[PATCH] slab: fix kmalloc regression with big constant allocations

[Glauber Costa]

kmalloc have a maximum allocation size, but we are currently not
respecting it. We create a list of kmalloc sizes and return an array
index up to 26, which may or may not be within our limits, since this is
architecture dependent.

This patch fix this by making the slab code do the same as slub does: An
explicit check for the maximum size before the call to kmalloc_index,
and the use of the kmalloc non-constant fallback function in that case.

Signed-off-by: Glauber Costa <glommer@xxxxxxxxxx>
---
 include/linux/slab_def.h | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/include/linux/slab_def.h b/include/linux/slab_def.h
index 113ec08..3a240c8 100644
--- a/include/linux/slab_def.h
+++ b/include/linux/slab_def.h
@@ -172,8 +172,10 @@ static __always_inline void *kmalloc_node(size_t size, gfp_t flags, int node)
 		if (!size)
 			return ZERO_SIZE_PTR;
 
-		i = kmalloc_index(size);
+		if (size > KMALLOC_MAX_SIZE)
+			goto not_found;
 
+		i = kmalloc_index(size);
 #ifdef CONFIG_ZONE_DMA
 		if (flags & GFP_DMA)
 			cachep = kmalloc_dma_caches[i];
@@ -183,6 +185,7 @@ static __always_inline void *kmalloc_node(size_t size, gfp_t flags, int node)
 
 		return kmem_cache_alloc_node_trace(cachep, flags, node, size);
 	}
+not_found:
 	return __kmalloc_node(size, flags, node);
 }
 
-- 
1.8.1.4


--------------000009000007080109040608--
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/