Messages in this thread Patch in this message | | | Subject | Re: [PATCH] slab: minor cleanup to kmem_cache_alloc_node | From | Pekka Enberg <> | Date | Mon, 21 Nov 2005 20:36:34 +0200 |
| |
On Mon, 2005-11-21 at 09:21 -0800, Christoph Lameter wrote: > On Mon, 21 Nov 2005, Pekka J Enberg wrote: > > > This patch gets rid of one if-else statement by moving current node allocation > > check at the beginning of kmem_cache_alloc_node(). > > The problem with this is that the numa_node may change if irqs are still > active and your patch moves the check for the numa node outside of the > section where irqs are enabled. > > You could move the check for -1 into the section where interrupts are > disabled.
So we could do something like the following. Unfortunately it does not seem much of an improvement... Thoughts?
Pekka
Index: 2.6/mm/slab.c =================================================================== --- 2.6.orig/mm/slab.c +++ 2.6/mm/slab.c @@ -2866,21 +2866,23 @@ void *kmem_cache_alloc_node(kmem_cache_t unsigned long save_flags; void *ptr; - if (nodeid == -1) - return __cache_alloc(cachep, flags); + cache_alloc_debugcheck_before(cachep, flags); + local_irq_save(save_flags); + + if (nodeid == -1 || nodeid == numa_node_id()) { + ptr = ____cache_alloc(cachep, flags); + goto out; + } if (unlikely(!cachep->nodelists[nodeid])) { /* Fall back to __cache_alloc if we run into trouble */ printk(KERN_WARNING "slab: not allocating in inactive node %d for cache %s\n", nodeid, cachep->name); - return __cache_alloc(cachep,flags); + ptr = ____cache_alloc(cachep,flags); + goto out; } - cache_alloc_debugcheck_before(cachep, flags); - local_irq_save(save_flags); - if (nodeid == numa_node_id()) - ptr = ____cache_alloc(cachep, flags); - else - ptr = __cache_alloc_node(cachep, flags, nodeid); + ptr = __cache_alloc_node(cachep, flags, nodeid); + out: local_irq_restore(save_flags); ptr = cache_alloc_debugcheck_after(cachep, flags, ptr, __builtin_return_address(0));
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |