On 10/28/2016 03:19 AM, Will Deacon wrote:
On Tue, Oct 25, 2016 at 02:31:00PM -0700, David Daney wrote:
From: David Daney <david.daney@xxxxxxxxxx>
On arm64 NUMA kernels we can pass "numa=off" on the command line to
disable NUMA. A side effect of this is that kmalloc_node() calls to
non-zero nodes will crash the system with an OOPS:
[ 0.000000] [<fffffc00081bba84>] __alloc_pages_nodemask+0xa4/0xe68
[ 0.000000] [<fffffc00082163a8>] new_slab+0xd0/0x57c
[ 0.000000] [<fffffc000821879c>] ___slab_alloc+0x2e4/0x514
[ 0.000000] [<fffffc000823882c>] __slab_alloc+0x48/0x58
[ 0.000000] [<fffffc00082195a0>] __kmalloc_node+0xd0/0x2e0
[ 0.000000] [<fffffc00081119b8>] __irq_domain_add+0x7c/0x164
[ 0.000000] [<fffffc0008b75d30>] its_probe+0x784/0x81c
[ 0.000000] [<fffffc0008b75e10>] its_init+0x48/0x1b0
.
.
.
This is caused by code like this in kernel/irq/irqdomain.c
domain = kzalloc_node(sizeof(*domain) + (sizeof(unsigned int) * size),
GFP_KERNEL, of_node_to_nid(of_node));
When NUMA is disabled, the concept of a node is really undefined, so
of_node_to_nid() should unconditionally return NUMA_NO_NODE.
Add __of_force_no_numa() to allow of_node_to_nid() to be forced to
return NUMA_NO_NODE.
The follow on patch will call this new function from the arm64 numa
code.
Reported-by: Gilbert Netzer <noname@xxxxxxxxxx>
Signed-off-by: David Daney <david.daney@xxxxxxxxxx>
---
drivers/of/of_numa.c | 15 +++++++++++++++
include/linux/of.h | 2 ++
2 files changed, 17 insertions(+)
diff --git a/drivers/of/of_numa.c b/drivers/of/of_numa.c
index f63d4b0d..2212299 100644
--- a/drivers/of/of_numa.c
+++ b/drivers/of/of_numa.c
@@ -150,12 +150,27 @@ static int __init of_numa_parse_distance_map(void)
return ret;
}
+static bool of_force_no_numa;
+
+void __of_force_no_numa(void)
+{
+ of_force_no_numa = true;
+}
+
int of_node_to_nid(struct device_node *device)
{
struct device_node *np;
u32 nid;
int r = -ENODATA;
+ /*
+ * If NUMA forced off, nodes are meaningless. Return
+ * NUMA_NO_NODE so that any node specific memory allocations
+ * can succeed from the default pool.
+ */
+ if (of_force_no_numa)
+ return NUMA_NO_NODE;
Why don't you just check if the nid you get back from the device is set in
numa_nodes_parsed and return NUMA_NO_NODE if not?
numa_nodes_parsed is __initdata. Perhaps node_possible_map would be better.
I will try that.
David.
Will
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html