From: [email protected]

Package: genext2fs
Version: 1.3-7
Severity: grave

genext2fs generates completely broken images on s390.

| $ genext2fs -d ./tmp/generic/tree -b 5876 -r 0 ./tmp/generic/initrd
| $ fsck.ext2 tmp/generic/initrd
| e2fsck 1.35 (28-Feb-2004)
| Note: if there is several inode or block bitmap blocks
| which require relocation, or one part of the inode table
| which must be moved, you may wish to try running e2fsck
| with the '-b 5881' option first. The problem may lie only
| with the primary block group descriptor, and the backup
| block group descriptor may be OK.
|
| Block bitmap for group 0 is not in group. (block 50331648)
| Relocate<y>? cancelled!
|
| Inode bitmap for group 0 is not in group. (block 67108864)
| Relocate<y>? cancelled!
|
| Inode table for group 0 is not in group. (block 83886080)
| WARNING: SEVERE DATA LOSS POSSIBLE.
| Relocate<y>? cancelled!
|
| Filesystem did not have a UUID; generating one.
|
| $ dumpe2fs tmp/generic/initrd
| dumpe2fs 1.35 (28-Feb-2004)
| Filesystem volume name: <none>
| Last mounted on: <not available>
| Filesystem UUID: <none>
| Filesystem magic number: 0xEF53
| Filesystem revision #: 0 (original)
| Filesystem features: (none)
| Default mount options: (none)
| Filesystem state: clean with errors
| Errors behavior: Unknown (continue)
| Filesystem OS type: Linux
| Inode count: 736
| Block count: 5876
| Reserved block count: 0
| Free blocks: 712
| Free inodes: 435
| First block: 1
| Block size: 1024
| Fragment size: 1024
| Blocks per group: 5880
| Fragments per group: 5880
| Inodes per group: 736
| Inode blocks per group: 92
| Last mount time: n/a
| Last write time: Mon Aug 16 14:37:59 2004
| Mount count: 0
| Maximum mount count: 20
| Last checked: Thu Jan 1 01:00:00 1970
| Check interval: 0 (<none>)
| Reserved blocks uid: 0 (user root)
| Reserved blocks gid: 0 (group root)
| ext2fs_read_bb_inode: Attempt to read block from filesystem resulted in short read
|
| Group 0: (Blocks 1-5875)
| Primary superblock at 1, Group descriptors at 2-2
| Block bitmap at 50331648 (+50331647), Inode bitmap at 67108864 (+67108863) | Inode table at 83886080-83886171 (+83886079)
| 51202 free blocks, 45825 free inodes, 16896 directories
|
| dumpe2fs: tmp/generic/initrd: error reading bitmaps: Can't read an block bitmap

This breaks d-i on s390.

Bastian

--
Conquest is easy. Control is not.
-- Kirk, "Mirror, Mirror", stardate unknown

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.5 (GNU/Linux)

iEYEARECAAYFAkEgq4MACgkQnw66O/MvCNGKgwCgmJpaSYG9KDKplHdKKxEybs2g JbUAn3uXq5cKN4H9ft6Ms8xey/SqnmtU
=qqNY
-----END PGP SIGNATURE-----

--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)

From: [email protected]

Hi Bastian,

I don't have an s390 to test on unfortunately, so would you please try
a couple of tests to help me isolate this problem.

- does compiling genext2fs generate any warnings?

- does the following generate fsck errors (apart from the lost+found not present warning, and the missing volume UUID)?

$ genext2fs -b 5876 -r 0 /tmp/whatever
$ fsck.ext2 /tmp/whatever

I tried this on my celeron, it works ok for me. If it works for you, I may
need the contents of the directory you were storing on the image using "-d ./tmp/generic/tree", if possible. If it works for you, there is a good
chance that the absurd positions of the bitmaps and inode tables in your
bug report are caused by the group descriptor getting hosed.

Regards,

Finn Thain


--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]

--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)

From: [email protected]

On Tue, Aug 17, 2004 at 11:38:38PM +1000, Finn Thain wrote:

I don't have an s390 to test on unfortunately, so would you please try
a couple of tests to help me isolate this problem.

You have, ask http://db.debian.org/ for the developer machines.

- does compiling genext2fs generate any warnings?

Ask http://buildd.debian.org/.

- does the following generate fsck errors (apart from the lost+found not present warning, and the missing volume UUID)?

