Commit 7bebcf09 authored Jul 24, 2011 by Anton Blanchard Committed by Benjamin Herrenschmidt Sep 20, 2011

powerpc/numa: Disable NEWIDLE balancing at node level



On big POWER7 boxes we see large amounts of CPU time in system
processes like workqueue and watchdog kernel threads.

We currently rebalance the entire machine each time a task goes
idle and this is very expensive on large machines. Disable newidle
balancing at the node level and rely on the scheduler tick to
rebalance across nodes.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

parent d4761ad2

arch/powerpc/include/asm/topology.h

+1 −1

Original line number	Original line	Diff line number	Diff line
	@@ -75,7 +75,7 @@ static inline int pcibus_to_node(struct pci_bus *bus)
	.forkexec_idx = 0, \		.forkexec_idx = 0, \
	\		\
	.flags = 1*SD_LOAD_BALANCE \		.flags = 1*SD_LOAD_BALANCE \
	\| 1*SD_BALANCE_NEWIDLE \		\| 0*SD_BALANCE_NEWIDLE \
	\| 1*SD_BALANCE_EXEC \		\| 1*SD_BALANCE_EXEC \
	\| 1*SD_BALANCE_FORK \		\| 1*SD_BALANCE_FORK \
	\| 0*SD_BALANCE_WAKE \		\| 0*SD_BALANCE_WAKE \