Re: [RFC PATCH v1 2/3] test/vsock: add big message test

From: Arseniy Krasnov
Date: Wed Nov 23 2022 - 11:29:34 EST


On 23.11.2022 18:40, Bobby Eshleman wrote:
> On Wed, Nov 23, 2022 at 7:22 AM Stefano Garzarella <sgarzare@xxxxxxxxxx>
> wrote:
>
>> On Mon, Nov 21, 2022 at 09:40:39PM +0000, Arseniy Krasnov wrote:
>>> On 21.11.2022 19:50, Arseniy Krasnov wrote:
>>>> On 21.11.2022 17:52, Stefano Garzarella wrote:
>>>>> On Tue, Nov 15, 2022 at 08:52:35PM +0000, Arseniy Krasnov wrote:
>>>>>> This adds test for sending message, bigger than peer's buffer size.
>>>>>> For SOCK_SEQPACKET socket it must fail, as this type of socket has
>>>>>> message size limit.
>>>>>>
>>>>>> Signed-off-by: Arseniy Krasnov <AVKrasnov@xxxxxxxxxxxxxx>
>>>>>> ---
>>>>>> tools/testing/vsock/vsock_test.c | 62 ++++++++++++++++++++++++++++++++
>>>>>> 1 file changed, 62 insertions(+)
>>>>>>
>>>>>> diff --git a/tools/testing/vsock/vsock_test.c
>> b/tools/testing/vsock/vsock_test.c
>>>>>> index 107c11165887..bb4e8657f1d6 100644
>>>>>> --- a/tools/testing/vsock/vsock_test.c
>>>>>> +++ b/tools/testing/vsock/vsock_test.c
>>>>>> @@ -560,6 +560,63 @@ static void test_seqpacket_timeout_server(const
>> struct test_opts *opts)
>>>>>> close(fd);
>>>>>> }
>>>>>>
>>>>>> +static void test_seqpacket_bigmsg_client(const struct test_opts
>> *opts)
>>>>>> +{
>>>>>> + unsigned long sock_buf_size;
>>>>>> + ssize_t send_size;
>>>>>> + socklen_t len;
>>>>>> + void *data;
>>>>>> + int fd;
>>>>>> +
>>>>>> + len = sizeof(sock_buf_size);
>>>>>> +
>>>>>> + fd = vsock_seqpacket_connect(opts->peer_cid, 1234);
>>>>>
>>>>> Not for this patch, but someday we should add a macro for this port
>> and maybe even make it configurable :-)
>>>>>
>>>>>> + if (fd < 0) {
>>>>>> + perror("connect");
>>>>>> + exit(EXIT_FAILURE);
>>>>>> + }
>>>>>> +
>>>>>> + if (getsockopt(fd, AF_VSOCK, SO_VM_SOCKETS_BUFFER_SIZE,
>>>>>> + &sock_buf_size, &len)) {
>>>>>> + perror("getsockopt");
>>>>>> + exit(EXIT_FAILURE);
>>>>>> + }
>>>>>> +
>>>>>> + sock_buf_size++;
>>>>>> +
>>>>>> + data = malloc(sock_buf_size);
>>>>>> + if (!data) {
>>>>>> + perror("malloc");
>>>>>> + exit(EXIT_FAILURE);
>>>>>> + }
>>>>>> +
>>>>>> + send_size = send(fd, data, sock_buf_size, 0);
>>>>>> + if (send_size != -1) {
>>>>>
>>>>> Can we check also `errno`?
>>>>> IIUC it should contains EMSGSIZE.
>>> Hm, seems current implementation is a little bit broken and returns
>> ENOMEM, because any negative value, returned by
>>> transport callback is always replaced to ENOMEM. I think i need this
>> patch from Bobby:
>>>
>> https://lore.kernel.org/lkml/d81818b868216c774613dd03641fcfe63cc55a45.1660362668.git.bobby.eshleman@xxxxxxxxxxxxx/
>>> May be i can include it to this patchset also fixing review comments(of
>> course keeping Bobby as author). Or more
>>> simple way is to check ENOMEM instead of EMSGSIZE in this test(simple,
>> but a little bit dumb i think).
>>
>> Maybe in this patch you can start checking ENOMEM (with a TODO comment),
>> and then Bobby can change it when sending his patch.
>>
>> Or you can repost it (I'm not sure if we should keep the legacy behavior
>> for other transports or it was an error, but better to discuss it on
>> that patch). However, I think we should merge that patch.
>>
>> @Bobby, what do you think?
>>
>> Thanks,
>> Stefano
>>
>>
> This sounds good to me. I removed it from the last rev because I decided not
> to complicate the patch by also including SO_SNDBUF support, so had no
> need. I think it makes sense overall though.
Ok! So I'll use Your patch(both af_vsock.c and transports related things - seems i'll
split it to several patches, I think for transport patches, it is better to ask Vmware
/Microsoft guys also). I'm going to send next version of my tests on this week.

Thank You
>
> Also, sorry for the delay (I promised last week to send out new rev). I was
> planning on sending out v4 with additional data for the non-nested virt
> case,
> but I've been having some IT troubles with the new phys box.
No problem, im currently rebasing my patches for zerocopy on v3.
>
> Best,
> Bobby
>
> PS. sorry if this email format comes out wacky. I'm not at my dev machine
> so just using Gmail's web app directly... Hopefully there is no HTML or
> anything weird.
>