Re: [PATCH] Cpuset: alloc_pages_node overrides cpuset constraints

[Andrew Morton]

Paul Jackson <pj@xxxxxxx> wrote:
>
> @@ -113,6 +114,11 @@ static inline struct page *alloc_pages_n
>   	/* Unknown node is current node */
>   	if (nid < 0)
>   		nid = numa_node_id();
>  +	/*
>  +	 * Specified (or implied by nid < 0) node overrides cpuset placement.
>  +	 * Various slab, page and device node specific allocations need this.
>  +	 */
>  +	gfp_mask |= __GFP_NOCPUSET;

You're kidding.  This adds more code to every page allocation on every
machine, cpusets or not.

I stuck this on top:

--- devel/include/linux/gfp.h~cpuset-alloc_pages_node-overrides-cpuset-constraints-speedup	2006-03-19 23:01:04.000000000 -0800
+++ devel-akpm/include/linux/gfp.h	2006-03-19 23:01:04.000000000 -0800
@@ -47,7 +47,11 @@ struct vm_area_struct;
 #define __GFP_ZERO	((__force gfp_t)0x8000u)/* Return zeroed page on success */
 #define __GFP_NOMEMALLOC ((__force gfp_t)0x10000u) /* Don't use emergency reserves */
 #define __GFP_HARDWALL   ((__force gfp_t)0x20000u) /* Enforce hardwall cpuset memory allocs */
+#ifdef CONFIG_CPUSETS
 #define __GFP_NOCPUSET	((__force gfp_t)0x40000u)/* Ignore cpuset constraints */
+#else
+#define __GFP_NOCPUSET	((__force gfp_t)0u)
+#endif
 
 #define __GFP_BITS_SHIFT 20	/* Room for 20 __GFP_FOO bits */
 #define __GFP_BITS_MASK ((__force gfp_t)((1 << __GFP_BITS_SHIFT) - 1))
_

But it's a bit ugly.
-
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/