[LWN Logo]

Date:	Mon, 17 Apr 2000 09:05:24 -0700 (PDT)
From:	Chris Mason <mason@suse.com>
To:	reiserfs@devlinux.com, linux-kernel@vger.rutgers.edu
Subject: PATCH new reiserfs code available


Hello everyone, 

ReiserFS 3.6.4 (beta) is available for download from:
ftp://ftp.devlinux.com/pub/namesys/2.3-beta/linux-2.3.99-pre5-reiserfs-3.6.4-patch.gz

Our vs-13042 bug (also seen as a BUG() from within reiserfs_fs.h) should
be fixed.  But, old files from the previous beta that triggered the bug
will still print vs-13042 warning messages, and cannot be read.  So, if
you hit this bug in the past, please do a ls -laR right after booting with
3.6.4.  Any vs-13042 warnings you see are from old files, save the
warnings and make sure any new ones you get aren't duplicates.  Either
way, the BUG() call has been removed, so these bad files should not cause
panics anymore.

If that didn't make sense, send me mail, I'll try again ;-)

We've fixed deadlocks in mount, unmount, remount, and from our GFP_KERNEL
kmalloc calls (switch to GFP_BUFFER to avoid nested transactions).

We've added block preallocation, which has been a big performance boost
for multiprocess benchmarks.  We've also changed block allocation to keep
the data blocks away from the tree nodes, which reduces our tree
fragmentation and makes everything faster.  Neither change involves disk
format changes.

Most of the VFS clean up is done here, except read_inode2 is still there.
Our I/O error path handling needs improvement, it still panics instead of
mounting readonly.  We've fixed a few VFS related races, many thanks to
Andi Kleen and Al Viro for helping to review the code.

Old format support is commented out, and actually needs a patch before you
should start testing it.  One of us will send out the patch on Tuesday.

Endian fixes are mostly there, I have a patch I'm putting on the ftp site
today to fix journal.c.  I wanted to keep this out of the general use code
until at least a few people have had success with it on big endian
systems. 

-chris




-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/