On 5/21/25 16:56, Stefano Garzarella wrote:
> On Wed, May 21, 2025 at 12:55:23AM +0200, Michal Luczaj wrote:
>> There was an issue with SO_LINGER: instead of blocking until all queued
>> messages for the socket have been successfully sent (or the linger timeout
>> has been reached), close() would block until packets were handled by the
>> peer.
>>
>> Add a test to alert on close() lingering when it should not.
>>
>> Signed-off-by: Michal Luczaj <m...@rbox.co>
>> ---
>> tools/testing/vsock/vsock_test.c | 49
>> ++++++++++++++++++++++++++++++++++++++++
>> 1 file changed, 49 insertions(+)
>>
>> diff --git a/tools/testing/vsock/vsock_test.c
>> b/tools/testing/vsock/vsock_test.c
>> index
>> f401c6a79495bc7fda97012e5bfeabec7dbfb60a..1040503333cf315e52592c876f2c1809b36fdfdb
>> 100644
>> --- a/tools/testing/vsock/vsock_test.c
>> +++ b/tools/testing/vsock/vsock_test.c
>> @@ -1839,6 +1839,50 @@ static void test_stream_linger_server(const struct
>> test_opts *opts)
>> close(fd);
>> }
>>
>> +static void test_stream_nolinger_client(const struct test_opts *opts)
>> +{
>> + bool nowait;
>> + time_t ns;
>> + int fd;
>> +
>> + fd = vsock_stream_connect(opts->peer_cid, opts->peer_port);
>> + if (fd < 0) {
>> + perror("connect");
>> + exit(EXIT_FAILURE);
>> + }
>> +
>> + enable_so_linger(fd);
>
> If we use a parameter for the linger timeout, IMO will be easy to
> understand this test, defining the timeout in this test, set it and
> check the value, without defining LINGER_TIMEOUT in util.h.
Yes, you're right. I'll fix that.
>> + send_byte(fd, 1, 0); /* Left unread to expose incorrect behaviour. */
>> + nowait = vsock_wait_sent(fd);
>> +
>> + ns = current_nsec();
>> + close(fd);
>> + ns = current_nsec() - ns;
>> +
>> + if (nowait) {
>> + fprintf(stderr, "Test skipped, SIOCOUTQ not supported.\n");
>> + } else if ((ns + NSEC_PER_SEC - 1) / NSEC_PER_SEC >= LINGER_TIMEOUT) {
>
> Should we define a macro for this conversion?
>
> Or just use DIV_ROUND_UP:
Arrgh, I was looking for that. If you don't care much for a new macro, I'll
explicitly use DIV_ROUND_UP for now.
Thanks!
Michal
