Re: 2.6.12-rc6-mm1 & 2K lun testing

[Andrew Morton <akpm@osdl.org>]

Badari Pulavarty <pbadari@us.ibm.com> wrote:
>
> > Try this:
>  > 
>  > diff -puN mm/page-writeback.c~a mm/page-writeback.c
>  > --- 25/mm/page-writeback.c~a	Thu Jun 16 13:36:29 2005
>  > +++ 25-akpm/mm/page-writeback.c	Thu Jun 16 13:36:54 2005
>  > @@ -501,6 +501,8 @@ void laptop_sync_completion(void)
>  >  
>  >  static void set_ratelimit(void)
>  >  {
>  > +	ratelimit_pages = 32;
>  > +	return;
>  >  	ratelimit_pages = total_pages / (num_online_cpus() * 32);
>  >  	if (ratelimit_pages < 16)
>  >  		ratelimit_pages = 16;
>  > _
>  > 
> 
>  Wow !! Reducing the dirty ratios and the above patch did the trick.
>  Instead of 100% sys CPU, now I have only 50% in sys.

It shouldn't be necessary to do both.  Either the patch or the tuning
should fix it.  Please confirm.

Also please determine whether the deep CFQ queue depth is a problem when
the VFS tuning/patching is in place.

IOW: let's work out which of these three areas needs to be addressed.