[Jfs-discussion] Re: attempt to access beyond end of device

[Anthony L. <an...@ne...>]

On Mon, Oct 29, 2001 at 08:04:08AM -0600, Jeff Martin wrote:
> I just re-read your post to jfs-discussion list and noticed that you
> were using 2.4.13 and 2.4.13-pre3.

oops typo here...it should be 2.4.14-pre3

> I had this problem beginning with 2.4.11 and found that Al Viro had
> submitted patches:  2.4.12-ac2 and 2.4.13-pre1.
> 
> You said that when you "reboot to the same kernel, the problem
> disappeared...".  Which kernel are you referring to there?

2.4.13-pre3. It goes like this:
  reboot -> fsck -> no error
  mount -> do nothing -> umount -> fsck -> error

Notice the error was triggered after fsck started and before you hit
te (Y/N) prompt.

> I'm interested in this problem because I have an open defect on this
> particular issue, so any extra info you can provide might help me find
> out where the problem still lies.

I did some more tests, and went back to 2.4.10 (was compiled with jfs
1.0.7 though), all problem had gone.

However, not only jfs will encounter this problem, doing a mkreiserfs
will also trigger the same "...access beyond end of device" problem.
Here is what Ive found so far:

kernel       filesystem         problem
---------------------------------------
2.4.10       ext2,jfs,reiserfs  no
2.4.13       ext2               no
2.4.13       jfs, reiserfs      YES
2.4.14-pre3  ext2               no
2.4.14-pre3  jfs, reiserfs      YES

I have read about the 2.4.11 partition problem and verified that the
partition patch already in the 2.4.14-pre3 tree.

This may not be JFS problem after all: my guess is the partition patch
for 2.4.11 messed things up even more.  May be it is time to have some
regression test procedure, don't you think?

Anthony.