Messages in this thread | | | Date | Fri, 28 Aug 2009 14:36:31 +0100 | From | Mel Gorman <> | Subject | Re: [PATCH 2/2] page-allocator: Maintain rolling count of pages to free from the PCP |
| |
On Fri, Aug 28, 2009 at 04:02:44PM +0300, Pekka Enberg wrote: > Hi Mel, > > On Fri, 2009-08-28 at 13:57 +0100, Mel Gorman wrote: > > On Fri, Aug 28, 2009 at 03:16:34PM +0300, Pekka Enberg wrote: > > > Hi Mel, > > > > > > On Fri, Aug 28, 2009 at 11:44 AM, Mel Gorman<mel@csn.ul.ie> wrote: > > > > - page = list_entry(list->prev, struct page, lru); > > > > - /* have to delete it as __free_one_page list manipulates */ > > > > - list_del(&page->lru); > > > > - trace_mm_page_pcpu_drain(page, 0, migratetype); > > > > - __free_one_page(page, zone, 0, migratetype); > > > > + do { > > > > + page = list_entry(list->prev, struct page, lru); > > > > + /* must delete as __free_one_page list manipulates */ > > > > + list_del(&page->lru); > > > > + __free_one_page(page, zone, 0, migratetype); > > > > + trace_mm_page_pcpu_drain(page, 0, migratetype); > > > > > > This calls trace_mm_page_pcpu_drain() *after* __free_one_page(). It's > > > probably not a good idea as __free_one_page() can alter the struct > > > page in various ways. > > > > > > > While true, does it alter the struct page in any way that matters? > > Page flags and order are probably interesting for tracing? >
This is PCPU draining. The flags are already clear of any values of interest and the order is always 0. I can follow up a fix-patch that reverses it just in case but I don't think it makes a major difference?
-- Mel Gorman Part-time Phd Student Linux Technology Center University of Limerick IBM Dublin Software Lab
| |