Skip to contentSkip to conversation

perf tests: Harden branch stack sampling test

PATCH Superseded Ian Rogers 3 participants 6 messages Mar 12, 2025 – Mar 18, 2025 perf events

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Ian Rogers

author reviewer
#0 lore raw

PATCH 1/1 perf tests: Harden branch stack sampling test

On continuous testing the perf script output can be empty, or nearly empty, causing tr/grep to exit and due to "set -e" the test traps and fails. Add some empty file handling that sets the test to skip and make grep and other text rewriting failures non-fatal by adding "|| true".

---
 tools/perf/tests/shell/test_brstack.sh | 79 +++++++++++++++++++-------
 1 file changed, 59 insertions(+), 20 deletions(-)
tools/perf/tests/shell/test_brstack.sh +59 -20
index e01df7581393..6f5ae227b3e8 100755
@@ -1,4 +1,4 @@
-#!/bin/sh
+#!/bin/bash
 # Check branch stack sampling
 
 # SPDX-License-Identifier: GPL-2.0
@@ -17,35 +17,54 @@ fi
 
 skip_test_missing_symbol brstack_bench
 
+set -x
+err=0
 TMPDIR=$(mktemp -d /tmp/__perf_test.program.XXXXX)
 TESTPROG="perf test -w brstack"
 
 cleanup() {
 	rm -rf $TMPDIR
+	trap - EXIT TERM INT
 }
 
-trap cleanup EXIT TERM INT
+trap_cleanup() {
+	set +e
+	echo "Unexpected signal in ${FUNCNAME[1]}"
+	cleanup
+	exit 1
+}
+trap trap_cleanup EXIT TERM INT
 
 test_user_branches() {
 	echo "Testing user branch stack sampling"
 
-	perf record -o $TMPDIR/perf.data --branch-filter any,save_type,u -- ${TESTPROG} > /dev/null 2>&1
-	perf script -i $TMPDIR/perf.data --fields brstacksym | tr -s ' ' '\n' > $TMPDIR/perf.script
+	perf record -o "$TMPDIR/perf.data" --branch-filter any,save_type,u -- ${TESTPROG} > "$TMPDIR/record.txt" 2>&1
+	perf script -i "$TMPDIR/perf.data" --fields brstacksym > "$TMPDIR/perf.script"
 
 	# example of branch entries:
 	# 	brstack_foo+0x14/brstack_bar+0x40/P/-/-/0/CALL
 
-	set -x
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/IND_CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_foo\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bar\+[^ ]*/brstack_foo\+[^ ]*/RET/.*$"		$TMPDIR/perf.script
-	grep -E -m1 "^brstack_foo\+[^ ]*/brstack_bench\+[^ ]*/RET/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_bench\+[^ ]*/COND/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack\+[^ ]*/brstack\+[^ ]*/UNCOND/.*$"		$TMPDIR/perf.script
-	set +x
-
+	expected=(
+		"^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/IND_CALL/.*$"
+		"^brstack_foo\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"
+		"^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/CALL/.*$"
+		"^brstack_bench\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"
+		"^brstack_bar\+[^ ]*/brstack_foo\+[^ ]*/RET/.*$"
+		"^brstack_foo\+[^ ]*/brstack_bench\+[^ ]*/RET/.*$"
+		"^brstack_bench\+[^ ]*/brstack_bench\+[^ ]*/COND/.*$"
+		"^brstack\+[^ ]*/brstack\+[^ ]*/UNCOND/.*$"
+	)
+	for x in "${expected[@]}"
+	do
+		if ! tr -s ' ' '\n' < "$TMPDIR/perf.script" | grep -E -m1 -q "$x"
+		then
+			echo "Branches missing $x"
+			if [ "x$err" == "x0" ]
+			then
+				err=2
+			fi
+		fi
+	done
 	# some branch types are still not being tested:
 	# IND COND_CALL COND_RET SYSCALL SYSRET IRQ SERROR NO_TX
 }
@@ -57,14 +76,31 @@ test_filter() {
 	test_filter_expect=$2
 
 	echo "Testing branch stack filtering permutation ($test_filter_filter,$test_filter_expect)"
-
-	perf record -o $TMPDIR/perf.data --branch-filter $test_filter_filter,save_type,u -- ${TESTPROG} > /dev/null 2>&1
-	perf script -i $TMPDIR/perf.data --fields brstack | tr -s ' ' '\n' | grep '.' > $TMPDIR/perf.script
+	perf record -o "$TMPDIR/perf.data" --branch-filter "$test_filter_filter,save_type,u" -- ${TESTPROG}  > "$TMPDIR/record.txt" 2>&1
+	perf script -i "$TMPDIR/perf.data" --fields brstack > "$TMPDIR/perf.script"
 
 	# fail if we find any branch type that doesn't match any of the expected ones
 	# also consider UNKNOWN branch types (-)
-	if grep -E -vm1 "^[^ ]*/($test_filter_expect|-|( *))/.*$" $TMPDIR/perf.script; then
-		return 1
+	if [ ! -s "$TMPDIR/perf.script" ]
+	then
+		echo "Empty script output"
+		if [ "x$err" == "x0" ]
+		then
+			err=2
+		fi
+		return
+	fi
+        # Look for lines not matching test_filter_expect ignoring issues caused
+        # by empty output
+	tr -s ' ' '\n' < "$TMPDIR/perf.script" | grep '.' | \
+          grep -E -vm1 "^[^ ]*/($test_filter_expect|-|( *))/.*$" \
+          > "$TMPDIR/perf.script-filtered" || true
+	if [ -s "$TMPDIR/perf.script-filtered" ]
+	then
+		echo "Unexpected branch filter in script output"
+		cat "$TMPDIR/perf.script"
+		err=1
+		return
 	fi
 }
 
@@ -80,3 +116,6 @@ test_filter "any_ret"	"RET|COND_RET|SYSRET|ERET"
 test_filter "call,cond"		"CALL|SYSCALL|COND"
 test_filter "any_call,cond"		"CALL|IND_CALL|COND_CALL|IRQ|SYSCALL|COND"
 test_filter "cond,any_call,any_ret"	"COND|CALL|IND_CALL|COND_CALL|SYSCALL|IRQ|RET|COND_RET|SYSRET|ERET"
+
+cleanup
+exit $err
Signature · 1 line
2.49.0.rc0.332.g42c0ae87b1-goog
#1 lore raw

Hi Ian,

Quoted 74 lines · Ian Rogers

On continuous testing the perf script output can be empty, or nearly empty, causing tr/grep to exit and due to "set -e" the test traps and fails. Add some empty file handling that sets the test to skip and make grep and other text rewriting failures non-fatal by adding "|| true".

---
 tools/perf/tests/shell/test_brstack.sh | 79 +++++++++++++++++++-------
 1 file changed, 59 insertions(+), 20 deletions(-)
tools/perf/tests/shell/test_brstack.sh +21 -15
index e01df7581393..6f5ae227b3e8 100755
@@ -1,4 +1,4 @@
-#!/bin/sh
+#!/bin/bash
 # Check branch stack sampling
 
 # SPDX-License-Identifier: GPL-2.0
@@ -17,35 +17,54 @@ fi
 
 skip_test_missing_symbol brstack_bench
 
+set -x
+err=0
 TMPDIR=$(mktemp -d /tmp/__perf_test.program.XXXXX)
 TESTPROG="perf test -w brstack"
 
 cleanup() {
 	rm -rf $TMPDIR
+	trap - EXIT TERM INT
 }
 
-trap cleanup EXIT TERM INT
+trap_cleanup() {
+	set +e
+	echo "Unexpected signal in ${FUNCNAME[1]}"
+	cleanup
+	exit 1
+}
+trap trap_cleanup EXIT TERM INT
 
 test_user_branches() {
 	echo "Testing user branch stack sampling"
 
-	perf record -o $TMPDIR/perf.data --branch-filter any,save_type,u -- ${TESTPROG} > /dev/null 2>&1
-	perf script -i $TMPDIR/perf.data --fields brstacksym | tr -s ' ' '\n' > $TMPDIR/perf.script
+	perf record -o "$TMPDIR/perf.data" --branch-filter any,save_type,u -- ${TESTPROG} > "$TMPDIR/record.txt" 2>&1
+	perf script -i "$TMPDIR/perf.data" --fields brstacksym > "$TMPDIR/perf.script"
 
 	# example of branch entries:
 	# 	brstack_foo+0x14/brstack_bar+0x40/P/-/-/0/CALL
 
-	set -x
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/IND_CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_foo\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bar\+[^ ]*/brstack_foo\+[^ ]*/RET/.*$"		$TMPDIR/perf.script
-	grep -E -m1 "^brstack_foo\+[^ ]*/brstack_bench\+[^ ]*/RET/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_bench\+[^ ]*/COND/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack\+[^ ]*/brstack\+[^ ]*/UNCOND/.*$"		$TMPDIR/perf.script
-	set +x
-
+	expected=(
+		"^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/IND_CALL/.*$"
+		"^brstack_foo\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"
+		"^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/CALL/.*$"
+		"^brstack_bench\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"
+		"^brstack_bar\+[^ ]*/brstack_foo\+[^ ]*/RET/.*$"
+		"^brstack_foo\+[^ ]*/brstack_bench\+[^ ]*/RET/.*$"
+		"^brstack_bench\+[^ ]*/brstack_bench\+[^ ]*/COND/.*$"

tools/perf/tests/shell/test_brstack.sh fi · line 64

Quoted 10 lines
tools/perf/tests/shell/test_brstack.sh +10 -0
+		"^brstack\+[^ ]*/brstack\+[^ ]*/UNCOND/.*$"
+	)
+	for x in "${expected[@]}"
+	do
+		if ! tr -s ' ' '\n' < "$TMPDIR/perf.script" | grep -E -m1 -q "$x"
+		then
+			echo "Branches missing $x"
+			if [ "x$err" == "x0" ]
+			then
+				err=2

Here it sets "err=2", as a result, if any grep command fails, the script exits while reporting to skip the test. This seems incorrect to me.

My understanding is the regular expressions above are mandatory to be matched, otherwise, it must be something is wrong. We should not skip the test in this case.

I can understand that 'perf record' cannot record all branch types, if this is the case, maybe we can improve the recording quality rather than reporting skip? E.g.,

cat <<EOF > "$TMPDIR/loop.sh"
for run in {1..5}; do perf test -w brstack; done EOF

  perf record -o "$TMPDIR/perf.data" --branch-filter any,save_type,u
    -- sh $TMPDIR/loop.sh

If we run the test for 5 times, should this can allow us to ensure the branch samples are recorded?

Quoted 54 lines
tools/perf/tests/shell/test_brstack.sh +28 -5
+			fi
+		fi
+	done
 	# some branch types are still not being tested:
 	# IND COND_CALL COND_RET SYSCALL SYSRET IRQ SERROR NO_TX
 }
@@ -57,14 +76,31 @@ test_filter() {
 	test_filter_expect=$2
 
 	echo "Testing branch stack filtering permutation ($test_filter_filter,$test_filter_expect)"
-
-	perf record -o $TMPDIR/perf.data --branch-filter $test_filter_filter,save_type,u -- ${TESTPROG} > /dev/null 2>&1
-	perf script -i $TMPDIR/perf.data --fields brstack | tr -s ' ' '\n' | grep '.' > $TMPDIR/perf.script
+	perf record -o "$TMPDIR/perf.data" --branch-filter "$test_filter_filter,save_type,u" -- ${TESTPROG}  > "$TMPDIR/record.txt" 2>&1
+	perf script -i "$TMPDIR/perf.data" --fields brstack > "$TMPDIR/perf.script"
 
 	# fail if we find any branch type that doesn't match any of the expected ones
 	# also consider UNKNOWN branch types (-)
-	if grep -E -vm1 "^[^ ]*/($test_filter_expect|-|( *))/.*$" $TMPDIR/perf.script; then
-		return 1
+	if [ ! -s "$TMPDIR/perf.script" ]
+	then
+		echo "Empty script output"
+		if [ "x$err" == "x0" ]
+		then
+			err=2
+		fi
+		return
+	fi
+        # Look for lines not matching test_filter_expect ignoring issues caused
+        # by empty output
+	tr -s ' ' '\n' < "$TMPDIR/perf.script" | grep '.' | \
+          grep -E -vm1 "^[^ ]*/($test_filter_expect|-|( *))/.*$" \
+          > "$TMPDIR/perf.script-filtered" || true
+	if [ -s "$TMPDIR/perf.script-filtered" ]
+	then
+		echo "Unexpected branch filter in script output"
+		cat "$TMPDIR/perf.script"
+		err=1
+		return
 	fi
 }
 
