Re: [RFC][PATCH 1/6] kconfigurable resources core changes

[Matthew Frost <artusemrys@sbcglobal.net>]

Vivek Goyal wrote:
> On Tue, May 09, 2006 at 02:33:48PM -0500, Kumar Gala wrote:
>> On May 5, 2006, at 12:28 PM, Vivek Goyal wrote:
>>
>>>
>>> o Core changes for Kconfigurable memory and IO resources. By  
>>> default resources
>>>  are 64bit until chosen to be 32bit.
>>>
>>> o Last time I posted the patches for 64bit memory resources but it  
>>> raised
>>>  the concerns regarding code bloat on 32bit systems who use 32 bit
>>>  resources.
>>>
>>> o This patch-set allows resources to be kconfigurable.
>>>
>>> o I have done cross compilation on i386, x86_64, ppc, powerpc,  
>>> sparc, sparc64
>>>  ia64 and alpha.
>>>
>>> Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
>>> ---
>> [snip]
>>
>> I didn't think the bloat was a big issue based on the numbers you  
>> reported.  I'd still prefer to see us just move to a 64-bit resource  
>> on all systems.
> 
> I had also thought that bloat was not a big issue but Andrew thinks
> otherwise. Here is the link to the thread.
> 
> http://marc.theaimsgroup.com/?l=linux-kernel&m=114626907106986&w=2
> http://marc.theaimsgroup.com/?l=linux-kernel&m=114635425606186&w=2
> 
> In the latest patches, 64bit resources are default but one can force
> these to be 32bit.
> 

This may sound like a dumb question, but aside from code bloat, what are 
the performance issues involved in running 64-bit resources on 32-bit 
systems?  AKA, does it kill us x86 users to not have this switch, and by 
what kind of margin?  You can point if it's been said and I just haven't 
found it; I didn't see performance discussion in the last submission.

Thanks!
Matt

> Thanks
> Vivek
> -
> 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/
>