[PATCH] percpu: fix pcpu_last_unit_cpu

[Tejun Heo]

pcpu_first/last_unit_cpu are used to track which cpu has the first and
last units assigned.  This in turn is used to determine the span of a
chunk for man/unmap cache flushes and whether an address belongs to
the first chunk or not in per_cpu_ptr_to_phys().

When the number of possible CPUs isn't power of two, a chunk may
contain unassigned units towards the end of a chunk.  The logic to
determine pcpu_last_unit_cpu was incorrect when there was an unused
unit at the end of a chunk.  It failed to ignore the unused unit and
assigned the unused marker NR_CPUS to pcpu_last_unit_cpu.

This was discovered through kdump failure which was caused by
malfunctioning per_cpu_ptr_to_phys() on a kvm setup with 50 possible
CPUs by CAI Qian.

Signed-off-by: Tejun Heo <tj at kernel.org>
Reported-by: CAI Qian <caiqian at redhat.com>
Cc: stable at kernel.org
---
This is pretty subtle and has the potential to cause mysterious
problems on vcache architectuers.  Thanks a lot for discovering it and
your patience.  I'll put it out in linux-next for a couple of days and
push it to Linus and stable.

Thanks.

 mm/percpu.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/mm/percpu.c b/mm/percpu.c
index 58c572b..c76ef38 100644
--- a/mm/percpu.c
+++ b/mm/percpu.c
@@ -1401,9 +1401,9 @@ int __init pcpu_setup_first_chunk(const struct pcpu_alloc_info *ai,

 			if (pcpu_first_unit_cpu == NR_CPUS)
 				pcpu_first_unit_cpu = cpu;
+			pcpu_last_unit_cpu = cpu;
 		}
 	}
-	pcpu_last_unit_cpu = cpu;
 	pcpu_nr_units = unit;

 	for_each_possible_cpu(cpu)
-- 
1.7.1