Re: [MMTests] IO metadata on XFS

2012-07-04 Thread Mel Gorman
On Wed, Jul 04, 2012 at 10:47:06AM +1000, Dave Chinner wrote: > > > > > > > > [c999a223: xfs: introduce an allocation workqueue] > > > > > > Which indicates that there is workqueue scheduling issues, I think. > > > The same amount of work is being done, but half of it is being > > > pushed off in

[MMTests] IO metadata on XFS

2012-07-04 Thread Mel Gorman
On Wed, Jul 04, 2012 at 10:47:06AM +1000, Dave Chinner wrote: > > > > > > > > [c999a223: xfs: introduce an allocation workqueue] > > > > > > Which indicates that there is workqueue scheduling issues, I think. > > > The same amount of work is being done, but half of it is being > > > pushed off in

[MMTests] IO metadata on XFS

2012-07-04 Thread Dave Chinner
On Tue, Jul 03, 2012 at 11:59:51AM +0100, Mel Gorman wrote: > On Tue, Jul 03, 2012 at 10:19:28AM +1000, Dave Chinner wrote: > > On Mon, Jul 02, 2012 at 08:35:16PM +0100, Mel Gorman wrote: > > > Adding dri-devel and a few others because an i915 patch contributed to > > > the regression. > > > > > >

Re: [MMTests] IO metadata on XFS

2012-07-03 Thread Dave Chinner
On Tue, Jul 03, 2012 at 11:59:51AM +0100, Mel Gorman wrote: > On Tue, Jul 03, 2012 at 10:19:28AM +1000, Dave Chinner wrote: > > On Mon, Jul 02, 2012 at 08:35:16PM +0100, Mel Gorman wrote: > > > Adding dri-devel and a few others because an i915 patch contributed to > > > the regression. > > > > > >

[MMTests] IO metadata on XFS

2012-07-03 Thread Daniel Vetter
On Tue, Jul 03, 2012 at 02:04:14PM +0100, Mel Gorman wrote: > On Mon, Jul 02, 2012 at 08:35:16PM +0100, Mel Gorman wrote: > > > > > > > > It was obvious very quickly that there were two distinct regression so I > > ran two bisections. One led to a XFS and the other led to an i915 patch > > that en

[MMTests] IO metadata on XFS

2012-07-03 Thread Daniel Vetter
On Tue, Jul 03, 2012 at 11:59:51AM +0100, Mel Gorman wrote: > On Tue, Jul 03, 2012 at 10:19:28AM +1000, Dave Chinner wrote: > > On Mon, Jul 02, 2012 at 08:35:16PM +0100, Mel Gorman wrote: > > > Adding dri-devel and a few others because an i915 patch contributed to > > > the regression. > > > > > >

[MMTests] IO metadata on XFS

2012-07-03 Thread Mel Gorman
On Tue, Jul 03, 2012 at 02:31:19PM +0200, Daniel Vetter wrote: > On Tue, Jul 03, 2012 at 11:59:51AM +0100, Mel Gorman wrote: > > On Tue, Jul 03, 2012 at 10:19:28AM +1000, Dave Chinner wrote: > > > On Mon, Jul 02, 2012 at 08:35:16PM +0100, Mel Gorman wrote: > > > > Adding dri-devel and a few others

[MMTests] IO metadata on XFS

2012-07-03 Thread Mel Gorman
On Mon, Jul 02, 2012 at 08:35:16PM +0100, Mel Gorman wrote: > > > > > It was obvious very quickly that there were two distinct regression so I > ran two bisections. One led to a XFS and the other led to an i915 patch > that enables RC6 to reduce power usage. > > [c999a223: xfs: introduce an alloc

[MMTests] IO metadata on XFS

2012-07-03 Thread Mel Gorman
On Tue, Jul 03, 2012 at 11:59:51AM +0100, Mel Gorman wrote: > > Can you run latencytop to see > > if there is excessive starvation/wait times for allocation > > completion? > > I'm not sure what format you are looking for. latencytop is shit for > capturing information throughout a test and it do

[MMTests] IO metadata on XFS

2012-07-03 Thread Mel Gorman
On Tue, Jul 03, 2012 at 10:19:28AM +1000, Dave Chinner wrote: > On Mon, Jul 02, 2012 at 08:35:16PM +0100, Mel Gorman wrote: > > Adding dri-devel and a few others because an i915 patch contributed to > > the regression. > > > > On Mon, Jul 02, 2012 at 03:32:15PM +0100, Mel Gorman wrote: > > > On Mo

[MMTests] IO metadata on XFS

2012-07-03 Thread Eugeni Dodonov
On Tue, Jul 3, 2012 at 9:31 AM, Daniel Vetter wrote: > Well, presuming I understand things correctly the cpu die only goes into > the lowest sleep state (which iirc switches off l3 caches and > interconnects) when both the cpu and gpu are in the lowest sleep state. > rc6 is that deep-sleep state

