[DavidSamir]

What

sceNetCtlGetInfo finds the host's default route in /proc/net/route, which macOS does not have, so on macOS every IPv4 query throws:

sceNetCtlGetInfo: the host routing table is unavailable

On macOS it now reads the IPv4 routes that have a gateway with sysctl({CTL_NET, PF_ROUTE, 0, AF_INET, NET_RT_FLAGS, RTF_GATEWAY}), the table netstat -rn shows, and keeps the routes to 0.0.0.0 that are up, neither rejecting nor host routes, on an interface that still exists. macOS keeps an unscoped default route, the one it sends traffic by, beside interface-scoped ones (RTF_IFSCOPE), which rank after it. Address, netmask, MTU and DNS servers already come from getifaddrs, SIOCGIFMTU and /etc/resolv.conf, which macOS has. Linux and Windows are unchanged.

The parsing is exported as NetCtlDefaultRoutes_nid_no_patch, like NetCtlReadByteCounter_nid_no_patch, so a test can feed it a routing table built by hand: an unscoped default route, a host route, a host route to 0.0.0.0, a rejecting default route, an interface-scoped one, one that is down, one on an interface that does not exist and a truncated last message.

sceNetCtlGetIfStat keeps throwing on macOS, now with a debt entry: an unprivileged process reads the interface byte counters, through getifaddrs and sysctl(NET_RT_IFLIST2) alike, reduced to their low 32 bits (en0 here: 426753024 received against 47671393472 in netstat -ib, 11 times 2^32 less), so a total would wrap every 4 GiB.

Commits: the macOS route lookup, its test, the debt entries.

Tested

macOS 26.7.1, Apple M2 on Wi-Fi, AppleClang 21.0.0, x86-64 build under Rosetta, Release.

main does not build libc and libkernel on macOS for other reasons as well (#4627, #4630, #6124 and mugurc's macOS branches address them), so the tests ran in a local tree of main ce991f1 with those merged.

• main: guest_net_ctl aborts with the error above.
• This branch: guest_net_ctl passes 5 runs of 5. The values it reads match the host: address 192.168.0.102, gateway 192.168.0.1 (route -n get default), MTU 1500, first IPv4 DNS server 192.168.0.1 (/etc/resolv.conf).
• The new case fails when host routes are kept (aborts) and when scoped routes are not ranked after the unscoped one (aborts).
• tools/check_conventions.py --base origin/main and git diff --check pass.

Windows 11 (WinLibs GCC 15.2.0 r7, UCRT, Release) and Ubuntu 24.04 under WSL2 (GCC 13.3.0, Release), head 80a9ef9e3 merged onto main 5d9dcc4: guest_net_ctl pass 3 runs of 3 with the change, and 3 runs of 3 with libSceNetCtl/Export.cpp as on main and the tests kept. It passes either way there: their code paths are unchanged and the new test case is compiled on macOS only.

Overlap: #6138 (callbacks) edits the declarations and the start of main in GuestNetCtl.cpp, and #6039, #6138 and #6227 edit the sceNetCtlGetInfo entry next to the new debt entries; none of them changes how the route is found. #6044 moves GuestNetCtl.cpp. Whichever lands second rebases.

Checklist

• Based on current main; no other open PR implements the same functions
• 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: none named; ProsperoLichess (PPSA99009) and Prospero Explorer (PPSA99109) import sceNetCtlGetInfo, but on macOS neither reached it in a 45 s run
• GPUs tested: n/a
• AI-assisted: yes

[github-actions]

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 6039	docs/dev/debt/libSceNetCtl.md
PR 6044	core/libs/tests/GuestNetCtl.cpp
PR 7201	core/libs/prx/libSceNetCtl/Export.cpp

[DavidSamir]

Windows and Linux results for 80a9ef9, merged onto main 5d9dcc4:

Host	Toolchain	With the change	Code as on main, new tests kept	Restored
Windows 11	WinLibs GCC 15.2.0 r7 (UCRT), Release	guest_net_ctl 3/3	3/3	1/1
Ubuntu 24.04 (WSL2)	GCC 13.3.0, Release	guest_net_ctl 3/3	3/3	1/1

Reverted for the middle column: libSceNetCtl/Export.cpp. The test passes either way on both hosts, as expected for a macOS-only routing-table path. Windows and Linux still build and pass with it.

AI-assisted: yes

[blawesom]

Rebase handoff — rebased onto current main (authorship unchanged). One docs-debt conflict, unioned.

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: 1338/1339 — 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/6685
git checkout <your branch>
git merge --ff-only FETCH_HEAD

Head is 90d1f458a.

[DavidSamir]

Thanks, @blawesom. I checked your rebase/6685 against this branch: the code commits are identical, and the debt-file commit differs only by the conflict resolution. Rather than force-pushing, I merged current main into the branch with the same resolution, so the history stays as reviewed. The PR's diff against main now matches your rebase. Head is 26bc5ce87, and it merges cleanly with main.

AI-assisted: yes

[adriadam10]

Checked the diff: the macOS path is confined to __APPLE__, the routing-message parser stays inside rtm_msglen/sa_len and the buffer size, the filters (up, not reject/host, destination 0.0.0.0, existing interface) and the scoped-route ranking match the description, the test pins values on a hand-built table including the truncated last message, and the debt entries are in place. Linux/Windows code paths are unchanged. macOS runtime behaviour relies on the author's run; merged.