Introduce IPv6, unit tests pass #3282
Conversation
scaprile
commented
Sep 23, 2025
scaprile
commented
- IPv4 keeps working, all tests pass
- IPv6 I can't test. Some address-related stuff has been refactored to be more "cesantish" and to accomodate link-layer independence
src/drivers/stm32h.h
Outdated
| #if defined(MG_ENABLE_IPV6) && MG_ENABLE_IPV6 | ||
| #define MG_IPV6_INIT(mif) \ | ||
| do { \ | ||
| memcpy(mif.ip6ll, (uint8_t[16]) MG_TCPIP_LINK_LOCAL, 16); \ |
There was a problem hiding this comment.
Are we going to keep this approach for initializing static configurations?
There was a problem hiding this comment.
I'd rather not, but let's start the game with something.
I added 64-bit htonll functions to help with other approaches.
|
|
||
| #if defined(MG_ENABLE_IPV6) && MG_ENABLE_IPV6 | ||
|
|
||
| #ifndef MG_TCPIP_GLOBAL |
There was a problem hiding this comment.
This is a weird macro name. What does it even mean? Shall we rename it?
| uint8_t hops; // Hop limit | ||
| uint8_t src[16]; // Source IP | ||
| uint8_t dst[16]; // Destination IP | ||
| uint64_t src[2]; // Source IP |
There was a problem hiding this comment.
Because comparisons are faster and sentences shorter and functions more readable with uint64_t
| #endif | ||
|
|
||
| static bool tx_udp(struct mg_tcpip_if *ifp, uint8_t *mac_dst, | ||
| struct mg_addr *ip_src, struct mg_addr *ip_dst, |
There was a problem hiding this comment.
I suggest to rename
struct mg_addr *ip_src
to
struct mg_addr *src_addr
becase ip_src->sport down below looks weird.
There was a problem hiding this comment.
IIRC, I maintained the same naming scheme across all functions. In fact, using 'addr' doesn't make sense: addr->port, ip does make sense, it is the IP layer information and it is the source endpoint. 'ip' doesn't mean "address", it means IP.
If I use 'l3' you will sure not like it... 'ip' is OK for me, 'addr' is not, ip->ip4 and ip->port make sense, addr->ip4 and addr->port do not
Maybe we should use l3, or net, or nwk, or "half-socket"...
