netdev - Re: [PATCH net-next v2 4/6] vsock/test: Add test for accept

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]

Message-ID: <dfc51f68-5013-4c7a-ba0d-d3969ac9ecc8@rbox.co>
Date: Mon, 16 Dec 2024 15:52:03 +0100
From: Michal Luczaj <mhal@...x.co>
To: Stefano Garzarella <sgarzare@...hat.com>
Cc: netdev@...r.kernel.org
Subject: Re: [PATCH net-next v2 4/6] vsock/test: Add test for accept_queue
 memory leak

On 12/16/24 15:35, Stefano Garzarella wrote:
> On Mon, Dec 16, 2024 at 01:01:00PM +0100, Michal Luczaj wrote:
>> Attempt to enqueue a child after the queue was flushed, but before
>> SOCK_DONE flag has been set.
>>
>> Test tries to produce a memory leak, kmemleak should be employed. Dealing
>> with a race condition, test by its very nature may lead to a false
>> negative.
>>
>> Fixed by commit d7b0ff5a8667 ("virtio/vsock: Fix accept_queue memory
>> leak").
>>
>> Signed-off-by: Michal Luczaj <mhal@...x.co>
>> ---
>> tools/testing/vsock/vsock_test.c | 51 ++++++++++++++++++++++++++++++++++++++++
>> 1 file changed, 51 insertions(+)
>>
>> diff --git a/tools/testing/vsock/vsock_test.c b/tools/testing/vsock/vsock_test.c
>> index 1ad1fbba10307c515e31816a2529befd547f7fd7..1a9bd81758675a0f2b9b6b0ad9271c45f89a4860 100644
>> --- a/tools/testing/vsock/vsock_test.c
>> +++ b/tools/testing/vsock/vsock_test.c
>> @@ -1474,6 +1474,52 @@ static void test_stream_cred_upd_on_set_rcvlowat(const struct test_opts *opts)
>> 	test_stream_credit_update_test(opts, false);
>> }
>>
>> +/* The goal of test leak_acceptq is to stress the race between connect() and
>> + * close(listener). Implementation of client/server loops boils down to:
>> + *
>> + * client                server
>> + * ------                ------
>> + * write(CONTINUE)
>> + *                       expect(CONTINUE)
>> + *                       listen()
>> + *                       write(LISTENING)
>> + * expect(LISTENING)
>> + * connect()             close()
>> + */
>> +#define ACCEPTQ_LEAK_RACE_TIMEOUT 2 /* seconds */
>> +
>> +#define CONTINUE	1
>> +#define DONE		0
> 
> I would add a prefix here, looking at the timeout, I would say
> ACCEPTQ_LEAK_RACE_CONTINUE and ACCEPTQ_LEAK_RACE_DONE.

I was hoping to make them useful for other tests (see failslab example in
patch 6/6). If CONTINUE/DONE is too generic, how about prefixing them with
something like LOOP_ or TEST_ or CONTROL_ ?