[MMTests] IO metadata on XFS

2012-07-03 Thread Dave Chinner
On Mon, Jul 02, 2012 at 08:35:16PM +0100, Mel Gorman wrote: > Adding dri-devel and a few others because an i915 patch contributed to > the regression. > > On Mon, Jul 02, 2012 at 03:32:15PM +0100, Mel Gorman wrote: > > On Mon, Jul 02, 2012 at 02:32:26AM -0400, Christoph Hellwig wrote: > > > > It i

Re: [MMTests] IO metadata on XFS

2012-07-03 Thread Mel Gorman
On Tue, Jul 03, 2012 at 02:31:19PM +0200, Daniel Vetter wrote: > On Tue, Jul 03, 2012 at 11:59:51AM +0100, Mel Gorman wrote: > > On Tue, Jul 03, 2012 at 10:19:28AM +1000, Dave Chinner wrote: > > > On Mon, Jul 02, 2012 at 08:35:16PM +0100, Mel Gorman wrote: > > > > Adding dri-devel and a few others

Re: [MMTests] IO metadata on XFS

2012-07-03 Thread Mel Gorman
On Mon, Jul 02, 2012 at 08:35:16PM +0100, Mel Gorman wrote: > > > > > It was obvious very quickly that there were two distinct regression so I > ran two bisections. One led to a XFS and the other led to an i915 patch > that enables RC6 to reduce power usage. > > [c999a223: xfs: introduce an alloc

Re: [MMTests] IO metadata on XFS

2012-07-03 Thread Mel Gorman
On Tue, Jul 03, 2012 at 11:59:51AM +0100, Mel Gorman wrote: > > Can you run latencytop to see > > if there is excessive starvation/wait times for allocation > > completion? > > I'm not sure what format you are looking for. latencytop is shit for > capturing information throughout a test and it do

Re: [MMTests] IO metadata on XFS

2012-07-03 Thread Daniel Vetter
On Tue, Jul 03, 2012 at 02:04:14PM +0100, Mel Gorman wrote: > On Mon, Jul 02, 2012 at 08:35:16PM +0100, Mel Gorman wrote: > > > > > > > > It was obvious very quickly that there were two distinct regression so I > > ran two bisections. One led to a XFS and the other led to an i915 patch > > that en

Re: [MMTests] IO metadata on XFS

2012-07-03 Thread Eugeni Dodonov
On Tue, Jul 3, 2012 at 9:31 AM, Daniel Vetter wrote: > Well, presuming I understand things correctly the cpu die only goes into > the lowest sleep state (which iirc switches off l3 caches and > interconnects) when both the cpu and gpu are in the lowest sleep state. > rc6 is that deep-sleep state

Re: [MMTests] IO metadata on XFS

2012-07-03 Thread Daniel Vetter
On Tue, Jul 03, 2012 at 11:59:51AM +0100, Mel Gorman wrote: > On Tue, Jul 03, 2012 at 10:19:28AM +1000, Dave Chinner wrote: > > On Mon, Jul 02, 2012 at 08:35:16PM +0100, Mel Gorman wrote: > > > Adding dri-devel and a few others because an i915 patch contributed to > > > the regression. > > > > > >

Re: [MMTests] IO metadata on XFS

2012-07-02 Thread Dave Chinner
On Mon, Jul 02, 2012 at 08:35:16PM +0100, Mel Gorman wrote: > Adding dri-devel and a few others because an i915 patch contributed to > the regression. > > On Mon, Jul 02, 2012 at 03:32:15PM +0100, Mel Gorman wrote: > > On Mon, Jul 02, 2012 at 02:32:26AM -0400, Christoph Hellwig wrote: > > > > It i

Re: [MMTests] IO metadata on XFS

2012-07-02 Thread Mel Gorman
Adding dri-devel and a few others because an i915 patch contributed to the regression. On Mon, Jul 02, 2012 at 03:32:15PM +0100, Mel Gorman wrote: > On Mon, Jul 02, 2012 at 02:32:26AM -0400, Christoph Hellwig wrote: > > > It increases the CPU overhead (dirty_inode can be called up to 4 > > > times

[MMTests] IO metadata on XFS

2012-07-02 Thread Mel Gorman
Adding dri-devel and a few others because an i915 patch contributed to the regression. On Mon, Jul 02, 2012 at 03:32:15PM +0100, Mel Gorman wrote: > On Mon, Jul 02, 2012 at 02:32:26AM -0400, Christoph Hellwig wrote: > > > It increases the CPU overhead (dirty_inode can be called up to 4 > > > times