@@ -80,3 +116,6 @@ test_filter "any_ret"	"RET|COND_RET|SYSRET|ERET"
 test_filter "call,cond"		"CALL|SYSCALL|COND"
 test_filter "any_call,cond"		"CALL|IND_CALL|COND_CALL|IRQ|SYSCALL|COND"
 test_filter "cond,any_call,any_ret"	"COND|CALL|IND_CALL|COND_CALL|SYSCALL|IRQ|RET|COND_RET|SYSRET|ERET"
+
+cleanup
+exit $err
Signature · 1 line
2.49.0.rc0.332.g42c0ae87b1-goog
maintainer
#2 lore raw

Hello,

Quoted 103 lines · Leo Yan

Hi Ian,

Quoted 74 lines · Ian Rogers

On continuous testing the perf script output can be empty, or nearly empty, causing tr/grep to exit and due to "set -e" the test traps and fails. Add some empty file handling that sets the test to skip and make grep and other text rewriting failures non-fatal by adding "|| true".

---
 tools/perf/tests/shell/test_brstack.sh | 79 +++++++++++++++++++-------
 1 file changed, 59 insertions(+), 20 deletions(-)
tools/perf/tests/shell/test_brstack.sh +21 -15
index e01df7581393..6f5ae227b3e8 100755
@@ -1,4 +1,4 @@
-#!/bin/sh
+#!/bin/bash
 # Check branch stack sampling
 
 # SPDX-License-Identifier: GPL-2.0
@@ -17,35 +17,54 @@ fi
 
 skip_test_missing_symbol brstack_bench
 
+set -x
+err=0
 TMPDIR=$(mktemp -d /tmp/__perf_test.program.XXXXX)
 TESTPROG="perf test -w brstack"
 
 cleanup() {
 	rm -rf $TMPDIR
+	trap - EXIT TERM INT
 }
 
-trap cleanup EXIT TERM INT
+trap_cleanup() {
+	set +e
+	echo "Unexpected signal in ${FUNCNAME[1]}"
+	cleanup
+	exit 1
+}
+trap trap_cleanup EXIT TERM INT
 
 test_user_branches() {
 	echo "Testing user branch stack sampling"
 
-	perf record -o $TMPDIR/perf.data --branch-filter any,save_type,u -- ${TESTPROG} > /dev/null 2>&1
-	perf script -i $TMPDIR/perf.data --fields brstacksym | tr -s ' ' '\n' > $TMPDIR/perf.script
+	perf record -o "$TMPDIR/perf.data" --branch-filter any,save_type,u -- ${TESTPROG} > "$TMPDIR/record.txt" 2>&1
+	perf script -i "$TMPDIR/perf.data" --fields brstacksym > "$TMPDIR/perf.script"
 
 	# example of branch entries:
 	# 	brstack_foo+0x14/brstack_bar+0x40/P/-/-/0/CALL
 
-	set -x
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/IND_CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_foo\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bar\+[^ ]*/brstack_foo\+[^ ]*/RET/.*$"		$TMPDIR/perf.script
-	grep -E -m1 "^brstack_foo\+[^ ]*/brstack_bench\+[^ ]*/RET/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_bench\+[^ ]*/COND/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack\+[^ ]*/brstack\+[^ ]*/UNCOND/.*$"		$TMPDIR/perf.script
-	set +x
-
+	expected=(
+		"^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/IND_CALL/.*$"
+		"^brstack_foo\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"
+		"^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/CALL/.*$"
+		"^brstack_bench\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"
+		"^brstack_bar\+[^ ]*/brstack_foo\+[^ ]*/RET/.*$"
+		"^brstack_foo\+[^ ]*/brstack_bench\+[^ ]*/RET/.*$"
+		"^brstack_bench\+[^ ]*/brstack_bench\+[^ ]*/COND/.*$"

tools/perf/tests/shell/test_brstack.sh fi · line 64

Quoted 10 lines
tools/perf/tests/shell/test_brstack.sh +10 -0
+		"^brstack\+[^ ]*/brstack\+[^ ]*/UNCOND/.*$"
+	)
+	for x in "${expected[@]}"
+	do
+		if ! tr -s ' ' '\n' < "$TMPDIR/perf.script" | grep -E -m1 -q "$x"
+		then
+			echo "Branches missing $x"
+			if [ "x$err" == "x0" ]
+			then
+				err=2

Here it sets "err=2", as a result, if any grep command fails, the script exits while reporting to skip the test. This seems incorrect to me.

