ARM: Stop irqsoff trace on return to user
PATCH 1/1 ARM: Stop irqsoff trace on return to user
If the irqsoff tracer is in use, stop tracing the interrupt disable interval when returning to userspace. Tracing userspace execution time as interrupts disabled time is not useful for kernel performance analysis purposes, and crowds out more helpful results with long userspace execution times. Some other architectures do something similar.
Only do so if the irqsoff tracer is enabled, to avoid overhead for lockdep, which doesn't care about userspace execution time with interrupts disabled.
--- arch/arm/kernel/entry-common.S | 6 ++++++ 1 files changed, 6 insertions(+), 0 deletions(-)
arch/arm/kernel/entry-common.S +6 -0
@@ -29,6 +29,9 @@ ret_fast_syscall: ldr r1, [tsk, #TI_FLAGS] tst r1, #_TIF_WORK_MASK bne fast_work_pending +#if defined(CONFIG_IRQSOFF_TRACER) + asm_trace_hardirqs_on +#endif /* perform architecture specific actions before user return */ arch_ret_to_user r1, lr
@@ -65,6 +68,9 @@ ret_slow_syscall: tst r1, #_TIF_WORK_MASK bne work_pending no_work_pending: +#if defined(CONFIG_IRQSOFF_TRACER) + asm_trace_hardirqs_on +#endif /* perform architecture specific actions before user return */ arch_ret_to_user r1, lr
Signature · 1 line
1.7.3.1
PATCH 1/1 ARM: Stop irqsoff trace on return to user
Quoted 10 lines · Todd Poynor
If the irqsoff tracer is in use, stop tracing the interrupt disable interval when returning to userspace. Tracing userspace execution time as interrupts disabled time is not useful for kernel performance analysis purposes, and crowds out more helpful results with long userspace execution times. Some other architectures do something similar.
Only do so if the irqsoff tracer is enabled, to avoid overhead for lockdep, which doesn't care about userspace execution time with interrupts disabled.
Ok, can you send this to the patch system please?