(yet another) attempt to access beyond end of device

(yet another) attempt to access beyond end of device

[Max Power]

Not looking for solutions, just delivering a bug report.

I have a 30GB Seagate with one partition on it formatted ReiserFS (version
3.6.25 apparently) on which some CDROM images reside. Kernel 2.4.18
(mandrake).

When I try to read the files I get the following type of messages in my
syslog:

Aug 20 19:28:28 stroke kernel: attempt to access beyond end of device
Aug 20 19:28:28 stroke kernel: 16:41: rw=0, want=471759240, (=0x1c1e7988), limit=471759240
Aug 20 19:33:57 stroke kernel: attempt to access beyond end of device
Aug 20 19:33:57 stroke kernel: 16:41: rw=0, want=623828640, (=0x252edea0), limit=623828640
Aug 20 19:36:19 stroke kernel: attempt to access beyond end of device
Aug 20 19:36:19 stroke kernel: 16:41: rw=0, want=1685807688, (=0x647b6248), limit=1685807688
Aug 21 00:33:34 stroke kernel: attempt to access beyond end of device
Aug 21 00:33:34 stroke kernel: 16:41: rw=0, want=1758259028, (=0x68cce754), limit=1758259028
Aug 21 00:33:38 stroke kernel: attempt to access beyond end of device
Aug 21 00:33:38 stroke kernel: 16:41: rw=0, want=389077160, (=0x1730d8a8), limit=389077160
Aug 21 00:33:46 stroke kernel: attempt to access beyond end of device
Aug 21 00:33:46 stroke kernel: 16:41: rw=0, want=1411977092, (=0x54290f84), limit=1411977092

I've tried all manor of reiserfscking repeatedly. Each time it seems to
fix something different. I've read about the issues with fdisk and not
rebooting and admit that this _could_ be my issue but that doesn't explain
why all the "want" and "limit" numbers are identical. I'm also curious why
upon umount/mounting the partition the location where the read fails
changes.

I was able to retreive my data but I couldn't tell you why it worked. To
do so, I umount/mount'd the partition then cat'd the iso file to
/dev/null. THEN I was able to cat the files and redirect them to new files
on a different disk. If I try to "copy" 2 files in a row without a
umount/mount I get the I/O errors right away. A plain old 'cp' never
worked right. Had to do the cat trick for every file.


Hope this helps. Lemme know if you want more data.


BTW, that problem with the fdisk and not rebooting... Is there any way you
can code reiserfsck so it will tell you that you screwed it up? How can I
know if I did it right or not at this point?


-Frank Johnson

-- 
Nobody snuggles with Max Power.  You strap yourself in and feel the "G"s!

Re: (yet another) attempt to access beyond end of device

[Oleg Drokin]

Hello!

On Wed, Aug 21, 2002 at 01:28:12AM -0700, Max Power wrote:
> Not looking for solutions, just delivering a bug report.
> I have a 30GB Seagate with one partition on it formatted ReiserFS (version
> 3.6.25 apparently) on which some CDROM images reside. Kernel 2.4.18
> (mandrake).
> 
> When I try to read the files I get the following type of messages in my
> syslog:
> Aug 20 19:28:28 stroke kernel: attempt to access beyond end of device
> Aug 20 19:28:28 stroke kernel: 16:41: rw=0, want=471759240, (=0x1c1e7988), limit=471759240
> Aug 20 19:33:57 stroke kernel: attempt to access beyond end of device
> Aug 20 19:33:57 stroke kernel: 16:41: rw=0, want=623828640, (=0x252edea0), limit=623828640
> Aug 20 19:36:19 stroke kernel: attempt to access beyond end of device
> Aug 20 19:36:19 stroke kernel: 16:41: rw=0, want=1685807688, (=0x647b6248), limit=1685807688
> Aug 21 00:33:34 stroke kernel: attempt to access beyond end of device
> Aug 21 00:33:34 stroke kernel: 16:41: rw=0, want=1758259028, (=0x68cce754), limit=1758259028
> Aug 21 00:33:38 stroke kernel: attempt to access beyond end of device
> Aug 21 00:33:38 stroke kernel: 16:41: rw=0, want=389077160, (=0x1730d8a8), limit=389077160
> Aug 21 00:33:46 stroke kernel: attempt to access beyond end of device
> Aug 21 00:33:46 stroke kernel: 16:41: rw=0, want=1411977092, (=0x54290f84), limit=1411977092

> I've tried all manor of reiserfscking repeatedly. Each time it seems to
> fix something different. I've read about the issues with fdisk and not
> rebooting and admit that this _could_ be my issue but that doesn't explain
> why all the "want" and "limit" numbers are identical. I'm also curious why

What do you mean by "identical"? I see they are different each time and this is
very strance since limit should be always the same.

> upon umount/mounting the partition the location where the read fails
> changes.

This is also very strange.
Are you sure your hardware correctly delivers data? Have you verified that?

> I was able to retreive my data but I couldn't tell you why it worked. To
> do so, I umount/mount'd the partition then cat'd the iso file to
> /dev/null. THEN I was able to cat the files and redirect them to new files
> on a different disk. If I try to "copy" 2 files in a row without a
> umount/mount I get the I/O errors right away. A plain old 'cp' never
> worked right. Had to do the cat trick for every file.

I'd suspect hardware problems first.

> BTW, that problem with the fdisk and not rebooting... Is there any way you
> can code reiserfsck so it will tell you that you screwed it up? How can I

Not easily.

