Hi,
Just in case you missed it, see this post on LKML -- explains radix
trees in detail which is very important for this project!
ftp://ftp.kernel.org/pub/linux/kernel/people/npiggin/patches/lockless/2.6.1=
6-rc5/radix-intro.pdf
Cheers,
Nitin
---------- Forwarded message ----------
From: Nick Piggin <np...@su...>
Date: Mar 10, 2006 8:48 PM
Subject: A lockless pagecache for Linux
To: Linux Kernel <lin...@vg...>, Linux Memory Management <
lin...@kv...>
Cc: Nick Piggin <np...@su...>
Hi,
I was waiting for 2.6.16 before releasing my patchset, but that got
boring.
ftp://ftp.kernel.org/pub/linux/kernel/people/npiggin/patches/lockless/2.6.1=
6-rc5/
Now I've used some clever subject lines on the subsequent patches
to make you think this isn't a big deal. Actually there are about
36 other "prep" patches before those, and PageReserved removal before
that (which are luckily now mostly in -mm or -linus, respectively).
What's more, there aren't 3 lockless pagecache patches, there are
5 -- but the last two are optimisations.
I'm writing some stuff about these patches, and I've uploaded a
**draft** chapter on the RCU radix-tree, 'radix-intro.pdf' in above
directory (note the bibliography didn't make it -- but thanks Paul
McKenney!)
If anyone would like to test or review it, I would be very happy.
Suggestions to the code or document would be very welcome... but
I'm still hoping nobody spots a fundamental flaw until after OLS.
Rollup of prep patches (5 posted patches apply to the top of this):
2.6.16-rc5-git14-prep.patch.gz
Rollup of prep+lockless patches (includes the 5 posted patches):
2.6.16-rc5-git14-lockless.patch.gz
Note: anyone interested in benchmarking should test prep+rollup vs
prep rather than vs mainline if possible, because there are various
other optimisations in prep.
Thanks,
Nick
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to maj...@kv.... For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=3Dmailto:"do...@kv..."> em...@kv... </a>
|
