Hi @ljanyst , thanks for this PR.

I am curious, what inspired you to add support for using EUI64 IPv6 addresses?
As I look at your code, it will seldomly be used, only in case a proposed address has a collision, ie. when bIPAddressInUse is true.

it will seldomly be used, only in case a proposed address has a collision

It is not the case at all. The state machine for router advertisements starts with bIPAddressInUse = pdTRUE_UNSIGNED. See: https://github.com/FreeRTOS/FreeRTOS-Plus-TCP/blob/main/source/FreeRTOS_RA.c#L416

The reasoning for the feature is explained here: #1150

It is not the case at all. The state machine for router advertisements starts with bIPAddressInUse = pdTRUE_UNSIGNED.

Sorry, you are right. It is a splendid solution! I tested it and it works on my LAN with RA router.

The only "but" would be that we change the signature:

BaseType_t FreeRTOS_CreateIPv6Address( IPv6_Address_t * pxIPAddress,
                                       const IPv6_Address_t * pxPrefix,
                                       size_t uxPrefixLength,
+                                      const IPv6_Address_t * pxHost )
-                                      BaseType_t xDoRandom )

If the function is called with a 0 or a 1 for and xDoRandom, can we detect that?

Otherwise I like it, and it works well!
Thanks

I guess that depends on the compiler and verbosity settings. 0 is a valid value for a pointer, but 1 is extremely unlikely to be one in a hosted environment. This function was used in the code exactly once and I don't think it's likely to have been used by a external caller. On top, the previous interface actually did not make much sense. An IPv6 address with all zeros in the host part is reserved for router anycasts by RFC4291.