fails to read stage1 file on cciss raid

[Debian Bug Importer]

Automatically imported from Debian bug report #341888 http://bugs.debian.org/341888

[Debian Bug Importer (debzilla)]

Automatically imported from Debian bug report #341888 http://bugs.debian.org/341888

[Debian Bug Importer (debzilla)]

Message-ID: <email address hidden>
Date: Sat, 3 Dec 2005 16:21:34 -0500
From: Joey Hess <email address hidden>
To: Debian Bug Tracking System <email address hidden>
Subject: fails to read stage1 file on cciss raid

--X1bOJ3K7DJ5YkBrT
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

Package: grub
Version: 0.97-1
Severity: serious
Tags: d-i

The main automated test machine for d-i is a proliant DL360 with a HP
Smart Array controller driven by the cciss driver. Root partition is on
/dev/cciss/c0d0p1.=20

As of this version of grub, grub-install fails

sh-3.00# /sbin/grub-install --recheck --no-floppy "(hd0)"
Probing devices to guess BIOS drives. This may take a long time.
The file /boot/grub/stage1 not read correctly. =20

Whatever bios calls this new version of grub is using to read stage1
must not be working. This makes installing debian on this machine fail.
Note that an otherwise identically configured install of debian that
installs testing (and grub version 0.95+cvs20040624-19) works fine.
Indeed, I can manually downgrade grub and then it works:

Unpacking replacement grub ... =20
Setting up grub (0.95+cvs20040624-19) ... =20
sh-3.00# /sbin/grub-install --recheck --no-floppy "(hd0)" =
=20
Probing devices to guess BIOS drives. This may take a long time. =
   =20
Installation finished. No error reported. =
   =20
This is the contents of the device map /boot/grub/device.map. =20
Check if this is correct or not. If any of the lines is incorrect, =20
fix it and re-run the script `grub-install'.
                                                                           =
   =20
(hd0) /dev/cciss/c0d0 =20

The problem is 100% reproducible, and make it impossible for me
to run automated installation tests with d-i and any version of debian
containing grub 0.97-1. :-(

-- System Information:
Debian Release: testing/unstable
  APT prefers unstable
  APT policy: (500, 'unstable'), (1, 'experimental')
Architecture: i386 (i686)
Shell: /bin/sh linked to /bin/bash
Kernel: Linux 2.4.27
Locale: LANG=3Den_US.UTF-8, LC_CTYPE=3Den_US.UTF-8 (charmap=3DUTF-8)

Versions of packages grub depends on:
ii libc6 2.3.5-8.1 GNU C Library: Shared librarie=
s an
ii libncurses5 5.5-1 Shared libraries for terminal =
hand

grub recommends no packages.

-- no debconf information

--=20
see shy jo

--X1bOJ3K7DJ5YkBrT
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: Digital signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)

iD8DBQFDkgxed8HHehbQuO8RAuOxAKCmd48k7tQCyK5++TUXpVP/mVbLEwCg32Qu
bXicYqGuHqSGb+hnXjtT9jI=
=7xi1
-----END PGP SIGNATURE-----

--X1bOJ3K7DJ5YkBrT--

[Sven Joachim (sven-joachim)]

Joey Hess wrote:

> sh-3.00# /sbin/grub-install --recheck --no-floppy "(hd0)"
> Probing devices to guess BIOS drives. This may take a long time.
> The file /boot/grub/stage1 not read correctly.

Is d-i still using a 2.4 kernel? If so, this may be the same problem
which I encountered in #341381. Also, look in /tmp, "grub-install" should
have left a log there.

Best regards,

--
Sven Joachim

[Debian Bug Importer (debzilla)]

Message-ID: <email address hidden>
Date: Sun, 04 Dec 2005 11:13:35 +0100
From: Sven Joachim <email address hidden>
To: Joey Hess <email address hidden>, <email address hidden>
Subject: Re: Bug#341888: fails to read stage1 file on cciss raid

Joey Hess wrote:

> sh-3.00# /sbin/grub-install --recheck --no-floppy "(hd0)"
> Probing devices to guess BIOS drives. This may take a long time.
> The file /boot/grub/stage1 not read correctly.

Is d-i still using a 2.4 kernel? If so, this may be the same problem
which I encountered in #341381. Also, look in /tmp, "grub-install" should
have left a log there.

