Re: [BUG] commit 3c517a61, slab: better fallback allocation behavior

[Christoph Lameter <clameter@sgi.com>]

Ahh. Fallback_alloc() does not do the check for GFP_WAIT as done in 
cache_grow(). Thus interrupts are disabled when we call kmem_getpages() 
which results in the failure.

Duplicate the handling of GFP_WAIT in cache_grow().

Jay could you try this patch?

Signed-off-by: Christoph Lameter <clameter@sgi.com>

Index: linux-2.6/mm/slab.c
===================================================================
--- linux-2.6.orig/mm/slab.c	2006-12-11 09:25:57.000000000 -0800
+++ linux-2.6/mm/slab.c	2006-12-11 09:34:08.000000000 -0800
@@ -3252,6 +3252,7 @@
 	struct zone **z;
 	void *obj = NULL;
 	int nid;
+	gfp_t local_flags = (flags & GFP_LEVEL_MASK);
 
 retry:
 	/*
@@ -3275,7 +3276,12 @@
 		 * We may trigger various forms of reclaim on the allowed
 		 * set and go into memory reserves if necessary.
 		 */
+		if (local_flags & __GFP_WAIT)
+			local_irq_enable();
+		kmem_flagcheck(cache, flags);
 		obj = kmem_getpages(cache, flags, -1);
+		if (local_flags & __GFP_WAIT)
+			local_irq_disable();
 		if (obj) {
 			/*
 			 * Insert into the appropriate per node queues