Re: [patch] kmalloc_percpu

[Rusty Russell <rusty@rustcorp.com.au>]

On Thu, 31 Oct 2002 21:36:40 +0530
Ravikiran G Thirumalai <kiran@in.ibm.com> wrote:

> Here's  kmalloc_percpu interfaces ported (...mostly rediffed) to 
> 2.5.45.  (One last try for 2.6).

IMHO this is a "when we need it" patch.  Baby steps...

> +#define this_cpu_ptr(ptr) per_cpu_ptr(ptr, smp_processor_id())

Probably want a get_cpu_ptr() & put_cpu_ptr() using get_cpu() and put_cpu()
(and this would become __get_cpu_ptr()).

And probably move them all to linux/percpu.h.

> +extern void *kmalloc_percpu(size_t size, int flags);
> +extern void kfree_percpu(const void *);
> +extern int percpu_interlaced_alloc(struct percpu_data *, size_t, int);
> +extern void percpu_interlaced_free(struct percpu_data *);
> +extern void percpu_data_sys_init(void);

Hmm... It'd be nice to remove these three from the header so that the
interface is clear (which may mean exposing some slab.c internals so
you can access them in mm/percpu_data.c).

> +	if(!blklist->cachep)

A space between the "if" and the "(" is traditional.

Cheers,
Rusty.
-- 
   there are those who do and those who hang on and you don't see too
   many doers quoting their contemporaries.  -- Larry McVoy