Need to send a dummy packet to establish connection using the TCP/IP stack

[zephyrbot]

Reported by Sergio Rodriguez:

In order to establish a connection to other element in the network using the current IP stack, some packet has to be sent, but this package is not being recognized by the other end, and is dropped, causing some some handshakes to fail (i.e. TLS) so in order to create the connection a dummy packet is send before the proper sequence, this might cause problems with other types of strict network negotiated protocols. this is related to GH-2105 so at least one byte

(Imported from Jira ZEP-612)

[zephyrbot]

by Ravi kumar Veeramally:

• tcp-conn-close.pcapng (size: 12700)

[zephyrbot]

by Ravi kumar Veeramally:

• tcp-bug.patch (size: 3138)

[zephyrbot]

by Ravi kumar Veeramally:

Hi Sergio,

Please find attached wireshark [^tcp-conn-close.pcapng] log, and also changes in echo_server application. [^tcp-bug.patch]

We have net tools in https://gerrit.zephyrproject.org/r/#/admin/projects/net-tools.

I tried echo-server running in qemu and echo-client on linux host. Attached sample tcp log from wireshark. It shows that we do not require any dummy packet to start connection.

Please try this.
Patch which I attached is only changes in sample test, nothing in core stack.

1. Apply the patch.
2. goto net tools and run ./loop-socat.sh
   and sudo ./loop-slip.sh
3. run echo-server with 'make qemu'
4. run echo-client from net tools
   sudo ./echo-client 2001:db8::2 -t

Let me know your findings.

[zephyrbot]

by Kuo-Lang Tseng:

Ravi kumar Veeramally Looks like the current work-around (1-bye dummy packet) is still needed.

[zephyrbot]

by Ravi kumar Veeramally:

Can you attach some kind of log with and without workaround and also sample app also to reproduce.

[zephyrbot]

by Andrei Laperie:

assigning to Kuo-Lang Tseng per Ravi kumar Veeramally comment

[zephyrbot]

by Kuo-Lang Tseng:

Sergio will provide more info once he is back.

[zephyrbot]

by Anas Nashif:

I do not get it why this bug is assigned back to the reporter instead of closing it as invalid or whatever. This way of doing things seems to be very common leaving a large number of bugs open for very long time without any resolution.

If you think a bug is invalid, just close it, if the reporter has some more information and a reason to believe the bug is still valid, reopen, this is how how bug systems work, re-assigning back and forth is not productive.

[zephyrbot]

by Anas Nashif:

please reopen if issue still exists

[zephyrbot]

by Mark Linkmeyer:

Sergio Rodriguez , as the Reporter, please move this to Verified then to Closed, or Reopen it with comments explaining why it can't be closed. Thx.

[zephyrbot]

by Sergio Rodriguez:

Since this has been indicated as won't do, and we have a current workaround, closing this jira item

[zephyrbot]

by Mark Linkmeyer:

Correcting the priority field

[zephyrbot]

Blocks GH-1969

[zephyrbot]

Blocks GH-2123

[zephyrbot]

Blocks GH-2124