[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1187127432.6281.135.camel@localhost>
Date: Tue, 14 Aug 2007 17:37:12 -0400
From: Lee Schermerhorn <Lee.Schermerhorn@...com>
To: "Serge E. Hallyn" <serge@...lyn.com>
Cc: Dhaval Giani <dhaval@...ux.vnet.ibm.com>, clameter@....com,
bob.picco@...com, nacc@...ibm.com, kamezawa.hiroyu@...fujitsu.com,
mel@...net.ie, akpm@...ux-foundation.org,
Balbir Singh <balbir@...ibm.com>,
Srivatsa Vaddagiri <vatsa@...ux.vnet.ibm.com>,
lkml <linux-kernel@...r.kernel.org>,
ckrm-tech <ckrm-tech@...ts.sourceforge.net>
Subject: Re: Regression in 2.6.23-rc2-mm2, mounting cpusets causes a hang
On Tue, 2007-08-14 at 15:49 -0500, Serge E. Hallyn wrote:
<snip>
>
> Here is a patch which fixes the bug on my system. As noted in
> the patch description, there are other calls to node_set_state() in
> mm/page_alloc.c which might also need to be switched to node_set(),
> if this is deemed the right solution.
Here's an update to that patch to handle the other 2 places I found
where node_set_state() would be a no-op when MAX_NUMNODES == 1.
I'm continuing to look at uses of node_online_map throughout the kernel.
I'll look at current usage of the node state wrappers as well. However,
you probably want to fix these other 2 places as well, or you might run
into other issues with empty 'NORMAL_MEMORY and 'CPU node state masks...
Lee
[PATCH 1/1] memoryless: actually set node_states[N_HIGH_MEMORY] - v2
v1 -> v2"
+ switch to node_set() to initialize N_NORMAL_MEMORY and N_CPU
node_states[] as well.
If CONFIG_NODES_SHIFT is unset, then MAX_NUMNODES=1, and
node_set_state() does nothing. This means that __build_all_zonelists()
does not actually set node_states[N_HIGH_MEMORY]. cpusets needs
node_states[N_HIGH_MEMORY] to be correctly set, or it will raise a
bug when a new cpuset is created.
Switch to using node_set() in __build_all_zonelists() to
make sure it is actually set.
Note there are other calls to node_set_state() in mm/page_alloc.c,
which perhaps should also be modified. Since this area is completely
foreign to me I thought I'd first make sure this is the right approach.
addressed in v2 --lts
Signed-off-by: Serge E. Hallyn <serue@...ibm.com>
Signed-off-by: Dhaval Giani <dhaval@...ux.vnet.ibm.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@...com>
mm/page_alloc.c | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)
Index: Linux/mm/page_alloc.c
===================================================================
--- Linux.orig/mm/page_alloc.c 2007-08-14 16:48:28.000000000 -0400
+++ Linux/mm/page_alloc.c 2007-08-14 16:57:24.000000000 -0400
@@ -2425,7 +2425,8 @@ static void check_for_regular_memory(pg_
for (zone_type = 0; zone_type <= ZONE_NORMAL; zone_type++) {
struct zone *zone = &pgdat->node_zones[zone_type];
if (zone->present_pages)
- node_set_state(zone_to_nid(zone), N_NORMAL_MEMORY);
+ node_set(zone_to_nid(zone),
+ node_states[N_NORMAL_MEMORY]);
}
#endif
}
@@ -2443,7 +2444,7 @@ static int __build_all_zonelists(void *d
/* Any memory on that node */
if (pgdat->node_present_pages)
- node_set_state(nid, N_HIGH_MEMORY);
+ node_set(nid, node_states[N_HIGH_MEMORY]);
check_for_regular_memory(pgdat);
}
return 0;
@@ -2793,7 +2794,7 @@ static int __cpuinit process_zones(int c
(zone->present_pages / percpu_pagelist_fraction));
}
- node_set_state(node, N_CPU);
+ node_set(node, node_states[N_CPU]);
return 0;
bad:
for_each_zone(dzone) {
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists