Fix incorrect state transition

[Fr3ya]

When a TCP connection in eCONNECT_SYN state times out after exceeding the maximum retry count, the socket was incorrectly transitioning to eCLOSE_WAIT state instead of eCLOSED.

According to the standard TCP state machine (RFC 793):

• eCONNECT_SYN is the state where a client has sent a SYN and is waiting for a SYN-ACK
• When connection establishment fails (timeout or RST), the socket should transition to eCLOSED state

This fix:

1. Changes the state transition in prvTCPSendPacket() from eCLOSE_WAIT to eCLOSED when SYN retries are exhausted
2. Updates vTCPStateChange() to handle the eCONNECT_SYN -> eCLOSED transition, ensuring connection state change callbacks are properly triggered.

This also aligns the timeout behavior with the existing RST handling code, which correctly uses eCLOSED state (see FreeRTOS_TCP_IP.c:859).

Files changed:

• source/FreeRTOS_TCP_Transmission.c: Fix state transition
• source/FreeRTOS_TCP_IP.c: Update vTCPStateChange to handle eCLOSED transition from connecting states

Description

Test Steps

Checklist:

• I have tested my changes. No regression in existing tests.
• I have modified and/or added unit-tests to cover the code changes in this Pull Request.

Related Issue

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[htibosch]

@Fr3ya , I have fixed the unit tests. I will send you an update later today.

[htibosch]

@Fr3ya , I made some changes to repair the unit testing and 100% coverage :

ut_changes_htibosch.patch

./source/FreeRTOS_TCP_IP.c
./test/unit-test/FreeRTOS_TCP_IP/FreeRTOS_TCP_IP_utest.c

Can you apply my patch and do a new commit?

Here are the plain files in a zip:

ut_changes_htibosch.zip
Most changes are found in test_vTCPStateChange_ClosedWaitState_PrvStateSyn().

[htibosch]

I checked the changes, repaired the unit tests, and the 100% coverage. @Fr3ya , can you please apply the patch that I posted here?

[Fr3ya]

Hi @htibosch, I have applied the patch that you posted, thanks!

[htibosch]

Hello, can anyone approve and merge this PR please?
Thanks, Hein

[rmn30]

I think this PR also fixes another issue I was seeing:

If you attempt to FreeRTOS_Connect to a remote socket which is closed you will immediate receive a RST packet. This causes the eTCPState to transition from eCONNECT_SYN to eCLOSED but there is no notification on the event group so the user thread doesn't get woken up until it times out. The change in vTCPStateChange fixes this :-)