[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CA+GJov5hCde3JAuwdZayzQ0HUkexbtgAVFrXbDeqpufV6OdC3g@mail.gmail.com>
Date: Tue, 27 Feb 2024 14:41:05 -0500
From: Rae Moar <rmoar@...gle.com>
To: Daniel Latypov <dlatypov@...gle.com>
Cc: shuah@...nel.org, davidgow@...gle.com, brendan.higgins@...ux.dev,
kevko@...gle.com, linux-kselftest@...r.kernel.org, kunit-dev@...glegroups.com,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH] kunit: tool: add parsing of all files in directory
On Thu, Feb 22, 2024 at 6:37 PM Daniel Latypov <dlatypov@...gle.com> wrote:
>
> On Thu, Feb 22, 2024 at 2:18 PM Rae Moar <rmoar@...gle.com> wrote:
> >
> > Add ability to parse all files within a directory. Additionally add the
> > ability to parse all results in the KUnit debugfs repository.
>
> Nice, I'd been hoping for this.
> It's enough to pull me back in for a bit :)
>
> <snip>
>
> > tools/testing/kunit/kunit.py | 45 ++++++++++++++++++++++++++----------
> > 1 file changed, 33 insertions(+), 12 deletions(-)
> >
> > diff --git a/tools/testing/kunit/kunit.py b/tools/testing/kunit/kunit.py
> > index bc74088c458a..827e6dac40ae 100755
> > --- a/tools/testing/kunit/kunit.py
> > +++ b/tools/testing/kunit/kunit.py
> > @@ -511,19 +511,40 @@ def exec_handler(cli_args: argparse.Namespace) -> None:
> >
> >
> > def parse_handler(cli_args: argparse.Namespace) -> None:
> > - if cli_args.file is None:
> > + parsed_files = []
>
> optional: can we annotate the type?
> parsed_files = [] # type: List[str]
Hi Daniel!
Yes, happy to make this change for the next version.
>
> > + total_test = kunit_parser.Test()
> > + total_test.status = kunit_parser.TestStatus.SUCCESS
> > + if cli_args.file_path is None:
> > sys.stdin.reconfigure(errors='backslashreplace') # type: ignore
> > kunit_output = sys.stdin # type: Iterable[str]
>
> This branch no longer does anything, since we only parse what's in
> `parsed_files`
>
> E.g. if you try
> $ kunit.py parse $FILENAME
> it'll work whereas
> $ kunit.py parse < $FILENAME
> will do nothing
>
> We'll need to rework the control flow somehow
Ahh I see. Thanks for bringing this to my attention! I will change
this for the next version.
>
> > - else:
> > - with open(cli_args.file, 'r', errors='backslashreplace') as f:
> > + elif cli_args.file_path == "debugfs":
> > + for (root, _, files) in os.walk("/sys/kernel/debug/kunit"):
> > + for file in files:
> > + if file == "results":
> > + parsed_files.append(os.path.join(root, file))
> > + elif os.path.isdir(cli_args.file_path):
> > + for (root, _, files) in os.walk(cli_args.file_path):
> > + for file in files:
> > + parsed_files.append(os.path.join(root, file))
>
> just a note here, we could make this a bit terser via
> parsed_files.extend(os.path.join(root, f) for f in files)
>
> and the debugfs branch could be rendered as
> parsed_files.extend(os.path.join(root, f) for f in files if f == "results")
>
Will do.
> > + elif os.path.isfile(cli_args.file_path):
> > + parsed_files.append(cli_args.file_path)
>
> nit: should there be an `else` here that prints a warning?
>
> Example that would trigger this case and silently do nothing
> $ mkfifo /tmp/example_fifo
> $ ./tools/testing/kunit/kunit.py parse /tmp/example_fifo
> <no output>
>
Yep you are definitely right I will add one here.
>
> <snip>
>
> > @@ -569,8 +590,8 @@ def main(argv: Sequence[str]) -> None:
> > help='Parses KUnit results from a file, '
> > 'and parses formatted results.')
> > add_parse_opts(parse_parser)
> > - parse_parser.add_argument('file',
> > - help='Specifies the file to read results from.',
> > + parse_parser.add_argument('file_path',
> > + help='Specifies the file path to read results from.',
>
> Should this mention that the make `debugfs` string works?
>
> > type=str, nargs='?', metavar='input_file')
>
> Tangent: would it be useful to allow the user to pass in multiple
> files now and set this to nargs='*'?
>
> E.g.
> $ kunit.py parse /my/dir/some_prefix*
>
> It would also let people implement their own version of the debugfs logic via
> $ find /other/debugfs/dir -name 'results' | xargs kunit.py parse
>
> That could be useful if the user has recursively copied off the
> debugfs from a test machine and wants to inspect it elsewhere, for
> example.
>
Oh this is an interesting idea. I will play around with it!
Thanks for looking this patch over!
-Rae
> Thanks,
> Daniel
Powered by blists - more mailing lists