The default output_res keeps the display's shape only while that shape has the
prompts' folder (cfg.c shape_keeps_prompts). After a switch to PlayStation
prompts on a 21:9 display the default was 1920x1080, so the start and the menu
asked only for the 16:9 folder and the game stayed 16:9. cfg_screen_size gives
the display's size while the player set no output_res.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
When the game resolution, the HUD width, the button prompts or the pad layout need a folder the install lacks (an
older setup, a failed setup step, a new display), totk.exe starts the setup's generator for it in the background
with the tools build-cache/rebuild.txt names, once per folder and process: at start for the settings in effect
(the resolution output_res asks for, also when the missing layout made this start 16:9), and from the settings
menu once a changed choice has stayed for 1.5 seconds or the menu closes. A missing 16:9 prompts folder is made too,
the PC prompts from totk.ini's bindings. The notes under Game resolution and Button prompts say the files are being
made and are used at the next start; when the tools did not start they say so and Try again starts them again.
The Prepare button is gone. Source builds without rebuild.txt and TOTK_ROMFS_PATCHES keep the old behaviour.
romfs_shape_missing and romfs_shape_prepare become romfs_folders_missing (a mask of the missing folders) and
romfs_folders_prepare; --probe-cfg and tests/romfs_shapes_check.py check the new decisions.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
tools/make_romfs_patches.py writes into <out>.part-<pid> and renames it to <out> once every file and the manifest
are there. With --background (for runs the game starts) it runs at below-normal priority, does nothing when <out>
already has a manifest, and drops its own result when another run finished <out> first, so two runs for one folder
do not mix their files and a folder a running game loaded is never written into. Without the flag <out> is replaced
whole, as the setup and source builds expect. A .part folder left by a stopped run goes once it is an hour old: the
next run for that folder or the setup's clean-up removes it. build_game.py copy_romfs_patches skips .part folders.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
prompts = playstation shows the face buttons as circle, cross, triangle and square, L1 R1 L2 R2, Options and Share.
The setup writes romfs-patches-playstation (pad_layout = nintendo) and romfs-patches-playstation-xbox (xbox, auto) on
every install, next to the Xbox folders, and the Shapes step and the settings menu make their shaped folders like the
other styles. playstation-xbox lays the action prompt's diamond out as xbox-label does, cross at the bottom.
Rebased onto dev from pull request #741 by fabiomode2: the shaped folders, hud_width, the Xbox diamond and the
keyboard prompts' text glyphs of dev are kept, and cfg.c's shape default uses input_prompts_folder.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
1. scan_pads took a controller Steam hid when the virtual pad was not up yet and kept it for the session, so Steam's
remapping was lost. While the open pad is a hidden one, pad_physical looks again every RESCAN_MS and scan_pads
moves the buttons to the virtual pad; the hidden controller stays open as motion_extra, so motion goes on with
the same bias and orientation.
2. pad_identity marked " (under Steam)" only on motion_extra; a hidden controller that became the pad lost the mark
and could take the bias calibrated with Steam closed. The mark now follows steam_hidden_id.
3. After pull request #1024: sensors_ready logs SDL_GetError when the pad has both sensors but enabling them failed,
instead of "has no gyro and accelerometer".
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A game Steam starts gets SDL_GAMECONTROLLER_IGNORE_DEVICES with the controllers Steam reads itself and sees Steam's
virtual gamepad, which has no motion sensors. input.c steam_reveal keeps that list and clears the hint before
SDL_Init (only with motion on); scan_pads still takes the virtual pad for buttons and sticks, so Steam's remapping
holds, and motion_source opens the hidden controller for its gyro and accelerometer only. A bias belongs to the pad
as Steam presents it: " (Steam Input)" or " (under Steam)" in gyro_bias_pad. docs/controls.md, "Steam Input and
motion", says the Steam layout must bind the gyro to something or Steam zeroes it on the Steam Controller.
The patch as posted against 6151ac4, with sensors_ready merged with #1274's restart. Tested by the author on Linux
(Steam Controller) and the Steam Deck; not on Windows.
Co-authored-by: St5sz2Jk <St5sz2Jk@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The calibration removes the bias measured once; a bias that moves as the controller warms up, or a controller never
calibrated, still turns the aim slowly. motion_fusion.c drift_track learns it from 256 ms windows in which the gyro
stays within 0.04 rad/s and the acceleration within 0.03 G peak to peak, after half a second of rest, with a time
constant of 0.25 s until the estimate settles and 30 s after. The estimate sits on top of the calibrated bias and is
cleared with a new controller, a new or loaded calibration and a change of gyro_calibration. New setting
gyro_auto_bias (live, default 1).
Taken from #380: the rest window by peak to peak, the two time constants, the per-axis limit. Left out: the
0.006 rad/s dead zone, which ate slow aiming; the one global state shared by every fusion instance; the reset at
every StartSixAxisSensor; the 0.5 G acceleration gate, which let a controller in motion count as still. The first
estimate is limited to 0.1 rad/s from zero instead of 0.15; a larger bias is the calibration's job.
Also: a sample without a reading at a restart (#1274) no longer returns before the calibration's log and save.
--probe-services checks the drift with the switch off and on, after a calibration whose bias then moved, a slow
aim in the hand, and the levelling of #1274 with a bias on every axis.
