- Restart bits: a group waits only when the edit touches it (next differs
from the live value) and the result differs from boot. Dynamic
resolution moves the live internal resolution off boot, so an in-game
texture filter change relaunched the game. test_menu_video_snapshot
covers both cases.
- Return to the frontend from a local game opens the launcher as at
boot; only a lobby match start (the runtime's pending-launch exit, or
the provider's lobby view asking once its match started) keeps the
launch and resumes the room.
- Default provider: with no title choice, the UI library's provider is
used when it ships one (recomp_ui_psx_menu.h), so recomp-ui stays the
default; PsxMenuHost.default_selection tells it to keep the title's
default presentation.
- Debug: menu_host reports provider, open, pause, capture and snapshot.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Review follow-up on #661 (#636):
- test_depth24_present_hold_vblank.py reads sources as UTF-8 (cp1252 on
native Windows Python raised UnicodeDecodeError).
- The present skip is now guest state: set at guest VBlank, read (not
consumed) by the present and packed into the GPU snap's hold word (bit 8;
older snaps derive it from the count). The executable test showed the
host-side latch made the present of a restored tick differ from live on
the last held VBlank (hold 0, skip 1), exposing stale trailing FMV
columns after a rollback load.
- Depth24 scanout staging and trailing-column blanking move from main.cpp
into gpu_depth24_stage_scanout (gpu.c), so the test renders through the
same code every CPU depth24 present runs.
- New depth24_present_hold_exec_test: drives real gpu.c through a 15-bit to
24-bit FMV, GP1(07h) retargets and mode transitions; compares hold, skip,
upload span and snapshot bytes across windowed/headless/deferred presents,
rollback replay from every tick and disk loads, and checks staged frames
never show stale movie pixels while the new movie stays visible.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
psxrecomp owns the contract between the runtime and any in-game menu
frontend. A provider (PsxMenuProvider: attach, event, is_open, set_open,
draw, detach) is registered by the title with psx_menu_set_provider();
none is registered by default, so a title's UI and pin are unchanged.
The runtime offers every provider the same host (PsxMenuHost): pause
(P1, refused in netplay), video get/apply on PsxHostVideoSettings (P4),
mod switch/option set (P5, via the new psx_mod_feature_enable /
psx_mod_feature_set_option shared with the launcher ABI provider),
reload bindings, exit (quit, back to the frontend loop in-process, or a
quick relaunch with the original arguments), disc/tray slots filled by
the tray policy (psx_menu_set_disc_ops), and optional helpers on the
shared launcher ABI both frontends already speak (game info with the
mod catalogue and lobby, settings get/apply with the handoff's filter
translations).
Policies live here, the same for every provider: the lobby is pumped
during play and a started match ends the local session and starts in
the frontend loop (pending launch, seats and signals kept, pumping
stopped so the match port stays free), the menu is closed and its pause
dropped first; the provider detaches before its GL context goes away at
session teardown and attaches to the next one. --launcher sets
RECOMP_LAUNCHER_PREGAME for frontends that boot into the game;
PSX_HIDDEN_WINDOW is a hidden-window test hook that keeps the frontend.
Nothing here includes or names a frontend's own headers or types.
Tests: psx_menu_host_test, host_overlay_test (relaunch quoting).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
psx_mod_set_option_live records a session override for a committed
feature's option and calls the plugin's new option-changed callback in
that plugin's context; psx_mod_option_value answers the override. It
returns 1 when applied live, 0 when the change waits for a restart, -1
during netplay (the plan is negotiated between peers). Nothing changes
unless an overlay calls it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
#655 landed the same PSX_TEXTURE_FILTER_* enum and translation in
video_texture_filter.h/.c; drop this PR's duplicate (link error) and
include that header.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Review (#601/#603): the live path and the launcher handoff both collapsed
the texture filter to a boolean (value ? 1 : 0), so neither could carry xBR
(#586) and they could drift from settings.toml. PSX_TEXTURE_FILTER_* is now
the one enum (equal to [video] texture_filtering / gr_set_texture_filter),
and psx_texture_filter_from_ui is the one translation used by the launcher
handoff (seed, UserSettings, live globals) and the in-game apply. Unit-
tested in host_overlay_test.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Review (#601):
- psx_video_apply_live had no netplay guard. In a match only presentation
(fullscreen, scanlines, linear present) applies live; dynamic resolution,
frame generation and the filters are saved and reported with the new
PSX_VIDEO_RESTART_NETPLAY bit (next session).
- internal_resolution and present_thread are written to UserSettings like
the other restart-only fields, so the choice survives a restart.
- The live-apply code sat inside #ifndef PSX_NO_DEBUG_TOOLS but is used by
the launcher path, so Release + launcher did not compile. It now lives
outside the debug-tools block.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
psx_video_apply_live applies the [video] keys a running game can change
(fullscreen, texture and FMV filter, scanlines, linear present, dynamic
resolution, frame generation) through the existing live setters, saves
them to settings.toml, and reports renderer / internal resolution /
thread changes as restart-only. Nothing calls it without a provider.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
psx_host_set_input_sink: an overlay provider sees host SDL events first,
in the event drain and the host pause loop, and may consume them.
psx_host_set_ui_capture: while on, local game input reads released and
centred (netplay included, so peers receive "nothing pressed") and the
external-input gate stays closed. Debug builds gain host_key, which pushes
a real SDL key event for scripted runs. Inert without a provider.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Review (#599): who keeps the callback context alive until render-thread
calls are quiescent. Draw calls now run inside an in-flight count, and
set_draw_cb (replace or NULL) swaps the record and waits until no call
that read the old one is running, then frees it. When it returns, the
registrant may free ctx and its GL objects; it must unregister before
destroying them or the context. A change from inside the callback does
not wait. Documented in host_overlay.h. Tests: a reader thread drawing
while the owner unregisters and frees (300 rounds; fails without the
wait), and self-unregister from the callback.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
psx_host_overlay_set_draw_cb: the OpenGL backend calls it after the game
frame and host OSD layers are composed, before present_shot capture and
the swap, on the thread that owns the GL context. Without a registered
callback nothing changes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
psx_host_pause_push/pop: a nesting, host-owned pause. While it is held the
runtime freezes the guest at the vblank present (as the save-state menu
does), keeps presenting the last frame and pauses audio. Push refuses
during netplay. Nothing pushes unless an overlay provider (or the debug
server's new host_pause command) asks, so behaviour is unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The batched scaled charge fell through to the scale-1 batch path, which
also defers while g_psx_cyc_bb_defer is nonzero and routes through the VLC
local sink. Both are host call-stack state (a longjmp escape leaks or
resets the depth), so which charge published the batch -- and so which
clock each device saw -- differed between hosts. R4 2P rollback Mac<->PC
with modern controls split by a few cycles in live play (clk/tim fork at
~3296 with cpu/ram equal), then baseline mismatches and a realign storm
dropped the match; with the batch off the same pair matched.
The scaled path now batches on the guest clock only (device deadline or
the 64-cycle soft cap). New check: the service log is identical with
bb_defer 0 or 338 and with a local sink installed (fails before). Re-stamp
the OpenBIOS emitter sha (psx_cyc.h is in the codegen hash).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With [timing] guest_cycle_scale active, every one-cycle psx_cyc_charge went
straight to psx_cpu_charge and psx_advance_cycles, bypassing the batch the
scale-1 path uses. [timing] guest_cycle_scale_batch (PSX_GCS_BATCH=0/1)
scales the charge in place, with the same fractional carry and order as
psx_cpu_charge, then batches it like the scale-1 path. It flushes at the same
observation points. Default off.
R4 exactness, frame_fingerprint off vs on, 9000 frames each: headless boot
through intro, title and attract; a 1P Grand Prix race; and a 2P VS race.
cyc, mmio, mc and qc are identical on every frame. sp, wr, pc and ws are
identical at the end. The only differences are single-frame VBlank straddles
of wc/sc (never two frames in a row). 0 dispatch misses.
Speed, release builds, races at 60 Hz:
- M4 (macOS sample): psx_cpu_charge 4.3 % -> 0 % of emulation-thread samples;
busy 46.0 -> 44.4 % (2 pairs).
- 9800X3D: no measurable change (30.0 vs 30.5 %; one core 44.6 vs 44.2 %).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The detector cannot prove a loop has no later counter-dependent exit, so
the extension now skips only at block-edge PCs the title lists after an
audit. Unlisted sites are never skipped; an empty list disables it. Docs
record R4's v_wait audit as the reference; tests cover unlisted sites.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Last-store tracking: memory.c owns the record and writes it only while
the extension is on (g_psx_track_last_store); main.cpp hands it and the
host writer (psx_host_write_word) to psx_cycles.c through
psx_idle_skip_set_store_tracker / _set_host_writer, so psx_cycles.c no
longer references memory.c (fixes the tests that link one without the
other).
- Exact from boot: the device distance is now measured from the current
cycle (devices report it from their last service, up to a few hundred
cycles behind), and several edges per pass are accepted only together
with a counter store. Before this, R4's boot diverged at frame 962 (an
OpenBIOS multi-edge poll and a late VBlank IRQ after a v_wait skip).
- tests/test_idle_skip_store_counter.c: a v_wait trace skips (whole
passes, counter lowered by the host writer, never to the timeout, lands
before the event); an extra store, an MMIO read, a decrement of 2, a
store to MMIO, the option off and no host writer do not.
- generated/OpenBIOS.emitter.sha re-stamped.
R4 frame_fingerprint off vs on, 9000 frames each: boot through intro and
attract, a 1P race and a 2P VS race: cyc, mmio, mc, sp, sc and qc
identical on every frame, 0 dispatch misses.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The dispatch template emitted since mod-guest-functions landed declares
extern g_psx_mod_guest_functions / psx_mod_dispatch_guest_function in
every BIOS dispatcher, but the module name-mapping block and
PsxBiosModuleCallbacks never carried them, so a freshly built retail
backend (e.g. SCPH1001) failed RTLD_NOW with an undefined symbol —
both in the builder self-check and in the real dlopen. Add the two
mappings, table entries, glue dummy/forwarder, and host wiring; bump
the ABI so stale caches cannot load. Verified: SCPH-1001 module builds,
self-checks, and boots SLPS-00756 on Linux/gcc.
Smooth motion lost most of its generated frames with bloom on (R4 race,
120 Hz: ~30% of bloom-off). Measured and fixed in order of cost:
- Depth-stencil blits (CPU milliseconds each on macOS) are gone:
generation surfaces no longer copy the marks and are not marked while
drawn (generated frames reuse the newer real frame's mask); the
native-wide centre copy is colour only again and the mask takes the
centre's marks from the hr surface; the mask itself is resolved in the
presented surface's own pixels through an R8 target that shares the
surface's depth-stencil renderbuffer, then mapped to the window.
- Marking needs one pass per batch, not two: 3D batches only set, 2D
opaque ones only clear, semi-transparent ones only set; only batches of
unknown make (queued replays, lines) still do both.
- No full-size copy of the image: `source_down` (a blit straight to the
pass's size) feeds the bright pass, and the composite screen-blends onto
the window (`blend screen`; unlit pixels are discarded, so they keep their
exact value). The debug split/bloom-only views go with the copy.
M4, R4 race, render thread + reprojection, PSX_FRAME_GEN_FORCE, a host
loaded by other work (absolute rates low): bloom pass on real / generated
frames 1.2 / 0.7 ms at 1080p, 1.3 / 0.9 ms at 1440p, 2.6 / 1.6 ms at 4K
(glFinish-bracketed); generated frames with bloom 79-80% of bloom-off at
all three (was ~30%). A 1/8-resolution blur chain was tried and kept no
more frames, so the 1/4 chain (the look) stays.
Tests: parser (blend, source_down, bloom preset reads no full copy), GL
fixture (source_down + blend screen chain exact; HUD and mask checks
unchanged), FXAA unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Review of #584: the default must not make guest-visible VRAM depend on the
internal scale. Mode 1 ("on", default; "scaled" kept as an alias) now
puts the PS1 pattern on the native VRAM grid; mode 2 ("fine") is the
elective per-internal-pixel pattern. gl_dithering_test checks the default
mode's native VRAM is identical at 4x and 1x (equal-colour Gouraud now
averages 15.50 in the default at every scale). Docs keep the
semi-transparent approximation explicit. OpenBIOS stamp regenerated.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The OpenGL renderer has always drawn shaded and texture-modulated pixels
at 8 bits per channel with no dither pattern (true colour). That stays the
default. 'on' applies the PS1 4x4 offset matrix and 5-bit truncation per
internal-resolution pixel; 'scaled' puts the pattern on the native grid,
so it reads like the console at any scale and the native VRAM is the same
at every internal resolution.
GP0(E1h) bit 9 reaches the backend through a new optional facade hook
(gr_set_dither) recorded for the render thread and Smooth motion redraws.
With the mode off nothing flushes and no uniform changes. Flat, raw and
other 5-bit-exact pixels are never touched.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Matt Stanley