General aarch64 improvements & Apple Silicon support

.github/workflows/build.yml

@@ -177,6 +177,9 @@ jobs:
 
   build-macos:
     runs-on: macos-14
+    strategy:
+      matrix:
+        arch: [x86_64, arm64]
     steps:
     - name: "Checkout repo"
       uses: actions/checkout@v4
@@ -236,7 +239,7 @@ jobs:
         cd build
         cmake .. ${{ env.BUILD_FLAGS }} \
         -DCMAKE_BUILD_TYPE=${{ env.BUILD_MODE }} \
-        -DCMAKE_OSX_ARCHITECTURES=x86_64 \
+        -DCMAKE_OSX_ARCHITECTURES=${{ matrix.arch }} \
         -DMACOS_BUNDLE=ON \
         -G Ninja
 
@@ -259,5 +262,5 @@ jobs:
     - name: Upload artifact
       uses: actions/upload-artifact@v4
       with:
-        name: cemu-bin-macos-x64
+        name: cemu-bin-macos-${{ matrix.arch }}
         path: ./bin/Cemu.dmg

CMakeLists.txt

@@ -222,7 +222,12 @@ endif()
 
 add_subdirectory("dependencies/ih264d" EXCLUDE_FROM_ALL)
 
-if(CMAKE_SYSTEM_PROCESSOR MATCHES "(aarch64)|(AARCH64)")
+if (CMAKE_OSX_ARCHITECTURES)
+	set(CEMU_ARCHITECTURE ${CMAKE_OSX_ARCHITECTURES})
+else()
+	set(CEMU_ARCHITECTURE ${CMAKE_SYSTEM_PROCESSOR})
+endif()
+if(CEMU_ARCHITECTURE MATCHES "(aarch64)|(AARCH64)|(arm64)|(ARM64)")
 	add_subdirectory("dependencies/xbyak_aarch64" EXCLUDE_FROM_ALL)
 endif()
 
@@ -231,4 +236,4 @@ if (NOT ZArchive_FOUND)
 	add_subdirectory("dependencies/ZArchive" EXCLUDE_FROM_ALL)
 endif()
 
-add_subdirectory(src)
+add_subdirectory(src)

dependencies/ih264d/CMakeLists.txt

@@ -183,6 +183,9 @@ target_sources(ih264d PRIVATE
 "decoder/arm/ih264d_function_selector.c"
 )
 target_compile_options(ih264d PRIVATE -DARMV8)
+if(APPLE)
+    target_sources(ih264d PRIVATE "common/armv8/macos_arm_symbol_aliases.s")
+endif()
 else()
 message(FATAL_ERROR "ih264d unknown architecture: ${IH264D_ARCHITECTURE}")
 endif()

dependencies/ih264d/common/armv8/ih264_intra_pred_chroma_av8.s

@@ -429,8 +429,13 @@ ih264_intra_pred_chroma_8x8_mode_plane_av8:
     rev64     v7.4h, v2.4h
     ld1       {v3.2s}, [x10]
     sub       x5, x3, #8
+#ifdef __APPLE__
+    adrp      x12, _ih264_gai1_intrapred_chroma_plane_coeffs1@GOTPAGE
+    ldr       x12, [x12, _ih264_gai1_intrapred_chroma_plane_coeffs1@GOTPAGEOFF]
+#else
     adrp      x12, :got:ih264_gai1_intrapred_chroma_plane_coeffs1
     ldr       x12, [x12, #:got_lo12:ih264_gai1_intrapred_chroma_plane_coeffs1]
+#endif
     usubl     v10.8h, v5.8b, v1.8b
     ld1       {v8.8b, v9.8b}, [x12]     // Load multiplication factors 1 to 8 into D3
     mov       v8.d[1], v9.d[0]
@@ -484,10 +489,13 @@ ih264_intra_pred_chroma_8x8_mode_plane_av8:
     zip1      v1.8h, v0.8h, v2.8h
     zip2      v2.8h, v0.8h, v2.8h
     mov       v0.16b, v1.16b
-
+#ifdef __APPLE__
+    adrp      x12, _ih264_gai1_intrapred_chroma_plane_coeffs2@GOTPAGE
+    ldr       x12, [x12, _ih264_gai1_intrapred_chroma_plane_coeffs2@GOTPAGEOFF]
+#else
     adrp      x12, :got:ih264_gai1_intrapred_chroma_plane_coeffs2
     ldr       x12, [x12, #:got_lo12:ih264_gai1_intrapred_chroma_plane_coeffs2]
-
+#endif
     ld1       {v8.2s, v9.2s}, [x12]
     mov       v8.d[1], v9.d[0]
     mov       v10.16b, v8.16b

dependencies/ih264d/common/armv8/ih264_intra_pred_luma_16x16_av8.s

@@ -431,10 +431,13 @@ ih264_intra_pred_luma_16x16_mode_plane_av8:
     mov       x10, x1                   //top_left
     mov       x4, #-1
     ld1       {v2.2s}, [x1], x8
-
+#ifdef __APPLE__
+    adrp      x7, _ih264_gai1_intrapred_luma_plane_coeffs@GOTPAGE
+    ldr       x7, [x7, _ih264_gai1_intrapred_luma_plane_coeffs@GOTPAGEOFF]
+#else
     adrp      x7, :got:ih264_gai1_intrapred_luma_plane_coeffs
     ldr       x7, [x7, #:got_lo12:ih264_gai1_intrapred_luma_plane_coeffs]
-
+#endif
     ld1       {v0.2s}, [x1]
     rev64     v2.8b, v2.8b
     ld1       {v6.2s, v7.2s}, [x7]

