Messages in this thread | | | Date | Tue, 26 Jan 1999 13:23:33 +0100 (CET) | From | Rik van Riel <> | Subject | Re: MM deadlock [was: Re: arca-vm-8...] |
| |
On Mon, 25 Jan 1999, Linus Torvalds wrote: > On Mon, 25 Jan 1999, Dr. Werner Fink wrote: > > > > This hypothetical bit should only be set if the page is read physical > > from the swap device/file. That means it would take one step more > > to swap out this page again (test_and_clear_bit of both > > PG_recently_swapped_in and PG_referenced). > > Ehh - it is already marked "accessed" in the page tables, which > essentially amounts to exactly that kind of two-level aging (the > PG_referenced bit only takes effect once the swapped-in page has > once more been evicted from the page tables)
With a bit of imagination, you might even be able to call our current scheme two-handed...
Even though it's a bit different, it seems like we have all advantages and none of the disadvantages of a two-handed system. The main difference is that we do I/O on the first hand and page eviction on the second. This gives us a buffer of ready-to-evict pages which we can easily free when we're in a hurry.
The only thing we really need now is a way to keep track of (and manage) that buffer of freeable pages. I believe Andrea has a patch for that -- we should check it out and incorporate something like that ASAP.
There are several reasons why we need it: - we should never run out of freeable pages because that can introduce too much latency and possibly even system instability - page aging only is effective/optimal when the freeable buffer is large enough - when the freeable buffer is too large, we might have too many soft pagefaults or other overhead (not very much of a concern, but still...) - keeping a more or less fixed distance between both hands could make the I/O less bursty and improve system I/O performance
Rik -- If a Microsoft product fails, who do you sue? +-------------------------------------------------------------------+ | Linux memory management tour guide. riel@nl.linux.org | | Scouting Vries cubscout leader. http://www.nl.linux.org/~riel | +-------------------------------------------------------------------+
- 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/
| |