datapath: backport: ovs: do not allocate memory from offline numa node
authorPravin B Shelar <pshelar@ovn.org>
Sun, 17 Jul 2016 16:52:11 +0000 (09:52 -0700)
committerPravin B Shelar <pshelar@ovn.org>
Sun, 17 Jul 2016 17:25:09 +0000 (10:25 -0700)
commit96e72484ee7ff811b47bc765535fea729bfef83a
tree7d525ece7ab467858e9554579d77da653969db3c
parent2250cb8b4fc4d25e97e1bd609a0e380dab73ab22
datapath: backport: ovs: do not allocate memory from offline numa node

Upstream commit:
    commit 598c12d0ba6de9060f04999746eb1e015774044b
    Author: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>

    ovs: do not allocate memory from offline numa node

    When openvswitch tries allocate memory from offline numa node 0:
    stats = kmem_cache_alloc_node(flow_stats_cache, GFP_KERNEL | __GFP_ZERO, 0)
    It catches VM_BUG_ON(nid < 0 || nid >= MAX_NUMNODES || !node_online(nid))
    [ replaced with VM_WARN_ON(!node_online(nid)) recently ] in linux/gfp.h
    This patch disables numa affinity in this case.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Pravin B Shelar <pshelar@ovn.org>
Acked-by: Jesse Gross <jesse@kernel.org>
datapath/flow_table.c