mmotm 2015-01-22-15-04: qemu failure due to 'mm: memcontrol: remove unnecessary soft limit tree node test'
diff mbox

Message ID 20150123173659.GB12036@phnom.home.cmpxchg.org
State New, archived
Headers show

Commit Message

Johannes Weiner Jan. 23, 2015, 5:36 p.m. UTC
On Fri, Jan 23, 2015 at 08:59:51AM -0800, Guenter Roeck wrote:
> On 01/23/2015 08:02 AM, Johannes Weiner wrote:
> >On Fri, Jan 23, 2015 at 09:17:44AM -0600, Christoph Lameter wrote:
> >>On Fri, 23 Jan 2015, Johannes Weiner wrote:
> >>
> >>>Is the assumption of this patch wrong?  Does the specified node have
> >>>to be online for the fallback to work?
> >>
> >>Nodes that are offline have no control structures allocated and thus
> >>allocations will likely segfault when the address of the controls
> >>structure for the node is accessed.
> >>
> >>If we wanted to prevent that then every allocation would have to add a
> >>check to see if the nodes are online which would impact performance.
> >
> >Okay, that makes sense, thank you.
> >
> >Andrew, can you please drop this patch?
> >
> Problem is that there are three patches.
> 
> 2537ffb mm: memcontrol: consolidate swap controller code
> 2f9b346 mm: memcontrol: consolidate memory controller initialization
> a40d0d2 mm: memcontrol: remove unnecessary soft limit tree node test
> 
> Reverting (or dropping) a40d0d2 alone is not possible since it modifies
> mem_cgroup_soft_limit_tree_init which is removed by 2f9b346.

("mm: memcontrol: consolidate swap controller code") gave me no issues
when rebasing, but ("mm: memcontrol: consolidate memory controller
initialization") needs updating.

So how about this one to replace ("mm: memcontrol: remove unnecessary
soft limit tree node test"):

---
From: Johannes Weiner <hannes@cmpxchg.org>
Subject: [patch] mm: memcontrol: simplify soft limit tree init code

- No need to test the node for N_MEMORY.  node_online() is enough for
  node fallback to work in slab, use NUMA_NO_NODE for everything else.

- Remove the BUG_ON() for allocation failure.  A NULL pointer crash is
  just as descriptive, and the absent return value check is obvious.

- Move local variables to the inner-most blocks.

- Point to the tree structure after its initialized, not before, it's
  just more logical that way.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
---
 mm/memcontrol.c | 17 ++++++++---------
 1 file changed, 8 insertions(+), 9 deletions(-)

Comments

Guenter Roeck Jan. 24, 2015, 2:02 a.m. UTC | #1
On 01/23/2015 09:36 AM, Johannes Weiner wrote:
> On Fri, Jan 23, 2015 at 08:59:51AM -0800, Guenter Roeck wrote:
>> On 01/23/2015 08:02 AM, Johannes Weiner wrote:
>>> On Fri, Jan 23, 2015 at 09:17:44AM -0600, Christoph Lameter wrote:
>>>> On Fri, 23 Jan 2015, Johannes Weiner wrote:
>>>>
>>>>> Is the assumption of this patch wrong?  Does the specified node have
>>>>> to be online for the fallback to work?
>>>>
>>>> Nodes that are offline have no control structures allocated and thus
>>>> allocations will likely segfault when the address of the controls
>>>> structure for the node is accessed.
>>>>
>>>> If we wanted to prevent that then every allocation would have to add a
>>>> check to see if the nodes are online which would impact performance.
>>>
>>> Okay, that makes sense, thank you.
>>>
>>> Andrew, can you please drop this patch?
>>>
>> Problem is that there are three patches.
>>
>> 2537ffb mm: memcontrol: consolidate swap controller code
>> 2f9b346 mm: memcontrol: consolidate memory controller initialization
>> a40d0d2 mm: memcontrol: remove unnecessary soft limit tree node test
>>
>> Reverting (or dropping) a40d0d2 alone is not possible since it modifies
>> mem_cgroup_soft_limit_tree_init which is removed by 2f9b346.
>
> ("mm: memcontrol: consolidate swap controller code") gave me no issues
> when rebasing, but ("mm: memcontrol: consolidate memory controller
> initialization") needs updating.
>
> So how about this one to replace ("mm: memcontrol: remove unnecessary
> soft limit tree node test"):
>
> ---
> From: Johannes Weiner <hannes@cmpxchg.org>
> Subject: [patch] mm: memcontrol: simplify soft limit tree init code
>
> - No need to test the node for N_MEMORY.  node_online() is enough for
>    node fallback to work in slab, use NUMA_NO_NODE for everything else.
>
> - Remove the BUG_ON() for allocation failure.  A NULL pointer crash is
>    just as descriptive, and the absent return value check is obvious.
>
> - Move local variables to the inner-most blocks.
>
> - Point to the tree structure after its initialized, not before, it's
>    just more logical that way.
>
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

The latest version in mmotm passes my ppc64 qemu test, so it works
at least in this context.

Tested-by: Guenter Roeck <linux@roeck-us.net>

--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch
diff mbox

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index fb9788af4a3e..88c67303d141 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -4537,24 +4537,23 @@  EXPORT_SYMBOL(parent_mem_cgroup);
 
 static void __init mem_cgroup_soft_limit_tree_init(void)
 {
-	struct mem_cgroup_tree_per_node *rtpn;
-	struct mem_cgroup_tree_per_zone *rtpz;
-	int tmp, node, zone;
+	int node;
 
 	for_each_node(node) {
-		tmp = node;
-		if (!node_state(node, N_NORMAL_MEMORY))
-			tmp = -1;
-		rtpn = kzalloc_node(sizeof(*rtpn), GFP_KERNEL, tmp);
-		BUG_ON(!rtpn);
+		struct mem_cgroup_tree_per_node *rtpn;
+		int zone;
 
-		soft_limit_tree.rb_tree_per_node[node] = rtpn;
+		rtpn = kzalloc_node(sizeof(*rtpn), GFP_KERNEL,
+				    node_online(node) ? node : NUMA_NO_NODE);
 
 		for (zone = 0; zone < MAX_NR_ZONES; zone++) {
+			struct mem_cgroup_tree_per_zone *rtpz;
+
 			rtpz = &rtpn->rb_tree_per_zone[zone];
 			rtpz->rb_root = RB_ROOT;
 			spin_lock_init(&rtpz->lock);
 		}
+		soft_limit_tree.rb_tree_per_node[node] = rtpn;
 	}
 }