From: Linus Torvalds <torvalds@osdl.org>
Date: Wed, 14 Sep 2005 22:50:53 GMT
Subject: Re: [PATCH 21/22] teach the merge algorithm about cache iterators
Message-ID: <Pine.LNX.4.58.0509141549270.26803@g5.osdl.org>
In-Reply-To: <4328A3F9.1010506@citi.umich.edu>



On Wed, 14 Sep 2005, Chuck Lever wrote:
> 
> oh, i see.  the hash table won't help cache_find_name find an insertion 
> point quickly if the name isn't already in the cache.

Note that almost all insertion tends to happen linearly.

In particular, read-tree always inserts things in order.

So probably _most_ of the file finding could actually use even a stupid 
linear search, if they just had a place to start from. And 99% of the 
time, it would be very close to where they wanted to be.

Hmm?

		Linus