linux-kernel - Re: [PATCH 0/4] Convert to TAP13 using ksft

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <600ca478-1094-c1b8-5234-281c73abb212@pitt.edu>
Date:   Sat, 1 Jul 2017 03:34:11 +0900
From:   Paul Elder <paul.elder@...t.edu>
To:     shuah@...nel.org, alice.ferrazzi@...il.com,
        gregkh@...uxfoundation.org, labath@...gle.com, panand@...hat.com,
        will.deacon@....com
Cc:     Shuah Khan <shuahkh@....samsung.com>,
        linux-kselftest@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 0/4] Convert to TAP13 using ksft_ var arg msg api

On 07/01/2017 02:06 AM, Shuah Khan wrote:
> On 06/30/2017 10:52 AM, Paul Elder wrote:
>> On 07/01/2017 01:33 AM, Shuah Khan wrote:
>>> On 06/30/2017 10:13 AM, Paul Elder wrote:
>>>> On 06/30/2017 08:18 AM, Shuah Khan wrote:
>>>>> This patch series converts breakpoint_test_arm64 to TAP13 output. Uses
>>>>> ksft_ var arg msg functions as needed. Changed to ensure parent is the
>>>>> one that increments test counters.
>>>>>
>>>>> breakpoint_test, step_after_suspend_test, membarrier tests are changed
>>>>> to use ksft_ var arg msg functions as needed.
>>>>>
>>>>> This patch series depends on Paul Elder's
>>>>>
>>>>> [PATCH v2 0/2] kselftest: make ksft_* output functions variadic
>>>>> The v2 series doesn't append newline and makes it easier to use.
>>>>>
>>>>> Shuah Khan (4):
>>>>>   selftests: breakpoint_test: use ksft_* var arg msg api
>>>>>   selftests: breakpoints: step_after_suspend_test use ksft_* var arg msg
>>>>>     api
>>>>>   selftests: breakpoints: breakpoint_test_arm64: convert test to use
>>>>>     TAP13
>>>>>   selftests: membarrier: use ksft_* var arg msg api
>>>>>
>>>>>  .../selftests/breakpoints/breakpoint_test.c        | 28 ++++---
>>>>>  .../selftests/breakpoints/breakpoint_test_arm64.c  | 94 +++++++++++++---------
>>>>>  .../breakpoints/step_after_suspend_test.c          | 59 ++++++++------
>>>>>  .../testing/selftests/membarrier/membarrier_test.c | 91 ++++++++++-----------
>>>>>  4 files changed, 148 insertions(+), 124 deletions(-)
>>>>>
>>>> Looks good to me.
>>>>
>>>> Although I am wondering why you said earlier that you were going to go with
>>>> my v1, while this depends on v2.
>>>>
>>>> Thank you,
>>>>
>>>> Paul
>>>>
>>>>
>>>
>>> Yeah my bad!
>>>
>>> I was playing with both v1 and v2 for my testing. I lost track of which one worked
>>> well. I had to start from square one and re-test.
>> Oh okay. Understood.
>>
>>>
>>> v2 by not appending newline, allows custom formats in individual test messages.
>>> For example, a test wants to use tabs between two ksft_print_msg() calls or _report_
>>> messages. v2 is flexible and works better and leds itself well. Users can just pass
>>> the format and expect it to stick.
>> I see. I didn't notice this advantage. Though if a test wants to separate two
>> ksft_print_msg()s with tabs, wouldn't there still be a # between them?
> 
> Sure. The important thing for me is that - passed in format sticking and not changing.
I see.

> 
>>
>>>
>>> If there are no issues with this series, I plan to apply your v2 series and this
>>> series later on today or early tomorrow to linux-kselftest next.
>> I have no objections.
>>
> 
> Thanks.
> 
>>>
>>> If we have this infrastructure in place, we will be able to make progress in the
>>> next release with the TAP13 conversion of other tests.> Indeed. Although I'm having trouble with tests that fork.
> 
> Please focus on tests that don't fork()/exec() first.
Okay.

> 
> I would recommend having parent print test counts - child will use ksft_print_msg()
> for all its output I was able to use that strategy for tests in this series.
I was thinking of that too. The parent could increment an anonymous test count after printing
just the test number? And then the child would print the test name without the #.

> 
> This might not work for all tests. We have to revisit the tests that fork() and exec().
> I plan to look at parent/child tests in the next couple of weeks.
Got it.

Thank you,

Paul

> 
> thanks,
> -- Shuah
>