lkml.org 
[lkml]   [1998]   [Jun]   [17]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: ext2 fs not properly updated upon dismount.
[SNIPPED all]
Statistics only confuse the issue. Disks are only mounted/unmounted
using two system calls. If the device was marked as properly dismounted,
which is what fsck checks and states, then the device was dismounted. You
don't care about version numbers of mount/dismount/...etc. You also
don't care about C-runtime libraries. The system-call was executed with
the proper device name as a parameter. It doesn't make any difference
how the name got there nor how the call was generated.

If corruption only occurs when disks are dismounted, which is what it
seems, then you can rule out drivers and physical devices. These
things don't know why a write occurred --many writes occur during
normal operation over many hours of use.

The ext2 file systems are cached. It is important that everything that
was buffered gets written to the physical device when it is dismounted.
It looks as though some cached inode data was not written. It also
looks as though this problem occurs when a greater amount of virtual RAM
is available because a swap file is installed.

There is more physical RAM available for caching when a swap file is
installed. There will probably be more dirty inodes cached under these
conditions. This is probably why the problem occurs when a swap file
is installed.

What needs to be done, is a check of the fs-cache code has to be made
to see if there is a "sneak-path" that could prevent all the cached data
from being written to the device during the dismount operation. A hint
of where to look is in the evidence about the installed swap file.

There are at least two experts on the ext2 file-system. I am pretty
sure that at least one is reviewing the reported problem. In the
meantime, including the '-f' option in the startup script that
mounts file-systems after a check, may prevent a creeping problems
that eventually cause errors that can't be readily repaired.


Cheers,
Dick Johnson
***** FILE SYSTEM MODIFIED *****
Penguin : Linux version 2.1.105 on an i586 machine (66.15 BogoMips).
Warning : It's hard to remain at the trailing edge of technology.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu

\
 
 \ /
  Last update: 2005-03-22 13:43    [W:0.101 / U:0.536 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site