Messages in this thread | | | Date | Thu, 21 Jun 2012 13:45:29 -0400 | From | Rik van Riel <> | Subject | Re: [PATCH -mm 4/7] mm: make page colouring code generic |
| |
On 06/21/2012 01:40 PM, Andrew Morton wrote: > On Thu, 21 Jun 2012 10:30:26 -0400 Rik van Riel<riel@redhat.com> wrote: > >> On 06/21/2012 07:20 AM, Peter Zijlstra wrote: >>> On Mon, 2012-06-18 at 18:05 -0400, Rik van Riel wrote: >>>> Fix the x86-64 page colouring code to take pgoff into account. >>> >>> Shouldn't that be a separate patch? >> >> My idea was that it would be easier to review >> these two nearly identical functions together. >> >> Andrew, do you have any strong opinions? > > It depends on the significance of the change. I suspect it's one of > things which speeds up many workloads by 1.5% and slows down a few > weird/important ones by 11%. Which makes it a thing to be put under > the microscope and poked at. Some people might end up reverting it, > making it tunable/configurable etc etc. > > If any of that is true then yes, I guess it should be a standalone thing.
Behaviour is not changed by this patch, except for taking pgoff into account - which should not matter a whole lot in practice, because mmap of files is generally done starting at offset 0.
| |