Commit Graph

1010 Commits

Author SHA1 Message Date
nmzik 207cef9602 renderer: report invalid texture upload layout 2026-08-03 00:35:02 +02:00
nmzik fb5ecec455 renderer: ignore stale stencil state for depth-only targets 2026-08-02 14:14:32 +02:00
nmzik fc8d2a3b83 guest_gpu: remove memory-unmap submission deadlock + remove legacy agc buffering 2026-08-02 13:26:29 +02:00
nmzik 59b8fad341 graphics: support 3D color render targets 2026-08-02 09:38:57 +02:00
nmzik b877b4be9c graphics: perf - batch GPU page watcher updates at 4 MiB granularity 2026-08-02 08:39:50 +02:00
nmzik 9da7fc5dd6 renderer: minor optimizations 2026-08-02 08:34:44 +02:00
nmzik da0d33224d renderer: eliminate extra copy for small streaming buffers 2026-08-02 08:34:44 +02:00
nmzik 4631b96178 perf(gpu): run dirty-page validation only in debug builds 2026-08-02 04:58:56 +02:00
nmzik 44d7f2a3e8 shader cfg: handle shared early exits
Duplicate small shared exit tails so each selection gets its own merge block. This keeps overlapping early-exit ladders on structured SPIR-V and adds a regression test.
2026-08-02 03:16:08 +02:00
nmzik 2dcb90066c shader cfg: normalize loop structure
Give loops one header and one continue path before SPIR-V generation. This handles conditional headers and multiple latches without falling back to a dispatcher.
2026-08-02 03:15:26 +02:00
nmzik 51a33cc363 shader cfg: handle loop control branches
Keep simple break, continue, and repeat branches in structured control flow. Split conflicting merge blocks and add regression tests for nested loop exits.
2026-08-02 03:14:39 +02:00
Claxten 43f30d3ab2 graphics: shader: ignore unused sampler border state
* Sampler dword 3 only matters when a clamp mode uses border color
  (values >= 4). When no border mode is active, dword 3 is unused
  but can still vary across loop iterations due to wave-lane spills.
  This makes resource tracking think the descriptor is dynamic and
  fail with "unsupported GPU selection".

* Fix by zeroing dword 3 when all clamp modes are non-border.

Signed-off-by: Claxten <claxten10@gmail.com>
2026-08-02 02:42:15 +02:00
Claxten c1a5927036 graphics: pm4: accept trailing PM4 type-2 packets
* A one-dword type-2 NOP is a valid packet tail. Parse it normally instead of aborting command-buffer dumps.

Signed-off-by: Claxten <claxten10@gmail.com>
2026-08-02 02:07:37 +02:00
Claxten bc436548a9 graphics: support packed 10-10-10-2 uint buffers
Signed-off-by: Claxten <claxten10@gmail.com>
2026-08-02 01:44:21 +02:00
nmzik a65d17a5d6 renderer: skip debug checks when stencil&depth is not active 2026-08-01 00:22:24 +02:00
nmzik c690aeea62 add HiS PM4 handler, accept Ngs2CustomMastering 2026-08-01 00:12:42 +02:00
nmzik f830d6b2e4 renderer: remove legacy code left from refactoring 2026-07-31 23:13:48 +02:00
nmzik d68a477276 renderer: broaden compatibility 2026-07-31 22:24:56 +02:00
nmzik 8977d4d2f0 shader: add descriptor log 2026-07-31 21:29:26 +02:00
nmzik 4b4e3bf3cf pm4: implement missing selectors 2026-07-31 19:45:47 +02:00
nmzik 846002c5eb shader: fix invalid texture descriptors and add missing format 2026-07-31 17:30:31 +02:00
nmzik d8a4c83cc7 format src and tests with clang-format 2026-07-31 11:36:12 +02:00
nmzik 68be13345a fix(shader): support multisampled depth image loads 2026-07-31 11:36:12 +02:00
nmzik 167da0abe0 shader: fix readlane/writelane for inactive host lanes 2026-07-31 11:36:12 +02:00
nmzik 48c31d61ee Implement VideoDec2 2026-07-31 11:36:12 +02:00
nmzik 212282d693 fix(shader): preserve packed UINT16 MRT exports 2026-07-31 11:36:12 +02:00
nmzik 6bca35d1f5 renderer: broaden compatibility 2026-07-31 11:36:12 +02:00
Nikos 3b75a5659a shader: specialize cube image descriptors (#134)
* shader: specialize cube image descriptors

Track whether image descriptors refer to cube maps during resource specialization, and apply the coordinate offset conversion when sampling cube maps as 2D image arrays in SPIR-V emission.

* shader: fix cube array coordinate lowering

---------

Co-authored-by: nmzik <Nmzik@mail.ru>
2026-07-31 03:59:10 +02:00
nmzik 2f5396c6a5 Rework guest memory tracking/virtual address space/direct and flexible memory (#135)
* Rework guest memory tracking

* add unknwon flag

* Fix macOS guest address-space reservation
2026-07-31 03:07:17 +02:00
nmzik d04938c88c Embedded fetch shader: Fix overlapping buffer loads (#133)
Fix overlapping buffer loads. Fixes many games
2026-07-30 05:08:38 +02:00
nmzik 3965d41d36 texture_cache: fix exact-match reuse across different tile modes 2026-07-30 00:10:40 +02:00
nmzik c508c4a9c0 shader_recompiler: allow GDS append/consume offsets 2026-07-30 00:10:40 +02:00
nmzik cc76827e63 Fix vertex buffer ranges crossing memory mappings 2026-07-29 21:05:24 +02:00
nmzik 832bc84100 fix(shader): stabilize scalar provenance phis in cyclic CFGs 2026-07-29 21:05:24 +02:00
nmzik b9ae2537ef renderer: broaden compatibility 2026-07-29 18:47:26 +02:00
nmzik 0b9edaa721 graphics: broaden storage image atomic compatibility 2026-07-29 18:47:24 +02:00
nmzik 8a244677d7 fix(renderer): resolve delayed GPU page faults through buffer and texture caches 2026-07-29 18:47:19 +02:00
nmzik f6e01e5403 Optimize bulk memory invalidation (#124)
* Per page -> per range search (optimization)
2026-07-29 05:09:36 +02:00
nmzik 861729fc6c Optimize texture cache tracking (#123)
Optimize texture cache page tracking
2026-07-29 03:37:09 +02:00
nmzik ec11f31aa6 fix graphical bug (PPSA17221) 2026-07-29 00:31:08 +02:00
nmzik 64845e2294 new ABI, broaden support 2026-07-29 00:31:08 +02:00
nmzik a6b61d7aa0 refactor(graphics): split renderer and recompiler into focused modules 2026-07-29 00:30:24 +02:00
Stefanos Costa 2b9cba457e Linux: port the emulator to a working state on Linux (#117)
Linux: port the emulator to a working state
2026-07-29 00:25:58 +02:00
M. Abdullah 4d79f19089 macOS (Apple Silicon) support: native build running under Rosetta 2 with MoltenVK (#102)
* macos: POSIX platform layer (host fault handler, virtual memory)

- hostException: Mach/POSIX signal-based host fault handler mirroring the
  Windows vectored handler (SIGSEGV/SIGBUS/SIGILL), behind
  '#elif defined(__APPLE__)'. Windows and Linux branches are untouched.
- sysLinuxVirtual: mach_vm_region-based is_mapped (no /proc/self/maps on
  macOS), PTHREAD_MUTEX_NORMAL, and a MAP_FIXED carve-in-place allocator that
  never leaves an unmapped hole for dyld/Rosetta/Metal to claim. All
  __APPLE__-guarded; the original Linux allocator path is preserved verbatim.
- sysLinuxDbg/sysLinuxFileIO: libgen.h for basename(), and a POSIX
  opendir/readdir implementation of SysFileGetDents (previously a stub).

* macos: guest kernel backing, threads, and POSIX libs

- memory/memoryAddressSpace: anonymous shm_open backing (macOS has no
  memfd_create) and re-reserve-on-unmap so a guest MAP_FIXED remap never
  destroys a host mapping (__APPLE__-guarded). Drops host <sys/mman.h> MAP_*
  macros so the guest's constexpr MAP_* constants compile (no-op on Windows).
- pthread: undef the host PTHREAD_STACK_MIN macro; on non-Windows, pin the
  stack-switch asm's guest rsp/rbp to callee-saved r14/r15 (the template
  clobbers r12/r13); __APPLE__ no-op for the absent pthread_condattr_setclock.
  The Windows stack-switch asm is byte-identical.
- network: define SOCKET/INVALID_SOCKET for the non-Windows paths (were
  undefined at 20+ use sites), and rename the non-Windows kernel_clock_*
  helpers to the CamelCase names the callers already use. Both repair the
  shared non-Windows build. Integer reinterpret_cast -> static_cast.

* loader: pin stack-switch asm operands to callee-saved registers

RunEntry switches to the guest stack with an asm template that clobbers
r12/r13 before consuming its inputs. With plain "r" constraints the compiler
may place func/guest_rsp/guest_rbp into r12/r13, so 'callq *func' jumps
through the saved host rsp -- a latent miscompile on every platform that any
unrelated codegen change can trigger. Pin the inputs to rbx/r14/r15, which the
template never touches and the SysV callee preserves.

General correctness fix (not macOS-specific); the same fix is applied to
pthread RunOnGuestStack.

* macos: Vulkan/MoltenVK graphics enablement

- pageManager: GPU write-tracking via Mach VM queries + mprotect (mprotect
  cannot report the previous protection, so the expected-old check is dropped),
  thread id via pthread_mach_thread_np, 4 KB page-size check -- __APPLE__-guarded.
- shaders/renderDraw/vulkanWindow: MoltenVK lacks VK_EXT_color_write_enable,
  VK_EXT_depth_clip_enable and depthBounds; fall back to static color-write
  masks and default depth clipping, request VK_KHR_portability_subset, and
  reuse the graphics queue for present when only one queue is exposed. Non-Apple
  pipeline/extension setup is unchanged.
- window: optional borderless window (KYTY_BORDERLESS) to sidestep a Rosetta
  NSException in macOS window chrome.

* graphics: macOS thread identity for region tracking

Region-ownership locks resolve the current thread via GetCurrentThreadId()
on Windows and EXIT elsewhere. Use the Mach thread port on macOS (nonzero
per-thread id; 0 stays the no-owner sentinel).

* macos: marshal AppKit window operations to the main thread

The present worker thread shows the window, updates its icon and title, and
recreates a lost Vulkan surface. AppKit traps with 'Must only be used from the
main thread' when these run off the main thread. Route them through a small
task queue drained by the SDL main loop (an SDL_USEREVENT wakes the loop when
it is blocked in SDL_WaitEvent). Title updates are fire-and-forget; showing
the window and surface recreation wait for completion.

Windows and Linux call the SDL functions directly, as before.

* build: ignore the _Build output directory

_Build is the conventional out-of-tree build location (only _Build/vscode-clang
was ignored); a stray git add could sweep build artifacts into a commit.

* macos: isolate stack-switch workaround

---------

Co-authored-by: nmzik <Nmzik@mail.ru>
2026-07-28 19:03:15 +02:00
nmzik 7351eae117 graphics: add 1D image descriptor support 2026-07-27 14:10:51 +02:00
nmzik 125287dc30 graphics: broaden depth compatibility 2026-07-27 10:29:37 +02:00
nmzik 142b3c20fa graphics: unify buffer bindings and depth target handling 2026-07-27 10:12:41 +02:00
nmzik f43567e12c shader: validate MSAA image descriptors during materialization 2026-07-27 10:12:41 +02:00
nmzik c86d4b1f0d Fix linux compilation 2026-07-27 10:12:41 +02:00
nmzik 31fb3ec5af graphics: broaden storage image and view format compatibility 2026-07-27 10:12:41 +02:00