My understanding is the regular expressions above are mandatory to be matched, otherwise, it must be something is wrong. We should not skip the test in this case.

I can understand that 'perf record' cannot record all branch types, if this is the case, maybe we can improve the recording quality rather than reporting skip? E.g.,

cat <<EOF > "$TMPDIR/loop.sh"
for run in {1..5}; do perf test -w brstack; done EOF

Quoted 5 lines
  perf record -o "$TMPDIR/perf.data" --branch-filter any,save_type,u
    -- sh $TMPDIR/loop.sh

If we run the test for 5 times, should this can allow us to ensure the branch samples are recorded?

The brstack (and other workload programs) can take an argument to control its duration. For brstack, it's the number of loop iteration and default is 999999.

Thanks,
Namhyung

#3 lore raw

Hi Namhyung,

On Thu, Mar 13, 2025 at 01:18:08PM -0700, Namhyung Kim wrote:

[...]

Quoted 59 lines
Quoted 59 lines
Quoted 30 lines

test_user_branches() {

tools/perf/tests/shell/test_brstack.sh +10 -13
 	echo "Testing user branch stack sampling"
 
-	perf record -o $TMPDIR/perf.data --branch-filter any,save_type,u -- ${TESTPROG} > /dev/null 2>&1
-	perf script -i $TMPDIR/perf.data --fields brstacksym | tr -s ' ' '\n' > $TMPDIR/perf.script
+	perf record -o "$TMPDIR/perf.data" --branch-filter any,save_type,u -- ${TESTPROG} > "$TMPDIR/record.txt" 2>&1
+	perf script -i "$TMPDIR/perf.data" --fields brstacksym > "$TMPDIR/perf.script"
 
 	# example of branch entries:
 	# 	brstack_foo+0x14/brstack_bar+0x40/P/-/-/0/CALL
 
-	set -x
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/IND_CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_foo\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bar\+[^ ]*/brstack_foo\+[^ ]*/RET/.*$"		$TMPDIR/perf.script
-	grep -E -m1 "^brstack_foo\+[^ ]*/brstack_bench\+[^ ]*/RET/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack_bench\+[^ ]*/brstack_bench\+[^ ]*/COND/.*$"	$TMPDIR/perf.script
-	grep -E -m1 "^brstack\+[^ ]*/brstack\+[^ ]*/UNCOND/.*$"		$TMPDIR/perf.script
-	set +x
-
+	expected=(
+		"^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/IND_CALL/.*$"
+		"^brstack_foo\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"
+		"^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/CALL/.*$"
+		"^brstack_bench\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"
+		"^brstack_bar\+[^ ]*/brstack_foo\+[^ ]*/RET/.*$"
+		"^brstack_foo\+[^ ]*/brstack_bench\+[^ ]*/RET/.*$"
+		"^brstack_bench\+[^ ]*/brstack_bench\+[^ ]*/COND/.*$"

tools/perf/tests/shell/test_brstack.sh fi · line 64

Quoted 10 lines
tools/perf/tests/shell/test_brstack.sh +10 -0
+		"^brstack\+[^ ]*/brstack\+[^ ]*/UNCOND/.*$"
+	)
+	for x in "${expected[@]}"
+	do
+		if ! tr -s ' ' '\n' < "$TMPDIR/perf.script" | grep -E -m1 -q "$x"
+		then
+			echo "Branches missing $x"
+			if [ "x$err" == "x0" ]
+			then
+				err=2

Here it sets "err=2", as a result, if any grep command fails, the script exits while reporting to skip the test. This seems incorrect to me.

My understanding is the regular expressions above are mandatory to be matched, otherwise, it must be something is wrong. We should not skip the test in this case.

