2.4.20rc1aa1

URL:

	http://www.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.20rc1aa1.gz
	http://www.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.20rc1aa1/

Diff between 2.4.20pre11aa1 and 2.4.20rc1aa1:

Only in 2.4.20rc1aa1: 00_elevator-backmerge-1

Allow merging at the end when latency is zero.

Only in 2.4.20pre11aa1: 00_extraversion-11 Only in 2.4.20rc1aa1: 00_extraversion-12

Rediffed.

Only in 2.4.20pre11aa1: 00_module-locking-fix-1 Only in 2.4.20rc1aa1: 00_module-locking-fix-2

Readd the BKL too, to serialize against module callbacks (thanks to Andrew for noticing).

Only in 2.4.20pre11aa1: 00_reduce-module-races-1 Only in 2.4.20rc1aa1: 20_reduce-module-races-1

Rename.

Only in 2.4.20pre11aa1: 08_qlogicfc-template-aa-4 Only in 2.4.20rc1aa1: 08_qlogicfc-template-aa-5

Fix qlogic compilation.

Only in 2.4.20pre11aa1: 20_sched-o1-fixes-5 Only in 2.4.20rc1aa1: 20_sched-o1-fixes-8

Further o1 sched fixes and improvements (on all the benchmarks).

Only in 2.4.20pre11aa1: 9900_aio-12.gz Only in 2.4.20rc1aa1: 9900_aio-13.gz

Further fixes, now fully functional with fs and rawio. More optimizations coming soon.

Only in 2.4.20pre11aa1: 9910_shm-largepage-5.gz Only in 2.4.20rc1aa1: 9910_shm-largepage-6.gz

Fixed, apparently ipc shm just couldn't use largepages at all in the original version (VM_BIGPAGE would never be set in the vma so it was only a fake that forbidden the vm to touch the SGA possibly showing improvements with bad VM algorithms, equivalent to mlock/SHM_LOCK), now ipcshm will use real large tlb entries too if the alignment/size suggested by userspace is correct (same as in shmfs).

Only in 2.4.20pre11aa1: 9920_kgdb-3.gz Only in 2.4.20rc1aa1: 9920_kgdb-4.gz

Allow only the admin to invoke kgdb (noticed by Andrew).

Only in 2.4.20rc1aa1: 9941_ocfs-20021012.gz Only in 2.4.20rc1aa1: 9942_ocfs-compile-1

Updated to a recent snapshot.

Andrea