Skip to contentSkip to conversation

ext3-2.4-0.9.9

DISCUSSION Andrew Morton 6 participants 8 messages Sep 7, 2001 – Sep 13, 2001

lore
Share & tools
lore
#0 lore raw

Patches against 2.4.10-pre4 and 2.4.9-ac9 are at

	http://www.uow.edu.au/~andrewm/linux/ext3/

It's a fairly large change. The most significant parts are

* the inclusion of Stephen's error-handling work, which is designed to remount the fs read-only in the presence of software and hardware errors, rather than forcing a panic.

* Stephen's fix for the journal_revoke assertion failure which three people have reported.

There have been two reports of a possible interaction problem with vfat where a readdir on the vfat mountpoint returns ENOTDIR when ext3 is present in the kernel. This has proved elusive and in fact has been observed on systems where ext3 was never used. It is probably a vfat bug.

At the above website there is also a patch from Ted Ts'o which will provide significant speedups for accessing large directories. Ted's equivalent patch for ext2 has already been incorporated into the official kernel(s). If possible, please test Ted's patch on top of ext3 0.9.9.

Detailed changelog:

0.9.7
-----

- Merge in a large batch of changes to allow ext3 to recover gracefully from fatal errors. If the fs is set to remount-readonly on error, then we should still be able to unwind cleanly and unmount the filesystem.

- Performance: don't write superblocks synchronously. This reduces a bottleneck in the VM.

Load the ext3 module with the parameter "do_sync_supers=1" to restore the previous behaviour.

- Performance: don't force a new transaction every time we sync (should prevent the writes previously happening every 5 seconds, allowing laptop drives to spin down again.)

0.9.8
-----

- Fix an NFS oops when doing a local delete on an active, nfs-exported file.

- Add proper log levels to a lot of kernel warnings when mounting a bad filesystem or a fs with errors

- Make sure we set the error flag both in the journal and fs superblocks on error (unless we're doing panic-on-error)

0.9.9
-----

- Fix the buffer-already-revoked assertion failure by looking up an aliased buffercache buffer and clearing the revoke bits in there as well as in the journalled data buffer.

- Reorganise page truncation code so we don't take the address of block_flushpage(). This is to simplify merging with Andrea's O_DIRECT patch, which turns block_flushpage() into a macro.

-

#1 lore raw

For reference, I am using 2.4.9-ac9 + ext3-0.9.9 + dir_speedup + kpreempt with no problem. I have my root mounted as ext3, type ordered, with a local journal.

I don't have any data points on performance gains, but I would be happy to provide any if you specify what. Everything feels good.

I see no reason not to merge this into Alan's tree. At least ext3-0.9.9, but the directory speedup seems reasonable enough too.

Signature · 3 lines
Robert M. Love
rml at ufl.edu
rml at tech9.net
#2 lore raw
In reply to Andrew Morton
Quoted 7 lines · Robert Love

For reference, I am using 2.4.9-ac9 + ext3-0.9.9 + dir_speedup + kpreempt with no problem. I have my root mounted as ext3, type ordered, with a local journal.

I don't have any data points on performance gains, but I would be happy to provide any if you specify what. Everything feels good.

OK, thanks.

Quoted 2 lines

I see no reason not to merge this into Alan's tree. At least ext3-0.9.9, but the directory speedup seems reasonable enough too.

Well, it's a fairly large diff, so it's best to not force it onto people until volunteer testers have had a week or so to try it out.

-

#3 lore raw
In reply to Andrew Morton
Quoted 1 line · Andrew Morton

Patches against 2.4.10-pre4 and 2.4.9-ac9 are at

One issue about 0.9.6, 0.9.9 shows no relevant changes: can I mount my root partition with data=journal without setting up an initrd?

#4 lore raw

I went ahead and rediffed ext3-0.9.9 against 2.4.9-ac10. Find it attached (65k, 17k gzipped).

Ted's ext3 directory speedup still applies cleanly, of course.

Still using ext3-0.9.9 + dir speed up, now on ac10, with no problems.

Signature · 3 lines
Robert M. Love
rml at ufl.edu
rml at tech9.net
#5 lore raw
In reply to Andrew Morton
Quoted 6 lines · Matthias Andree
Quoted 1 line · Andrew Morton

Patches against 2.4.10-pre4 and 2.4.9-ac9 are at

One issue about 0.9.6, 0.9.9 shows no relevant changes: can I mount my root partition with data=journal without setting up an initrd?

If you pass rootflags=data=journal, yes. I _think_ that's the flag.

Signature · 2 lines
Tom Rini (TR1265)
http://gate.crashing.org/~trini/
#6 lore raw
In reply to Andrew Morton
Quoted 7 lines · Robert Love

I went ahead and rediffed ext3-0.9.9 against 2.4.9-ac10. Find it attached (65k, 17k gzipped).

Ted's ext3 directory speedup still applies cleanly, of course.

Still using ext3-0.9.9 + dir speed up, now on ac10, with no problems.

Can someone point me to that dir speed up patch? I've looked on kernel newbies for Tso's page, google searches found old dev archives for ext2, and a patch against 2.4.4 for directory indexing....

Is there anything I should know before trying this?

Mike

#7 lore raw
In reply to Andrew Morton
Quoted 6 lines · Mike Fedyk
Quoted 5 lines · Robert Love

I went ahead and rediffed ext3-0.9.9 against 2.4.9-ac10. Find it attached (65k, 17k gzipped).

Ted's ext3 directory speedup still applies cleanly, of course.

Quoted 8 lines
Quoted 2 lines

Still using ext3-0.9.9 + dir speed up, now on ac10, with no problems.

Can someone point me to that dir speed up patch? I've looked on kernel newbies for Tso's page, google searches found old dev archives for ext2, and a patch against 2.4.4 for directory indexing....

Is there anything I should know before trying this?

You can download it at the ext3 for 2.4 page:

http://www.uow.edu.au/~andrewm/linux/ext3/

Only thing you should know is that it hasn't been that heavily tested on ext3.

-Dave