perf: test: Speed up running brstack test

PATCH Applied Rob Herring (Arm) 5 participants 6 messages Dec 13, 2024 – Jan 13, 2025 perf events

lore

Share & tools

lore

Rob Herring (Arm)

author

2024-12-13 23:19 UTC #0 lore raw

PATCH 1/1 perf: test: Speed up running brstack test

From: James Clark <[email protected]>

The brstack test runs quite slowly in software models. Part of the reason is "xargs -n1" is quite inefficient in replacing spaces with newlines. While that's not noticeable on normal machines, it is on software models. Use "tr -s ' ' '\n'" instead which can do the same transformation, but is much faster. For comparison on an M1 Macbook Pro:

$ time seq -s ' ' 10000 | xargs -n1 > /dev/null

real    0m2.729s
user    0m2.009s
sys     0m0.914s
$ time seq -s ' ' 10000 | tr -s ' ' '\n' | grep '.' > /dev/null

real    0m0.002s
user    0m0.001s
sys     0m0.001s

The "grep '.'" is also needed to remove any remaining blank lines.

Signed-off-by James Clark <[email protected]>

Signed-off-by Anshuman Khandual <[email protected]>

[robh: Drop changing loop iterations on arm64. Squash blank line fix and redo commit msg]

Signed-off-by Rob Herring (Arm) <[email protected]>

---
Originally part of this series[1], but I've dropped any Arm specifics, and it stands on its own. No reason this needs to wait on Arm BRBE support (which I'm working on now). I don't expect to have other changes to this test related to BRBE anymore.

[1] https://lore.kernel.org/all/[email protected]/

 tools/perf/tests/shell/test_brstack.sh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

tools/perf/tests/shell/test_brstack.sh +2 -2

index 5f14d0cb013f..e01df7581393 100755
@@ -30,7 +30,7 @@ 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 | xargs -n1 > $TMPDIR/perf.script
+	perf script -i $TMPDIR/perf.data --fields brstacksym | tr -s ' ' '\n' > $TMPDIR/perf.script
 
 	# example of branch entries:
 	# 	brstack_foo+0x14/brstack_bar+0x40/P/-/-/0/CALL

@@ -59,7 +59,7 @@ test_filter() {
 	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 | xargs -n1 > $TMPDIR/perf.script
+	perf script -i $TMPDIR/perf.data --fields brstack | tr -s ' ' '\n' | grep '.' > $TMPDIR/perf.script
 
 	# fail if we find any branch type that doesn't match any of the expected ones
 	# also consider UNKNOWN branch types (-)

Signature · 1 line

2.45.2

James Clark

reviewer Reviewed-by on v1 1/1

2024-12-17 14:42 UTC #1 lore raw

Quoted 25 lines · Rob Herring (Arm)

From: James Clark <[email protected]>
The brstack test runs quite slowly in software models. Part of the reason is "xargs -n1" is quite inefficient in replacing spaces with newlines. While that's not noticeable on normal machines, it is on software models. Use "tr -s ' ' '\n'" instead which can do the same transformation, but is much faster. For comparison on an M1 Macbook Pro:
$ time seq -s ' ' 10000 | xargs -n1 > /dev/null
real    0m2.729s
user    0m2.009s
sys     0m0.914s
$ time seq -s ' ' 10000 | tr -s ' ' '\n' | grep '.' > /dev/null
real    0m0.002s
user    0m0.001s
sys     0m0.001s
The "grep '.'" is also needed to remove any remaining blank lines.
Signed-off-by James Clark <[email protected]>
Signed-off-by Anshuman Khandual <[email protected]>
[robh: Drop changing loop iterations on arm64. Squash blank line fix and redo commit msg]
Signed-off-by Rob Herring (Arm) <[email protected]>

Quoted 8 lines

---
Originally part of this series[1], but I've dropped any Arm specifics, and it stands on its own. No reason this needs to wait on Arm BRBE support (which I'm working on now). I don't expect to have other changes to this test related to BRBE anymore.
[1] https://lore.kernel.org/all/[email protected]/

Reviewed-by James Clark <[email protected]>

Quoted 25 lines

  tools/perf/tests/shell/test_brstack.sh | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)

tools/perf/tests/shell/test_brstack.sh +2 -2

index 5f14d0cb013f..e01df7581393 100755
@@ -30,7 +30,7 @@ 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 | xargs -n1 > $TMPDIR/perf.script
+	perf script -i $TMPDIR/perf.data --fields brstacksym | tr -s ' ' '\n' > $TMPDIR/perf.script
 
