Re: [PATCH] signal_struct slab cache

[Tigran Aivazian]

On Mon, 7 Aug 2000, Mark Hemment wrote:
> Not all caches are system-wide (and the sigact_cachep in my patch _is_not_
> system-wide).  Why add to all the polution?

Of course, I know that you know about slab.h - you wrote it! :)

I was talking about the new more recent usage of slab.h as something more
than just interface header to your SLAB allocator. Namely it now serves as
a header for declarations of SLAB caches used by more than one compilation
unit. 

And if it is used by more than one compilation unit (and your
sigact_cachep is) then it is visible system wide, i.e. it is not static to
that compilation unit. It was a pollution to have single (or in pairs)
"extern kmem_cache_t *bla_cachep" all over the place so I put some of them
in slab.h next to the rest which were already there and that was accepted
by Linus as cleanup (rather than adding to pollution).

Regards,
Tigran