On Fri, 19 May 2000, Andrea Arcangeli wrote:
> I'm not sure about the problems you're talking about. And I'm
> not convinced that waiting I/O completation during shink_mmap
> will be successfully.
Skipping pages on which async I/O is in progress will cause
shrink_mmap() to scan the list and evict clean pages which
shouldn't be evicted from memory yet, leading to calls to
swap_out() which will mean the eviction of more pages which
shouldn't be evicted from memory yet.
This bad page replacement, in turn, will cause a lot of
smaller processes to get pagefaults they never should have
gotten and system performance goes down the drain.
Of course this is only an issue with large amounts of dirty
pages, but I'm told that audio and video streaming are pretty
common workloads so we really want to support them ;)
[The old code would skip dirty and/or locked buffers, leading
to the fact that memory could be filled with dirty and locked
buffers while the multimedia program itself would be paged out]
regards,
Rik
-- The Internet is not a network of computers. It is a network of people. That is its real strength.Wanna talk about the kernel? irc.openprojects.net / #kernelnewbies http://www.conectiva.com/ http://www.surriel.com/
- 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/
This archive was generated by hypermail 2b29 : Tue May 23 2000 - 21:00:17 EST