I merged some bits from mainline to trigger swapout earlier to see if it
helps , but still no anon pages in the lru, it's easy to add them back
in the lru as soon as we verify the largemem works ok too.
As usual it is been tested on high end hardware kindly provided by
www.osdlab.org.
ftp://ftp.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.14pre3aa2/
ftp://ftp.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.14pre3aa2.bz2
Only in 2.4.14pre3aa1: 00_rcu-poll-1
Only in 2.4.14pre3aa2: 00_rcu-poll-2
Don't waste an additional cacheline, move the quiescent sequence number
in the schedule_data so rcu become zerocost in terms of cpu cache and
memory bandwith in the fast path.
Only in 2.4.14pre3aa1: 10_numa-sched-12
Only in 2.4.14pre3aa2: 10_numa-sched-13
rediffed due rcu-poll changes.
Only in 2.4.14pre3aa1: 10_vm-5
Only in 2.4.14pre3aa2: 10_vm-6
Merged max_mapped*10 logic from Linus, still not adding anon pages
from the lru, so we can see the difference, I can readd the anon
pages to the lru anytime as soon as it's clear what's better.
Andrea
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Wed Oct 31 2001 - 21:00:34 EST