lkml.org 
[lkml]   [2009]   [Apr]   [24]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 02/22] Do not sanity check order in the fast path
On Thu, Apr 23, 2009 at 12:45:30PM -0700, Dave Hansen wrote:
> On Thu, 2009-04-23 at 12:26 -0700, Dave Hansen wrote:
> > I'm going to retry this with a NUMA config.
>
> Oddly enough, removing the alloc_pages() check didn't do anything to
> text on a NUMA=y config, either:
>
> dave@kernel:~/work/mhp-build$ size i386-numaq-sparse.?/vmlinux
> text data bss dec hex filename
> 3460792 449336 1626112 5536240 5479f0 i386-numaq-sparse.0/vmlinux
> 3460792 449336 1626112 5536240 5479f0 i386-numaq-sparse.1/vmlinux
> 3460776 449336 1626112 5536224 5479e0 i386-numaq-sparse.2/vmlinux
>
> Here's bloatmeter from removing the alloc_pages_node() check:
>
> dave@kernel:~/work/mhp-build$ ../linux-2.6.git/scripts/bloat-o-meter
> i386-numaq-sparse.{1,2}/vmlinux
> add/remove: 0/0 grow/shrink: 9/16 up/down: 81/-99 (-18)
> function old new delta
> st_int_ioctl 2600 2624 +24
> tcp_sendmsg 2153 2169 +16
> diskstats_show 739 753 +14
> iov_shorten 49 58 +9
> unmap_vmas 1653 1661 +8
> sg_build_indirect 449 455 +6
> ahc_linux_biosparam 251 253 +2
> nlmclnt_call 557 558 +1
> do_mount 1533 1534 +1

It doesn't make sense at all that text increased in size. Did you make clean
between each .config change and patch application? Are you using distcc
or anything else that might cause confusion? I found I had to eliminate
distscc and clean after each patch application because sometimes
net/ipv4/udp.c would sometimes generate different assembly when
accessing struct zone. Not really sure what was going on there.

--
Mel Gorman
Part-time Phd Student Linux Technology Center
University of Limerick IBM Dublin Software Lab


\
 
 \ /
  Last update: 2009-04-24 11:25    [W:0.128 / U:0.468 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site