On Tue, 17 Nov 1998 07:42:12 +0100 (CET), Rik van Riel
<H.H.vanRiel@phys.uu.nl> said:
> I meant the page aging that occurs in vmscan.c, where we
> decide on which page to unmap from a program's address
> space.
For the last time, NO IT DOES NOT. Read the source. Linus removed it.
We do not use page->age AT ALL in vmscan.c in current 2.1 kernels.
> There we do aging while we don't age pages from files that are read().
For the last time, YES WE DO. shrink_mmap() for the page cache in
mm/filemap.c still uses page aging in current 2.1 kernels. Read() uses
the page cache.
This is a problem.
> OK, I can (and have for quite a while) agree with this.
> Kernels with this feature and enough memory will run great,
> maybe small machines (<16M) will have a bit of trouble
> keeping up readahead performance (since kswapd will have
> made it's round a bit fast) but those machines will have
> sucky performance anyway :)
This change improves low memory performance very measurably in all tests
I have tried so far.
--Stephen.
-
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/