Co-authored-by: airquark <97342227+airquark@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
It acts only with sync_ahead in effect and is on by default there, so a player who turns on sync_ahead gets it; 0 and check stay for comparing. docs/settings.md keeps the row.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The new line is written for wb_eager 0, 1 and check alike, so the stall wb_eager removes can be compared between the settings. It costs a clock read only on the blocking path.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
dlaa_effects_after moves to the Screen tab under the DLSS rows, to [display] in totk.ini and totk.ini.example next
to dlss_char_mask, and vk_dlaa_frame reads it after the dl_active check, so a game without DLSS does no new work
per frame.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With sync_ahead on, the fence write-back of small GPU-drawn targets that the game reads back still blocked the game thread behind the running frame, about every ninth frame. wb_eager copies those targets into staging at the end of the submission that rendered them, so the write-back at the fence wait needs no GPU wait.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
build_game.py used to reuse build/<ver>'s generated objects only as a whole and only for generated/<ver>, so a private
lift of a patch change (gen.py --out generated/<ver>-prN, build_game.py --gen) compiled all 421 chunks, about 5 min on
12 jobs, for one changed chunk. main_objects now decides per object: same compiler, flags and rule in the main
build.ninja (with the main runtime/ and generated dir in place of ours), main build up to date, and every file the main
compile read according to its Ninja deps log byte-identical on our side and unchanged since the object was written
(NTFS ChangeTime, st_ctime elsewhere), the compiler binaries too. A file new on our side whose name those compiles
included or probed blocks the reuse. Anything unknown compiles. The build prints "reused N of M generated objects",
--verbose lists the rest with the reason; obj/reused.txt lets a later build delete a copied object before compiling it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
recomp/patches-v121.txt gets set fov on the two camera routines the aspect set already patches, at the load of the
camera's vertical view angle (+0x13c): camera update main+0xcb4ad8, camera setup main+0x1252194. The update site matches
1.4.3's instruction for instruction around the load; the setup site copies the camera's fields with other instructions
and registers before it and matches 1.4.3's from the load on. Both call sinf, cosf and tanf in the same order as on
1.4.3. Each version fills sead's perspective projection this way in ten places, each with its counterpart in the other
version: these two camera copies, which are the only places 1.4.3 patches, and eight others (inlined copies, projection
routines and four projections on the stack), none patched. The replacement words are those of the 1.1.0 and 1.4.3 sets
(x10 kept on the stack, s8 overwritten by the next instruction), so no register liveness is assumed. #621 (ipavlin98,
closed by its author) chose the same two sites with x8 and s2 as scratch registers.
The setting's text in totk.ini.example and in the four menu languages names 1.2.1, and the comment of
runtime/resolution.c res_fov_live names the patch files of all three versions.
Co-Authored-By: Claude <noreply@anthropic.com>
The game thread hands its sound commands to the audio thread in lists: 16, of which 4 are always held, so at most 12
wait in the command ring. The game thread submits at most one per update (1.2.1 main+0x9bd144), the audio thread
takes one per 5 ms audio frame (after a list, main+0x883780 goes on past the ring's check). Above about 200 updates a
second the ring stays full. The game's guard (main+0x9bcdc4) counts the updates that find the ring of free lists
empty and exactly 12 lists in the command ring (0x9bce1c cmp w9, #0xc); the count returns to 0 only when a free list
is there. At every multiple of 300 (10 s at 30 fps, 1.25 s at 240) it clears the sound command switch
(main+0x2268bfc) and stops every active voice (main+0x226f29c). Only the switch's constructor sets it to 1
(main+0xf31ce8, from the engine's init main+0xf31530), so the game is silent until it is restarted.
Measured on 1.2.1 (Linux, RX 6750 XT), same save and scripted input, mix dumped without a device: with fps_limit 60,
144, 180, 200 and 201 the mix has sound; with 205, 210, 220, 230, 239 and 240 it goes silent after the boot logo and
stays silent, also at the 130-148 fps of the scene that follows. The guard is read from the code; the switch going
to 0 has not been observed in a run.
The set audiodrain (TOTK_AUDIO_DRAIN_FIX, audio_drain_fix, on by default) sends the audio thread back to the ring's
check after each list, so it takes every queued list in the pass: one site per version (1.2.1 main+0x883780, 1.4.3
main+0x149244), each going to a target the game's own code already branches to in the same way (1.2.1 main+0x883c18,
1.4.3 main+0x149700). The setting is in runtime/cfg.c, totk.ini.example and docs/settings.md.
At fps_limit 240, 120 s runs of one build per game version with the set switched, seconds with sound in the mix: on
1.2.1 5 of 118 with the set off and 108 of 118 with it on; on 1.4.3 2 of 118 off and 110 of 119 on. Not run:
Windows. Not covered: 1.1.0, whose main we do not have, so its table has no site and it keeps the game's code.
Co-Authored-By: Claude <noreply@anthropic.com>
plugin_init.c reads TOTK_PLUGIN_TRACE through cfg_bool, which stops the game when the key is unset and metas[] has no
row. Every test run set the key, so only a plain start hit it. tests/settings_doc_check.py now fails on any
cfg_value/cfg_int/cfg_double/cfg_bool reading without a metas[] row that is not guarded by cfg_get of the same key.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
tools/build_game.py --portable writes build.ninja paths on the tools' drive relative to the kit, also through '..'
(the generated C and objects in build-cache/work), names the compiler, windres and ar the same way (backslashes:
CreateProcess does not find a relative program with forward slashes), and passes clang -resource-dir and --sysroot
relative to the kit, because clang names its own headers by its absolute folder whatever argv[0] says. The Ninja deps
log then holds no absolute path. installer/steps.py build_args passes --portable for the build kit.
Measured on an install moved after a portable build: the rebuild compiled modules.c only, 3.5 s; before, the move cost
a full compile of 430 objects (335 s on 12 cores). Without --portable build.ninja is byte-identical to before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>