Re: [patch] radix-tree: avoid atomic allocations for preloaded insertions

[Peter Zijlstra <peterz@infradead.org>]

On Thu, 2007-11-08 at 21:37 +0100, Nick Piggin wrote:
> On Thu, Nov 08, 2007 at 12:57:02PM +0100, Peter Zijlstra wrote:
> > On Wed, 2007-11-07 at 20:12 -0800, Andrew Morton wrote:
> > 
> > > <looks at fs/nfs/write.c>
> > > 
> > > again: unreliable, remembers to test for failure, would be better to use
> > > radix_tree_preload().
> > 
> > http://lkml.org/lkml/2007/10/30/271
> 
> Ah, missed that. See my subsequent patch too, slightly different. It only
> preloads if a request isn't already found (is this a good idea?, if it wasn't
> relatively common, they'd just be checking for -EEXIST in the insertion?).
> 
> Anyway we can also simplify the code because the insertion can't fail with a
> preload.

Yeah, saw that, didn't get round to verifying the logic. Patch looked
good on first glance.

> NFS can also use GFP_NOFS for the preload (at least, for upstream).

Agreed, the NOIO comes from me swapping over it.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>