Only in 2.4.8aa1: 00_alpha-ds20l-support-1 Only in 2.4.8aa1: 00_alpha-illegal-irq-1 Only in 2.4.8aa1: 00_alpha-illop-1 Only in 2.4.8aa1: 00_alpha-iommu_fixup-1 Only in 2.4.8aa1: 00_alpha-mips-rtc-1 Only in 2.4.8aa1: 00_alpha-numa-initrd-1 Only in 2.4.8aa1: 00_alpha-sg_fill-1 Only in 2.4.8aa1: 00_alpha-smp_tune_scheduling-1 Only in 2.4.8aa1: 00_alpha-srm-2.4.6-pre1-1 Only in 2.4.8aa1: 00_drm_vm-1 Only in 2.4.8aa1: 00_module-oops-1 Only in 2.4.8aa1: 00_rawio-down_read-1 Merged in mainline. Only in 2.4.8aa1: 00_3c59x-zerocopy-2 Only in 2.4.9aa1: 00_3c59x-zerocopy-3 Rediffed. The highlevel will take care to linearize (bounce) when needed at runtime when highmem is enabled. Only in 2.4.9aa1: 00_alpha-pc_keyb-1 Fixed compilation trouble on alpha. Only in 2.4.8aa1: 00_cachelinealigned-in-smp-1 Only in 2.4.9aa1: 00_cachelinealigned-in-smp-2 Rediffed for trivial rejects. Only in 2.4.8aa1: 00_create_bounces-sleeps-1 Dropped, alternate patch is in mainline (seems we want to avoid kmap when we don't sleep). Only in 2.4.8aa1: 00_debug-vm-lru-mm-corruption-1 Dropped (it was random memory corruption, probably not related to the vm but related to the eepro100 driver). Only in 2.4.9aa1: 00_lvm-0.9.1_beta7-6_min-1 Fixed compilation trouble. Only in 2.4.8aa1: 00_o_direct-12 Only in 2.4.9aa1: 00_o_direct-13 Fixed a some minor bug about negative count, reads beyond the end of the file and similar non fatal issues, spotted by Janet Morgan. Only in 2.4.9aa1: 00_poll-nfds-1 Limit the nfds in poll to the rlimit value to avoid an unprivilegied user to allocate dozen of mbytes per process of unswappable ram. Only in 2.4.8aa1: 40_blkdev-pagecache-12 Only in 2.4.9aa1: 40_blkdev-pagecache-13 Rediffed in function of the updated o_direct patch. Only in 2.4.8aa1: 50_uml-2.4.7-5.bz2 Only in 2.4.9aa1: 50_uml-patch-2.4.9.bz2 Picked new uml patch from sourceforge. Only in 2.4.8aa1: 60_pagecache-atomic-2 Only in 2.4.9aa1: 60_pagecache-atomic-3 Rediffed due trivial rejects. Only in 2.4.9aa1: 70_mmap-rb-3 Rewritten the core vma lookup engine with an rbtree instead of an avl for faster rebalance and for checkpointing and in turn avoiding many unneded browse of the tree. WARNING: this is experimental so make sure to backout this patch called "70_mmap-rb-3" for any production box, it must be tested more and benchmarked carefully. The reason I included it even if it's not well tested is just to get more testing and in turn more feedback. This patch should also do all the clever merging for the anonymous mappings for mmap/mremap/mprotect like 2.2 was used to do. It should be an overall noticeable improvement.