Re: [PING PATCH] perf/test: fix perf ftrace test on s390

[James Clark]

On 14/11/2024 9:01 am, Thomas Richter wrote:
> On s390 the perf test case ftrace sometimes fails as follows:
>
>    # ./perf test ftrace
>    79: perf ftrace tests    : FAILED!
>    #
>
> The failure depends on the kernel .config file. Some configurarions
> always work fine, some do not.  The ftrace profile test mostly fails,
> because the ring buffer was not large enough, and some lines
> (especially the interesting ones with nanosleep in it) where dropped.
>
> To achieve success for all our tested kernel configurations, enlarge
> the buffer to store the traces complete without wrapping.
> The default buffer size is too small  for all kernel configurations.
> Set the buffer size of /sys/kernel/tracing/buffer_size_kb to 16 MB
>
>
> Output after:
>    # ./perf test ftrace
>    79: perf ftrace tests     : Ok
>    #
>
> Signed-off-by: Thomas Richter <tmricht@xxxxxxxxxxxxx>
> Suggested-by: Sven Schnelle <svens@xxxxxxxxxxxxx>
> ---
>   tools/perf/tests/shell/ftrace.sh | 11 +++++++++++
>   1 file changed, 11 insertions(+)
>
> diff --git a/tools/perf/tests/shell/ftrace.sh b/tools/perf/tests/shell/ftrace.sh
> index a6ee740f0d7e..742d6b8f34d3 100755
> --- a/tools/perf/tests/shell/ftrace.sh
> +++ b/tools/perf/tests/shell/ftrace.sh
> @@ -80,10 +80,21 @@ test_ftrace_profile() {
>       echo "perf ftrace profile test  [Success]"
>   }
>   
> +if [ "$(uname -m)" = "s390x" ]
> +then
> +	ftrace_size=$(cat /sys/kernel/tracing/buffer_size_kb)
> +	echo 16384 > /sys/kernel/tracing/buffer_size_kb
> +fi
> +

It probably wouldn't be terrible to do this for all platforms to reduce 
fragmentation. It doesn't do any harm and it only added a few seconds to 
the test time even on a small Arm box, maybe it will prevent flakes 
everywhere else in the future. But I don't feel too strongly about this one.

>   test_ftrace_list
>   test_ftrace_trace
>   test_ftrace_latency
>   test_ftrace_profile
>   
> +if [ "$(uname -m)" = "s390x" ]
> +then
> +	echo $ftrace_size > /sys/kernel/tracing/buffer_size_kb
> +fi
> +

Restoring the value should be in cleanup() so it works on interrupt too.

With that:

Reviewed-by: James Clark <james.clark@xxxxxxxxxx>

>   cleanup
>   exit 0