From: Dimitri Sivanich Date: Wed, 15 Apr 2009 15:56:25 +0000 (-0500) Subject: ia64: smp_flush_tlb_mm() should only send IPI's to cpus in cpu_vm_mask Patch-mainline: 2.6.30-rc3 Git-commit: edb91dc01a216e84b78721b71a06db1e0db141b7 References: bnc#497807 [IA64] smp_flush_tlb_mm() should only send IPI's to cpus in cpu_vm_mask Having flush_tlb_mm->smp_flush_tlb_mm() send an IPI to every cpu on the system is occasionally triggering spin_lock contention in generic_smp_call_function_interrupt(). Follow x86 arch's lead and only sends IPIs to the cpus in mm->cpu_vm_mask. Experiments with this change have shown significant improvement in this contention issue. Signed-off-by: Dimitri Sivanich Signed-off-by: Tony Luck Acked-by: Jeff Mahoney --- arch/ia64/kernel/smp.c | 13 +++++-------- 1 file changed, 5 insertions(+), 8 deletions(-) --- a/arch/ia64/kernel/smp.c +++ b/arch/ia64/kernel/smp.c @@ -300,15 +300,12 @@ smp_flush_tlb_mm (struct mm_struct *mm) return; } + smp_call_function_mask(mm->cpu_vm_mask, + (void (*)(void *))local_finish_flush_tlb_mm, mm, 1); + local_irq_disable(); + local_finish_flush_tlb_mm(mm); + local_irq_enable(); preempt_enable(); - /* - * We could optimize this further by using mm->cpu_vm_mask to track which CPUs - * have been running in the address space. It's not clear that this is worth the - * trouble though: to avoid races, we have to raise the IPI on the target CPU - * anyhow, and once a CPU is interrupted, the cost of local_flush_tlb_all() is - * rather trivial. - */ - on_each_cpu((void (*)(void *))local_finish_flush_tlb_mm, mm, 1); } void arch_send_call_function_single_ipi(int cpu)