On Wed, Feb 22, 2012 at 1:53 PM, Rafael Aquini <aquini@xxxxxxxxxx> wrote:
> Adds a new sysctl, 'oom_dump_slabs', that enables the kernel to produce a
> dump of all eligible system slab caches when performing an OOM-killing.
> Information includes per cache active objects, total objects, object size,
> cache name and cache size.
>
> The eligibility for being reported is given by an auxiliary sysctl,
> 'oom_dump_slabs_ratio', which express (in percentage) the memory committed
> ratio between a particular cache size and the total slab size.
>
> This, alongside with all other data dumped in OOM events, is very helpful
> information in diagnosing why there was an OOM condition specially when
> kernel code is under investigation.
>
> Signed-off-by: Rafael Aquini <aquini@xxxxxxxxxx>
Makes sense. Do you have an example how an out-of-memory slab cache
dump looks like?
> diff --git a/mm/slab.c b/mm/slab.c
> index f0bd785..c2b5d14 100644
> --- a/mm/slab.c
> +++ b/mm/slab.c
> @@ -4629,3 +4629,75 @@ size_t ksize(const void *objp)
> return obj_size(virt_to_cache(objp));
> }
> EXPORT_SYMBOL(ksize);
> +
> +/**
> + * oom_dump_slabs - dump top slab cache users
> + * @ratio: memory committed ratio between a cache size and the total slab size
> + *
> + * Dumps the current memory state of all eligible slab caches.
> + * State information includes cache's active objects, total objects,
> + * object size, cache name, and cache size.
> + */
> +void oom_dump_slabs(int ratio)
> +{
> + struct kmem_cache *cachep;
> + struct kmem_list3 *l3;
> + struct slab *slabp;
> + unsigned long active_objs, num_objs, free_objects, cache_size;
> + unsigned long active_slabs, num_slabs, slab_total_mem;
> + int node;
[snip]
> + list_for_each_entry(cachep, &cache_chain, next) {
> + num_objs = 0;
> + num_slabs = 0;
> + active_objs = 0;
> + free_objects = 0;
> + active_slabs = 0;
Minor style nit: just define the zeroed variables in this block.
> +
> + for_each_online_node(node) {
> + l3 = cachep->nodelists[node];
> + if (!l3)
> + continue;
> +void oom_dump_slabs(int ratio)
> +{
> + unsigned long cache_size, slab_total_mem;
> + unsigned long nr_objs, nr_free, nr_inuse;
> + struct kmem_cache *cachep;
> + int node;
> +
> + slab_total_mem = (global_page_state(NR_SLAB_RECLAIMABLE) +
> + global_page_state(NR_SLAB_UNRECLAIMABLE)) << PAGE_SHIFT;
> +
> + if (ratio < 0)
> + ratio = 0;
> +
> + if (ratio > 100)
> + ratio = 100;
> +
> + pr_info("--- oom_dump_slabs:\n");
> + pr_info("<active_objs> <num_objs> <objsize> <cache_name>\n");
> + down_read(&slub_lock);
> + list_for_each_entry(cachep, &slab_caches, list) {
> + nr_objs = 0;
> + nr_free = 0;
ditto.
> +
> + for_each_online_node(node) {
> + struct kmem_cache_node *n = get_node(cachep, node);
> + if (!n)
> + continue;
> +
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href