[PATCH v2] kunit: tool: Default --jobs to number of CPUs

[PATCH v2] kunit: tool: Default --jobs to number of CPUs

[David Gow]

The --jobs parameter for kunit_tool currently defaults to 8 CPUs,
regardless of the number available. For systems with significantly more
(or less), this is not as efficient. Instead, default --jobs to the
number of CPUs available to the process: while there are as many
superstitions as to exactly what the ideal jobs:CPU ratio is, this seems
sufficiently sensible to me.

A new helper function to get the default number of jobs is added:
get_default_jobs() -- this is used in kunit_tool_test instead of a
hardcoded value, or an explicit call to len(os.sched_getaffinity()), so
should be more flexible if this needs to change in the future.

Signed-off-by: David Gow <davidgow@google.com>
---

Changes since v1:
https://lore.kernel.org/linux-kselftest/20211211084928.410669-1-davidgow@google.com/
- Use len(os.sched_getaffinity()) instead of os.cpu_count(), which gives
  the number of available processors (to this process), rather than the
  total.
- Fix kunit_tool_test.py, which had 8 jobs hardcoded in a couple of
  places.
- Thanks to Daniel Latypov for these suggestions.

---

 tools/testing/kunit/kunit.py           | 5 ++++-
 tools/testing/kunit/kunit_tool_test.py | 5 +++--
 2 files changed, 7 insertions(+), 3 deletions(-)

diff --git a/tools/testing/kunit/kunit.py b/tools/testing/kunit/kunit.py
index 68e6f461c758..6b0ddd6d0115 100755
--- a/tools/testing/kunit/kunit.py
+++ b/tools/testing/kunit/kunit.py
@@ -264,6 +264,9 @@ def massage_argv(argv: Sequence[str]) -> Sequence[str]:
 		return  f'{arg}={pseudo_bool_flag_defaults[arg]}'
 	return list(map(massage_arg, argv))
 
+def get_default_jobs() -> int:
+	return len(os.sched_getaffinity(0))
+
 def add_common_opts(parser) -> None:
 	parser.add_argument('--build_dir',
 			    help='As in the make command, it specifies the build '
@@ -310,7 +313,7 @@ def add_build_opts(parser) -> None:
 	parser.add_argument('--jobs',
 			    help='As in the make command, "Specifies  the number of '
 			    'jobs (commands) to run simultaneously."',
-			    type=int, default=8, metavar='jobs')
+			    type=int, default=get_default_jobs(), metavar='jobs')
 
 def add_exec_opts(parser) -> None:
 	parser.add_argument('--timeout',
diff --git a/tools/testing/kunit/kunit_tool_test.py b/tools/testing/kunit/kunit_tool_test.py
index 9c4126731457..512936241a56 100755
--- a/tools/testing/kunit/kunit_tool_test.py
+++ b/tools/testing/kunit/kunit_tool_test.py
@@ -419,7 +419,7 @@ class KUnitMainTest(unittest.TestCase):
 	def test_build_passes_args_pass(self):
 		kunit.main(['build'], self.linux_source_mock)
 		self.assertEqual(self.linux_source_mock.build_reconfig.call_count, 0)
-		self.linux_source_mock.build_kernel.assert_called_once_with(False, 8, '.kunit', None)
+		self.linux_source_mock.build_kernel.assert_called_once_with(False, kunit.get_default_jobs(), '.kunit', None)
 		self.assertEqual(self.linux_source_mock.run_kernel.call_count, 0)
 
 	def test_exec_passes_args_pass(self):
@@ -525,8 +525,9 @@ class KUnitMainTest(unittest.TestCase):
 
 	def test_build_builddir(self):
 		build_dir = '.kunit'
+		jobs = kunit.get_default_jobs()
 		kunit.main(['build', '--build_dir', build_dir], self.linux_source_mock)
-		self.linux_source_mock.build_kernel.assert_called_once_with(False, 8, build_dir, None)
+		self.linux_source_mock.build_kernel.assert_called_once_with(False, jobs, build_dir, None)
 
 	def test_exec_builddir(self):
 		build_dir = '.kunit'
-- 
2.34.1.173.g76aa8bc2d0-goog

Re: [PATCH v2] kunit: tool: Default --jobs to number of CPUs

[Daniel Latypov]

On Sat, Dec 11, 2021 at 7:32 PM David Gow <davidgow@google.com> wrote:
>
> The --jobs parameter for kunit_tool currently defaults to 8 CPUs,
> regardless of the number available. For systems with significantly more
> (or less), this is not as efficient. Instead, default --jobs to the
> number of CPUs available to the process: while there are as many
> superstitions as to exactly what the ideal jobs:CPU ratio is, this seems
> sufficiently sensible to me.
>
> A new helper function to get the default number of jobs is added:
> get_default_jobs() -- this is used in kunit_tool_test instead of a
> hardcoded value, or an explicit call to len(os.sched_getaffinity()), so
> should be more flexible if this needs to change in the future.
>
> Signed-off-by: David Gow <davidgow@google.com>

Reviewed-by: Daniel Latypov <dlatypov@google.com>

Looks good to me.
I played around with a few commands like
$ taskset 0x3 ./tools/testing/kunit/kunit.py run
and saw it pick the pass the expected --jobs value to make.

