at sector 1
 gpt: rld0d: GPT partition: type=83bd6b9d-7f41-11dc-be0b-001560b84f0f, start=34, size=128
 gpt: error: bogus map
 gpt: unable to open device 'rld0d': Undefined error: 0


 The quickest solution for me is simply clearing disk and re-installing system,
 (it isn't a problem for me) but - IMVHO - this case should be covered in install 
 notes (or, better, at installer level - at least as BIG warning in partitioning 
 menu "Warning: GPT partition is already present..." etc).

 -- 
 Piotr 'aniou' Meyer

Home	PR Database Search