powerpc/pseries: RE-enable Virtual Processor Home Node updating
authorJesse Larrew <jlarrew@linux.vnet.ibm.com>
Wed, 24 Apr 2013 06:05:22 +0000 (06:05 +0000)
committerBenjamin Herrenschmidt <benh@kernel.crashing.org>
Fri, 26 Apr 2013 06:08:25 +0000 (16:08 +1000)
The new PRRN firmware feature provides a more convenient and event-driven
interface than VPHN for notifying Linux of changes to the NUMA affinity of
platform resources. However, for practical reasons, it may not be feasible
for some customers to update to the latest firmware. For these customers,
the VPHN feature supported on previous firmware versions may still be the
best option.

The VPHN feature was previously disabled due to races with the load
balancing code when accessing the NUMA cpu maps, but the new stop_machine()
approach protects the NUMA cpu maps from these concurrent accesses. It
should be safe to re-enable this feature now.

Signed-off-by: Nathan Fontenot <nfont@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
arch/powerpc/mm/numa.c

index f298a7d4b2ce578eeaee509c2e2153efd9c3ccdc..908699fb61c9ccf6196ccc2f5eb748220c3b5990 100644 (file)
@@ -1572,9 +1572,8 @@ int start_topology_update(void)
                        vphn_enabled = 0;
                        rc = of_reconfig_notifier_register(&dt_update_nb);
                }
-       } else if (0 && firmware_has_feature(FW_FEATURE_VPHN) &&
+       } else if (firmware_has_feature(FW_FEATURE_VPHN) &&
                   get_lppaca()->shared_proc) {
-               /* Disabled until races with load balancing are fixed */
                if (!vphn_enabled) {
                        prrn_enabled = 0;
                        vphn_enabled = 1;