[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20241122043840.217453-11-howardchu95@gmail.com>
Date: Thu, 21 Nov 2024 20:38:40 -0800
From: Howard Chu <howardchu95@...il.com>
To: peterz@...radead.org
Cc: mingo@...hat.com,
acme@...nel.org,
namhyung@...nel.org,
mark.rutland@....com,
alexander.shishkin@...ux.intel.com,
jolsa@...nel.org,
irogers@...gle.com,
adrian.hunter@...el.com,
kan.liang@...ux.intel.com,
linux-perf-users@...r.kernel.org,
linux-kernel@...r.kernel.org,
Howard Chu <howardchu95@...il.com>,
James Clark <james.clark@...aro.org>,
Arnaldo Carvalho de Melo <acme@...hat.com>
Subject: [PATCH v9 10/10] perf test: Add direct off-cpu test
Why is there a --off-cpu-thresh 2000000?
We collect an off-cpu period __ONLY ONCE__, either in direct sample form,
or in accumulated form (in BPF stack trace map).
If I don't add --off-cpu-thresh 200000, the sample in the original test
goes into the ring buffer instead of the BPF stack trace map.
Additionally, when using -e dummy, the ring buffer is not open, causing
us to lose a sample.
Signed-off-by: Howard Chu <howardchu95@...il.com>
Cc: Adrian Hunter <adrian.hunter@...el.com>
Cc: Alexander Shishkin <alexander.shishkin@...ux.intel.com>
Cc: Ian Rogers <irogers@...gle.com>
Cc: Ingo Molnar <mingo@...hat.com>
Cc: James Clark <james.clark@...aro.org>
Cc: Jiri Olsa <jolsa@...nel.org>
Cc: Kan Liang <kan.liang@...ux.intel.com>
Cc: Mark Rutland <mark.rutland@....com>
Cc: Namhyung Kim <namhyung@...nel.org>
Cc: Peter Zijlstra <peterz@...radead.org>
Link: https://lore.kernel.org/r/20241108204137.2444151-11-howardchu95@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@...hat.com>
---
tools/perf/tests/builtin-test.c | 1 +
tools/perf/tests/shell/record_offcpu.sh | 35 ++++++++++++++++++++++++-
tools/perf/tests/tests.h | 1 +
tools/perf/tests/workloads/Build | 1 +
tools/perf/tests/workloads/offcpu.c | 16 +++++++++++
5 files changed, 53 insertions(+), 1 deletion(-)
create mode 100644 tools/perf/tests/workloads/offcpu.c
diff --git a/tools/perf/tests/builtin-test.c b/tools/perf/tests/builtin-test.c
index 8dcf74d3c0a3..43dc04075ecb 100644
--- a/tools/perf/tests/builtin-test.c
+++ b/tools/perf/tests/builtin-test.c
@@ -146,6 +146,7 @@ static struct test_workload *workloads[] = {
&workload__brstack,
&workload__datasym,
&workload__landlock,
+ &workload__offcpu,
};
#define workloads__for_each(workload) \
diff --git a/tools/perf/tests/shell/record_offcpu.sh b/tools/perf/tests/shell/record_offcpu.sh
index 678947fe69ee..8719130fbf70 100755
--- a/tools/perf/tests/shell/record_offcpu.sh
+++ b/tools/perf/tests/shell/record_offcpu.sh
@@ -6,6 +6,10 @@ set -e
err=0
perfdata=$(mktemp /tmp/__perf_test.perf.data.XXXXX)
+TEST_PROGRAM="perf test -w offcpu"
+
+ts=$(printf "%u" $((~0 << 32))) # OFF_CPU_TIMESTAMP
+dummy_timestamp=${ts%???} # remove the last 3 digits to match perf script
cleanup() {
rm -f ${perfdata}
@@ -39,7 +43,11 @@ test_offcpu_priv() {
test_offcpu_basic() {
echo "Basic off-cpu test"
- if ! perf record --off-cpu -e dummy -o ${perfdata} sleep 1 2> /dev/null
+ # We collect an off-cpu period __ONLY ONCE__, either in direct sample form, or in accumulated form
+ # (in BPF stack trace map). Without the --off-cpu-thresh 200000 below, the sample will go into the
+ # ring buffer instead of the BPF stack trace map. Additionally, when using -e dummy, the ring
+ # buffer is not enabled, resulting in a lost sample.
