Daniel LynchandClaude Opus 5
1f3dc40c07
Repo hygiene round following a full review. No shim behaviour changes. Leak risk: - .gitignore: ignore CLAUDE.md (personal assistant-lane config, was one `git add -A` away from a public commit) and scratch_obj/. Docs vs. reality: - shim/README.md: rewritten. It described a pre-implementation skeleton with "core fns are TODO stubs returning -1005", three mutually inconsistent stub counts, and four completed milestones listed as open. Now carries the verified breakdown: 438/438 exports = 371 generated stubs + 46 core + 7 layers + 2 Vulkan queries + 12 passthru trampolines. - TESTING.md: dropped the self-contradicting "NOT yet" block (5 of 6 items were done or misstated, and contradicted the same file 45 lines above). Path B now points at tools/desktop-harness, which exists, instead of the orphaned shim/tests/harness.c. Path A prereqs marked as the record they are. - HOST.md: corrected the runtime assumption. The OpenXR runtime inside Lepton is SteamVR (vendor/etc/openxr/1/active_runtime.json -> vrclient.so), not Monado. Favourable: SteamVR emulates Oculus Touch by default and advertises the XR_FB_foveation family, so the existing input and foveation paths should carry over. The old "remaining unknowns" are resolved by Lepton's published source and replaced with the items to check before a first Frame boot. - README.md: same runtime correction. - docs/research/RECON.md: the four passages prescribing an entitlement NOP/stub/bypass are corrected in place rather than merely disclaimed by the top banner, which they contradicted. Build correctness: - shim/build_android.sh: missing patchelf is now fatal. It warned and exited 0, producing a .so that cannot resolve the OpenXR loader at runtime. - scripts/fetch_deps.sh + packaging/build_openxr_loader.sh: pin the OpenXR and Vulkan header versions (were tracking `main`), overridable via OPENXR_TAG / VULKAN_HEADERS_TAG; require cmake for the loader build. - packaging/steamframe_patches.sh: use the apktool.jar that fetch_deps.sh downloads. Its prereq check demanded an `apktool` binary on PATH that the documented setup never provides, so it could not run after a clean setup. - shim/gen_stubs.sh: it reads all_exports.txt, not shim_surface.txt; comment and emitted banner corrected. stubs.c regenerated (banner line only). - shim/src/core.c: split seven `if (out) ...; return ...;` one-liners. Host build now compiles with zero warnings, down from seven. Verified: host build 0 warnings; gen_stubs.sh output identical on regeneration; bash -n clean on all edited scripts; pinned header/tarball URLs return 200 and the tag tarball extracts to the expected directory name. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
4.0 KiB
libOVRPlugin.so shim (OVRPlugin v1.51 ABI -> OpenXR)
Drop-in replacement for RE4 VR's libOVRPlugin.so. It re-exports the ovrp_* C ABI the
game calls and implements it on OpenXR + Vulkan instead of Meta's libvrapi.so.
Status: working. RE4 VR boots and is playable on a Quest 2 through this shim on Meta's
OpenXR runtime — stereo rendering, head and controller tracking, buttons, grips, haptics,
and save loading. Steam Frame is the next target; see ../HOST.md.
Export surface
The artifact exports 438/438 ovrp_* symbols, matching ../analysis/all_exports.txt
exactly (0 missing, 0 extra), so any dlsym of any ovrp_ resolves even for the ~199
entry points RE4 never references.
| Source | Exports | What |
|---|---|---|
src/stubs.c (generated) |
371 | 103 Unsupported (-1004), 67 no-op Success (0), 201 NotYetImplemented (-1005) |
src/core.c |
46 | lifecycle, frame loop, poses, controller, system info |
src/layers.c |
7 | SetupLayer / GetLayerTexture2 / stage count / eye layer desc |
src/vk_session.c |
2 | the Vulkan instance/device extension queries |
src/passthru.c |
12 | trampolines for calls the game makes that need no real impl |
gen_stubs.sh regenerates src/stubs.c from ../analysis/all_exports.txt; functions
prototyped in the header are skipped there (67 of them) and implemented in core.c.
Regeneration is deterministic — re-running it on a clean tree produces no diff.
Layout
include/ovrplugin_shim.h— theovrp_*C ABI: types, structs, core prototypes. Independently written against the observed ABI, with_Static_asserts pinning the binary-verified struct sizes so a layout mistake fails the build rather than the device.src/xr_runtime.{c,h}— the OpenXR engine: instance/session lifecycle, the frame loop, swapchains, layer composition, foveation, perf levels.src/vk_session.c— Vulkan graphics binding, image enumeration, the tile-memory flush.src/layers.c— layer setup and eye layer geometry.src/xr_input.c— action sets and controller/haptics mapping.src/android_init.c— JavaVM/activity capture,xrInitializeLoaderKHR, Android instance create-info chaining.src/passthru.c— the author-only A/B diagnostic (see below).src/stubs.c— generated.
Roughly 48 distinct OpenXR entry points are used.
Build
The deployable artifact is Android arm64 — the game runs inside Horizon OS or Lepton:
../scripts/fetch_deps.sh # OpenXR + Vulkan headers, NDK, JDK, build-tools
./build_android.sh # -> build/arm64/libOVRPlugin.so
Needs the NDK (r27c) and patchelf (to add NEEDED libopenxr_loader.so). Then
../packaging/repack.sh swaps it into your own dumped APK.
Host build is compile-validation plus the desktop OpenXR harness, not a shipping artifact:
./build_host.sh # -> ../build/host/{libOVRPlugin.so,harness}
../tools/desktop-harness/run.sh
Needs libvulkan and an OpenXR loader; without the loader it builds the objects and stops.
Debug: passthru forwarding (author-only diagnostic)
debug.re4vr.passthru=1 hands the whole OVRPlugin session to a real, SONAME-patched
libOVRPlugin_real.so if one is present in the APK — an A/B harness for diffing native
runtime behaviour against the shim. No such binary is included or distributed by this
repo; you would have to place one from your own device dump, and without it the flag is
inert. All other debug.re4vr.* props are likewise diagnostics.
Known gaps
- Depth layer submission is gated behind
debug.re4vr.depthand off by default; Meta's runtime accepts but does not use plain KHR depth for reprojection. - Input binds
/interaction_profiles/oculus/touch_controlleronly. - Several
debug.re4vr.*paths are documented dead ends kept for reference (copy-ring, render-ahead, submithook); see../docs/research/for what each one ruled out.
See ../docs/research/SHIM-SCOPE.md for how the surface was scoped, ../TESTING.md for the
test plan, and ../docs/README.md for the research trail.