Re: rbtree for uvm_pagelookup()

[yamt%mwd.biglobe.ne.jp@localhost (YAMAMOTO Takashi)]

> On Mon, Jun 02, 2008 at 03:17:33PM +0100, Mindaugas Rasiukevicius wrote:
> 
> > Andrew Doran <ad%netbsd.org@localhost> wrote:
> > > This replaces the global page hash with a per-object red-black tree. 
> > > Memory
> > > consumption should be lower than with the hash and it seems to speed up
> > > build.sh slightly on a MP system.
> > > 
> > >   http://www.netbsd.org/~ad/pagelookup.diff
> > > 
> > > Thoughts?

i like it.
a tree-based lookup is useful for range queries in genfs_putpages, too.

> > Looks good!
> > 
> > By the way, is not FreeBSD going to use kind of RADIX tree for this?
> 
> I hadn't heard that, is there is a post somewhere about it?
> 
> Andrew

http://www.freebsd.org/projects/ideas/#p-vmalgorithm

although the above page doesn't have enough details,
i guess it means something similar to what linux does.
the linux radix-tree has tagging functionality, which is useful to
fix fsync performance problems.
a disadvantage is that it will complicate code due to memory allocations
for intermediate nodes.

YAMAMOTO Takashi