Re: [RFC][PATCH v2] slub: Keep page and object in sync in slab_alloc_node()

[Steven Rostedt <rostedt@goodmis.org>]

On Fri, 2013-01-18 at 13:42 +0900, Joonsoo Kim wrote:

> diff --git a/include/linux/slub_def.h b/include/linux/slub_def.h
> index 9db4825..b54dffa 100644
> --- a/include/linux/slub_def.h
> +++ b/include/linux/slub_def.h
> @@ -46,6 +46,9 @@ enum stat_item {
>  struct kmem_cache_cpu {
>  	void **freelist;	/* Pointer to next available object */
>  	unsigned long tid;	/* Globally unique transaction id */
> +#ifdef CONFIG_NUMA
> +	int node;

Note, you put an int between a long and a pointer, which will waste 4
bytes on 64bit machines.

> +#endif
>  	struct page *page;	/* The slab from which we are allocating */
>  	struct page *partial;	/* Partially allocated frozen slabs */
>  #ifdef CONFIG_SLUB_STATS



> @@ -2038,10 +2049,10 @@ static void flush_all(struct kmem_cache *s)
>   * Check if the objects in a per cpu structure fit numa
>   * locality expectations.
>   */
> -static inline int node_match(struct page *page, int node)
> +static inline int node_match(struct kmem_cache_cpu *c, int node)
>  {
>  #ifdef CONFIG_NUMA
> -	if (node != NUMA_NO_NODE && page_to_nid(page) != node)
> +	if (node != NUMA_NO_NODE && c->node != node)

We still have the issue of cpu fetching c->node before c->tid and
c->freelist.

I still believe the only solution is to prevent the task from migrating
via a preempt disable.

-- Steve


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>