[lkml]   [2001]   [Sep]   [22]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
This one adds some tweak to the VM compared to plain pre14 that should
make swap smoother. Linus's vm changes seems good here, however dbench
performance decreased significantly but I suspect dbench performance
increase when we're very unfair and not ok for general purpose and I'm
happy with pre14 too.

Still uml fails on vmalloc (swap map) it sorted out being the
flush_tlb_all missing from vmalloc() that is hurting uml, will be fixed
in the next one somehow.

Only in 2.4.10pre13aa1: 00_GFP_ATOMIC-2

Included, also a further fix by Linus is in mainline.

Only in 2.4.10pre14aa1: 00_blkdev-part-1
Only in 2.4.10pre14aa1: 00_blkdev-part-2
Only in 2.4.10pre14aa1: 00_blkdev-part-3
Only in 2.4.10pre14aa1: 00_blkdev-part-4
Only in 2.4.10pre14aa1: 00_blkdev-part-5
Only in 2.4.10pre14aa1: 00_blkdev-part-6

blkdev vfs updates from Alexander Viro.

Only in 2.4.10pre14aa1: 00_bufcache-aging-1

Allow the buffer cache to use mark_page_accessed.

Only in 2.4.10pre14aa1: 00_loop-sem-1

Fix from -ac: remeber to grab the i_sem before writing to the loop

Only in 2.4.10pre13aa1: 00_ramdisk-secure-2

Included in mainline an updated version from Al.

Only in 2.4.10pre13aa1: 00_reiserfs-IO-1

Dropped as it would screwup the bufcache-aging patch
logic, plus the inactive cache should be bigger now
making less likely we run into such trouble.

Only in 2.4.10pre13aa1: 00_start-aggressive-readahead-1
Only in 2.4.10pre14aa1: 00_start-aggressive-readahead-2

Added gfp_mask parameter to specify in which kind of memory the
readahead will be allocated.

Only in 2.4.10pre14aa1: 00_swapin-race-1

Fix an race condition with threads that could corrupt the swapcache.

Only in 2.4.10pre13aa1: 00_unmap-dirty-pte-1
Only in 2.4.10pre14aa1: 00_unmap-dirty-pte-2

Converted to a BUG() after the appreciated infos from Ben.

Only in 2.4.10pre14aa1: 00_vm-swap-1

VM tweaks, the swapout behaviour in pre14 isn't that good, this
should make it better. In particular keeping constantly mapped
pages in the inactive was wasting too much cpu. Plus moved
the mark_page_accessed in the page fault rather than in munmap.
Now also aging the anonymous memory. Disabled swapin readahead
unless there's some memory free, otherwise I suspect we're going to
free the readahead memory before we have a chance to make an use if it
at least by turning down the page-cluster it looked more responsive,
but of course we definitely want swapin readahead after the swap storm
completed, so this looked an acceptable solution. And a few other
minor bits.

Only in 2.4.10pre14aa1: 00_write-aging-1

Mark written pages as referenced, as the swapins (in vm-swap-1 above)
we don't allow them to use the mark_page_accessed logic, so they
cannot go into the active cache but this ages them a little,
we must try to avoid read-modify-write cycles.

Only in 2.4.10pre13aa1: 50_uml-patch-2.4.9-7-1.bz2
Only in 2.4.10pre14aa1: 50_uml-patch-2.4.9-8-1.bz2

Picked last update from sourceforge (was supposed to fix swapon but
it still crashes for me).

Only in 2.4.10pre13aa1: 60_tux-2.4.9-ac10-J9
Only in 2.4.10pre14aa1: 60_tux-2.4.9-ac10-K7

Picked last update from .

Only in 2.4.10pre14aa1: 60_tux-create_child-1
Only in 2.4.10pre13aa1: 60_tux-dprintk-1
Only in 2.4.10pre14aa1: 60_tux-dprintk-2

Defined DTprintk and added create_child callback to the sock for
the tux update.

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
More majordomo info at
Please read the FAQ at

 \ /
  Last update: 2005-03-22 13:03    [W:0.026 / U:0.280 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site