From a2bc4e88b723c36d4d8ff16311b9b94d9a60f5e2 Mon Sep 17 00:00:00 2001 From: Allen Yu Date: Fri, 11 Sep 2015 10:06:20 -0700 Subject: watchdog: t18x: refactor WDT affinity and on/off 1. Use cluster ID instead of CPU ID for WDT instances since CPU index can vary depending on the configuration. Essentially, each WDT instance is responsible for the associated cluster. 2. Bind WDT IRQ and WDT enablement to the first CPU of associated cluster. The CPU is representive of cluster health and responsible for WDT petting. 3. Remove the cpu notifier. To be safe, WDTs should always be active to detect any system lockup condition, irrespective of cpu/cluster power states. Deep sleep is an exception since WDT/timer blocks are powered off in this state. Bug 1617276 Change-Id: I9d1b2d8dedb62f9cb9d07d70801c8e51e46a66c1 Signed-off-by: Allen Yu Reviewed-on: http://git-master/r/796054 GVS: Gerrit_Virtual_Submit Reviewed-by: Laxman Dewangan --- include/linux/platform/tegra/tegra18_cpu_map.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/linux') diff --git a/include/linux/platform/tegra/tegra18_cpu_map.h b/include/linux/platform/tegra/tegra18_cpu_map.h index ee6924483..9b79700e1 100644 --- a/include/linux/platform/tegra/tegra18_cpu_map.h +++ b/include/linux/platform/tegra/tegra18_cpu_map.h @@ -1,6 +1,7 @@ #include #include #include +#include static inline u8 tegra18_logical_to_cluster(u8 cpu) { return MPIDR_AFFINITY_LEVEL(cpu_logical_map(cpu), 1); -- cgit v1.2.2