Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
Message-ID: <d36e7dc7-e063-365f-2bf4-8cac590b5802@linux.ibm.com>
Date:   Mon, 2 May 2022 11:39:44 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.8.0
Subject: Re: [PATCH] perf test: Add skip to --per-thread test
Content-Language: en-US
To:     Ian Rogers <irogers@google.com>, linux-kernel@vger.kernel.org,
        linux-perf-users@vger.kernel.org, acme@kernel.org,
        Jiri Olsa <jolsa@kernel.org>,
        Namhyung Kim <namhyung@kernel.org>
Cc:     svens@linux.ibm.com, gor@linux.ibm.com, sumanthk@linux.ibm.com,
        hca@linux.ibm.com
References: <20220429061336.1184848-1-irogers@google.com>
From:   Thomas Richter <tmricht@linux.ibm.com>
Organization: IBM
In-Reply-To: <20220429061336.1184848-1-irogers@google.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
MIME-Version: 1.0
Precedence: bulk

On 4/29/22 08:13, Ian Rogers wrote:
> As reported in:
> https://lore.kernel.org/linux-perf-users/20220428122821.3652015-1-tmricht@linux.ibm.com/
> the 'instructions:u' event may not be supported. Add a skip using 'perf
> stat'.
> 
> Switch some code away from pipe to make the failures clearer.
> 
> Reported-by: Thomas Richter <tmricht@linux.ibm.com>
> Signed-off-by: Ian Rogers <irogers@google.com>
> ---
>  tools/perf/tests/shell/record.sh | 44 +++++++++++++++++++++++++++++---
>  1 file changed, 41 insertions(+), 3 deletions(-)
> 
> diff --git a/tools/perf/tests/shell/record.sh b/tools/perf/tests/shell/record.sh
> index cd1cf14259b8..efc0cb07a929 100755
> --- a/tools/perf/tests/shell/record.sh
> +++ b/tools/perf/tests/shell/record.sh
> @@ -5,11 +5,43 @@
>  set -e
>  
>  err=0
> +perfdata=$(mktemp /tmp/__perf_test.perf.data.XXXXX)
> +
> +cleanup() {
> +  rm -f ${perfdata}
> +  rm -f ${perfdata}.old
> +  trap - exit term int
> +}
> +
> +trap_cleanup() {
> +  cleanup
> +  exit 1
> +}
> +trap trap_cleanup exit term int
> +
>  test_per_thread() {
>    echo "Basic --per-thread mode test"
> -  perf record -e instructions:u --per-thread -o- true 2> /dev/null \
> -    | perf report -i- -q \
> -    | egrep -q true
> +  if ! perf stat -e instructions:u --quiet true
> +  then
> +    echo "Per-thread record [Skipped instructions:u not supported]"
> +    if [ $err -ne 1 ]
> +    then
> +      err=2
> +    fi
> +    return
> +  fi
> +  if ! perf record -e instructions:u --per-thread -o ${perfdata} true 2> /dev/null
> +  then
> +    echo "Per-thread record of instructions:u [Failed]"
> +    err=1
> +    return
> +  fi
> +  if ! perf report -i ${perfdata} -q | egrep -q true
> +  then
> +    echo "Per-thread record [Failed missing output]"
> +    err=1
> +    return
> +  fi
>    echo "Basic --per-thread mode test [Success]"
>  }
>  
> @@ -18,6 +50,10 @@ test_register_capture() {
>    if ! perf list | egrep -q 'br_inst_retired.near_call'
>    then
>      echo "Register capture test [Skipped missing instruction]"
> +    if [ $err -ne 1 ]
> +    then
> +      err=2
> +    fi
>      return
>    fi
>    if ! perf record --intr-regs=\? 2>&1 | egrep -q 'available registers: AX BX CX DX SI DI BP SP IP FLAGS CS SS R8 R9 R10 R11 R12 R13 R14 R15'
> @@ -39,4 +75,6 @@ test_register_capture() {
>  
>  test_per_thread
>  test_register_capture
> +
> +cleanup
>  exit $err

Ian,

thanks very much for looking into this.
I have tested your patch on several configurations and to be perfect, please
change this line:
# git diff
diff --git a/tools/perf/tests/shell/record.sh b/tools/perf/tests/shell/record.sh
index efc0cb07a929..9cb8494f314d 100755
--- a/tools/perf/tests/shell/record.sh
+++ b/tools/perf/tests/shell/record.sh
@@ -21,7 +21,7 @@ trap trap_cleanup exit term int
 
 test_per_thread() {
   echo "Basic --per-thread mode test"
-  if ! perf stat -e instructions:u --quiet true
+  if ! perf record -e instructions:u --quiet true 2> /dev/null
   then
     echo "Per-thread record [Skipped instructions:u not supported]"
     if [ $err -ne 1 ]
# 

This runs fine, just use record instead of stat, you use perf record later
on anyway.

With this change you have my
Tested-by: Thomas Richter <tmricht@linux.ibm.com>
-- 
Thomas Richter, Dept 3303, IBM s390 Linux Development, Boeblingen, Germany
--
Vorsitzender des Aufsichtsrats: Gregor Pillen
Geschäftsführung: David Faller
Sitz der Gesellschaft: Böblingen / Registergericht: Amtsgericht Stuttgart, HRB 243294