XFS Release 1.2 pre3 is available

--=-i/jyPSHfVzN1c6pm0vOL
Content-Type: text/plain
Content-Transfer-Encoding: 7bit

ftp://oss.sgi.com/projects/xfs/Release-1.2pre3/

The latest in the 1.2 test kernels and cmds are now available.


--=-i/jyPSHfVzN1c6pm0vOL
Content-Description: 
Content-Disposition: attachment; filename=CHANGES-kernel
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain; charset=ISO-8859-1

Changes in 1.2-pre3

* pagebuf fixes for null pointer dereference

* drop lock in pagebuf before calling run_task_queue(&tq_disk)

* fix pagebuf io completion count handling=20

* rework the read-only mount code again,=20
	don't clean up orphaned blocks under a read only mount.
	don't allow quota updates to a read only file system, but allow
	quota inode creation to a r-o device.

* call remove_inode_hash before calling make_bad_inode.
	XFS would go into shutdown if it found an inode that was
	marked bad, removing it from the hash before marking it bad
	prevents XFS from finding it again.

* Fix data corruption when file system block size is less than a page size
	A page in this scenario would have multiple buffer heads.
        The case where a page has been written to via mmap=20
	AND buffered writes the buffer heads were in different states
	and were not always being written out when writepage was called.=20


Changes in 1.2-pre2:

*  Backport faster 8-byte swap from 2.5

*  Prevent blocks from being put on XFS's "busy list" twice.

*  Rearrange how xfs deals with read-only mounts vs. read-only devices.
   Check device for read-only (rather than mount flag)
   before doing xlog_clear_stale_blocks().  We were skipping
   this on read-only mounts, which was not a good thing.
   Remove the vfs read-only checks for various xfs operations,
   just check whether the underlying device is read-only

*  Symlinks are created by default with mode 777 now, old behavior is still
   accessible through sysctl through.  irixsgid mount option deprecated and
   it too is still accessible through sysctl.
  =20
*  fix alloc busy list off by one errors, and simplify the bounds
   check code.

*  fix a couple of nasty log problems. When a transaction crosses
   multiple iclogs, the async transaction code needs to force the
   log up until the last iclog. We need to record this lsn in the
   transaction so we can do a log flush on it.
   Secondly, there was a sleep/wakeup pair between flushing the
   log and log write completions which was a) incorrect, and b)
   no longer needed. This could result in early wakeups of
   threads waiting for log flushes.

--=-i/jyPSHfVzN1c6pm0vOL--