2006-11-22 11:48:49

by David Binderman

[permalink] [raw]
Subject: arch/x86_64/mm/numa.c(124): remark #593: variable "bootmap_size" was set but nev


Hello there,

I just tried to compile Linux kernel 2.6.18.3 with the Intel C
C compiler.

The compiler said

arch/x86_64/mm/numa.c(124): remark #593: variable "bootmap_size" was set but
never used

The source code is

unsigned long start_pfn, end_pfn, bootmap_pages, bootmap_size,
bootmap_start;

I have checked the source code and I agree with the compiler.
Suggest delete local variable.


Regards

David Binderman

_________________________________________________________________
Find Singles In Your Area Now With Match.com! msnuk.match.com


2006-11-22 12:43:53

by Andi Kleen

[permalink] [raw]
Subject: Re: arch/x86_64/mm/numa.c(124): remark #593: variable "bootmap_size" was set but nev

"d binderman" <[email protected]> writes:

> Hello there,
>
> I just tried to compile Linux kernel 2.6.18.3 with the Intel C
> C compiler.
>
> The compiler said
>
> arch/x86_64/mm/numa.c(124): remark #593: variable "bootmap_size" was
> set but never used

Actually it looks like a real bug -- probably added recently with the
new bootmap code.

The bootmap should be reserved based on that size.

-Andi

2006-11-22 13:29:46

by Mel Gorman

[permalink] [raw]
Subject: Re: arch/x86_64/mm/numa.c(124): remark #593: variable "bootmap_size" was set but nev

On Wed, 22 Nov 2006, Andi Kleen wrote:

> "d binderman" <[email protected]> writes:
>
>> Hello there,
>>
>> I just tried to compile Linux kernel 2.6.18.3 with the Intel C
>> C compiler.
>>
>> The compiler said
>>
>> arch/x86_64/mm/numa.c(124): remark #593: variable "bootmap_size" was
>> set but never used
>
> Actually it looks like a real bug -- probably added recently with the
> new bootmap code.
>
> The bootmap should be reserved based on that size.
>

I checked the arch-independent sizing patches submitted related to x86_64
and no line is changed that does anything with bootmap_size so that ruled
out an obvious candidate. I did a diff on a grep for bootmap_size in
arch/x86_64 and found that there is no difference in usage between 2.6.18
and 2.6.19-rc6. Both call reserve_bootmem(bootmap, bootmap_size) which has
changed. However, the changes seem to be more cosmetic than anything else
using PFN_DOWN and PFN_UP instead of some_value/PAGE_SIZE so it's not
super-clear where the warning is coming from.

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

2006-11-22 14:58:04

by Adrian Bunk

[permalink] [raw]
Subject: Re: arch/x86_64/mm/numa.c(124): remark #593: variable "bootmap_size" was set but nev

On Wed, Nov 22, 2006 at 01:43:40PM +0100, Andi Kleen wrote:
> "d binderman" <[email protected]> writes:
>
> > Hello there,
> >
> > I just tried to compile Linux kernel 2.6.18.3 with the Intel C
> > C compiler.
> >
> > The compiler said
> >
> > arch/x86_64/mm/numa.c(124): remark #593: variable "bootmap_size" was
> > set but never used
>
> Actually it looks like a real bug -- probably added recently with the
> new bootmap code.

No, this unused assignment is in all 2.6 kernels starting with 2.6.0 and
even in 2.4 .

> The bootmap should be reserved based on that size.
>
> -Andi

cu
Adrian

--

"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed