Only in 2.4.7aa1: 00_gcc-30-extern-static-1
Only in 2.4.7aa1: 00_gcc-30-reiserfs-1

	minor gcc-3_0-branch compilation fixes.

Only in 2.4.7aa1: 00_i_blocks-pipe-socket-1

	Avoid fstat to return garbage in st_blocks for socket and pipe
	filedescriptors (the bug generate userspace troubles).
	Thanks to Andreas Schwab and various other suse guys for 
	spotting it.

Only in 2.4.7aa1: 00_net_rx_softirq-optimize-1

	Microoptimization (__cpu_raise_softirq is required to be
	atomic against irqs, [side note: it is not required to
	be atomic against SMP]).

Only in 2.4.7aa1: 00_paride-max_sectors-1

	Merged s/max_segments/max_sectors/ fix from Sergij Kolisnyk
	to avoid generating requests with more segments than "cluster".

Only in 2.4.7aa1: 00_sd-max_sectors-1

	Allow sym driver to generate 512k scsi commands. Set max_sectors
	of the sd majors to 8*MAX_SEGMENTS, and the SHpnt->max_sectors
	of the sym to 8*MAX_SEGMENTS. All other devices should get
	backwards compatibility because when the ->max_sectors isn't
	specified by the lowlevel driver, it defaults to MAX_SECTORS.

Only in 2.4.7pre8aa1: 40_blkdev-pagecache-6
Only in 2.4.7aa1: 40_blkdev-pagecache-7

	Further fixes: serialize the cache coherency protocol
	against ->read_super and ->remount_sb to make sure
	the MS_RDONLY doesn't go away under us while we synchronize
	the caches (see update_buffers()) and also automatically serializes
	against the journal reply on a read only mount. Thanks to Chris Mason
	for spotting and helping fixing those races.

	Found also a race condition in invalidate_devices, we must lock
	down the super before messing with the fs, to ensure the fs doesn't
	go away under us. s_op->clear_inode then is not allowed anymore to
	lock_down the super, and if somebody does better to deadlock than to
	crash ;).

Only in 2.4.7aa1: 50_uml-2.4.7-1.bz2
Only in 2.4.7pre8aa1: 50_uml-arch-2.4.5ac17.bz2
Only in 2.4.7pre8aa1: 50_uml-common-1.bz2
Only in 2.4.7pre8aa1: 51_uml-ac-to-aa-3.bz2
Only in 2.4.7aa1: 51_uml-ac-to-aa-4
Only in 2.4.7aa1: 53_uml-gcc-3.0-1

	uml updates from Jeff Dike (in sync with 2.4.7-1 plus some minor fix
	just submitted).

Known problems:

	For some reason lvm is not working correctly on alpha (not on x86!). Strange
	because nothing seems changed on the lvm side, it shouldn't be related
	to the sd_max-sectors changes either since the harddisks works otherwise ok.
	Since it is reproducible only on alpha so far it will be fixed in an
	later update (probably next week).