[alexpournaras]

What

Accept and store SO_ONESBCAST (0x10000) at the SOL_SOCKET level in sceNetSetsockopt, and read it back in sceNetGetsockopt.

Call of Duty: Black Ops (PPSA34501) sets this option when it starts its local server for campaign and zombies.

Previously, sceNetSetsockopt failed with EOPNOTSUPP, and the game later crashed in its Server thread while the mission loaded:

FATAL: unhandled exception 0xc0000005 on thread 6812 'Server'
  rip 0x00000001404700f0 app.exe+0x4700f0
  read of 0x000000000000001b

The implementation now:

• Stores the value per socket, defaulting to 0.
• Returns the stored value from sceNetGetsockopt, as shadPS4 does.
• Keeps the existing checks: a value shorter than an int fails with EINVAL, and other unknown options still fail with EOPNOTSUPP.
• Adds a guest test covering the default, set and read back, per-socket isolation, and a too-short value.
• Documents the behaviour under libSceNet technical debt.

Sends ignore the option, so this is a compatibility fallback, not the broadcast behaviour itself. FreeBSD 11.4's ip(4) documents the like-named IP_ONESBCAST. With it, a send to an interface's directed broadcast address, such as 192.168.2.255, leaves through that interface with its destination rewritten to 255.255.255.255. Here host routing picks the interface and the destination is not rewritten. shadPS4 stores the option the same way and does not change sends either. Its TODO reads the option the other way round (255.255.255.255 converted to the interface broadcast address) and notes that "hosts deliver it on the LAN anyway". The exact PS5 semantics, including the value read back, remain unverified.

Tested

Windows 11 Pro, Ryzen 9 9900X, RTX 2060.

• Without the changes: crashes in the Server thread when campaign or zombies game loads.
• With the changes: campaign and zombies start and load without crashing.
• Build and test results: Build succeeded; guest_sce_net, guest_net_ctl and guest_netctl_ap_dialog pass, including the new CheckOnesBroadcast.

Checklist

• Based on current main; no other open PR implements the same functions (no open PR handles SO_ONESBCAST; fix(libSceNet): read SO_RCVTIMEO and SO_SNDTIMEO as a timeval in sceNetSetsockopt #5967, fix(libSceNet): set an option on a libkernel socket through sceNetSetsockopt #6142 and fix(libSceNet): stop Sackboy loading retries on peer sockets #5836 change sceNetSetsockopt/sceNetGetsockopt for other options)
• 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 under Silent stubs in their library's technical debt file
• No notes, investigation .md files or images added to the repository (attach them to this PR)
• New third-party code is a submodule built from source
• Generic behaviour, not specific to one title
• Depends on: none
• Unblocks: PPSA34501 Call of Duty: Black Ops (PPSA34501) #5389
• GPUs tested: n/a
• AI-assisted: yes

[github-actions]

Other open pull requests touch the same code. Merging one will break the other, or both implement the same thing. If this one builds on them, list them in Depends on.

Pull request	Conflicting files (lines on main)	Same exports	Same new files
PR 5836	core/libs/prx/libSceNet/Export.cpp L89
PR 6142	core/libs/tests/GuestSceNet.cpp L1085

Some comparisons could not be completed because a pull request conflicts with the base branch. Review these shared files manually before concluding that there is no overlap.

Pull request	Shared files not compared
PR 6044	core/libs/tests/GuestSceNet.cpp
PR 6138	core/libs/tests/GuestSceNet.cpp
PR 6447	core/libs/prx/libSceNet/Export.cpp docs/dev/debt/libSceNet.md

[jotitadev]

First Linux run (the PR was tested on Windows only).

Machine: Linux x86_64, i7-8700, Ubuntu 26.04, GCC 15.2, Release, BUILD_TESTING=ON.
Head a51d7dcce merged onto main 66e290f1c.

• guest_sce_net, guest_net_ctl, guest_netctl_ap_dialog: pass.
• Test-only on main (only GuestSceNet.cpp from the PR): guest_sce_net aborts. With the CheckOnesBroadcast() call commented out, the same binary passes, so the abort comes from the new check (EOPNOTSUPP on 0x10000).
• Full non-GPU suite (-E "^agc|gpu|vulkan|video_out"): 648/650, same as base. The two failures (file_writer_binary/file_writer_text) are known on this host and also fail on base. One run showed a guest_host_standard_error timeout under load; it passed on rerun and in a second full-suite run.

No Linux-specific difference found.

Written with AI assistance (Claude Code). I ran the builds and tests above on my machine.

[blawesom]

Rebase handoff — rebased onto current main (authorship unchanged). One conflict in docs/dev/debt/libSceNet.md: your SO_ONESBCAST entry sits alongside main's newer SO_SNDTIMEO/SO_RCVTIMEO entry, both kept.

Verified on Linux (CachyOS, kernel 7.2.9, GCC 16.2.1, Ninja, Release, BUILD_TESTING=ON, AMD RX 9070 XT / RADV):

• builds; tools/check_conventions.py --base origin/main passes
• full suite: 1335/1336 — the only failure is the known main regression (agc_driver_tessellation segfault, feat(shader): read hull inputs at a control-point index computed from data #4727), not this PR

To pick it up:

git fetch https://github.com/blawesom/AnyPS5 rebase/6886
git checkout <your branch>
git merge --ff-only FETCH_HEAD

Head is 4b656882b.

[alexpournaras]

Thanks for the rebase and the full-suite run. I pushed 4b656882b as is. 🚀