|
From: Jeremy F. <je...@go...> - 2002-11-21 00:48:29
|
On Wed, 2002-11-20 at 14:20, Julian Seward wrote:
> A questions: What's the deal on the LRU mechanism when
> chaining is going, now? Do chained translations get marked when they
> get used, now? Not afaics ... but this is going to present something
> of a problem when LRUing because chained ones will appear to not have
> been used, and so will be much more likely to be dumped. Perhaps we can
> do something clever based on the idea that if a T has been chained
> it must have been used at least once since the last LRU pass.
I did notice the problem and made a small attempt to address it:
patch_me updates the epoch field in the same way as the dispatch loop.
It doesn't actually help much, but it looks like I paid attention to it.
I was thinking of a scheme like an OS VM scheme: have a loop go through
the TTEs at some low rate and unchaining blocks; if on the 2nd pass the
block is still unchained and still hasn't updated the LRU entry, then
consider it old.
J
|