> know if I did it right or not at this point?

If fdisk told you ioctl to reread partition table failed, you need to reboot.
Hard to tell it later.

Bye,
    Oleg

Re: (yet another) attempt to access beyond end of device

[Max Power]

On Wed, 21 Aug 2002, Oleg Drokin wrote:

# Hello!
#
# On Wed, Aug 21, 2002 at 01:28:12AM -0700, Max Power wrote:
# > Not looking for solutions, just delivering a bug report.
# > I have a 30GB Seagate with one partition on it formatted ReiserFS (version
# > 3.6.25 apparently) on which some CDROM images reside. Kernel 2.4.18
# > (mandrake).
# >
# > When I try to read the files I get the following type of messages in my
# > syslog:
# > Aug 20 19:28:28 stroke kernel: attempt to access beyond end of device
# > Aug 20 19:28:28 stroke kernel: 16:41: rw=0, want=471759240, (=0x1c1e7988), limit=471759240
# > Aug 20 19:33:57 stroke kernel: attempt to access beyond end of device
# > Aug 20 19:33:57 stroke kernel: 16:41: rw=0, want=623828640, (=0x252edea0), limit=623828640
# > Aug 20 19:36:19 stroke kernel: attempt to access beyond end of device
# > Aug 20 19:36:19 stroke kernel: 16:41: rw=0, want=1685807688, (=0x647b6248), limit=1685807688
# > Aug 21 00:33:34 stroke kernel: attempt to access beyond end of device
# > Aug 21 00:33:34 stroke kernel: 16:41: rw=0, want=1758259028, (=0x68cce754), limit=1758259028
# > Aug 21 00:33:38 stroke kernel: attempt to access beyond end of device
# > Aug 21 00:33:38 stroke kernel: 16:41: rw=0, want=389077160, (=0x1730d8a8), limit=389077160
# > Aug 21 00:33:46 stroke kernel: attempt to access beyond end of device
# > Aug 21 00:33:46 stroke kernel: 16:41: rw=0, want=1411977092, (=0x54290f84), limit=1411977092
#
# > I've tried all manor of reiserfscking repeatedly. Each time it seems to
# > fix something different. I've read about the issues with fdisk and not
# > rebooting and admit that this _could_ be my issue but that doesn't explain
# > why all the "want" and "limit" numbers are identical. I'm also curious why
#
# What do you mean by "identical"? I see they are different each time and this is
# very strance since limit should be always the same.

I was referring to the per-line sameness. In some of the mail list
archives I looked thru the "want" number was different than the "limit"
number each time. Mine are the same. And I don't get why if I "want" X and
the "limit" is X then why can't I have X?

# > upon umount/mounting the partition the location where the read fails
# > changes.
#
# This is also very strange.
# Are you sure your hardware correctly delivers data? Have you verified that?

The main files I was trying to copy were tar.gz's and they passed a "gzip
-t" after I did the cat maneuver. So yes, the data is moving. From my
perspective it's the FS. But I'm neither a hard-disk nor FS expert so
don't take my word for it. The iso files I cat'd across mounted up ok too.
Haven't tried burning one to disk yet. I tried running "find /mnt/disk
-type f | xargs cat > /dev/null" assuming that it would actually cat every
file within the iso image. No errors were displayed for those operations.

I've been using this drive quite reliably for about a year and a half. The
only problems seem to be with the very-large (~300MB - 2GB) files.

# > I was able to retreive my data but I couldn't tell you why it worked. To
# > do so, I umount/mount'd the partition then cat'd the iso file to
# > /dev/null. THEN I was able to cat the files and redirect them to new files
# > on a different disk. If I try to "copy" 2 files in a row without a
# > umount/mount I get the I/O errors right away. A plain old 'cp' never
# > worked right. Had to do the cat trick for every file.
#
# I'd suspect hardware problems first.
#
# > BTW, that problem with the fdisk and not rebooting... Is there any way you
# > can code reiserfsck so it will tell you that you screwed it up? How can I
#
# Not easily.
#
# > know if I did it right or not at this point?
#
# If fdisk told you ioctl to reread partition table failed, you need to reboot.
# Hard to tell it later.
#
# Bye,
#     Oleg
#

Thanks for the quick response!
-Frank

-- 
Nobody snuggles with Max Power.  You strap yourself in and feel the "G"s!

Re: (yet another) attempt to access beyond end of device

[Oleg Drokin]

Hello!

On Wed, Aug 21, 2002 at 02:26:18AM -0700, Max Power wrote:

> # > Aug 21 00:33:46 stroke kernel: attempt to access beyond end of device
> # > Aug 21 00:33:46 stroke kernel: 16:41: rw=0, want=1411977092, (=0x54290f84), limit=1411977092
> # > I've tried all manor of reiserfscking repeatedly. Each time it seems to
> # > fix something different. I've read about the issues with fdisk and not
> # > rebooting and admit that this _could_ be my issue but that doesn't explain
> # > why all the "want" and "limit" numbers are identical. I'm also curious why
> # What do you mean by "identical"? I see they are different each time and this is
> # very strance since limit should be always the same.
> I was referring to the per-line sameness. In some of the mail list
> archives I looked thru the "want" number was different than the "limit"
> number each time. Mine are the same. And I don't get why if I "want" X and
> the "limit" is X then why can't I have X?

Ah, I see now. Very strange indeed. Can you try some stock kernel (say 2.4.19)
and see what happens there because I am not familiar with what is in
Mandrake kernels.

Bye,
    Oleg