+ if ! perf record --off-cpu --off-cpu-thresh 2000000 -e dummy -o ${perfdata} sleep 1 2> /dev/null
then
echo "Basic off-cpu test [Failed record]"
err=1
@@ -88,6 +96,27 @@ test_offcpu_child() {
echo "Child task off-cpu test [Success]"
}
+test_offcpu_direct() {
+ echo "Direct off-cpu test"
+
+ # dump off-cpu samples for task blocked for more than 1.999999s
+ # -D for initial delay, to enable evlist
+ if ! perf record -e dummy -D 500 --off-cpu --off-cpu-thresh 1999999 -o ${perfdata} ${TEST_PROGRAM} 2> /dev/null
+ then
+ echo "Direct off-cpu test [Failed record]"
+ err=1
+ return
+ fi
+ # Direct sample's timestamp should be lower than the dummy_timestamp of the at-the-end sample.
+ if ! perf script -i ${perfdata} -F time,period | sed "s/[\.:]//g" | \
+ awk "{ if (\$1 < ${dummy_timestamp} && \$2 > 1999999999) exit 0; else exit 1; }"
+ then
+ echo "Direct off-cpu test [Failed missing direct sample]"
+ err=1
+ return
+ fi
+ echo "Direct off-cpu test [Success]"
+}
test_offcpu_priv
@@ -99,5 +128,9 @@ if [ $err = 0 ]; then
test_offcpu_child
fi
+if [ $err = 0 ]; then
+ test_offcpu_direct
+fi
+
cleanup
exit $err
diff --git a/tools/perf/tests/tests.h b/tools/perf/tests/tests.h
index cb58b43aa063..2e655a617b30 100644
--- a/tools/perf/tests/tests.h
+++ b/tools/perf/tests/tests.h
@@ -217,6 +217,7 @@ DECLARE_WORKLOAD(sqrtloop);
DECLARE_WORKLOAD(brstack);
DECLARE_WORKLOAD(datasym);
DECLARE_WORKLOAD(landlock);
+DECLARE_WORKLOAD(offcpu);
extern const char *dso_to_test;
extern const char *test_objdump_path;
diff --git a/tools/perf/tests/workloads/Build b/tools/perf/tests/workloads/Build
index 5af17206f04d..0e78fd01eaf1 100644
--- a/tools/perf/tests/workloads/Build
+++ b/tools/perf/tests/workloads/Build
@@ -7,6 +7,7 @@ perf-test-y += sqrtloop.o
perf-test-y += brstack.o
perf-test-y += datasym.o
perf-test-y += landlock.o
+perf-test-y += offcpu.o
CFLAGS_sqrtloop.o = -g -O0 -fno-inline -U_FORTIFY_SOURCE
CFLAGS_leafloop.o = -g -O0 -fno-inline -fno-omit-frame-pointer -U_FORTIFY_SOURCE
diff --git a/tools/perf/tests/workloads/offcpu.c b/tools/perf/tests/workloads/offcpu.c
new file mode 100644
index 000000000000..57cee201a4c3
--- /dev/null
+++ b/tools/perf/tests/workloads/offcpu.c
@@ -0,0 +1,16 @@
+#include <linux/compiler.h>
+#include <unistd.h>
+#include "../tests.h"
+
+static int offcpu(int argc __maybe_unused, const char **argv __maybe_unused)
+{
+ /* get past the initial delay */
+ sleep(1);
+
+ /* what we want to collect as a direct sample */
+ sleep(2);
+
+ return 0;
+}
+
+DEFINE_WORKLOAD(offcpu);
--
2.43.0
Powered by blists - more mailing lists