Best regards,

--
Sven Joachim

[Joey Hess (joeyh)]

Sven Joachim wrote:
> Is d-i still using a 2.4 kernel? If so, this may be the same problem
> which I encountered in #341381. Also, look in /tmp, "grub-install" should
> have left a log there.

No, this is with 2.6.14.

--
see shy jo

[Debian Bug Importer (debzilla)]

Message-ID: <email address hidden>
Date: Sun, 4 Dec 2005 12:33:59 -0500
From: Joey Hess <email address hidden>
To: Sven Joachim <email address hidden>
Cc: <email address hidden>
Subject: Re: Bug#341888: fails to read stage1 file on cciss raid

--jI8keyz6grp/JLjh
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

Sven Joachim wrote:
> Is d-i still using a 2.4 kernel? If so, this may be the same problem
> which I encountered in #341381. Also, look in /tmp, "grub-install" should
> have left a log there.

No, this is with 2.6.14.

--=20
see shy jo

--jI8keyz6grp/JLjh
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: Digital signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)

iD8DBQFDkyiGd8HHehbQuO8RAj5VAKDnO1M3AuHjRQu1whEI/fTvqqUYuACglDS4
8jEaky2awlvC/jd5dOvIskA=
=rx+i
-----END PGP SIGNATURE-----

--jI8keyz6grp/JLjh--

[Joey Hess (joeyh)]

My other mail was wrong, this is with 2.4.27. I've also experienced
earlier problems with 2.6 kernels and grub-install on this hardware, but
I had given up on seeing that working. Having it also break in 2.4.27
now is a suprise. Bug #341381 could be it.

Here is the grub-install log file from /tmp:

/target/tmp # cat grub2vQFGZ

    GNU GRUB version 0.97 (640K lower / 3072K upper memory)

       [ Minimal BASH-like line editing is supported. For
         the first word, TAB lists possible command
         completions. Anywhere else TAB lists the possible
         completions of a device/filename. ]
grub> dump (hd0,0)/boot/grub/stage1 /tmp/grubdAukBZ

Error 21: Selected disk does not exist
grub> quit

--
see shy jo

[Debian Bug Importer (debzilla)]

Message-ID: <email address hidden>
Date: Wed, 7 Dec 2005 16:51:38 -0500
From: Joey Hess <email address hidden>
To: <email address hidden>
Subject: update

--C7zPtVaVf+AK4Oqc
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

My other mail was wrong, this is with 2.4.27. I've also experienced
earlier problems with 2.6 kernels and grub-install on this hardware, but
I had given up on seeing that working. Having it also break in 2.4.27
now is a suprise. Bug #341381 could be it.

Here is the grub-install log file from /tmp:

/target/tmp # cat grub2vQFGZ

    GNU GRUB version 0.97 (640K lower / 3072K upper memory)
                                            =20
       [ Minimal BASH-like line editing is supported. For
         the first word, TAB lists possible command
         completions. Anywhere else TAB lists the possible
         completions of a device/filename. ]
grub> dump (hd0,0)/boot/grub/stage1 /tmp/grubdAukBZ =20
                 =20
Error 21: Selected disk does not exist
grub> quit =20

--=20
see shy jo

--C7zPtVaVf+AK4Oqc
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: Digital signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)

iD8DBQFDl1lqd8HHehbQuO8RAvc4AKCo/TCROl23XuNukGnT9Dgspl6i9gCeN+Jd
kEQ3rJdF+iAhPAHRP0CuOro=
=f6LK
-----END PGP SIGNATURE-----

--C7zPtVaVf+AK4Oqc--

[Otavio Salvador (otavio)]

# Automatically generated email from bts, devscripts version 2.9.10
package grub grub-disk grub-doc
tags 341888 + pending

[Debian Bug Importer (debzilla)]

Message-Id: <20051213222554.E09231C0B6AE@localhost>
Date: Tue, 13 Dec 2005 20:25:54 -0200
From: Otavio Salvador <email address hidden>
To: <email address hidden>
Subject: setting package to grub grub-disk grub-doc, tagging 341888

# Automatically generated email from bts, devscripts version 2.9.10
package grub grub-disk grub-doc
tags 341888 + pending

[Chuck Short (zulcss)]

Grub for dapper has ccis support.

[Otavio Salvador (otavio)]

Source: grub
Source-Version: 0.97-2

We believe that the bug you reported is fixed in the latest version of
grub, which is due to be installed in the Debian FTP archive:

grub-disk_0.97-2_all.deb
  to pool/main/g/grub/grub-disk_0.97-2_all.deb
grub-doc_0.97-2_all.deb
  to pool/main/g/grub/grub-doc_0.97-2_all.deb
grub_0.97-2.diff.gz
  to pool/main/g/grub/grub_0.97-2.diff.gz
grub_0.97-2.dsc
  to pool/main/g/grub/grub_0.97-2.dsc
grub_0.97-2_i386.deb
  to pool/main/g/grub/grub_0.97-2_i386.deb

A summary of the changes between this version and the previous one is
attached.

Thank you for reporting the bug, which will now be closed. If you
have further comments please address them to <email address hidden>,
and the maintainer will reopen the bug report if appropriate.

Debian distribution maintenance software
pp.
Otavio Salvador <email address hidden> (supplier of updated grub package)

(This message was generated automatically at their request; if you
believe that there is a problem with it please contact the archive
administrators by mailing <email address hidden>)

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Format: 1.7
Date: Thu, 22 Dec 2005 15:57:17 -0200
Source: grub
Binary: grub-disk grub grub-doc
Architecture: source i386 all
Version: 0.97-2
Distribution: unstable
Urgency: low
Maintainer: Grub Maintainers <email address hidden>
Changed-By: Otavio Salvador <email address hidden>
Description:
 grub - GRand Unified Bootloader
 grub-disk - GRUB bootable disk image
 grub-doc - Documentation for GRand Unified Bootloader
Closes: 302359 340849 341888 342259
Changes:
 grub (0.97-2) unstable; urgency=low
 .
   * Update debian/control. This caused the wrong NMU identification :(
   * Disable fix_amd64_compile.diff since it broke amd64 building. (closes: #340849)
   * Add message to show that grub-reboot isn't supported in this release anymore.
   * Drop odirect.diff since it broke RAID disk detection. (closes: #341888)
   * Applied patch from Piotr Roszatycki <email address hidden> to
     fix grub-install detection of RAID root device. (closes: #302359)
   * Applied patch from Sven Joachim <email address hidden> to fix some
     typos on grub-floppy manpage. (closes: #342259)
Files:
 6e35ff2218490a0d7674140a48f5fb73 932 admin optional grub_0.97-2.dsc
 ba9a6ac15d25a33a801bdfa11998ef9b 58935 admin optional grub_0.97-2.diff.gz
 7bbbdc31899255138320379b4021d4df 357132 admin optional grub_0.97-2_i386.deb
 1f6b33f2c18b1f6a1fcbab1d50417aef 231128 admin optional grub-disk_0.97-2_all.deb
 7cf382dec3ace109d22e2a4c68fcd763 265772 doc optional grub-doc_0.97-2_all.deb

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)

iD8DBQFDquxDLqiZQEml+FURAofmAKCnZCjLfNNYIEDmi39/qZAOpYr4aACeN8lJ
luPxiOTOLeaEtBFmH/IgivQ=
=y1n1
-----END PGP SIGNATURE-----

[Debian Bug Importer (debzilla)]

Message-Id: <email address hidden>
Date: Thu, 22 Dec 2005 10:17:10 -0800
From: Otavio Salvador <email address hidden>
To: <email address hidden>
Subject: Bug#341888: fixed in grub 0.97-2

Source: grub
Source-Version: 0.97-2

We believe that the bug you reported is fixed in the latest version of
grub, which is due to be installed in the Debian FTP archive:

grub-disk_0.97-2_all.deb
  to pool/main/g/grub/grub-disk_0.97-2_all.deb
grub-doc_0.97-2_all.deb
  to pool/main/g/grub/grub-doc_0.97-2_all.deb
grub_0.97-2.diff.gz
  to pool/main/g/grub/grub_0.97-2.diff.gz
grub_0.97-2.dsc
  to pool/main/g/grub/grub_0.97-2.dsc
grub_0.97-2_i386.deb
  to pool/main/g/grub/grub_0.97-2_i386.deb

A summary of the changes between this version and the previous one is
attached.

Thank you for reporting the bug, which will now be closed. If you
have further comments please address them to <email address hidden>,
and the maintainer will reopen the bug report if appropriate.

