[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <1e560ea5-3eb2-40a8-33aa-514eb026de3a@arm.com>
Date: Thu, 13 Jul 2023 15:32:19 +0100
From: Ryan Roberts <ryan.roberts@....com>
To: Mark Brown <broonie@...nel.org>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
Shuah Khan <shuah@...nel.org>,
Jérôme Glisse <jglisse@...hat.com>,
David Hildenbrand <david@...hat.com>,
John Hubbard <jhubbard@...dia.com>,
Florent Revest <revest@...omium.org>,
"Liam R. Howlett" <Liam.Howlett@...cle.com>,
linux-kernel@...r.kernel.org, linux-mm@...ck.org,
linux-kselftest@...r.kernel.org
Subject: Re: [PATCH v1 1/9] selftests: Line buffer test program's stdout
On 13/07/2023 15:16, Mark Brown wrote:
> On Thu, Jul 13, 2023 at 02:54:32PM +0100, Ryan Roberts wrote:
>> The selftests runner pipes the test program's stdout to tap_prefix. The
>> presence of the pipe means that the test program sets its stdout to be
>> fully buffered (as aposed to line buffered when directly connected to
>> the terminal). The block buffering means that there is often content in
>> the buffer at fork() time, which causes the output to end up duplicated.
>> This was causing problems for mm:cow where test results were duplicated
>> 20-30x.
>>
>> Solve this by using `stdbuf`, when available to force the test program
>> to use line buffered mode. This means previously printf'ed results are
>> flushed out of the program before any fork().
>
> This is going to be useful in general since not all selftests use the
> kselftest helpers but it'd probably also be good to make
> ksft_print_header() also make the output unbuffered
Yeah sounds reasonable.
> so that if setbuf
> isn't installed on the target system or the tests are run standalone we
> don't run into issues there. Even if the test isn't corrupting data
> having things unbuffered is going to be good for making sure we don't
> drop any output if the test dies.
Note that currently I've set stdbuf to encourage line buffering rather than no
buffering. Are you saying no buffering is preferred? I took the view that line
buffering is a good middle ground, and and aligns with what people see when
developing and running the program manually in the terminal.
>
>> + if [ -x /usr/bin/stdbuf ]; then
>> + stdbuf="/usr/bin/stdbuf --output=L "
>> + fi
>
> Might be more robust to use type -p to find stdbuf in case it's in /bin
> or something?
Yep good idea.
Powered by blists - more mailing lists