UDPSOCKET_ECHOTEST change to tolerate duplicate packets

[mtomczykmobica]

Description

UDP should ignore duplicate packets, as some networks may produce those.

Pull request type

[x] Fix
[ ] Refactor
[ ] Target update
[ ] Functionality change
[ ] Docs update
[ ] Test update
[ ] Breaking change

Reviewers

@SeppoTakalo
@AnttiKauppila
@michalpasztamobica

Release Notes

[ciarmcom]

@mtomczykmobica, thank you for your changes.
@AnttiKauppila @SeppoTakalo @michalpasztamobica @ARMmbed/mbed-os-test @ARMmbed/mbed-os-ipcore @ARMmbed/mbed-os-maintainers please review.

[0xc0170]

Please fix styling (see travis astyle failure)

[0xc0170]

Once Travis passing

[mtomczykmobica]

Please fix styling (see travis astyle failure)

Done

[SeppoTakalo]

I don't like this logic.

You assume that all duplicates will be received in order.

This might be true for ECHOTEST, but might not be true for ECHOTEST_BURST.

So while re-working this, can you change logic so that it tolerates re-ordering as well.

I would assume this kind of logic might work:

// List of packet sizes to send
static const int pkt_sizes[PKTS] = {...);
// List of boolean flags to mark each size as "received"
static bool pkt_received[PKTS] = {false};

// ... in the test case
for (int retry_cnt = 0; retry_cnt <= 2; retry_cnt++) {
    ...
    sent = sock.sendto(udp_addr, tx_buffer, pkt_s);
    if (...) {
        // errors..
    }
    do {
        recvd = sock.recvfrom(...);
        // Check that packet is what we sent
        if (recvd == pkt_sizes[s_idx] && memcmp(rx_last, rx_buffer, recvd) == 0)
            pkt_received[s_idx] = true;
        }
    } while(pkt_received[s_idx] == false);
}

So just accept what we were waiting for, ignore the rest.

[mtomczykmobica]

I've changed logic. Now responses for all sends and duplicates of this responses are accepted. Other packets with incorrect size are interpreted as error.

[SeppoTakalo]

This looks OK.

One thing I'm wondering, do we set the socket timeout, or does it hang if there is packet losses?

[0xc0170]

CI started

[mbed-ci]

Test run: SUCCESS

Summary: 5 of 5 test jobs passed
Build number : 1
Build artifacts