linux-kernel - Re: [PATCH 2/2] selftests: lib: prime

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>] [day] [month] [year] [list]

Message-ID: <CAO_48GHZ+Qb-5NzaYGB3nRu2WdRG6S05kz4X5qBDyMus_KQWJA@mail.gmail.com>
Date:   Mon, 19 Jun 2017 20:41:33 +0530
From:   Sumit Semwal <sumit.semwal@...aro.org>
To:     Chris Wilson <chris@...is-wilson.co.uk>
Cc:     shuah@...nel.org, linux-kselftest@...r.kernel.org, lukas@...ner.de,
        LKML <linux-kernel@...r.kernel.org>,
        Daniel Vetter <daniel.vetter@...ll.ch>,
        Joonas Lahtinen <joonas.lahtinen@...ux.intel.com>
Subject: Re: [PATCH 2/2] selftests: lib: prime_numbers: update presence check

On 19 June 2017 at 20:27, Chris Wilson <chris@...is-wilson.co.uk> wrote:
> Quoting Sumit Semwal (2017-06-19 15:46:20)
>> Hi Chris,
>>
>> On 19 June 2017 at 19:21, Chris Wilson <chris@...is-wilson.co.uk> wrote:
>> > Quoting Sumit Semwal (2017-06-19 14:44:32)
>> >> The test for prime numbers doesn't differentiate between missing
>> >> prime_numbers.ko and failure in prime_numbers.ko.
>> >>
>> >> Update it to check for presence of the file itself to skip, therefore
>> >> correctly exercising the test failure case.
>> >
>> > modprobe -r shouldn't be executing the module? But you still need to
>> > unload the module before you can load it with the selftest module
>> > parameters. If you can't unload the module due to an earlier failure,
>> > you cannot discern whether or not the module itself is at fault, so
>> > still want to SKIP.
>>
>> My bad here: I missed the '-r' in the first modprobe.
>>
>> I am wondering if 'modprobe -q -n' won't suffice, as it is a dry-run
>> only, but will still search for the module? Unless of course, there's
>> something specific about '-q -r' that seems better still?
>
> I think there are two things to be tested here, both causing a SKIP.
>
>  - If the module doesn't exist at all; modprobe -q -n seems sensible for
>    querying its existence.
>
>  - If the module cannot be [un]loaded; for which I was using the
>    modprobe -q -r. If we can't unload the module, then we can't test :)

:) Right; then the question is, for prime_numbers.ko, do we need to
differentiate between these 2 SKIPs?

The unloading of the prime_numbers module before running the test is
required since it isn't a standalone test module - unlike the
test_bitmap and test_printf ones.

So then, for prime_numbers: if distinguishing between the two cases
you mentioned above isn't important, we can just keep your original
code. If it is important to distinguish, I can add the -q -n test to
query existence separately.

For test_bitmap and test_printf tests, I think I will just go ahead
with -q -n itself, since we can assume that the test modules will only
be loaded/unloaded via these tests I guess?

>
> -Chris

Best,
Sumit.