linux-kernel - Re: [PATCH v5 08/10] kunit: string-stream: Add tests for freeing resource-managed string

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

Date:   Fri, 25 Aug 2023 10:41:38 +0100
From:   Richard Fitzgerald <rf@...nsource.cirrus.com>
To:     David Gow <davidgow@...gle.com>
CC:     <brendan.higgins@...ux.dev>, <rmoar@...gle.com>,
        <linux-kselftest@...r.kernel.org>, <kunit-dev@...glegroups.com>,
        <linux-kernel@...r.kernel.org>, <patches@...nsource.cirrus.com>
Subject: Re: [PATCH v5 08/10] kunit: string-stream: Add tests for freeing
 resource-managed string_stream

On 25/08/2023 07:49, David Gow wrote:
> On Thu, 24 Aug 2023 at 22:32, 'Richard Fitzgerald' via KUnit
> Development <kunit-dev@...glegroups.com> wrote:
>>
>> string_stream_managed_free_test() allocates a resource-managed
>> string_stream and tests that kunit_free_string_stream() calls
>> string_stream_destroy().
>>
>> string_stream_resource_free_test() allocates a resource-managed
>> string_stream and tests that string_stream_destroy() is called
>> when the test resources are cleaned up.
>>
>> The old string_stream_init_test() has been split into two tests,
>> one for kunit_alloc_string_stream() and the other for
>> alloc_string_stream().
>>
>> Signed-off-by: Richard Fitzgerald <rf@...nsource.cirrus.com>
>> ---
>> Changes since V4:
>> - Added test case for kunit_free_string_stream().
>> - Split the initialization test into separate tests for managed and
>>    unmanaged allocations.
>> ---
> 
> Looking over this again, I'm not convinced the streams are actually
> getting freed. Once the stub has finished, the stream is removed from
> the list of deferred actions / resources.
> 

Argh, I think you're right. My original version stashed the stream into
the private data and freed it in a test exit() function so that it was
guaranteed to be freed even if the resource cleanup wasn't called and
the test function aborted before it could do a manual cleanup.
I decided to simplify that but actually that original implementation was
better.