Messages in this thread | | | Date | Thu, 06 Jul 2000 15:54:31 +0200 | From | Roman Zippel <> | Subject | Re: [PATCH] 2.2.17pre7 VM enhancement Re: I/O performance on2.4.0-test2 |
| |
Hi,
Andrea Arcangeli wrote:
> So basically we'll have these completly different lists: > > lru_swap_cache > lru_cache > lru_mapped > > The three caches have completly different importance that is implicit by > the semantics of the memory they are queuing. Shrinking swap_cache first > is vital for performance under swap for example (and I can just do that in > recent classzone patches). Shrinking lru_cache first is vital for > performance under streaming I/O but without low on freeable memory > scenario.
How do you want to synchronize and balance these caches? Do you expect that these are never used at the same time? What happens with disk blocks that end up in different caches? IMO the problem gets worse, if we want better direct i/o support especially on systems where fs block size is different from page size.
bye, Roman
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |