cpumask: fix spurious cpumask_of_node() on non-NUMA multi-node configs
authorTejun Heo <tj@kernel.org>
Mon, 28 Aug 2017 21:51:27 +0000 (14:51 -0700)
committerBen Hutchings <ben@decadent.org.uk>
Sat, 11 Nov 2017 13:34:35 +0000 (13:34 +0000)
commitf26ddcdac427e02459ed6ec84dfde76719233756
treedf0e71e6f762a9247bff831d7dcc73d9e965484a
parent58ca58151b6b7c80d1b5628c6068f3b9c02b15ab
cpumask: fix spurious cpumask_of_node() on non-NUMA multi-node configs

commit b339752d054fb32863418452dff350a1086885b1 upstream.

When !NUMA, cpumask_of_node(@node) equals cpu_online_mask regardless of
@node.  The assumption seems that if !NUMA, there shouldn't be more than
one node and thus reporting cpu_online_mask regardless of @node is
correct.  However, that assumption was broken years ago to support
DISCONTIGMEM and whether a system has multiple nodes or not is
separately controlled by NEED_MULTIPLE_NODES.

This means that, on a system with !NUMA && NEED_MULTIPLE_NODES,
cpumask_of_node() will report cpu_online_mask for all possible nodes,
indicating that the CPUs are associated with multiple nodes which is an
impossible configuration.

This bug has been around forever but doesn't look like it has caused any
noticeable symptoms.  However, it triggers a WARN recently added to
workqueue to verify NUMA affinity configuration.

Fix it by reporting empty cpumask on non-zero nodes if !NUMA.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-and-tested-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
include/asm-generic/topology.h