Re: [PATCH,RFC] faster kmalloc lookup

[Manfred Spraul <manfred@colorfullife.com>]

Nikita Danilov wrote:

>Marcus Alanen writes:
> > >Most kmalloc calls get constant size argument (usually
> > >sizeof(something)). So, if switch() is used in stead of loop (and
> > >kmalloc made inline), compiler would be able to optimize away
> > >cache_sizes[] selection completely. Attached (ugly) patch does this.
> > 
> > Perhaps a compile-time test to check if the argument is
> > a constant, and only in that case call your new kmalloc, otherwise
> > a non-inline kmalloc call? With your current patch, a non-constant
> > size argument to kmalloc means that the function is inlined anyway,
> > leading to unnecessary bloat in the resulting image.
>
>Yes, exactly.
>  
>
I agree, I have an old patch that does that.

http://www.colorfullife.com/~manfred/slab/patch-km_div

Please ignore the part about fixed point division, it's not needed - the 
'div' instructions is now outside of the hot path.

The problem is that the -mm tree contains around 10 slab patches, I want 
to see them in Linus' tree before I add further patches.

--
    Manfred