$ genext2fs -b 5876 -r 0 /tmp/whatever
$ fsck.ext2 /tmp/whatever

| $ genext2fs -b 5876 -r 0 /tmp/whatever
| $ fsck.ext2 /tmp/whatever
| e2fsck 1.35 (28-Feb-2004)
| Note: if there is several inode or block bitmap blocks
| which require relocation, or one part of the inode table
| which must be moved, you may wish to try running e2fsck
| with the '-b 5881' option first. The problem may lie only
| with the primary block group descriptor, and the backup
| block group descriptor may be OK.
|
| Block bitmap for group 0 is not in group. (block 50331648)
| Relocate<y>? cancelled!
|
| Inode bitmap for group 0 is not in group. (block 67108864)
| Relocate<y>? cancelled!
|
| Inode table for group 0 is not in group. (block 83886080)
| WARNING: SEVERE DATA LOSS POSSIBLE.
| Relocate<y>? cancelled!
|
| Filesystem did not have a UUID; generating one.

I tried this on my celeron, it works ok for me.

You tried it on a 32bit little-endian machine. I try it on a
64bit/31bit big-endian machine.

Bastian

--
Actual war is a very messy business. Very, very messy business.
-- Kirk, "A Taste of Armageddon", stardate 3193.0

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.5 (GNU/Linux)

iEYEARECAAYFAkEiPMoACgkQnw66O/MvCNEuhACeONdGN2Qb23nqCyDExnjywHg8 wB4An3t065XfqaSSK8LDGem3spWO2tYt
=qPHe
-----END PGP SIGNATURE-----

--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)

From: [email protected]

On Tue, 17 Aug 2004, Bastian Blank wrote:

On Tue, Aug 17, 2004 at 11:38:38PM +1000, Finn Thain wrote:

I don't have an s390 to test on unfortunately, so would you please try
a couple of tests to help me isolate this problem.

You have, ask http://db.debian.org/ for the developer machines.

Cool. David, what's the procedure for obtaining access to raptor?

- does compiling genext2fs generate any warnings?

Ask http://buildd.debian.org/.

Very nice. I had no idea that site existed. Interesting to note that make
check fails, but that is not suprising -- it fails where fsck succeeds on
my machines. I'd better look at that too.

- does the following generate fsck errors (apart from the lost+found not present warning, and the missing volume UUID)?

$ genext2fs -b 5876 -r 0 /tmp/whatever
$ fsck.ext2 /tmp/whatever

| $ genext2fs -b 5876 -r 0 /tmp/whatever
| $ fsck.ext2 /tmp/whatever
| e2fsck 1.35 (28-Feb-2004)
| Note: if there is several inode or block bitmap blocks
| which require relocation, or one part of the inode table
| which must be moved, you may wish to try running e2fsck
| with the '-b 5881' option first. The problem may lie only
| with the primary block group descriptor, and the backup
| block group descriptor may be OK.
|
| Block bitmap for group 0 is not in group. (block 50331648)
| Relocate<y>? cancelled!
|
| Inode bitmap for group 0 is not in group. (block 67108864)
| Relocate<y>? cancelled!
|
| Inode table for group 0 is not in group. (block 83886080)
| WARNING: SEVERE DATA LOSS POSSIBLE.
| Relocate<y>? cancelled!
|
| Filesystem did not have a UUID; generating one.

I tried this on my celeron, it works ok for me.

You tried it on a 32bit little-endian machine. I try it on a
64bit/31bit big-endian machine.

I also tested on my alpha (64-bit little endian) and it works ...
interesting.

Looks like I'm gonna need that s390 account. Thanks for your help.

-F

Bastian

--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]

--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)

From: [email protected]

This message is in MIME format. The first part should be readable text,
while the remaining parts are likely unreadable without MIME-aware tools.
Send mail to [email protected] for more info.

Attached is a patch for the make check failure. This s390 bug is one bug
that make check should be able to detect (unless of course it is an e2fsck
bug ;)

David, would it be possible to release this patch to the build servers in
order that the build logs may tell us whether the s390 endianness bug also afflicts the 32-bit big-endian arch's?

(As I mentioned, this bug doesn't afflict my little-endian 64-bit Alpha,
nor my little-endian 32-bit x86.)

-F
[SoupGate killed MIME-encoded file genext2fs-1.3-7-fix-make-check.patch (449 bytes)]

--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)