qualcomm-runtime-r3-4b8c7f84-qa (DroidDeck-Components run 40244231868) was
built from this branch's patch series: only the Gamescope binary and its
source archive changed; Zink and the WSI layer rebuilt byte-identical to r2.
The libhybris bridge and WSI archives now come from that release, mirrored
unchanged from wallentx/DroidDeck (same SHA-256 and size; the original URLs
stay in mirrored_from).
runtime.json now records the patch series the Gamescope was built from, and
a test fails when a Gamescope patch changes without a rebuild, so the
Qualcomm Gamescope cannot silently fall behind the regular one.
On the X11 SHM path Gamescope's WSI layer is disabled, so its two
Performance mode parts (more swapchain images, the present-check cache)
cannot apply; the unpaced release and Wine's window-check cache still do.
With zero-copy every part applies. The log now says which case a session is in.
Every Qualcomm session started the driver through libhybris twice before
Steam: droiddeck-qualcomm-caps (up to 20 s) and glxinfo for the browser's
OpenGL level (up to 15 s). Both answers only change with the driver, so
they are now saved under $XDG_CACHE_HOME/droiddeck and reused while the
driver file (SHA-256), the selected driver, the bridge build, the layers
and the check script are unchanged. Gamescope's linux-dmabuf global is
still checked every session, failed or empty results are never saved, and
the log marks a reused result.
The Qualcomm port stopped pinning Gamescope to the Steam client's cores for
every session. Outside Qualcomm sessions Gamescope does not copy frames
through X11 SHM, and a user who restricts the client's cores still expects
Gamescope on them, as since #152. Only Qualcomm sessions skip it now.
0120 (nested swapchain capability fallback) and 0124 (SDL explicit refresh)
only matter for gamescope's SDL backend, which only Qualcomm sessions use.
In tools/gamescope/patches they would go into the next regular gamescope
release for every user. They now live in tools/qualcomm/gamescope-patches;
build_gamescope.py reads both folders in name order, so the Qualcomm build
applies the same series as before.
0121 and 0123 only change anything on PowerVR (vendor 0x1010) and are dropped.
0122 is renamed 0122-dmabuf-import-memory-type: it keeps the part every
vendor uses (the imported DMA-BUF's memory type must also be allowed by
vkGetMemoryFdPropertiesKHR) and drops the PowerVR-only query retry.
The full series still applies to gamescope 8f212644 (3.16.29).
The builds come from the DroidDeck-Components qualcomm-runtime release, built on ARM64 CI from this branch, so CI and fresh checkouts can stage the Qualcomm runtime without building it locally. The pins now match those Linux builds.
(cherry picked from commit 2ed529842d)
wallentx's libhybris runtime builder and patches, prebuilt bridge and WSI
releases, system-driver staging, android_wlegl buffer receiver and
Gamescope integration patches are what this backend is built on.
JeezDisReez identified turning off the image-bounds rewrite on Adreno as
the fix for the red graphics in DRAGON BALL: Sparking! ZERO.
(cherry picked from commit 09eb999a3a)
Older Adreno drivers such as the 8 Gen 2's 0.676.1 lack
VK_EXT_external_memory_dma_buf, DRM modifiers and calibrated timestamps.
They failed device creation in the WSI layer, failed Proton games on the
zero-copy path, and corrupted the Steam UI under GPU compositing.
- WSI patch 0005 makes the Wayland platform optional per device, so a
device without its extensions stays usable for X11 swapchains instead
of failing vkCreateDevice.
- droiddeck-qualcomm-caps checks DMA-BUF export and import, the swapchain
extensions and Gamescope's linux-dmabuf once per session; Proton games
use the zero-copy WSI path only when it passes and stay on X11 SHM
otherwise. The session log records the verdict.
- The Steam browser uses software compositing when Zink offers less than
OpenGL 4.0, with BL_STEAM_CEF_SOFTWARE as the manual override.
Verified on the Adreno 740 (falls back to SHM and software compositing,
UI correct) and the Adreno 840 (zero-copy and GPU compositing unchanged).
(cherry picked from commit d8d0b10cced1f21a1e9bc7dca00efd89b389adb3)
(cherry picked from commit 00775f0511)
VK_LAYER_HYBRIS_compat enables a Mali+DXVK workaround by default that
replaces the zero operand of the next OpSelect after any image sample or
fetch. On Adreno it collapses DXVK's comparison selects to constants,
which turned Sparking Zero's main-menu scene red.
(cherry picked from commit 8962b4f6f8c6d46178596dcfb02f98543135744d)
(cherry picked from commit d9c84afa8b)
Games in Qualcomm sessions now hand Gamescope LINEAR DMA-BUFs exported by
the driver instead of copying frames through X11 SHM. The ARM WSI layer is
rebuilt with its Wayland backend and the private Gamescope gains a
linux-dmabuf global without a DRM device. BL_QUALCOMM_SHM=1 keeps the
shared-memory path; Turnip sessions are unchanged.
(cherry picked from commit 1bd4c04f193f48bdaf4d4aad374010abaf48765d)
(cherry picked from commit d24b824f25)
* fix: Reuse the prefix the client made for a shortcut id
droiddeck-proton-run builds the prefix path as steamapps/compatdata/<appid>,
but the client names that folder with the signed 32-bit form of the shortcut
id. A shortcut whose id has the top bit set therefore got a second, empty
prefix, and the first launch after that ran wineboot from scratch: the
session sat on a black screen for the length of a full prefix init, and the
old prefix was never reused.
Match the folder by its unsigned value instead, the way the client does:
int(name) % 2 ** 32 == appid % 2 ** 32. prefix_of() also accepts a negative
folder name now, and reports its unsigned id so the Steam environment it sets
up stays valid.
* fix: Reject folder names int() cannot read, and skip files
lstrip("-").isdigit() accepts "--1" and superscript digits, which int() then
raises on, and one such entry in compatdata would stop shortcut resolution.
Match r"-?[0-9]+" instead. A matching name is also only used when it is a
directory, so a stray file cannot take STEAM_COMPAT_DATA_PATH.
* fix: reuse signed shortcut compatdata prefix
* fix: prefer initialized Proton shortcut prefixes
---------
Co-authored-by: xXJsonDeruloXx <danielhimebauch@gmail.com>
Co-authored-by: Kurt Himebauch <136133082+xXJSONDeruloXx@users.noreply.github.com>
Adds a Performance mode switch under Mode settings > Frame rate, off by default and stored per mode like the frame limit. Off keeps today's behaviour. On:
- with no frame limit, the compositor hands buffers back to the game as soon as they are replaced instead of once per refresh, and gamescope's WSI layer is asked for at least 8 swapchain images; a chosen frame limit still paces
- DROIDDECK_WSI_BYPASS_CHECK_MS=100 lets gamescope's WSI layer reuse its present-time Xwayland bypass check (patch 0117, gamescope-3.16.29-p13); WINE_OFFSCREEN_CHECK_MS=100 does the same for winex11 once a DroidDeck-Proton release carries that fix. Off sets both to 0.
The gamescope pack now ships the patched WSI layer and SessionFiles stages it over the runtime's stock copy. Two per-game options, off by default: PROTON_DXVK_D3D8 and WINE_D3D_CONFIG renderer=vulkan|gl.
AIO Graphics Test on an AYANEO Pocket FIT (Adreno 750), off -> on: Vulkan 204 -> 1787, OpenGL 296 -> 2169, D3D12 308 -> 3563, D3D11 1133 -> 2874, D3D9 1253 -> 3098 fps; DiRT Showdown 92.2 -> 93.4 (GPU-bound).
* feat: adjustable RAM and VRAM caps for the container
Performance > Memory gets two sliders:
- RAM cap (soft, Steam sessions): when the session is resident above the
cap, the session watcher compresses the Steam client into zRAM, waits for
its page-out acknowledgment, then compresses the rest of the session
(processes that loaded libblsession) if still over. Nothing is closed.
Measured at most every 30 s; the client is read back once usage drops
below the cap. Invalid values leave the cap off. Needs zRAM swap.
- VRAM cap: sets dxgi.maxDeviceMemory ahead of the game's own DXVK_CONFIG,
so Direct3D games budget for less video memory. A game profile still
wins. Applies from the next session.
* fix: back off the RAM cap when paging out cannot get under it
A cap below the game's working set was never met, so the watcher paged out
the client and the whole session every 30 s for the rest of the session.
Usage is still measured every 30 s, but after a page-out that leaves the
session over the cap the next one waits twice as long (60 s up to 10 min),
resetting once usage is under the cap. The client step is skipped while the
client is already under 512 MB resident.
* fix: RAM cap covers desktop Steam, signals each group once, no idle churn
- Steam opened from the desktop runs the same session watcher, so the cap
is exported whenever the Steam client runs, not only in Steam mode.
- The second page-out goes to the rest of the session only; Steam was
signalled again even when its own step had been skipped as already small.
- A cap page-out while no game runs stays paged out. Reading the client back
right away pushed it over the cap again and repeated every minute; only a
game closing while over the cap still defers a read-back.
* fix: RAM cap marks the client packed only when it compressed it
When the cap skipped the client as already small during a game, it still
recorded the client as packed, so the regular in-game client page-out
waited up to 300 s for a page-out that never happened.
---------
Co-authored-by: jazzyjames <182448902+jazzyjamesgames@users.noreply.github.com>
GE-based Protons, the default with the x86-64 Steam client, run umu-protonfixes. Its Black Ops III fix switches the prefix to Wine's ALSA driver, which has no device here and replaces DirectAudio, and its Dark Souls Remastered fix runs the VC++ 2017 installer, which fails under ARM64 Proton.
Xbox Cloud Gaming: stream from the store and Games tab, native player with DroidDeck frame generation
Stream Xbox games with Xbox Cloud Gaming (Game Pass) from DroidDeck, outside
the Linux session.
- Backend: `xodus-cli cloud-token` (tools/xbox/droiddeck-cloud-token.patch)
gets the streaming tokens; they stay in memory.
- Web player (default): the Greenlight player page (tools/xcloud-player, MIT)
in a WebView, plus Microsoft's own xbox.com/play player. In-stream menu with
screen effects (sharpen, colour, deband, FXAA, CRT, NTSC, Toon, fps cap),
stream quality, performance overlay (level, layout, top right / top left)
and rumble.
- Native player (opt-in, Android 10+, Player engine: Native): Google's WebRTC
(io.github.webrtc-sdk 125.6422.07, BSD) with the Greenlight protocol ported
to Kotlin, in its own process (:xcloudnative) with its own instance of
DroidDeck's compositor. Decoded frames reach the compositor as
AHardwareBuffers (new stream mode, VK_ANDROID_external_memory_android_
hardware_buffer import), so DDFG 2x and LSFG 2x/3x/4x run on them. Live
frame generation, frame-rate target (presets and a 10-165 fps slider, even
pacing), zero-copy picture, small network buffer, fast decoder, instant
controller input, overlay with engine and on-screen fps, background end,
restart and web fallback. It takes the main process's tokens through
XboxCloudTokenService; they are fetched ahead while a cloud game is shown.
- The compositor requests the dma-buf import extensions only where the driver
has them (the system driver has none).
- Catalog: cloud availability, Game Pass plan CFQ7TTC0K6L8 counted, library
filter All / Owned / Game Pass / Cloud, Stream and Add to Games on the store
page, stream-only Games-tab entries (Stream + Download / Play + Stream).
- Xbox games get VC++ 2015-2022 and the ARM64 runtime registry entries
(vcredist2022_arm64, built in until the catalog has it) before each launch.
- Controller focus reaches pickers opened from a game page (install location,
uninstall); Add / Remove from Games keep the focus.
Credits: Greenlight by UnknownSKL (Jim Kroon), MIT, player and protocol;
Better xCloud for the resolution / OS mapping; MGamer404's Bannerlator #627
for the ARM64 runtime approach. Ported from the412banner's Bannerlator.
On some phones (tester: Honor BKQ-N49, Android 17) Xbox sign-in stopped with
"the Xbox helper stopped with exit code 1" before Microsoft's page opened.
The backend asks the app to read and save credentials over a socket and waits
10 s for each reply. The app sealed credentials with a StrongBox-backed
Keystore key, and each StrongBox operation took 7-15 s on that phone, so the
backend gave up. The app still finished those saves afterwards, leaving a
record without a username that every later attempt had to read back slowly.
- Seal new store credentials with a TEE Keystore key (v2 envelopes); v1
envelopes still open with the old key and move to v2 on their next write.
- A sign-in discards an Xbox record without a username (an unfinished
sign-in) before starting the backend.
- The backend waits 30 s for the app's credential reply.
- The backend names the failing step instead of only credential-io (socket
connect/exchange, session lock, missing socket name/port/key), and falls
back to a keyed 127.0.0.1 port when the abstract socket is refused.
- The app's credential listener logs and retries accept errors, serves each
connection on its own thread, and logs slow requests with per-stage times.
Refs #1337, #1300
Co-authored-by: KeegantheVeganat0r <58369898+KeegantheVeganat0r@users.noreply.github.com>