2019-02-12 11:27:16

by Mars Cheng

Subject: [PATCH 1/1] arm64: Use _rcuidle for smp_cross_call() tracepoints

fix similar issue for aarch64 as the commit 7c64cc0531fa
("arm: Use _rcuidle for smp_cross_call() tracepoints") fixed
in aarch32

Signed-off-by: Mars Cheng <[email protected]>
---
arch/arm64/kernel/smp.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm64/kernel/smp.c b/arch/arm64/kernel/smp.c
index 1598d6f..c1cf8f9 100644
--- a/arch/arm64/kernel/smp.c
+++ b/arch/arm64/kernel/smp.c
@@ -758,7 +758,7 @@ void __init set_smp_cross_call(void (*fn)(const struct cpumask *, unsigned int))

static void smp_cross_call(const struct cpumask *target, unsigned int ipinr)
{
- trace_ipi_raise(target, ipi_types[ipinr]);
+ trace_ipi_raise_rcuidle(target, ipi_types[ipinr]);
__smp_cross_call(target, ipinr);
}

--
1.7.9.5

2019-02-13 14:36:21

by Will Deacon

Subject: Re: [PATCH 1/1] arm64: Use _rcuidle for smp_cross_call() tracepoints

On Tue, Feb 12, 2019 at 07:09:25PM +0800, Mars Cheng wrote:
> fix similar issue for aarch64 as the commit 7c64cc0531fa
> ("arm: Use _rcuidle for smp_cross_call() tracepoints") fixed
> in aarch32

I'm not convinced about this justification. As far as I can tell, the commit
mentioned above fixed a cross-call from the idle path as a result of
ARCH_NEEDS_CPU_IDLE_COUPLED, but this never appears to be selected on arm64
afaict.

Are you actually seeing a problem on arm64 in practice?

Will