Re: slab-alignment-rework.patch in -mc

[Andrew Morton <akpm@osdl.org>]

Manfred Spraul <manfred@colorfullife.com> wrote:
>
> Andrea Arcangeli wrote:
> 
> >On Tue, Apr 20, 2004 at 12:24:23AM -0700, Andrew Morton wrote:
> >  
> >
> >>So I do think that we should either make "align=0" translate to "pack them
> >>densely" or do the big sweep across all kmem_cache_create() callsites.
> >>    
> >>
> >
> >agreed.
> >  
> >
> What about this proposal:
> SLAB_HWCACHE_ALIGN clear: align to max(sizeof(void*), align).
> SLAB_HWCACHE_ALIGN set: align to max(cpu_align(), align).
> 
> cpu_align is the cpu cache line size - either runtime or compile time.
> 
> Or are there users that want an alignment smaller than sizeof(void*)?

I doubt if this is likely to cause problems, and in cases where we expect
to have really large numbers of objects we could explicitly select an
alignment of 4 anyway.

But why would you choose to make the "SLAB_HWCACHE_ALIGN clear" case use
sizeof(void*) rather than sizeof(int)?