[patch 1/2] mm: page_alloc: reset aging cycle with GFP_THISNODE

2014-02-25 Thread Johannes Weiner
Jan Stancek reports manual page migration encountering allocation failures after some pages when there is still plenty of memory free, and bisected the problem down to 81c0a2bb515f ("mm: page_alloc: fair zone allocator policy"). The problem is that page migration uses GFP_THISNODE and this makes t

Re: [patch 1/2] mm: page_alloc: reset aging cycle with GFP_THISNODE

2014-02-25 Thread Rik van Riel
On 02/25/2014 03:27 PM, Johannes Weiner wrote: > Jan Stancek reports manual page migration encountering allocation > failures after some pages when there is still plenty of memory free, > and bisected the problem down to 81c0a2bb515f ("mm: page_alloc: fair > zone allocator policy"). > > The proble

Re: [patch 1/2] mm: page_alloc: reset aging cycle with GFP_THISNODE

2014-02-26 Thread Mel Gorman
On Tue, Feb 25, 2014 at 03:27:01PM -0500, Johannes Weiner wrote: > Jan Stancek reports manual page migration encountering allocation > failures after some pages when there is still plenty of memory free, > and bisected the problem down to 81c0a2bb515f ("mm: page_alloc: fair > zone allocator policy"

Re: [patch 1/2] mm: page_alloc: reset aging cycle with GFP_THISNODE

2014-02-26 Thread Johannes Weiner
On Wed, Feb 26, 2014 at 09:54:22AM +, Mel Gorman wrote: > On Tue, Feb 25, 2014 at 03:27:01PM -0500, Johannes Weiner wrote: > > Jan Stancek reports manual page migration encountering allocation > > failures after some pages when there is still plenty of memory free, > > and bisected the problem

Re: [patch 1/2] mm: page_alloc: reset aging cycle with GFP_THISNODE

2014-02-26 Thread Johannes Weiner
On Wed, Feb 26, 2014 at 12:12:06PM -0500, Johannes Weiner wrote: > On Wed, Feb 26, 2014 at 09:54:22AM +, Mel Gorman wrote: > > How about special casing the (alloc_flags & ALLOC_WMARK_LOW) check in > > get_page_from_freelist to also ignore GFP_THISNODE? The NR_ALLOC_BATCH > > will go further neg

Re: [patch 1/2] mm: page_alloc: reset aging cycle with GFP_THISNODE

2014-02-27 Thread Rik van Riel
On 02/26/2014 03:13 PM, Johannes Weiner wrote: Would this be an acceptable replacement for 1/2? Looks reasonable to me. This should avoid the issues that were observed with NUMA migrations. --- From: Johannes Weiner Subject: [patch 1/2] mm: page_alloc: exempt GFP_THISNODE allocations from

Re: [patch 1/2] mm: page_alloc: reset aging cycle with GFP_THISNODE

2014-02-28 Thread Mel Gorman
On Wed, Feb 26, 2014 at 03:13:33PM -0500, Johannes Weiner wrote: > On Wed, Feb 26, 2014 at 12:12:06PM -0500, Johannes Weiner wrote: > > On Wed, Feb 26, 2014 at 09:54:22AM +, Mel Gorman wrote: > > > How about special casing the (alloc_flags & ALLOC_WMARK_LOW) check in > > > get_page_from_freelis