Re: [RFC][PATCH 0/10] 64 bit resources

[Kumar Gala <galak@kernel.crashing.org>]

On Mar 24, 2006, at 12:05 PM, Vivek Goyal wrote:

> On Fri, Mar 24, 2006 at 01:12:17AM -0800, Andrew Morton wrote:
> > Vivek Goyal <vgoyal@in.ibm.com> wrote:
> > > Hi,
> > >
> > > Here is an attempt to implement support for 64 bit resources.  
> > > This will
> > > enable memory more than 4G to be exported through /proc/iomem,  
> > > which is used
> > > by kexec/kdump to determine the physical memory layout of the  
> > > system.
> > >
> > > ...
> > >
> > > We used "make allyesconfig" with CONFIG_DEBUG_INFO=n on 2.6.16-mm1.
> > >
> > > i386
> > > ----
> > >
> > > vmlinux size without patch: 40191425
> > > vmlinux size with path: 40244677
> > > vmlinux size bloat: 52K (.13%)
> >
> > ugh, that's actually a surprising amount of growth.  Could you  
> > look into it
> > a bit more please?  Where's it coming from?  text?  data?
>
>
> Andrew, most of it seems to be coming from .text. I have pasted few  
> results
> below.
>
> > A bit of growth in drivers is probably OK, as all machines load a  
> > tiny
> > subset of them.  But if it's core kernel, not so good.  What is  
> > the effect
> > on allnoconfig?
>
> Here are more compilation results with allnoconfig, allmodconfig and
> allyesconfig on i386. I have picked section sizes from the output  
> of readelf.
>
> allnoconfig
> ----------
>
> vmlinux bloat: 0
>
> .text bloat: 1008 bytes
> .data bloat: 672 bytes.
> .init.text bloat: 128 bytes
> .init.data bloat: 0 bytes
>
> (Not sure why vmlinux size difference is zero, given the fact that few
>  sections are showing bloated size)
>
>
> allmodconfig (CONFIG_DEBUG_INFO=n)
> ------------
>
> vmlinux bloat:4096 bytes
>
> .text bloat: 4064 bytes
> .init.text bloat: 470 bytes
> .data bloat: 640 bytes
>
>
> allyesconfig  (CONFIG_DEBUG_INFO=n)
> -----------
>
> vmlinux size bloat: 52K
>
> .text bloat: 28.5K
> .init_text bloat: 5K
> .eh_frame bloat: 16K  (What's that. Looks big)
> .rodata bloat: 152 bytes
> .data bloat: 768 bytes

So the bloat seems be in the drivers as expected.

Vivek, mind updating these against -mm2 also, can you fixup arch/ 
powerpc/kernel/pci_32.c.

Andrew, any issues in merging into -mm?

- kumar
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/