On Fri, 17 Aug 2018 01:43:20 +0900
Masami Hiramatsu <mhiramat@xxxxxxxxxx> wrote:
> Add a testcase for tracing_cpumask with function tracer.
>
> Signed-off-by: Masami Hiramatsu <mhiramat@xxxxxxxxxx>
> ---
> .../selftests/ftrace/test.d/ftrace/func_cpumask.tc | 34 ++++++++++++++++++++
> 1 file changed, 34 insertions(+)
> create mode 100644 tools/testing/selftests/ftrace/test.d/ftrace/func_cpumask.tc
>
> diff --git a/tools/testing/selftests/ftrace/test.d/ftrace/func_cpumask.tc b/tools/testing/selftests/ftrace/test.d/ftrace/func_cpumask.tc
> new file mode 100644
> index 000000000000..37420e355445
> --- /dev/null
> +++ b/tools/testing/selftests/ftrace/test.d/ftrace/func_cpumask.tc
> @@ -0,0 +1,34 @@
> +#!/bin/sh
> +# SPDX-License-Identifier: GPL2.0
> +# description: ftrace - function trace with cpumask
> +
> +NP=`grep "^processor" /proc/cpuinfo | wc -l`
A better way to find the number of CPUs is to either use "nproc" or
just look at /sys/devices/system/cpu/cpu[0-9]*. Because what I learned
from experience is that searching for strings in /proc/cpuinfo is not
cross arch compatible. For example, other archs don't use "processor"
in the stings and would come up with a box with 0 CPUs. Something we've
been working on for some time ;-)
-- Steve
> +
> +if [ $NP -eq 1 ] ;then
> + echo "We can not test cpumask on UP environment"
> + exit_unresolved
> +fi
> +
> +do_reset() {
> + echo ffff > tracing_cpumask
Why ffff? Should we save what was in tracing_cpumask first and just
reuse it?
-- Steve
> +}
> +
> +echo 0 > tracing_on
> +echo > trace
> +: "Bitmask only record on CPU1"
> +echo 2 > tracing_cpumask
> +MASK=0x`cat tracing_cpumask`
> +test `printf "%d" $MASK` -eq 2 || do_reset
> +
> +echo function > current_tracer
> +echo 1 > tracing_on
> +(echo "forked")
> +echo 0 > tracing_on
> +
> +: "Check CPU1 events are recorded"
> +grep -q -e "\[001\]" trace || do_reset
> +
> +: "There should be No other cpu events"
> +! grep -qv -e "\[001\]" -e "^#" trace || do_reset
> +
> +do_reset
