linux-kernel - Re: [PATCH v2] selftests/rtc:Fix a resource leak

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: <423013a8-3973-402c-b835-543ae6a29546@nvidia.com>
Date: Tue, 16 Jul 2024 10:58:28 -0700
From: John Hubbard <jhubbard@...dia.com>
To: Shuah Khan <skhan@...uxfoundation.org>,
 Alexandre Belloni <alexandre.belloni@...tlin.com>,
 Zhu Jun <zhujun2@...s.chinamobile.com>
Cc: linux-kernel@...r.kernel.org, linux-kselftest@...r.kernel.org,
 linux-rtc@...r.kernel.org, shuah@...nel.org
Subject: Re: [PATCH v2] selftests/rtc:Fix a resource leak

On 7/10/24 9:00 AM, Shuah Khan wrote:
> On 7/10/24 01:55, Alexandre Belloni wrote:
>> On 10/07/2024 00:43:09-0700, Zhu Jun wrote:
>>> The opened file should be closed before exit, otherwise resource leak
>>> will occur that this problem was discovered by code reading
>>
>> The question is still why should it be closed before exit as it will be
>> closed on exit?
>>
> 
> Zhu Jun,
> 
> +1 on this. I have responded to your other patches that do the same
> in other tests. There is no need to make such changes.
> 
> thanks,
> -- Shuah
> 

Yes. What Shuah and Alexandre said. And the same reasoning also applies
to memory buffers: they are also released by the kernel at program exit.

This blind adherence to "good programming practices" is not helpful here,
because it turns out that selftests are small, short-running, and simple.
These are all desirable properties, because remember, there are no
tests to verify the selftests, so we rely on keeping them simple and
very easy to quickly read and understand.

"Good programming practices" is not a "one approach to all programs"
thing.

So adding line count to free something a few lines before program exit,
for tiny programs, is actually counter productive in this environment.

Zhu, if you're going to keep working on selftests, I think you would
really benefit from some of the literature that goes into the design
philosophy for unit tests. Google has the best (and most recent) work
IMHO [1], but there are others of course.

[1] Software Engineering at Google (Apr 7, 2020), by Titus Winters

thanks,
-- 
John Hubbard
NVIDIA