2022-07-06 21:00:25

by Steven Rostedt

Subject: [PATCH] ftrace: Be more specific about arch impact when function tracer is enabled

From: Steven Rostedt (Google) <[email protected]>

It was brought up that on ARMv7, that because the FUNCTION_TRACER does not
use nops to keep function tracing disabled because of the use of a link
register, it does have some performance impact.

The start of functions when -pg is used to compile the kernel is:

push {lr}
bl 8010e7c0 <__gnu_mcount_nc>

When function tracing is tuned off, it becomes:

push {lr}
add sp, sp, #4

Which just puts the stack back to its normal location. But these two
instructions at the start of every function does incur some overhead.

Be more honest in the Kconfig FUNCTION_TRACER description and specify that
the overhead being in the noise was x86 specific, but other architectures
may vary.

Link: https://lore.kernel.org/all/[email protected]/

Reported-by: Sascha Hauer <[email protected]>
Signed-off-by: Steven Rostedt (Google) <[email protected]>
---
diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
index debbbb083286..ccd6a5ade3e9 100644
--- a/kernel/trace/Kconfig
+++ b/kernel/trace/Kconfig
@@ -194,7 +194,8 @@ config FUNCTION_TRACER
sequence is then dynamically patched into a tracer call when
tracing is enabled by the administrator. If it's runtime disabled
(the bootup default), then the overhead of the instructions is very
- small and not measurable even in micro-benchmarks.
+ small and not measurable even in micro-benchmarks (at least on
+ x86, but may have impact on other architectures).

config FUNCTION_GRAPH_TRACER
bool "Kernel Function Graph Tracer"

2022-07-11 11:29:46

by Sascha Hauer

Subject: Re: [PATCH] ftrace: Be more specific about arch impact when function tracer is enabled

On Wed, Jul 06, 2022 at 04:12:31PM -0400, Steven Rostedt wrote:
> From: Steven Rostedt (Google) <[email protected]>
>
> It was brought up that on ARMv7, that because the FUNCTION_TRACER does not
> use nops to keep function tracing disabled because of the use of a link
> register, it does have some performance impact.
>
> The start of functions when -pg is used to compile the kernel is:
>
> push {lr}
> bl 8010e7c0 <__gnu_mcount_nc>
>
> When function tracing is tuned off, it becomes:
>
> push {lr}
> add sp, sp, #4
>
> Which just puts the stack back to its normal location. But these two
> instructions at the start of every function does incur some overhead.
>
> Be more honest in the Kconfig FUNCTION_TRACER description and specify that
> the overhead being in the noise was x86 specific, but other architectures
> may vary.
>
> Link: https://lore.kernel.org/all/[email protected]/
>
> Reported-by: Sascha Hauer <[email protected]>
> Signed-off-by: Steven Rostedt (Google) <[email protected]>
> ---

Thanks Steven

Acked-by: Sascha Hauer <[email protected]>

Sascha

> diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
> index debbbb083286..ccd6a5ade3e9 100644
> --- a/kernel/trace/Kconfig
> +++ b/kernel/trace/Kconfig
> @@ -194,7 +194,8 @@ config FUNCTION_TRACER
> sequence is then dynamically patched into a tracer call when
> tracing is enabled by the administrator. If it's runtime disabled
> (the bootup default), then the overhead of the instructions is very
> - small and not measurable even in micro-benchmarks.
> + small and not measurable even in micro-benchmarks (at least on
> + x86, but may have impact on other architectures).
>
> config FUNCTION_GRAPH_TRACER
> bool "Kernel Function Graph Tracer"
>

--
Pengutronix e.K. | |
Steuerwalder Str. 21 | http://www.pengutronix.de/ |
31137 Hildesheim, Germany | Phone: +49-5121-206917-0 |
Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |