Support distributor DLL packages and validate Windows player mode

docs/windows-gpu-support.md

@@ -2,8 +2,11 @@
 
 The shim, diagnostics, native K/M kernels and game installer accept the
 targets below, with one package per actual ISA. **Only RX 9070 XT has physical
-validation here.** All other targets are experimental and compile-tested;
-working DLSS, interop and performance on them remain unverified.
+validation here.** Community RX 9060 XT / gfx1200 logs now show K/M backend
+execution in Cyberpunk and Dawnwalker with all required native layers and
+finite output. Visual issues remain reported and are not resolved by those
+checks. All other targets are experimental and compile-tested; their DLSS,
+interop and performance remain unverified.
 
 | Family | Target | Examples / scope |
 | --- | --- | --- |
@@ -68,6 +71,28 @@ The new packages/preflight address the identified failures. They do not prove
 those games work on RX 9060 XT; a new native-D3D12 run on that GPU is required.
 Raw community logs and personal directory paths stay local.
 
+## Community retest, 2026-10-03
+
+[Comment 5962444244](https://github.com/countervolts/d4r/issues/10#issuecomment-5962444244)
+uses the corrected `52d9872` package / `3d62b98` binaries on RX 9060 XT
+(`gfx1200`, 16 GB). Cyberpunk K/M complete 4,365 / 1,751 session frames;
+Dawnwalker K/M complete 6,713 / 1,690. Every frame has a finite-output GPU
+check, all required native layers execute and all four sessions exit normally.
+The logs record zero backend failures, previous-frame outputs and CPU image
+copies. This is community backend coverage, not a reference-image comparison:
+the tester reports texture/LOD and motion-related artifacts. All four runs use
+the HDR network input/output path, unlike the local Cyberpunk LDR validation.
+
+Spider-Man 2 crashes with `0xc0000005` before any completed DLSS frame;
+its nonstandard launch is not evidence of a transformer kernel failure.
+Assassin's Creed exits with zero frames and Streamline logs repeated internal
+minidumps; its clean process exit is still inconclusive. Their causes remain
+open without a stack trace. The Witcher 3 log in
+[comment 5962808629](https://github.com/countervolts/d4r/issues/10#issuecomment-5962808629)
+selects the gfx1200 d4r shim instead of the original NGX core (its SHA256
+matches the shipped shim exactly), and the installer correctly refuses
+before modifying the game. It does not establish a Witcher runtime failure.
+
 ## Build
 
 Prepare dependencies following `windows-rdna4-port.md`. Default: gfx1201.

docs/windows-quick-test.txt

@@ -9,14 +9,16 @@ The package detects the real GPU automatically. Do not rename any DLLs.
    Already installed an older test? Close the game and run RESTORE-GAME.cmd
    from that old package first, then extract this version to a new folder.
 
-2. CLOSE YOUR GAME. Have these TWO ORIGINAL NVIDIA files ready:
+2. CLOSE YOUR GAME. If your distributor included the validated NVIDIA DLL pair,
+   it is selected automatically; you only need to choose the game executable.
+   Otherwise have these TWO ORIGINAL NVIDIA files ready:
    _nvngx.dll       version 32.0.16.1714
    nvngx_dlss.dll   version 310.9.1
-   NVIDIA DLLs are not included. Other versions are rejected.
+   The xdfnx-dev public test ZIP does not include them. Other versions are rejected.
    The small _nvngx.dll inside this package is d4r, NOT the original NVIDIA file.
 
 3. DOUBLE-CLICK START-K.cmd. Pick the actual game .exe when asked.
-   Pick your two NVIDIA DLLs. These choices are remembered for the next run.
+   If asked, pick your two NVIDIA DLLs. These choices are remembered for the next run.
    Next time, press Enter for the same game, or C to choose another.
    If Steam/Epic is needed, keep it open. Choose a D3D12 game without anti-cheat.
    If a launcher opens, type C and choose the real game executable next time.
@@ -53,7 +55,8 @@ Colour, depth, motion and output remain in VRAM in the runtime fast path.
 
 Runtime files are in files/. Cache/settings are created in work/.
 Licenses are in licenses/. Corresponding sources are a SEPARATE release download;
-testers do not need to download or build them. No NVIDIA proprietary DLLs are shipped.
+testers do not need to download or build them. NVIDIA DLLs, when supplied by a
+distributor, keep their separate license and are not part of the GPL source archive.
 
 Report: https://github.com/countervolts/d4r/issues/10
 Source: https://github.com/xdfnx-dev/d4r

docs/windows-rdna4-port.md

@@ -19,6 +19,76 @@ explicit limits; the validated package keeps conservative arithmetic.
 
 ## Milestone and gates
 
+2026-10-03 upstream Windows refresh: imported ayo's `e7e5c44` (LF patch
+attributes) and `54c201a` (drag-in installation / per-game `d4r.ini`) while
+preserving their authorship. The target-independent NGX loader selects the
+shim using the game's D3D12/HIP LUID. Player mode derives paths and the
+validated codegen settings from `d4r.ini`, pins NVIDIA DLL identities and
+exposes K/M, async interop, logging, validation and cache settings. OptiScaler
+patch `0006` resolves relative library directories. These changes do not
+replace the transformer arithmetic or establish a performance improvement.
+
+The new packaging fixture passes 3/3 cases. Player config/loader probes pass
+2/2 in both PowerShell 5 and PowerShell 7, including real gfx1201 LUID
+selection without executing GPU kernels. Found and fixed a PowerShell 7
+argument-quoting regression in `test-player-config.ps1`: only Legacy/PS5
+passing needs escaped embedded quotes. Both new DLL targets compile through
+CMake/Ninja, and all 20 RX 9070 XT CTest gates pass (6.21 seconds).
+Reports are in `test-results/upstream-player-54c201a-ps5/`,
+`upstream-player-54c201a-ps7/` and `upstream-release-54c201a/`.
+The drag-in installation has not yet received a real-game K/M test, and
+OptiScaler patch `0006` is not in the previously published frontend binary.
+The existing public quick-test archive is unchanged and continues to use the
+game-tested installer layout. Do not label it as the new drag-in release.
+
+2026-10-03 community retest ([comment 5962444244](https://github.com/countervolts/d4r/issues/10#issuecomment-5962444244)):
+seven new diagnostic bundles are read without running their contents. On
+RX 9060 XT / gfx1200, Cyberpunk K/M complete **4365 / 1751** frames and
+Dawnwalker K/M complete **6713 / 1690**. Each frame has a finite-output GPU
+check; all required native transformer layers execute, exit codes are zero,
+and backend failures / previous-frame output / CPU image-copy counters are
+zero. These sessions use HDR input/output helpers; the local Cyberpunk tests
+below used LDR. This establishes additional community backend execution,
+not numerical or visual equivalence. The tester reports texture/LOD and motion
+artifacts, which remain an explicit correctness gate before performance work.
+Do not claim that NaN/Inf scans establish image correctness.
+
+Mean recorded `interop_ngx_ms` is 18.623 / 50.237 ms for Cyberpunk K/M and
+8.126 / 32.632 ms for Dawnwalker K/M. These are CPU completion intervals in
+the synchronous diagnostic, including scheduling and validation, not isolated
+GPU kernel times or a controlled cross-game/preset benchmark. K performance
+remains open. Extracted logs, original ZIP hashes and parsed reports stay local
+under `build/issue-10-latest/`; personal user paths are not committed.
+
+The new Spider-Man 2 log exits `0xc0000005` before a DLSS frame, without a
+usable crash stack. Assassin's Creed's process exits zero with no DLSS frames,
+but Streamline reports repeated internal minidumps before capability discovery;
+it remains inconclusive. Neither report establishes a native WMMA failure.
+[The Witcher 3 report](https://github.com/countervolts/d4r/issues/10#issuecomment-5962808629)
+selects the gfx1200 d4r shim instead of the original NGX core; the submitted
+SHA256 matches that shim exactly. It is rejected before installation, rather
+than failing a GPU workload. Launcher/Streamline crash diagnosis remains open.
+
+Distributor-supplied NVIDIA DLL support: quick-test packaging accepts optional
+`-NvidiaDirectory <directory>` and `-NvidiaLicensePath <applicable-license>`.
+It stages the exact validated originals under `files/nvidia/`, adds both DLLs
+and the separate license to integrity metadata and selects them automatically.
+Explicit developer overrides still have strict SHA256 checks; bundled files
+take precedence over stale remembered file selections. Restore remains
+independent of DLL/GPU selection. This option is for local testing or a
+distributor with appropriate permission; matching hashes are not a grant of
+redistribution rights. The current public xdfnx-dev release remains unchanged
+and contains no NVIDIA runtime binaries or private NVIDIA-derived kernels.
+The [current NVIDIA SDK license](https://github.com/NVIDIA/DLSS/blob/main/LICENSE.txt)
+has distribution conditions and a supplement limiting DLSS/NGX development
+to systems with NVIDIA GPUs; the Linux bundle's contents alone do not resolve
+permission for this Windows package. A local-only bundled package with the
+user-provided pair passes eight isolated installer/restore/negative fixtures,
+including stale-choice recovery and altered bundled-core rejection before any
+game write. The ordinary no-DLL package also passes all six existing fixtures;
+reports are in `test-results/windows-bundled-fixtures/` and
+`windows-unbundled-fixtures/`. Real games are untouched by these fixture tests.
+
 2026-10-02 issue #10 follow-up ([comment 5958769640](https://github.com/countervolts/d4r/issues/10#issuecomment-5958769640)):
 four new logs confirm that RX 9060 XT / gfx1200 HIP discovery succeeds. The
 Cyberpunk K/M runs never launch the transformer: numeric CUDA NGX capabilities