Debian distribution maintenance software
pp.
Otavio Salvador <email address hidden> (supplier of updated grub package)

(This message was generated automatically at their request; if you
believe that there is a problem with it please contact the archive
administrators by mailing <email address hidden>)

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Format: 1.7
Date: Thu, 22 Dec 2005 15:57:17 -0200
Source: grub
Binary: grub-disk grub grub-doc
Architecture: source i386 all
Version: 0.97-2
Distribution: unstable
Urgency: low
Maintainer: Grub Maintainers <email address hidden>
Changed-By: Otavio Salvador <email address hidden>
Description:
 grub - GRand Unified Bootloader
 grub-disk - GRUB bootable disk image
 grub-doc - Documentation for GRand Unified Bootloader
Closes: 302359 340849 341888 342259
Changes:
 grub (0.97-2) unstable; urgency=low
 .
   * Update debian/control. This caused the wrong NMU identification :(
   * Disable fix_amd64_compile.diff since it broke amd64 building. (closes: #340849)
   * Add message to show that grub-reboot isn't supported in this release anymore.
   * Drop odirect.diff since it broke RAID disk detection. (closes: #341888)
   * Applied patch from Piotr Roszatycki <email address hidden> to
     fix grub-install detection of RAID root device. (closes: #302359)
   * Applied patch from Sven Joachim <email address hidden> to fix some
     typos on grub-floppy manpage. (closes: #342259)
Files:
 6e35ff2218490a0d7674140a48f5fb73 932 admin optional grub_0.97-2.dsc
 ba9a6ac15d25a33a801bdfa11998ef9b 58935 admin optional grub_0.97-2.diff.gz
 7bbbdc31899255138320379b4021d4df 357132 admin optional grub_0.97-2_i386.deb
 1f6b33f2c18b1f6a1fcbab1d50417aef 231128 admin optional grub-disk_0.97-2_all.deb
 7cf382dec3ace109d22e2a4c68fcd763 265772 doc optional grub-doc_0.97-2_all.deb

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)

iD8DBQFDquxDLqiZQEml+FURAofmAKCnZCjLfNNYIEDmi39/qZAOpYr4aACeN8lJ
luPxiOTOLeaEtBFmH/IgivQ=
=y1n1
-----END PGP SIGNATURE-----

[Sven Joachim (sven-joachim)]

In the new grub package 0.97-2, I no longer see the dreadful message

Error 21: Selected disk does not exist

when running grub unter a 2.4 kernel. AFICT, the only change that can be
responsible for this welcome change is the removal of the odirect.diff
patch, which was supposed to close #341888. And in the manpage for open(2)
I found the following statement about a difference between 2.4 and 2.6
kernelswhich supports this conjecture:

 O_DIRECT
  Try to minimize cache effects of the I/O to and from this file.
  In general this will degrade performance, but it is useful in
  special situations, such as when applications do their own
  caching. File I/O is done directly to/from user space buffers.
  The I/O is synchronous, i.e., at the completion of a read(2) or
  write(2), data is guaranteed to have been transferred. Under
  Linux 2.4 transfer sizes, and the alignment of user buffer and
  file offset must all be multiples of the logical block size of
  the file system. Under Linux 2.6 alignment to 512-byte boundaries
  suffices.

So it seems that the odirect.diff patch introduced the problem, and since
it has been removed I'm closing this bug.

[Debian Bug Importer (debzilla)]

Message-ID: <email address hidden>
Date: Tue, 27 Dec 2005 14:07:42 +0100
From: Sven Joachim <email address hidden>
To: <email address hidden>
CC: <email address hidden>, <email address hidden>
Subject: Problem is gone in version 0.97-2

In the new grub package 0.97-2, I no longer see the dreadful message

Error 21: Selected disk does not exist

when running grub unter a 2.4 kernel. AFICT, the only change that can be
responsible for this welcome change is the removal of the odirect.diff
patch, which was supposed to close #341888. And in the manpage for open(2)
I found the following statement about a difference between 2.4 and 2.6
kernelswhich supports this conjecture:

 O_DIRECT
  Try to minimize cache effects of the I/O to and from this file.
  In general this will degrade performance, but it is useful in
  special situations, such as when applications do their own
  caching. File I/O is done directly to/from user space buffers.
  The I/O is synchronous, i.e., at the completion of a read(2) or
  write(2), data is guaranteed to have been transferred. Under
  Linux 2.4 transfer sizes, and the alignment of user buffer and
  file offset must all be multiples of the logical block size of
  the file system. Under Linux 2.6 alignment to 512-byte boundaries
  suffices.

So it seems that the odirect.diff patch introduced the problem, and since
it has been removed I'm closing this bug.