linux-kernel - Re: [PATCH] kernel/sysctl-test: Unregister sysctl table after test completion

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 for Android: free password hash cracker in your pocket

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

Message-ID: <d2c93db4-6406-47ec-9096-479aa7d7fd23@roeck-us.net>
Date: Fri, 23 May 2025 08:54:44 -0700
From: Guenter Roeck <linux@...ck-us.net>
To: Joel Granados <joel.granados@...nel.org>
Cc: Kees Cook <kees@...nel.org>, linux-kernel@...r.kernel.org,
 linux-fsdevel@...r.kernel.org, Wen Yang <wen.yang@...ux.dev>
Subject: Re: [PATCH] kernel/sysctl-test: Unregister sysctl table after test
 completion

On 5/23/25 08:01, Joel Granados wrote:
> On Thu, May 22, 2025 at 11:53:15AM -0700, Guenter Roeck wrote:
>> On Wed, May 21, 2025 at 06:32:11PM -0700, Guenter Roeck wrote:
>>> One of the sysctl tests registers a valid sysctl table. This operation
>>> is expected to succeed. However, it does not unregister the table after
>>> executing the test. If the code is built as module and the module is
>>> unloaded after the test, the next operation trying to access the table
>>> (such as 'sysctl -a') will trigger a crash.
>>>
>>> Unregister the registered table after test completiion to solve the
>>> problem.
>>>
>>
>> Never mind, I just learned that a very similar patch has been submitted
>> last December or so but was rejected, and that the acceptable (?) fix seems
>> to be stalled.
>>
>> Sorry for the noise.
>>
>> Guenter
> 
> Hey Guenter
> 
> It is part of what is getting sent for 6.16 [1]
> That test will move out of kunit into self-test.
> 

Yes, I was pointed to that. The version I have seen seems to assume that
the test is running as module, because the created sysctl entry is removed
in the module exit function. If built into the kernel, it would leave
the debug entry in place after the test is complete. Also, it moves
the affected set of tests out of the kunit infrastructure. Is that accurate
or a misunderstanding on my side ?

No criticism or objection, I am just trying to understand the direction
of unit testing and specifically of the kunit infrastructure going forward.

Thanks,
Guenter