  	# example of branch entries:
  	# 	brstack_foo+0x14/brstack_bar+0x40/P/-/-/0/CALL

@@ -59,7 +59,7 @@ test_filter() {
  	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 | xargs -n1 > $TMPDIR/perf.script
+	perf script -i $TMPDIR/perf.data --fields brstack | tr -s ' ' '\n' | grep '.' > $TMPDIR/perf.script
 
  	# fail if we find any branch type that doesn't match any of the expected ones
  	# also consider UNKNOWN branch types (-)

Anshuman Khandual

Reviewed-by on v1 1/1

2024-12-18 03:32 UTC #2 lore raw

In reply to Rob Herring (Arm)

Quoted 19 lines · Rob Herring (Arm)

From: James Clark <[email protected]>
The brstack test runs quite slowly in software models. Part of the reason is "xargs -n1" is quite inefficient in replacing spaces with newlines. While that's not noticeable on normal machines, it is on software models. Use "tr -s ' ' '\n'" instead which can do the same transformation, but is much faster. For comparison on an M1 Macbook Pro:
$ time seq -s ' ' 10000 | xargs -n1 > /dev/null
real    0m2.729s
user    0m2.009s
sys     0m0.914s
$ time seq -s ' ' 10000 | tr -s ' ' '\n' | grep '.' > /dev/null
real    0m0.002s
user    0m0.001s
sys     0m0.001s

Quoted 6 lines

The "grep '.'" is also needed to remove any remaining blank lines.
Signed-off-by James Clark <[email protected]>
Signed-off-by Anshuman Khandual <[email protected]>
[robh: Drop changing loop iterations on arm64. Squash blank line fix and redo commit msg]
Signed-off-by Rob Herring (Arm) <[email protected]>

Reviewed-by Anshuman Khandual <[email protected]>

Quoted 33 lines

[1] https://lore.kernel.org/all/[email protected]/

 tools/perf/tests/shell/test_brstack.sh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

tools/perf/tests/shell/test_brstack.sh +2 -2

index 5f14d0cb013f..e01df7581393 100755
@@ -30,7 +30,7 @@ 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 | xargs -n1 > $TMPDIR/perf.script
+	perf script -i $TMPDIR/perf.data --fields brstacksym | tr -s ' ' '\n' > $TMPDIR/perf.script
 
 	# example of branch entries:
 	# 	brstack_foo+0x14/brstack_bar+0x40/P/-/-/0/CALL

@@ -59,7 +59,7 @@ test_filter() {
 	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 | xargs -n1 > $TMPDIR/perf.script
+	perf script -i $TMPDIR/perf.data --fields brstack | tr -s ' ' '\n' | grep '.' > $TMPDIR/perf.script
 
 	# fail if we find any branch type that doesn't match any of the expected ones
 	# also consider UNKNOWN branch types (-)

Namhyung Kim

maintainer Reviewed-by on v1 1/1

2024-12-19 05:48 UTC #3 lore raw

In reply to Rob Herring (Arm)

Quoted 19 lines · Rob Herring (Arm)

From: James Clark <[email protected]>
The brstack test runs quite slowly in software models. Part of the reason is "xargs -n1" is quite inefficient in replacing spaces with newlines. While that's not noticeable on normal machines, it is on software models. Use "tr -s ' ' '\n'" instead which can do the same transformation, but is much faster. For comparison on an M1 Macbook Pro:
$ time seq -s ' ' 10000 | xargs -n1 > /dev/null
real    0m2.729s
user    0m2.009s
sys     0m0.914s
$ time seq -s ' ' 10000 | tr -s ' ' '\n' | grep '.' > /dev/null
real    0m0.002s
user    0m0.001s
sys     0m0.001s

Quoted 6 lines

The "grep '.'" is also needed to remove any remaining blank lines.
Signed-off-by James Clark <[email protected]>
Signed-off-by Anshuman Khandual <[email protected]>
[robh: Drop changing loop iterations on arm64. Squash blank line fix and redo commit msg]
Signed-off-by Rob Herring (Arm) <[email protected]>

Reviewed-by Namhyung Kim <[email protected]>

Thanks,
Namhyung

Quoted 36 lines

[1] https://lore.kernel.org/all/[email protected]/

 tools/perf/tests/shell/test_brstack.sh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

tools/perf/tests/shell/test_brstack.sh +2 -2

index 5f14d0cb013f..e01df7581393 100755
@@ -30,7 +30,7 @@ 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 | xargs -n1 > $TMPDIR/perf.script
+	perf script -i $TMPDIR/perf.data --fields brstacksym | tr -s ' ' '\n' > $TMPDIR/perf.script
 