I can understand that 'perf record' cannot record all branch types, if this is the case, maybe we can improve the recording quality rather than reporting skip? E.g.,

cat <<EOF > "$TMPDIR/loop.sh"
for run in {1..5}; do perf test -w brstack; done EOF

  perf record -o "$TMPDIR/perf.data" --branch-filter any,save_type,u
    -- sh $TMPDIR/loop.sh
Quoted 6 lines
Quoted 2 lines

If we run the test for 5 times, should this can allow us to ensure the branch samples are recorded?

The brstack (and other workload programs) can take an argument to control its duration. For brstack, it's the number of loop iteration and default is 999999.

Sorry I did not dig into the brstack workload program.

If the workload has run for a large number of loops, the question is:
why isn't the test capturing the expected branch stacks?

Thanks,
Leo

author reviewer
#4 lore raw
Quoted 69 lines · Leo Yan

Hi Namhyung,

On Thu, Mar 13, 2025 at 01:18:08PM -0700, Namhyung Kim wrote:

[...]

Quoted 62 lines
Quoted 61 lines
Quoted 30 lines

test_user_branches() {
echo "Testing user branch stack sampling"

- perf record -o $TMPDIR/perf.data --branch-filter any,save_type,u -- ${TESTPROG} > /dev/null 2>&1
- perf script -i $TMPDIR/perf.data --fields brstacksym | tr -s ' ' '\n' > $TMPDIR/perf.script
+ perf record -o "$TMPDIR/perf.data" --branch-filter any,save_type,u -- ${TESTPROG} > "$TMPDIR/record.txt" 2>&1
+ perf script -i "$TMPDIR/perf.data" --fields brstacksym > "$TMPDIR/perf.script"

  # example of branch entries:
  #       brstack_foo+0x14/brstack_bar+0x40/P/-/-/0/CALL

- set -x
- grep -E -m1 "^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/IND_CALL/.*$" $TMPDIR/perf.script
- grep -E -m1 "^brstack_foo\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$" $TMPDIR/perf.script
- grep -E -m1 "^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/CALL/.*$" $TMPDIR/perf.script
- grep -E -m1 "^brstack_bench\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$" $TMPDIR/perf.script
- grep -E -m1 "^brstack_bar\+[^ ]*/brstack_foo\+[^ ]*/RET/.*$" $TMPDIR/perf.script
- grep -E -m1 "^brstack_foo\+[^ ]*/brstack_bench\+[^ ]*/RET/.*$" $TMPDIR/perf.script
- grep -E -m1 "^brstack_bench\+[^ ]*/brstack_bench\+[^ ]*/COND/.*$" $TMPDIR/perf.script
- grep -E -m1 "^brstack\+[^ ]*/brstack\+[^ ]*/UNCOND/.*$" $TMPDIR/perf.script
- set +x

tools/perf/tests/shell/test_brstack.sh +8 -1
-
+ expected=(
+         "^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/IND_CALL/.*$"
+         "^brstack_foo\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"
+         "^brstack_bench\+[^ ]*/brstack_foo\+[^ ]*/CALL/.*$"
+         "^brstack_bench\+[^ ]*/brstack_bar\+[^ ]*/CALL/.*$"
+         "^brstack_bar\+[^ ]*/brstack_foo\+[^ ]*/RET/.*$"
+         "^brstack_foo\+[^ ]*/brstack_bench\+[^ ]*/RET/.*$"
+         "^brstack_bench\+[^ ]*/brstack_bench\+[^ ]*/COND/.*$"

tools/perf/tests/shell/test_brstack.sh fi · line 64

Quoted 10 lines
tools/perf/tests/shell/test_brstack.sh +10 -0
+         "^brstack\+[^ ]*/brstack\+[^ ]*/UNCOND/.*$"
+ )
+ for x in "${expected[@]}"
+ do
+         if ! tr -s ' ' '\n' < "$TMPDIR/perf.script" | grep -E -m1 -q "$x"
+         then
+                 echo "Branches missing $x"
+                 if [ "x$err" == "x0" ]
+                 then
+                         err=2

Here it sets "err=2", as a result, if any grep command fails, the script exits while reporting to skip the test. This seems incorrect to me.

My understanding is the regular expressions above are mandatory to be matched, otherwise, it must be something is wrong. We should not skip the test in this case.

