Messages in this thread | | | From | Nikita Danilov <> | Date | Tue, 8 Nov 2005 00:13:51 +0300 | Subject | Re: [PATCH 3/3] vm: writeout watermarks |
| |
Marcelo Tosatti writes: > > Nikita has a customer using large percentage of RAM for > a kernel module, which results in get_dirty_limits() misbehaviour > since > > unsigned long available_memory = total_pages; > > It should work on the amount of cacheable pages instead. > > He's got a patch but I dont remember the URL. Nikita?
http://linuxhacker.ru/~nikita/patches/2.6.14-rc5/09-throttle-against-free-memory.patch
It changes balance_dirty_pages() to calculate threshold not from total amount of physical pages, but from the maximal amount of pages that can be consumed by the file system cache. This amount is approximated by total size of LRU list plus free memory (across all zones).
This has a downside of starting write-out earlier, so patch should probably be accompanied by some tuning of default thresholds.
Nikita.
> > On Tue, Nov 01, 2005 at 04:21:15PM +1100, Nick Piggin wrote: > > 3/3 > > > > -- > > SUSE Labs, Novell Inc. > > > > > Slightly change the writeout watermark calculations so we keep background > > and synchronous writeout watermarks in the same ratios after adjusting them. > > This ensures we should always attempt to start background writeout before > > synchronous writeout. > > > > Signed-off-by: Nick Piggin <npiggin@suse.de> > > > > > > Index: linux-2.6/mm/page-writeback.c > > =================================================================== > > --- linux-2.6.orig/mm/page-writeback.c 2005-11-01 13:41:39.000000000 +1100 > > +++ linux-2.6/mm/page-writeback.c 2005-11-01 14:29:27.000000000 +1100 > > @@ -165,9 +165,11 @@ get_dirty_limits(struct writeback_state > > if (dirty_ratio < 5) > > dirty_ratio = 5; > > > > - background_ratio = dirty_background_ratio; > > - if (background_ratio >= dirty_ratio) > > - background_ratio = dirty_ratio / 2; > > + /* > > + * Keep the ratio between dirty_ratio and background_ratio roughly > > + * what the sysctls are after dirty_ratio has been scaled (above). > > + */ > > + background_ratio = dirty_background_ratio * dirty_ratio/vm_dirty_ratio; > > > > background = (background_ratio * available_memory) / 100; > > dirty = (dirty_ratio * available_memory) / 100; - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |