COBALT_BOOT_STATUS_2026-11-17b.md
24548 bytes · Original / download
Cobalt boot: second session (17/11/2026)
Test title: YTVCB0001 (diagnostic only, NOT a functional app).
Runner: tools/test_cobalt_boot_vita3k.ps1 [-Seconds N] [-TitleId T -Vpk V]
(syncs every changed VPK file into the installed title, captures cobalt_boot.log,
starboard15.log and vita3k.log to tools/capture/cobalt-boot-<timestamp>/,
reports SELF_EXIT/EXIT_CODE, and never touches other Vita3K instances).
Root causes found and fixed (in order)
| # | Symptom | Cause | Fix |
|---|---|---|---|
| 1 | Start hung after eglCreateContext |
SbEglInterface.eglCreatePbufferSurface (and 21 other slots) were null; Cobalt's GraphicsContextEGL calls it right after creating the context, so it jumped to 0 |
port_graphics.cpp: real EGL 1.4 layer over VitaGL: pbuffer registry, per-thread current context (EGL_BAD_ACCESS across threads), per-thread errors, every slot filled (real call or proper EGL error) |
| 2 | GXM was never initialized | VitaGL's eglInitialize is a facade; vglInit was never called |
eglInitialize brings up VitaGL once (RAM pool capped at 48 MiB) |
| 3 | Rasterizer spun on null+0x64 | Skia rejected the GL interface: glInsertEventMarkerEXT missing although VitaGL advertises GL_EXT_debug_marker; GLES2 core funcs missing (glBlendColor, glGetShaderPrecisionFormat, ...) |
Our eglGetProcAddress serves marker + EGL overrides; GLES2 core slots implemented (shader precision = SGX fp32; renderbuffer/texture parameter queries tracked) |
| 4 | abort() during font init | Evergreen CRX ships an empty fonts.xml; platform must give system fonts | tools/stage_vita_system_fonts.py stages Cobalt's official 'limited' package (65 files, 2.9 MiB) to system_fonts/; SbSystemGetPath(FontDirectory/FontConfigurationDirectory) = app0:/system_fonts |
| 5 | glLinkProgram hung, then Vita3K crashed | VitaGL's GLSL preprocessor hangs on Cobalt's #pragma array (a build-time directive of generate_shader_impl.py) |
glShaderSource bridge drops pragmas GLSL ES requires drivers to ignore (keeps #pragma STDGL). Reproduced/validated in graphics test |
| 6 | V8 fatal "ignored" | SbMemoryMap(flags=0) (kSbMemoryMapProtectReserved) rejected, and partial SbMemoryUnmap (musl alignment trim) rejected |
Reserve = memblock + PROT_NONE; page-granular unmap with bitmap; map table 64 → 1024 |
| 7 | Memory budget | Evergreen's malloc is musl over SbMemoryMap, not newlib | newlib heap 8 → 64 MiB (uses ~3-4 MiB), main stack 4 MiB, default thread stack 1 MiB; ~105 MiB left for Cobalt maps |
| 8 | TLS rejected every server (no CA lookup) | SbDirectoryGetNext required a buffer of 256 but Cobalt passes kSbFileMaxName = 255, so ssl/certs listed as empty |
accept >= kSbFileMaxName |
Proven in Vita3K (build 4242)
SbEventHandle(Start)returns (COBALT_EVENT_START_RETURN); event pump runs.- 74/74 constructors, EGL context made current on main and rasterizer threads, Cobalt's 8 GL programs compile and link (shaders genuinely run through VitaGL/SceShaccCg).
- Cobalt resolves
www.youtube.com, completes the TLS handshake with its own trust store (140 CA files opened) and exchanges HTTP/2 traffic (~20 KB received). - Graphics test 31/31, Starboard regression 179 (+1 new test pending a run), 0 failures in the last complete run.
Not achieved yet
- No frame presented yet (eglSwapBuffers never called); no YouTube UI.
- New blocker: after the YouTube response arrives, V8 hits a fatal CHECK (official build, message stripped) on thread 691; heuristic backtrace starts at ELF 0x1c07d7c → 0x15d7b44 (V8_Fatal) → 0x1c065b8 → 0x1be5d3c → 0x1302714. Memory is fine (105 MiB free, no map failures). Next: identify that V8 check.
- SbPlayer still unavailable.
Diagnostics left in the code
YTV_GL_TRACECMake option (OFF by default; the current build dir was configured ON).ux0:data/youtubevita/cobalt_args.txt: extra Cobalt switches (e.g.--url=).ux0:data/youtubevita/diag_hide_neon.txt: hides NEON (diagnostic; ruled out as TLS cause).- Heartbeat, VM, net, socket-waiter, SSL-file and backtrace logs.
Update (same day, later): first frames and the YouTube device gate
Fixed:
- GLES2 pixel store: VitaGL's glPixelStorei only accepts GL_UNPACK_ROW_LENGTH; GL_UNPACK/PACK_ALIGNMENT raised GL_INVALID_ENUM and Cobalt's GL_CALL aborted (it was a rasterizer CHECK, not V8). Bridge now keeps alignment/row-length state and repacks uploads/readbacks.
- GL_DITHER (glEnable/glDisable/glIsEnabled), GL_IMPLEMENTATION_COLOR_READ_FORMAT/TYPE (RGBA/UNSIGNED_BYTE) and GL_NUM_PROGRAM_BINARY_FORMATS (0: VitaGL never names its format) answered by the bridge.
- With a diagnostic wrapper that checks glGetError after every GLES call (YTV_GL_TRACE): 0 GL errors in a full Cobalt run.
- Graphics test 33/33 (pixel-store coverage added). Probe pump duration configurable via ux0:data/youtubevita/pump_seconds.txt.
Observed:
- Cobalt presents frames (eglSwapBuffers from the browser). GL trace of frame 1/2: clear with 0.157,0.157,0.157 (= page background #282828), splash images 426x96 and 540x540, glyph atlas 2048x2048 + text draws — i.e. the guest renders the real page.
- The Vita3K window nevertheless shows plain white (OpenGL backend). Guest-side readback of display buffers returns zeros in Vita3K, so it cannot be used as evidence. Vulkan could not be selected (--backend-renderer Vulkan and the yml key were not applied by build 4242 in these runs). Pending: the Vita3K presentation fix the user applied for Darkest Dungeon; validation on real hardware.
- YouTube device gate: www.youtube.com/tv answers this honest device identity (Sony / PlayStation Vita / GAME_CONSOLE) with the page "O YouTube não está disponível neste dispositivo" (QR to yt.be/tv-devices); the TV app bundle is never requested. Reproduced from the PC with a Cobalt 25/Starboard 15 style User-Agent. Getting the real TV app would require presenting as a certified device; NOT done (user decision, service terms).
Tools: tools/capture_vita3k_window.ps1 (PrintWindow of the test title's window), runner options -Config, -Backend.
Update: official YouTube TV UI rendered in Vita3K
- Capture: Windows screen capture of Vita3K's OpenGL window shows white.
tools/opengl_capture(from Black Reliquary Vita; SwapBuffers hook + glReadPixels, injected only into the test's own Vita3K) shows the real framebuffer. Runner:-CaptureFrames -CaptureIntervalMs N -CaptureMax N;tools/opengl_capture/bmp_to_png.py. - Device identity (owner decision):
YTV_DEVICE_IDENTITY_PS4=1→ model "PlayStation 4", year 2013, firmware 12.00 (friendly name stays PlayStation Vita; no certification scope/keys). /tv then serves "YouTube na TV" (TVHTML5) instead of "not available on this device". - Result: Cobalt loads the TV app, fetches i.ytimg.com thumbnails, googlevideo.com, and renders the onboarding screen ("Watch the latest videos, just for you" / Get started). GIF:
tools/capture/cobalt-boot-20261117-111921/youtube_vita_boot_vita3k.gif. - VitaGL splash removed:
third_party/vitagl-nosplash= VitaGL ed51b21 (same as the VitaSDK image) built with NO_SPLASHSCREEN=1; CMake optionYTV_VITAGL_NOSPLASH(ON). Graphics test 33/33 with it. - Next: controller input → SbEvent input events (pad_adapter exists but is not pumped yet); SbPlayer for video.
Update: controller input and navigation
port_input.cpp: SceCtrl polled on the main thread each loop iteration, translated by pad_adapter (Starboard gamepad keys), dispatched as kSbEventTypeInput. SbInputData layout verified (76 bytes, key@16, wchar_t 4 bytes; SbEvent 24).- Diagnostic automation:
ux0:data/youtubevita/input_script.txt(" - Reached in Vita3K: Get started → sign-in options → guest Home (signed-out home has no feed) → sidebar → Shorts (thumbnail loads, playback fails "This video format is not supported" = SbPlayer unavailable) → Music tab with real thumbnail rows and focus navigation.
- Test VPK
build-starboard15/YouTubeVita_teste_v0.18.vpk(YTVCB0001, "YouTube Vita (teste)"): GL trace OFF, event loop unlimited unless pump_seconds.txt exists.
v0.19 (after first real-hardware test)
Hardware report: distorted audio, very low UI FPS, slow input, ~225/256 MB used, black screen at start when logged in, login not kept.
- LiveArea from the owner (assets/livearea → sce_sys/livearea/contents), TITLE "YouTube", APP_VER 00.19.
- Memory: param.sfo ATTRIBUTE2=12 (extended memory), SbSystemGetTotalCPUMemory 265 MiB.
- Audio: one SceAudioOut MAIN port per sink (no singleton), high-priority kernel feeder thread, continuous output (silence on underrun), 0 dB volume, 2048-frame min buffer.
- Performance: main loop event-driven (all due tasks, wake on SbEventSchedule, 8 ms input cadence; was fixed 16.7 ms sleep with 24-task cap); clocks 444/222/222/166; Starboard thread priorities mapped to kernel priorities; all threads on all user cores; 4th core via CapUnlocker when the kernel accepts SCE_KERNEL_CPU_MASK_SYSTEM (auto-probed, logged as YTV_LIFECYCLE_CPU_CORES); YTV_* logs only with ux0:data/youtubevita/verbose.txt.
- Touch: front panel → kSbInputDeviceTypeTouchScreen press/move/unpress (verified in Vita3K: tap on Get started works). Script command tap X Y.
- Login persistence: storage saves no longer blocked forever by .new/.backup leftovers of a killed app; several records may be open.
- Tests: Starboard 183/183, graphics 33/33, Cobalt touch navigation in Vita3K.
v0.20
- V8 JIT enabled: Cobalt reads CobaltEnableJit() from the
dev.cobalt.extension.ConfigurationStarboard extension and defaults to--jitlesswhen the platform provides none (ours returned nullptr). Port now provides the extension (v3): JIT on, compiled-JS cache on, bigger caches (image 32 MiB, encoded image 4 MiB, Skia 8 MiB, remote fonts 8 MiB, JS GC threshold 16 MiB). V8 uses W^X (flags=5 protects, no rejects). - V8's ARM FlushICache is a Linux
cacheflushsyscall (svc 0, r7=0xF0002) at ELF 0xD35198; the loader redirects it to kuKernelFlushCaches (verified original bytes first). - Profiling in Vita3K: rasterizer GL cost per second ≈1 ms draws, ~0 ms swap, 0-3 ms uploads, ~20 draws/frame → GPU/VitaGL is not the bottleneck; frames are limited by the browser's JS/layout work (CPU). Moving VitaGL → raw GXM would not address it.
- Overrides file
ux0:data/youtubevita/cobalt_config.txt(rasterizer=, jit=, swap_interval=);YTV_FPSper-second profile;tools/bench_cobalt_vita3k.sh. - Audio: Cobalt sink = float32 interleaved 48 kHz stereo, capacity 8192. PCM dump (verbose) shows peak 47 % FS, 0 clipped samples, no discontinuities in Vita3K → hardware distortion attributed to underruns from CPU starvation; minimum buffer raised to 8192 frames (~170 ms).
- Touch: 12 px tap dead zone (finger jitter no longer turns taps into drags);
tapjjittered scripted tap verified. - RAM: budget measured at start under ATTRIBUTE2=12, capped at 365 MiB.
- Tests: Starboard 183/183, graphics 33/33.
- Still missing: SbPlayer (video playback).
v0.21: video playback, microphone, startup
- SbPlayer (
port_player.cpp, replaces port_player_unavailable.cpp). SceAvPlayer needs a container file; Cobalt (MSE) delivers elementary samples, so the player drives the same hardware decoders AvPlayer uses: SceAvcdec H.264 → YUV420 planar (Vita3K does not implement RGBA output: it only bumps numOfOutput) in phycont memory → 3 GL_ALPHA textures → kSbDecodeTargetFormat3PlaneYUVI420; SceAudiodec AAC with ADTS headers → PCM ring → BGM port at stream rate; audio-driven clock. Media policy advertises H.264 Baseline/Main/High ≤ 854x480@30 and AAC-LC; refuses VP9/AV1/Opus/DRM. - Verified in Vita3K: youtube.com/tv#/watch?v=jNQXAC9IVRw plays (H.264 320x240 + AAC 44.1 kHz), 22-28 fps, no decode errors. VitaGL keeps 16 MiB phycont for the decoder.
- Microphone (
port_microphone.cpp): sceAudioIn VOICE 16 kHz mono via capture thread + ring (DeltaruneVita model). Test: opened, 512 bytes captured in Vita3K. Whether YouTube voice search accepts it is not verified yet. - Startup: VitaGL rebuilt with HAVE_SHADER_CACHE=1; 22 precompiled GXP shaders shipped in app0:/shader_cache and seeded to ux0:data/shader_cache/YTVCB0001 on first run. SbEventHandle(Start): 13.1-14.0 s → 2.6-2.8 s (cold and warm). Per-constructor log lines removed; loader lines timestamped.
- Tests: Starboard 192/192 (player, codecs, mic), graphics 33/33.
v0.22 (in progress): audio fix in SbPlayer, MP4 recording
- Correction to v0.21: "plays" was only checked from the UI frame rate. Logging showed the AAC decoder output 0 PCM and the audio-driven clock never moved, so playback froze about 4 s after it started.
- Cause: Cobalt's MP4 parser already hands over AAC access units with ADTS headers. port_player.cpp added a second header, so SceAudiodec consumed 8 bytes per frame and decoded nothing. Fix: detect the ADTS syncword and pass those samples through unchanged; raw access units still get a header.
- Verified in Vita3K: every frame decodes (1024 samples per AAC frame). Media time runs 0 → 19.06 s, the player reports EndOfStream, and YouTube autoplays the next video.
- Diagnostics added:
- YTV_PLAYER_STATUS heartbeat, SEEK/WRITE/EOS_IN and AAC traces (verbose only);
- record_audio.txt → rec_player.raw (exact PCM sent to sceAudioOut);
- capture DLL stream mode (VITA3K_CAPTURE_STREAM=1);
- runner flags -CaptureStream and -RecordAudio.
- Recording: tools/capture/YouTubeVita_Vita3K_me_at_the_zoo.mp4 (43 s, 960x544, 30 fps, AAC).
- v0.22 VPK packaged (APP_VER 00.22): build-starboard15/YouTubeVita_teste_v0.22.vpk.
- Player controls verified in Vita3K:
- X shows the overlay; X again pauses (SetPlaybackRate 0) and resumes (1).
- The down button reaches the shelf.
- Tapping CC shows captions (rendered by YouTube's JS).
- Playback speed is not supported: SetPlaybackRate rejects anything other than 0 and 1.
- Frame-rate investigation in Vita3K (shelf scroll with the video paused):
- 15-19 fps;
- per second: raster (first draw → swap) 12 ms, idle gap 996 ms;
- input to next frame: 4-60 ms.
- What limits the frame rate: the MainWebModule thread (JS + style + layout), not GL. Rasterizer and swap settings make no difference (direct-gles 25, hardware 23, swap_interval=0 24 fps median during video).
- Thread priorities: MainWebModule runs at 160, the same as the network and storage threads, V8 workers and media_pipeline. The Rasterizer runs at 112.
- Timer accuracy: a 16.667 ms wait wakes after 18.3 ms on average.
- sceKernelGetThreadInfo runClocks is always 0 in Vita3K, so YTV_THREAD_CPU only gives data on hardware.
v0.23: resolutions, zero-copy video, touch gestures
Video
- Zero copy. Each frame's Y/U/V planes are shown straight from decoder memory: the pool is GPU-mapped and each texture's GXM descriptor is rebuilt as linear on the plane. vitaGL textures are swizzled by default, which is why the first attempt showed scrambled colors. upload_ms went from a CPU copy per frame to 0.
- Frame lifetime. A frame leaves the pool only 2 vitaGL frames after it stops being shown, so the GPU never reads a frame being overwritten. Destroy runs glFinish, deletes the textures, then unmaps and frees the pool.
- Decoder capacity tiers. The largest tier that fits is used: 1920x1088 (non-game decoder), 1280x720, 864x480, 640x368, 432x240. Each frame's layout follows the stream's current size (stream_info of each sample), so YouTube's adaptive switches inside one player work.
- Vita3K over-read workaround. Vita3K copies frameWidth x frameHeight from FFmpeg's frame and crashes when the buffer is larger than the stream.
- Hardware limit. SceAvcdec for game apps accepts at most 1280x720 pixels. 1080p needs sceAvcdec*Nongameapp, which the kernel only allows for non-game programs. 1080p is advertised only if that decoder answers and there is enough device memory. In Vita3K it does not (phycont 16 MiB free, CDRAM 0 because vitaGL takes it all), so 1080p is refused there.
- Media policy.
- H.264 up to 1080p60 (level 4.2), subject to the 1080p check.
- Invalid probe attributes are refused: eotf other than bt709, tunnelmode=true, decode-to-texture other than true/false, experimental other than "allowed", cryptoblockformat other than "subsample", bitrate above 40 Mbps.
- Video buffer budget: 12 MiB up to 480p, 20 MiB for 720p, 30 MiB for 1080p.
- Verified in Vita3K.
- jNQXAC9IVRw at 240p plays to EOS on the 720p tier.
- LXb3EKWsInQ: YouTube picks H.264 1280x720 at 30 fps by itself (100 samples in 3.33 s). It accepts 720p60, but its ABR chose 30.
- A full-screen pre-roll ad plays.
Navigation
- MainWebModule and ScrollEngineThread raised from priority 160 to 128.
- No measurable gain in Vita3K: shelf scroll stays at 18 fps. The JS/layout thread is the limit (raster 12 ms/s, gap 994 ms/s).
Touch
- Tap → pointer click, sent on release.
- Swipe → one D-pad step per 72 px; content follows the finger.
- Two-finger tap → Back.
- Script command for tests: "secs swipe X0 Y0 X1 Y1".
- Verified in Vita3K: tap shows the player overlay, swipe up reaches the shelf, swipes left walk the thumbnails.
Tests
- Starboard 193/193. The pool is now 6 frames; new policy checks added.
v0.24: hardware crash at startup (v0.22/v0.23)
- Hardware report: v0.22 crashed before opening. The logs and the psp2core dump are in logs-psv/.
- Dump analysis:
- Crashed thread: ytv_heartbeat, raw kernel thread, stop reason 0x30004 (data abort).
- The PC is inside pthread-embedded (pte_threadStart / pte_new, text base 0x81028000).
- The last boot log line is "verified official Cobalt SbEventHandle", which comes right before the heartbeat start.
- Cause: the v0.22 timer probe ran SbMutex/SbConditionVariable (pthread) on a thread created with sceKernelCreateThread. On hardware, pthread-embedded crashes when used from threads it did not create. Vita3K tolerates it.
- Fixes:
- Timer probe removed.
- The player's audio output thread and the microphone capture thread, which also lock pthread mutexes, are now created with pthread_create. They set their kernel priority (64 and 96) at entry.
- The heartbeat thread now uses only sceKernel* calls and file logging.
- Also seen in the hardware log: CapUnlocker active (YTV_LIFECYCLE_CPU_CORES 4, mask 0xf0000). vitaGL takes all CDRAM (117 MiB → 0).
- Verified in Vita3K: video with sound plays to EOS (PCM 840704 frames). Starboard tests 193/193, including the microphone.
v0.25: hardware report for v0.24 (1 fps, freezes, harsh intro sound, sign-in lost)
The hardware log (logs-psv/v024) gave real per-thread CPU (runClocks) every 2 s: - MainWebModule: ~2000 ms of 2000 (one core saturated). - Rasterizer: up to 1500 ms. RasterzrSubDisp: up to 1085 ms. - NetworkModule: ~250 ms constant. - Idle threads (Watchdog, ServiceWorkerRegistry): ~62 ms each. - Main loop: ~90 ms. - CapUnlocker active (4 cores).
Fixes: 1. Condition variables. VitaSDK pthread-embedded pthread_cond_wait / pthread_cond_timedwait use pte_osSemaphoreCancellablePend, which polls sceKernelWaitSema every 500 us. Every idle thread woke ~2000 times per second. SbMutex and SbConditionVariable are now native SceKernelLwMutex / SceKernelLwCond (port_sync.cpp): - created lazily from the zero static initializer; - signals from a thread that does not hold the mutex retry under a try-lock if the kernel refuses them; - errors logged as YTV_LIFECYCLE_SYNC_ERROR. The scheduled-event queue (main loop wait) and the player worker now use these primitives too. 2. Network waiter. sceNetEpollWait's timeout is in microseconds (confirmed in Vita3K's epoll). The waiter passed 10 meaning milliseconds, so it waited 10 us and spun. Now: - timeouts are in microseconds; - WakeUp sends a byte to a loopback UDP socket kept in the same epoll set, so waits block, with 1 s slices; - without that socket, 4 ms slices. Verified in Vita3K: wake_fd created, navigation and network work. 3. Sign-in / cookies not persisted. Cobalt opens its savegame record with the name "aHR0cHM6Ly93d3cueW91dHViZS5jb20vdHY=" (base64 of the TV URL, with '='). name_path refused any byte outside [A-Za-z0-9_-], so SbStorageOpenRecord returned null and nothing was ever saved. Names are now sanitized, with an FNV-1a hash suffix when changed; '..', '/', '\' and ':' are still refused. Verified in Vita3K: storage .dat written (YTV_LIFECYCLE_STORAGE_OPEN / WRITE are always logged). 4. UI audio sink. It now pre-buffers about 85 ms before playing, and again after running dry, instead of alternating 512-frame blocks with silence. A Vita3K dump of the intro had 3 gaps (167 ms) in 2.8 s. The intro did not play again in later runs, so the change is not verified. 5. Heartbeat. Now runs for the whole session (5 s steps after the first minute). It adds GL_STATS: frames, swap/raster ms, draws, shader compiles and links with their time. Vita3K shows 4.4 s spent on the first frame (vitaGL shader translation/compile).
Tests: Starboard 193/193 with the new sync primitives. Vita3K: video with sound plays, search UI renders.
v0.26: freeze at 100% CPU, crash at first sign-in, no video on hardware
- Hardware log (logs-psv/v025): "YTV_LIFECYCLE_SYNC_ERROR create_lwmutex 0x8002450D" = SCE_KERNEL_ERROR_UID_RL_OVERFLOW.
- One LwMutex/LwCond per SbMutex/SbConditionVariable exhausted the process's kernel UIDs.
- Lock creation then failed, and MainWebModule, Rasterizer, ThreadPool and ServiceWorkerRegistry spun at about 11 s of CPU per 12 s: the freeze while browsing or searching.
- Code running without its lock is the likely cause of the first-sign-in crash (data abort inside libcobalt, no symbols to name the function).
- pthread mutexes also own a kernel semaphore each, so v0.24's freeze is probably the same limit.
- Fix: port_sync.cpp is now a futex emulation.
- SbMutex is one atomic word (Drepper mutex3); SbConditionVariable is one atomic sequence counter.
- Waits park on a fixed 64-bucket lot of LwMutex+LwCond (128 UIDs in total).
- No kernel object per lock; create and destroy are free; an uncontended lock is one atomic op.
- Before the freeze, idle CPU on hardware showed the v0.25 fixes working: Watchdog/ServiceWorker 0 ms, network 0-4 ms, MainWebModule ~170 ms per 5 s.
- Player:
- sceVideodecInitLibrary now asks for 1280x720 (the game-app limit), falling back to 864x480. v0.23-v0.25 asked for 1920x1088, which can leave no video decoder on a real Vita.
- Decoder tiers are bounded by the library size.
- Player creation, decoder tier, memory and failures are logged without verbose (YTV_LIFECYCLE_PLAYER_*).
- ShinoKai: its player sources inside "Shinokai-Vita-v0.9.5-r5-3 (1).zip" are corrupt (video_player.cpp extracts as 0 bytes). Its header and docs show the same design: hardware AVC via SceVideodec (FFmpeg h264_vita) with a 1280x720 area limit and NV12 output, sampled by GXM.
- Open issue: an 854x480 (864x480 decoded) ad displayed upside down with block corruption in Vita3K. 320x240 and 1280x720 were upright. To be handled in the GXM renderer.
- Vita3K crash seen once under UI stress with video: sceGxmCreateRenderTarget SCE_GXM_ERROR_DRIVER (vitaGL render targets).
- Tests: Starboard 193/193. Vita3K: video with sound plays to EOS.
- Release layout: final VPK in current/ (YouTubeVita_v0.26.vpk); older versions in current/anteriores/.
v0.27: LiveArea assets
- New LiveArea from the owner, flat in assets/livearea: icon0.png 128x128, labg.png 840x500, gate.png 280x158 (all 8-bit palette PNG), plus template.xml. CMake paths updated (no contents/ subfolder).
- Packaging bug found:
- vita_create_vpk is a macro, so VITA_PACK_VPK_FLAGS and resources accumulated from the earlier test VPKs in CMakeLists.txt.
- The YouTube VPK got the legacy icon0.png first (so every release up to v0.26 shipped the old icon), plus README, test videos, cacert.pem, cobalt-manifest.json and a duplicate libcobalt.so.
- Fixed by resetting both variables before the call.
- VPK is now 18.2 MB (was 20.0). The four LiveArea files match the sources byte for byte. param.sfo: APP_VER 00.27, ATTRIBUTE2 12, TITLE "YouTube", TITLE_ID YTVCB0001.
- The removed files were used only by other test VPKs (cacert.pem by tls_test).
- Vita3K smoke test: boots, a player is created.