[xXPBSXx]

What

bind, connect and sendto on a libkernel socket checked the guest sockaddr length in Address() (Udp.cpp) with rules of their own: a length below 2 failed with EFAULT (14), and an AF_INET address below 16 or an AF_INET6 address below 28 bytes with EAFNOSUPPORT (47). FreeBSD 11.4 answers these differently:

• getsockaddr fails a length above SOCK_MAXADDRLEN (255) with ENAMETOOLONG (63) and a length below offsetof(struct sockaddr, sa_data) (2) with EINVAL (22).
• The protocols compare the length (sa_len) with sizeof(struct sockaddr_in) / sockaddr_in6 and fail a mismatch with EINVAL (tcp_usr_connect, in_pcbconnect_setup). A null address with a length of 2 to 255 is EFAULT (copyin), as before; a null address with a length below 2 now fails with EINVAL and one above 255 with ENAMETOOLONG, because getsockaddr checks the length before it copies (read in release/11.4.0).

Address() now returns these three errors. A full-size address is unchanged. Not changed, because the console's answer is not known: an address longer than sockaddr_in / sockaddr_in6 is still accepted (FreeBSD says EINVAL), and a different family still fails with EAFNOSUPPORT. Both are in the TechnicalDebt entry.

No title evidence: found by writing failure-path tests from the FreeBSD source (see Tested), no title I ran passes a short address. This is FreeBSD conformance.

Overlap: the overlap bot lists no pull request for this one, and git merge-tree of the head 9fc03d33a against main bf8dc610e is clean (checked, not rebuilt). The test registration sits before guest_socket_read_write in core/libs/prx/libkernel/CMakeLists.txt and the TechnicalDebt entry after the send, sendto entry in libkernel.md, away from the places the other socket PRs use. #5565 (Dripl1x, TCP_NODELAY, SO_NOSIGPIPE and SO_NBIO) conflicts with this PR in docs/dev/debt/libkernel.md (git merge-tree of the two heads, re-checked against main ce5c83953; both heads are clean against main); whichever lands second rebases the debt entry.

Tested

Windows 10 (19045), MinGW-w64 GCC 15.2.0 (WinLibs r7), Release, on main c2565f014; moved since to a newer main (git merge-tree against main bf8dc610e is clean; not rebuilt there) with its test registration and debt entry at their own places:

• guest_socket_address_length (new) passes: lengths 0, 1, 3, 8, 15 (AF_INET), 20 (AF_INET6) fail with EINVAL, 256 with ENAMETOOLONG, a null address with EFAULT, a full sockaddr_in still binds.
• It fails without the change to Udp.cpp (connect with a length below the size of sa_family must fail with EINVAL (22) (errno 14)).
• The existing socket tests (guest_udp, guest_tcp, guest_sce_net, guest_poll, guest_dup, guest_select, guest_socket*): pass together with the new one.
• python tools/check_conventions.py --base upstream/main passes.
• Measured on a combined build of the twelve open branches of this series (this PR and the others) merged onto main 0f2683cb7: guest_socket_address_length takes 0.21 s (ctest, serial, Windows). The full guest_* / libc / libkernel group passes (391 of 391, 0 failed; the build also contains the other branches of this series).
• Second commit 2e553b97c (test only): guest_socket_address_length now checks the result of socket(AF_INET6, ...) and closes that socket (before, a host without IPv6 would have answered EBADF instead of EINVAL, and the descriptor stayed open). Rebuilt and run on the new head (base 1e7f4967e, Windows 10, MinGW GCC 15.2.0, Release, -j4): guest_socket_address_length passes; with Udp.cpp taken back to the base it fails (connect with a length below the size of sa_family must fail with EINVAL (22) (errno 14)); python tools/check_conventions.py --base upstream/main passes.
• Not tested: Linux (one source for both hosts, only Windows was built), a console, a title run.

Checklist

• Based on current main; no other open PR implements the same functions (overlaps described above)
• One topic per PR; follow-ups go in a new PR
• No comments in code except technical debt
• Unimplemented paths throw (NotImplemented_nid_no_patch); silent stubs are listed in TechnicalDebt (docs/dev/debt/libkernel.md)
• No notes, investigation .md files or images added to the repository
• New third-party code is a submodule built from source (none added)
• Generic behaviour, not specific to one title
• Depends on: none
• Unblocks: none
• AI-assisted: yes

AI-assisted (Claude Code).

🤖 Generated with Claude Code

[oneandonlydean]

Closing: no title is known to reach this. With the queue at 700+ open PRs, please hold off on new PRs that don't name a title needing the change until it's down. If a title stops on this on current main, reopen with the title ID and the log line.