Re: Regression in 2.6.23-rc2-mm2, mounting cpusets causes a hang

[Lee Schermerhorn]

Andrew:  this one should go with the other memoryless node patches.

Lee

---------------------------------------

PATCH	Initialize N_*_MEMORY and N_CPU masks for non-NUMA config

Against:  2.6.23-rc2-mm2

The N_*_MEMORY and N_CPU node state masks are not being initialized
for !NUMA because node_set_state() becomes a no-op for that config.
This causes one to hit a BUG_ON() attempting to mount cpusets to partition
cpus.  Discovered and diagnosed by Serge Hallyn and Dhaval Giani.

Statically initialize the N_*_MEMORY and N_CPU node state masks
for !NUMA configurations.  This static initialization is required
because the node_set_state() function becomes a no-op for !NUMA.
Other generic code assumes that these masks are set correctly.

Note that in NUMA configurations, these masks will be populated
correctly, so don't bother with static initialization.  No sense
in making assumptions that could be broken down the road, resulting
in extra work for someone to debug.  Unlikely, perhaps, but who
needs the aggravation...

Tested on i386 !NUMA by Dhaval Giani.  Fixes cpuset BUG.

Signed-off-by:  Lee Schermerhorn <lee.schermerhorn@xxxxxx>

 mm/page_alloc.c |    9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

Index: Linux/mm/page_alloc.c
===================================================================
--- Linux.orig/mm/page_alloc.c	2007-08-15 10:01:23.000000000 -0400
+++ Linux/mm/page_alloc.c	2007-08-15 10:05:41.000000000 -0400
@@ -52,7 +52,14 @@
  */
 nodemask_t node_states[NR_NODE_STATES] __read_mostly = {
 	[N_POSSIBLE] = NODE_MASK_ALL,
-	[N_ONLINE] = { { [0] = 1UL } }
+	[N_ONLINE] = { { [0] = 1UL } },
+#ifndef CONFIG_NUMA
+	[N_NORMAL_MEMORY] = { { [0] = 1UL } },
+#ifdef CONFIG_HIGHMEM
+	[N_HIGH_MEMORY] = { { [0] = 1UL } },
+#endif
+	[N_CPU] = { { [0] = 1UL } },
+#endif	/* NUMA */
 };
 EXPORT_SYMBOL(node_states);
 


-
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/