On Thu, 8 Nov 2018 09:12:04 +0800 Wei Yang <richard.weiyang@xxxxxxxxx> wrote:
> for_each_zone_zonelist() iterates the zonelist one by one, which means
> it will iterate on zones on the same node. While get_partial_node()
> checks available slab on node base instead of zone.
>
> This patch skip a node in case get_partial_node() fails to acquire slab
> on that node.
This is rather hard to follow.
I *think* the patch is a performance optimization: prevent
get_any_partial() from checking a node which get_partial_node() has
already looked at?
Could we please have a more complete changelog?
> --- a/mm/slub.c
> +++ b/mm/slub.c
> @@ -1873,7 +1873,7 @@ static void *get_partial_node(struct kmem_cache *s, struct kmem_cache_node *n,
> * Get a page from somewhere. Search in increasing NUMA distances.
> */
> static void *get_any_partial(struct kmem_cache *s, gfp_t flags,
> - struct kmem_cache_cpu *c)
> + struct kmem_cache_cpu *c, int except)
> {
> #ifdef CONFIG_NUMA
> struct zonelist *zonelist;
> @@ -1882,6 +1882,9 @@ static void *get_any_partial(struct kmem_cache *s, gfp_t flags,
> enum zone_type high_zoneidx = gfp_zone(flags);
> void *object;
> unsigned int cpuset_mems_cookie;
> + nodemask_t nmask = node_states[N_MEMORY];
> +
> + node_clear(except, nmask);
And please add a comment describing what's happening here and why it is
done. Adding a sentence to the block comment over get_any_partial()
would be suitable.