> ---
>
> Changes since v1:
> https://lore.kernel.org/linux-kselftest/20211211084928.410669-1-davidgow@google.com/
> - Use len(os.sched_getaffinity()) instead of os.cpu_count(), which gives
>   the number of available processors (to this process), rather than the
>   total.
> - Fix kunit_tool_test.py, which had 8 jobs hardcoded in a couple of
>   places.
> - Thanks to Daniel Latypov for these suggestions.
>
> ---
>
>  tools/testing/kunit/kunit.py           | 5 ++++-
>  tools/testing/kunit/kunit_tool_test.py | 5 +++--
>  2 files changed, 7 insertions(+), 3 deletions(-)
>
> diff --git a/tools/testing/kunit/kunit.py b/tools/testing/kunit/kunit.py
> index 68e6f461c758..6b0ddd6d0115 100755
> --- a/tools/testing/kunit/kunit.py
> +++ b/tools/testing/kunit/kunit.py
> @@ -264,6 +264,9 @@ def massage_argv(argv: Sequence[str]) -> Sequence[str]:
>                 return  f'{arg}={pseudo_bool_flag_defaults[arg]}'
>         return list(map(massage_arg, argv))
>
> +def get_default_jobs() -> int:
> +       return len(os.sched_getaffinity(0))
> +
>  def add_common_opts(parser) -> None:
>         parser.add_argument('--build_dir',
>                             help='As in the make command, it specifies the build '
> @@ -310,7 +313,7 @@ def add_build_opts(parser) -> None:
>         parser.add_argument('--jobs',
>                             help='As in the make command, "Specifies  the number of '
>                             'jobs (commands) to run simultaneously."',
> -                           type=int, default=8, metavar='jobs')
> +                           type=int, default=get_default_jobs(), metavar='jobs')
>
>  def add_exec_opts(parser) -> None:
>         parser.add_argument('--timeout',
> diff --git a/tools/testing/kunit/kunit_tool_test.py b/tools/testing/kunit/kunit_tool_test.py
> index 9c4126731457..512936241a56 100755
> --- a/tools/testing/kunit/kunit_tool_test.py
> +++ b/tools/testing/kunit/kunit_tool_test.py
> @@ -419,7 +419,7 @@ class KUnitMainTest(unittest.TestCase):
>         def test_build_passes_args_pass(self):
>                 kunit.main(['build'], self.linux_source_mock)
>                 self.assertEqual(self.linux_source_mock.build_reconfig.call_count, 0)
> -               self.linux_source_mock.build_kernel.assert_called_once_with(False, 8, '.kunit', None)
> +               self.linux_source_mock.build_kernel.assert_called_once_with(False, kunit.get_default_jobs(), '.kunit', None)
>                 self.assertEqual(self.linux_source_mock.run_kernel.call_count, 0)
>
>         def test_exec_passes_args_pass(self):
> @@ -525,8 +525,9 @@ class KUnitMainTest(unittest.TestCase):
>
>         def test_build_builddir(self):
>                 build_dir = '.kunit'
> +               jobs = kunit.get_default_jobs()
>                 kunit.main(['build', '--build_dir', build_dir], self.linux_source_mock)
> -               self.linux_source_mock.build_kernel.assert_called_once_with(False, 8, build_dir, None)
> +               self.linux_source_mock.build_kernel.assert_called_once_with(False, jobs, build_dir, None)
>
>         def test_exec_builddir(self):
>                 build_dir = '.kunit'
> --
> 2.34.1.173.g76aa8bc2d0-goog
>

Re: [PATCH v2] kunit: tool: Default --jobs to number of CPUs

[Brendan Higgins]

On Sat, Dec 11, 2021 at 10:32 PM David Gow <davidgow@google.com> wrote:
>
> The --jobs parameter for kunit_tool currently defaults to 8 CPUs,
> regardless of the number available. For systems with significantly more
> (or less), this is not as efficient. Instead, default --jobs to the
> number of CPUs available to the process: while there are as many
> superstitions as to exactly what the ideal jobs:CPU ratio is, this seems
> sufficiently sensible to me.
>
> A new helper function to get the default number of jobs is added:
> get_default_jobs() -- this is used in kunit_tool_test instead of a
> hardcoded value, or an explicit call to len(os.sched_getaffinity()), so
> should be more flexible if this needs to change in the future.
>
> Signed-off-by: David Gow <davidgow@google.com>

Reviewed-by: Brendan Higgins <brendanhiggins@google.com>