 	# example of branch entries:
 	# 	brstack_foo+0x14/brstack_bar+0x40/P/-/-/0/CALL

@@ -59,7 +59,7 @@ test_filter() {
 	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 | xargs -n1 > $TMPDIR/perf.script
+	perf script -i $TMPDIR/perf.data --fields brstack | tr -s ' ' '\n' | grep '.' > $TMPDIR/perf.script
 
 	# fail if we find any branch type that doesn't match any of the expected ones
 	# also consider UNKNOWN branch types (-)

Signature · 1 line

2.45.2

Rob Herring

author

2025-01-13 14:26 UTC #4 lore raw

In reply to Rob Herring (Arm)

Quoted 32 lines · Rob Herring (Arm)

From: James Clark <[email protected]>
The brstack test runs quite slowly in software models. Part of the reason is "xargs -n1" is quite inefficient in replacing spaces with newlines. While that's not noticeable on normal machines, it is on software models. Use "tr -s ' ' '\n'" instead which can do the same transformation, but is much faster. For comparison on an M1 Macbook Pro:
$ time seq -s ' ' 10000 | xargs -n1 > /dev/null
real    0m2.729s
user    0m2.009s
sys     0m0.914s
$ time seq -s ' ' 10000 | tr -s ' ' '\n' | grep '.' > /dev/null
real    0m0.002s
user    0m0.001s
sys     0m0.001s
The "grep '.'" is also needed to remove any remaining blank lines.
Signed-off-by James Clark <[email protected]>
Signed-off-by Anshuman Khandual <[email protected]>
[robh: Drop changing loop iterations on arm64. Squash blank line fix and redo commit msg]
Signed-off-by Rob Herring (Arm) <[email protected]>
---
Originally part of this series[1], but I've dropped any Arm specifics, and it stands on its own. No reason this needs to wait on Arm BRBE support (which I'm working on now). I don't expect to have other changes to this test related to BRBE anymore.

Quoted 4 lines

[1] https://lore.kernel.org/all/[email protected]/
 tools/perf/tests/shell/test_brstack.sh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Ping!

Quoted 26 lines

tools/perf/tests/shell/test_brstack.sh +2 -3

index 5f14d0cb013f..e01df7581393 100755
@@ -30,7 +30,7 @@ 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 | xargs -n1 > $TMPDIR/perf.script
+       perf script -i $TMPDIR/perf.data --fields brstacksym | tr -s ' ' '\n' > $TMPDIR/perf.script
 
        # example of branch entries:
        #       brstack_foo+0x14/brstack_bar+0x40/P/-/-/0/CALL

@@ -59,7 +59,7 @@ test_filter() {
        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 | xargs -n1 > $TMPDIR/perf.script
+       perf script -i $TMPDIR/perf.data --fields brstack | tr -s ' ' '\n' | grep '.' > $TMPDIR/perf.script
 
        # fail if we find any branch type that doesn't match any of the expected ones
        # also consider UNKNOWN branch types (-)
--

2.45.2

Arnaldo Carvalho de Melo

maintainer Applied

2025-01-13 14:59 UTC #5 lore raw

Quoted 33 lines · Rob Herring

Quoted 32 lines · Rob Herring (Arm)
From: James Clark <[email protected]>
The brstack test runs quite slowly in software models. Part of the reason is "xargs -n1" is quite inefficient in replacing spaces with newlines. While that's not noticeable on normal machines, it is on software models. Use "tr -s ' ' '\n'" instead which can do the same transformation, but is much faster. For comparison on an M1 Macbook Pro:
$ time seq -s ' ' 10000 | xargs -n1 > /dev/null
real    0m2.729s
user    0m2.009s
sys     0m0.914s
$ time seq -s ' ' 10000 | tr -s ' ' '\n' | grep '.' > /dev/null
real    0m0.002s
user    0m0.001s
sys     0m0.001s
The "grep '.'" is also needed to remove any remaining blank lines.
Signed-off-by James Clark <[email protected]>
Signed-off-by Anshuman Khandual <[email protected]>
[robh: Drop changing loop iterations on arm64. Squash blank line fix and redo commit msg]
Signed-off-by Rob Herring (Arm) <[email protected]>
---
Originally part of this series[1], but I've dropped any Arm specifics, and it stands on its own. No reason this needs to wait on Arm BRBE support (which I'm working on now). I don't expect to have other changes to this test related to BRBE anymore.

Quoted 6 lines

Quoted 4 lines
[1] https://lore.kernel.org/all/[email protected]/
 tools/perf/tests/shell/test_brstack.sh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
Ping!

Thanks, applied.

- Arnaldo