e2fsprogs bug fix update

Updated e2fsprogs packages that fix several bugs are now available.

The e2fsprogs packages contain a number of utilities for creating,
checking, modifying, and correcting any inconsistencies in second and third
extended (ext2/ext3) file systems. These utilities include e2fsck, mke2fs,
debugfs, and tune2fs.

These updated packages fix the following bugs:

* if a block's first four bytes consisted of the JFS_MAGIC_NUMBER
(0xc03b3998), these bytes were replaced with zeros when the block was
written into the file system's journal. If the file system was subsequently
recovered, a typo, associated with the Linux Journaling Block Device (JBD),
could result in data corruption on recovery. The Linux kernel typo has been
corrected but the fix was not carried over to e2fsprogs, where the
equivalent problem could present during an e2fsck log replay. It has now
been carried over and data corruption will not occur in the circumstances
described above.

Note: the data corruption potential was always low. It is unlikely ext3
metadata blocks will ever contain the JFS_MAGIC_NUMBER in their first four
bytes and the "data=journaled" mode is rarely used.

* time-based 128-bit UUIDs contain the system time at microsecond
resolution and a clock sequence to ensure uniqueness between values
generated in the same calendric microsecond. If such UUIDs are generated at
a sufficiently high rate on an SMP machine, there will not be enough time
for the implementation of libuuid to generate clock sequence values quickly
enough to ensure that each UUID is unique. E2fsprogs now contains a new
subpackage, "uuidd", which provides a daemon that facilitates unique UUID
creation at high rates be enforcing an increment in clock sequence values
between the generation of new UUIDs. If the daemon is not available,
libuuid falls back to generating UUIDs without it, as it did before.

* The blkid command did not report the labels of swap partitions. Some
end-users may have found this result surprising and been misled into
thinking that there was a problem with the partitioning of the device.
Blkid now reports labels of swap partitions, just as it reports the labels
of other partitions.

* a flaw in the code caused e2fsck to fail for directories larger than 32
MB and output warning messages that "Inode XXXXX is too big" and "Block
#XXXX (XXXXX) causes directory to be too big." If fsck was run manually
with the "-y" option, it would remove the data that was incorrectly
identified as broken. The flaw in the code has been corrected, and e2fsck
now supports directory sizes of up to 2 GB, as originally intended.

* ext2online canonicalized path names passed in from the command line
before searching for devices in /etc/mtab. In the case of multipath
devices, this would fail because the canonical path is not referenced in
/etc/mtab, and attempts to pass in the canonical path from the command line
would fail for the same reason. Ext2online could not therefore resize
multipath devices. Ext2online now searches /etc/mtab for the path name
passed in from the command line and will only search for the canonical name
if the first search fails. Since ext2online can now find multipath devices,
it can therefore resize them.

Users of e2fsprogs are advised to upgrade to these updated packages, which
resolve these issues.

Before applying this update, make sure that all previously-released
errata relevant to your system have been applied.

This update is available via Red Hat Network. Details on how to use
the Red Hat Network to apply this update are available at
http://kbase.redhat.com/faq/docs/DOC-11259

450767 - e2fsck: potential data corruption on log replay
467679 - blkid command does not report swap partition's label in rhel4.7-as-i386
480338 - ext2online does not resize multipath devices because it looks for the wrong device name in /etc/mtab
481466 - [RHEL4.8]: latest e2fsprogs causes install failure