Re: [PATCH bpf-next v2 5/9] selftests/bpf: Add selftest for sockmap/hashmap redirection

[Michal Luczaj <mhal@rbox.co>]

On 4/11/25 16:31, Jiayuan Chen wrote:
> On Fri, Apr 11, 2025 at 01:32:41PM +0200, Michal Luczaj wrote:
>> +static void test_send_redir_recv(int sd_send, int send_flags, int sd_peer,
>> +				 int sd_in, int sd_out, int sd_recv,
>> +				 struct maps *maps, int status)
>> +{
>> +	unsigned int drop, pass;
>> +	char *send_buf = "ab";
>> +	char recv_buf = '\0';
>> +	ssize_t n, len = 1;
>> +	/* Zero out the verdict map */
>> +	if (xbpf_map_update_elem(maps->verd, &u32(SK_DROP), &u32(0), BPF_ANY) ||
>> +	    xbpf_map_update_elem(maps->verd, &u32(SK_PASS), &u32(0), BPF_ANY))
>> +		return;
>> +
>> +	if (xbpf_map_update_elem(maps->in, &u32(0), &u64(sd_in), BPF_NOEXIST))
>> +		return;
>> +
>> +	if (xbpf_map_update_elem(maps->out, &u32(0), &u64(sd_out), BPF_NOEXIST))
>> +		goto del_in;
>> +
>> +	/* Last byte is OOB data when send_flags has MSG_OOB bit set */
>> +	if (send_flags & MSG_OOB)
>> +		len++;
>> +	n = send(sd_send, send_buf, len, send_flags);
>> +	if (n >= 0 && n < len)
>> +		FAIL("incomplete send");
>> +	if (n < 0) {
>> +		/* sk_msg redirect combo not supported? */
>> +		if (status & SUPPORTED || errno != EACCES)
>> +			FAIL_ERRNO("send");
>> +		goto out;
>> +	}
>> +
>> +	if (!(status & SUPPORTED)) {
>> +		handle_unsupported(sd_send, sd_peer, sd_in, sd_out, sd_recv,
>> +				   maps->verd, status);
>> +		goto out;
>> +	}
>> +
>> +	errno = 0;
>> +	n = recv_timeout(sd_recv, &recv_buf, 1, 0, IO_TIMEOUT_SEC);
>> +	if (n != 1) {
>> +		FAIL_ERRNO("recv_timeout()");
>> +		goto out;
>> +	}
> I prefer multiple send and receive operations, or implementing a loop at
> the outer level.

If you referring to MSG_OOB that "emulates" multiple send(), that's quite
deliberate: to exercise what is actually happening on
unix_stream_sendmsg(, MSG_OOB). And, well, to follow the selftests logic
in sockmap_listen.c:pairs_redir_to_connected().

Would you rather have it split anyway?

Thanks,
Michal