Rieser4 and 2.6.32

Message

maurs · Post by **maurs** » Wed Jan 06, 2010 4:17 pm

Edward has not released the reiser4 for 2.6.32 patch yes on kernel.org. There is some unofficial patch?

Tnx you in advance.

Post by **asturm** » Wed Jan 06, 2010 4:33 pm

There seem to be upstream issues with reiser4 in 2.6.32 as I recalled this: http://forums.gentoo.org/viewtopic-p-61 ... ml#6101458

cheater1034 wrote:Reiser4 is not good to use on 2.6.32 under all circumstances, i'm not sure what the broken conditions are, but someone should really get in contact with edward (not me, because i'm not using reiser4 and am not totally sure of the problem)

Dont Panic · Post by **Dont Panic** » Wed Jan 06, 2010 5:46 pm

I have also been trying to follow this issue, and everything I have seen so far agrees with genstorm's post.

There doesn't seem to be a robustly working reiser4 for 2.6.32 right now.

I've only located three respectable sources (two of which are really the same) of reiser4 patches:

Edward Shishkin's (seemingly the official source right now).

Andrew Morton's patched mm sources (here's a set hosted at zen kernel).

And the Zen kernel itself (http://zen-kernel.org/), which uses Anderw Morton's patch set to the best of my knowledge.

I've seen several sources warning off the use of reiser4 in the 2.6.32 zen kernel (for example, see the release notes for the 2.6.32-zen4-maggie release)

Let us know if you run across anything.

d2_racing · Post by **d2_racing** » Wed Jan 06, 2010 8:57 pm

For kernel 2.6.32, one of my friends add the reiser4 support inside the gentoo-sources.

More info here : http://www.gentoo-quebec.org/forum/view ... f=4&t=2380

This package is inside the gentoo-quebec overlay.

overlay/sys-kernel/mathieu-sources/mathieu-sources-2.6.32-r1.ebuild

cheater1034 · Post by **cheater1034** » Wed Jan 06, 2010 9:59 pm

The problem with reiser4 is in the upstream code, the only correct place to get reiser4 for 2.6.32 is currently in mmotm (the 2.6.31 patch requires additional work to even compile)

The code hasn't been updated since december 10th and the bug is still there (unless somebody has a patch

), we've thrown out a couple but they havent worked.

dusanc · Post by **dusanc** » Sun Jan 10, 2010 9:44 pm

R4 in -mm is buggy and unusable. Doesn't commit on sync etc. Patches went in without ack from maintainer

There's been some kernel changes, so that R4 for .32 has to be modified. That may take some time

Either use R4 and 2.6.30, or take a chance with your data

dusanc · Post by **dusanc** » Mon Jan 11, 2010 1:38 pm

2.6.32 has a new per-bdi flushing interface and related VFS changes,
which are incompatible with reiser4. Zen stuff works incorrectly, as
actually we need to fix VFS at first. I am working on this at leisure-time,
so wait..

Thanks,
Edward.

http://marc.info/?l=reiserfs-devel&m=12 ... 312842&w=2

Added your Q to Reiser4 FAQ

Dont Panic · Post by **Dont Panic** » Sat Jan 16, 2010 7:35 pm

Good News, Bad News!

First, the good news.

There's some movement in Reiser4 kernel patches (although these are for a mmotm 2.6.33-rc4 kernel).

FYI r4 patches

But now for the bad news.

Edward Shishkin doesn't feel like these patches will work.

Re: FYI r4 patches

kernelOfTruth · Post by **kernelOfTruth** » Sat Jan 16, 2010 11:17 pm

seems he's dedicating most of his time to btrfs now

which on the one side is a good thing (TM) - on the other reiser4 suffers

d2_racing · Post by **d2_racing** » Sun Jan 17, 2010 1:44 am

In fact, but BRTFS maybe the next big thing in terms of file system.

If they are working so hard on that project, there a reason behind it.

yngwin · Post by **yngwin** » Sun Jan 17, 2010 1:56 pm

Yes, brtfs is very promising, but it isn't mature, while reiser4 is. And until it is, reiser4 users are left out in the cold without updated patches.

Dont Panic · Post by **Dont Panic** » Tue Feb 02, 2010 4:48 am

It looks like there is some movement on new Reiser4 patches.

Edward Shishkin has sent a series of patches against the linux-2.6.33-rc5-mm1 kernel to the reiserfs mailing list.

[patch 0/7] per-bdi flushing model improvements. reiser4

kernelOfTruth · Post by **kernelOfTruth** » Tue Feb 02, 2010 11:04 am

Dont Panic wrote:It looks like there is some movement on new Reiser4 patches.

Edward Shishkin has sent a series of patches against the linux-2.6.33-rc5-mm1 kernel to the reiserfs mailing list.

[patch 0/7] per-bdi flushing model improvements. reiser4

lot of that stuff seems to be reverts of previously broken stuff but also improvements to the per-bdi flushing model fixes for writeback, jnode and VFS regarding writeback

that seems to speed up reiser4 again - let's see

nice !

Dont Panic · Post by **Dont Panic** » Thu Feb 11, 2010 2:00 pm

I finally see some 2.6.32 patches for a vanilla kernel.

http://www.kernel.org/pub/linux/kernel/ ... 4-for-2.6/

hirakendu · Post by **hirakendu** » Sat Feb 13, 2010 2:22 am

sweet.

d2_racing · Post by **d2_racing** » Sat Feb 13, 2010 2:58 pm

Yeah nice

Dont Panic · Post by **Dont Panic** » Sun Feb 14, 2010 12:41 am

I was doing some comparison of the 2.6.32 patch provided at http://www.kernel.org/pub/linux/kernel/ ... 4-for-2.6/ and the series of patches posted for the 2.6.33 MMOTM kernel at [patch 0/7] per-bdi flushing model improvements. reiser4

Both are published by Edward Shishkin, but I noticed the proposed patches for the 2.6.33 MMOTM kernel contain much more modification to fs/fs-writeback.c than the official 2.6.32 patch.

If I read the mailing list discussion correctly, Edward is hoping for some performance enhancements with the more extensive changes that haven't shown up in the 'official' patch yet.

So, stay tuned! There may be more changes in the offing.