I can understand that 'perf record' cannot record all branch types, if this is the case, maybe we can improve the recording quality rather than reporting skip? E.g.,

cat <<EOF > "$TMPDIR/loop.sh"
for run in {1..5}; do perf test -w brstack; done EOF

  perf record -o "$TMPDIR/perf.data" --branch-filter any,save_type,u
    -- sh $TMPDIR/loop.sh

If we run the test for 5 times, should this can allow us to ensure the branch samples are recorded?

Quoted 8 lines
Quoted 3 lines

The brstack (and other workload programs) can take an argument to control its duration. For brstack, it's the number of loop iteration and default is 999999.

Sorry I did not dig into the brstack workload program.

If the workload has run for a large number of loops, the question is:
why isn't the test capturing the expected branch stacks?

On our testing skipped == failed, I can change 2 to 1 above but I'd made it 2 as it wasn't clear to me all branch filter types would be supported by perf record and skipping/2 was a less terrible error message.
I'm keen to land the pulling apart of the perf command from the tr/grep as if we hit say an asan error currently that is hidden by code like:

```
perf record -o $TMPDIR/perf.data --branch-filter any,save_type,u --
${TESTPROG} > /dev/null 2>&1
```

where all the output is sent to /dev/null but the asan error code will cause the "set -e" to fail. If this code fails with asan then currently the first thing to do is start pulling apart the expressions.

Code like:

```
perf script -i $TMPDIR/perf.data --fields brstack | tr -s ' ' '\n' |
grep '.' > $TMPDIR/perf.script
```
is problematic as again we lose the asan like errors. Running the previous:
```
 if grep -E -vm1 "^[^ ]*/($test_filter_expect|-|( *))/.*$"
$TMPDIR/perf.script; then
```

could fail because of an unexpected branch filter type, but was failing for me just because there were blank or similar lines in the output. The new code doesn't change this but allows the output to be dumped for later diagnostics. The '|| true' in the expression means we get to dumping the diagnostics and dump just fail because some sub-command mismatched its input.

Thanks,
Ian

#5 lore raw

On Mon, Mar 17, 2025 at 08:38:44AM -0700, Ian Rogers wrote:

[...]

Quoted 4 lines

On our testing skipped == failed, I can change 2 to 1 above but I'd made it 2 as it wasn't clear to me all branch filter types would be supported by perf record and skipping/2 was a less terrible error message.

My concern is changing from returned val from 1 to 2 will lead to CI to never report errors.

Quoted 11 lines

I'm keen to land the pulling apart of the perf command from the tr/grep as if we hit say an asan error currently that is hidden by code like:

```
perf record -o $TMPDIR/perf.data --branch-filter any,save_type,u --
${TESTPROG} > /dev/null 2>&1
```

where all the output is sent to /dev/null but the asan error code will cause the "set -e" to fail. If this code fails with asan then currently the first thing to do is start pulling apart the expressions.

Agreed the most part in this patch, except returns err=2 in test_user_branches() and test_filter().

Quoted 10 lines

Code like:

```
perf script -i $TMPDIR/perf.data --fields brstack | tr -s ' ' '\n' |
grep '.' > $TMPDIR/perf.script
```
is problematic as again we lose the asan like errors. Running the previous:
```
 if grep -E -vm1 "^[^ ]*/($test_filter_expect|-|( *))/.*$"
$TMPDIR/perf.script; then
```
Quoted 3 lines

could fail because of an unexpected branch filter type, but was failing for me just because there were blank or similar lines in the output.

I saw you changed the command as:

  tr -s ' ' '\n' < "$TMPDIR/perf.script" | grep '.' | \
      grep -E -vm1 "^[^ ]*/($test_filter_expect|-|( *))/.*$" \
      > "$TMPDIR/perf.script-filtered" || true

This is a good change for me. After we removed the noises caused by spaces and empty lines, for an empty script output, I still think we should report error, as the test fails to capture any branch stack.

Quoted 4 lines

The new code doesn't change this but allows the output to be dumped for later diagnostics. The '|| true' in the expression means we get to dumping the diagnostics and dump just fail because some sub-command mismatched its input.

Yeah, agreed '|| true' is a good improvement.

Thanks,
Leo