lkml.org 
[lkml]   [2009]   [Mar]   [11]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 1/2] mm: use list.h for vma list

* Alexey Dobriyan <adobriyan@gmail.com> wrote:

> On Wed, Mar 11, 2009 at 11:55:48AM +0200, Daniel Lowengrub wrote:
> > Use the linked list defined list.h for the list of vmas that's stored
> > in the mm_struct structure. Wrapper functions "vma_next" and
> > "vma_prev" are also implemented. Functions that operate on more than
> > one vma are now given a list of vmas as input.
>
> > Signed-off-by: Daniel Lowengrub
>
> That's not how S-o-b line should look like.
>
> > --- linux-2.6.28.7.vanilla/arch/alpha/kernel/osf_sys.c
> > +++ linux-2.6.28.7/arch/alpha/kernel/osf_sys.c
> > @@ -1197,7 +1197,7 @@
> > if (!vma || addr + len <= vma->vm_start)
> > return addr;
> > addr = vma->vm_end;
> > - vma = vma->vm_next;
> > + vma = vma_next(vma);
>
> Well, this bloats both mm_struct and vm_area_struct.

here's some hard numbers from an earlier submission:

| I made a script that runs 'time ./mmap-perf' 100 times and
| outputs the average.
|
| The output on the standard kernel was:
|
| real: 1.022600
| user: 0.135900
| system: 0.852600
|
| The output after the patch was:
|
| real: 0.815400
| user: 0.113200
| system: 0.622200

Which is a 25% speedup in MM performance - which looks very
significant.

Note: i have not repeated the measurements, and it still looks a
bit too good - i'd have expected if there's such a low hanging
25% fruit in the MM we'd have fixed it already.

( Daniel - please preserve such measurements in changelogs. )

Ingo


\
 
 \ /
  Last update: 2009-03-11 11:55    [W:0.052 / U:1.320 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site