Re: kmem_cache_attr (was Re: [PATCH 04/36] usercopy: Prepare for usercopy whitelisting)

[Christopher Lameter]

On Tue, 16 Jan 2018, Matthew Wilcox wrote:

> On Tue, Jan 16, 2018 at 12:17:01PM -0600, Christopher Lameter wrote:
> > Draft patch of how the data structs could change. kmem_cache_attr is read
> > only.
>
> Looks good.  Although I would add Kees' user feature:

Sure I tried to do this quickly so that the basic struct changes are
visible.

> And I'd start with
> +struct kmem_cache *kmem_cache_create_attr(const kmem_cache_attr *);
>
> leaving the old kmem_cache_create to kmalloc a kmem_cache_attr and
> initialise it.

Well at some point we should convert the callers by putting the
definitions into const kmem_cache_attr initializations. That way
the callbacks function pointers are safe.

> Can we also do something like this?
>
> -#define KMEM_CACHE(__struct, __flags) kmem_cache_create(#__struct,\
> -		sizeof(struct __struct), __alignof__(struct __struct),\
> -		(__flags), NULL)
> +#define KMEM_CACHE(__struct, __flags) ({				\
> +	const struct kmem_cache_attr kca ## __stringify(__struct) = {	\
> +		.name = #__struct,					\
> +		.size = sizeof(struct __struct),			\
> +		.align = __alignof__(struct __struct),			\
> +		.flags = (__flags),					\
> +	};								\
> +	kmem_cache_create_attr(&kca ## __stringify(__struct));		\
> +})
>
> That way we won't need to convert any of those users.

Yep thats what I was planning.