dependencies/ih264d/common/armv8/ih264_intra_pred_luma_8x8_av8.s

@@ -1029,9 +1029,13 @@ ih264_intra_pred_luma_8x8_mode_horz_u_av8:
     mov       v3.d[0], v2.d[1]
     ext       v4.16b, v2.16b , v2.16b , #1
     mov       v5.d[0], v4.d[1]
-
+#ifdef __APPLE__
+    adrp      x12, _ih264_gai1_intrapred_luma_8x8_horz_u@GOTPAGE
+    ldr       x12, [x12, _ih264_gai1_intrapred_luma_8x8_horz_u@GOTPAGEOFF]
+#else
     adrp      x12, :got:ih264_gai1_intrapred_luma_8x8_horz_u
     ldr       x12, [x12, #:got_lo12:ih264_gai1_intrapred_luma_8x8_horz_u]
+#endif
     uaddl     v20.8h, v0.8b, v2.8b
     uaddl     v22.8h, v1.8b, v3.8b
     uaddl     v24.8h, v2.8b, v4.8b

dependencies/ih264d/common/armv8/ih264_weighted_bi_pred_av8.s

@@ -142,14 +142,22 @@ ih264_weighted_bi_pred_luma_av8:
     sxtw      x4, w4
     sxtw      x5, w5
     stp       x19, x20, [sp, #-16]!
+#ifndef __APPLE__
     ldr       w8, [sp, #80]             //Load wt2 in w8
     ldr       w9, [sp, #88]             //Load ofst1 in w9
-    add       w6, w6, #1                //w6  = log_WD + 1
-    neg       w10, w6                   //w10 = -(log_WD + 1)
-    dup       v0.8h, w10                //Q0  = -(log_WD + 1) (32-bit)
     ldr       w10, [sp, #96]            //Load ofst2 in w10
     ldr       w11, [sp, #104]           //Load ht in w11
     ldr       w12, [sp, #112]           //Load wd in w12
+#else
+    ldr       w8, [sp, #80]             //Load wd in w8
+    ldr       w9, [sp, #84]             //Load ht in w9
+    ldr       w10, [sp, #88]            //Load offst2 in w10
+    ldr       w11,  [sp, #92]           //Load offst1 in w11
+    ldr       w12,  [sp, #96]           //Load offst1 in w12
+#endif
+    add       w6, w6, #1                //w6  = log_WD + 1
+    neg       w10, w6                   //w10 = -(log_WD + 1)
+    dup       v0.8h, w10                //Q0  = -(log_WD + 1) (32-bit)
     add       w9, w9, #1                //w9 = ofst1 + 1
     add       w9, w9, w10               //w9 = ofst1 + ofst2 + 1
     mov       v2.s[0], w7
@@ -424,17 +432,24 @@ ih264_weighted_bi_pred_chroma_av8:
     sxtw      x5, w5
     stp       x19, x20, [sp, #-16]!
 
-
+#ifndef __APPLE__
     ldr       w8, [sp, #80]             //Load wt2 in w8
+    ldr       w9, [sp, #88]             //Load ofst1 in w9
+    ldr       w10, [sp, #96]            //Load ofst2 in w10
+    ldr       w11, [sp, #104]           //Load ht in w11
+    ldr       w12, [sp, #112]           //Load wd in w12
+#else
+    ldr       w8, [sp, #80]             //Load wd in w8
+    ldr       w9, [sp, #84]             //Load ht in w9
+    ldr       w10, [sp, #88]            //Load offst2 in w10
+    ldr       w11,  [sp, #92]           //Load offst1 in w11
+    ldr       w12,  [sp, #96]           //Load offst1 in w12
+#endif
     dup       v4.4s, w8                 //Q2 = (wt2_u, wt2_v) (32-bit)
     dup       v2.4s, w7                 //Q1 = (wt1_u, wt1_v) (32-bit)
     add       w6, w6, #1                //w6  = log_WD + 1
-    ldr       w9, [sp, #88]             //Load ofst1 in w9
-    ldr       w10, [sp, #96]            //Load ofst2 in w10
     neg       w20, w6                   //w20 = -(log_WD + 1)
     dup       v0.8h, w20                //Q0  = -(log_WD + 1) (16-bit)
-    ldr       w11, [sp, #104]           //Load ht in x11
-    ldr       w12, [sp, #112]           //Load wd in x12
     dup       v20.8h, w9                //0ffset1
     dup       v21.8h, w10               //0ffset2
     srhadd    v6.8b, v20.8b, v21.8b

dependencies/ih264d/common/armv8/macos_arm_symbol_aliases.s

@@ -0,0 +1,185 @@
+// macOS clang compilers append preceding underscores to function names, this is to prevent
+// mismatches with the assembly function names and the C functions as defined in the header.
+
+.global _ih264_deblk_chroma_horz_bs4_av8
+_ih264_deblk_chroma_horz_bs4_av8 = ih264_deblk_chroma_horz_bs4_av8
+
+.global _ih264_deblk_chroma_horz_bslt4_av8
+_ih264_deblk_chroma_horz_bslt4_av8 = ih264_deblk_chroma_horz_bslt4_av8
+
+.global _ih264_deblk_chroma_vert_bs4_av8
+_ih264_deblk_chroma_vert_bs4_av8 = ih264_deblk_chroma_vert_bs4_av8
+
+.global _ih264_deblk_chroma_vert_bslt4_av8
+_ih264_deblk_chroma_vert_bslt4_av8 = ih264_deblk_chroma_vert_bslt4_av8
+
+.global _ih264_deblk_luma_horz_bs4_av8
+_ih264_deblk_luma_horz_bs4_av8 = ih264_deblk_luma_horz_bs4_av8
+
+.global _ih264_deblk_luma_horz_bslt4_av8
+_ih264_deblk_luma_horz_bslt4_av8 = ih264_deblk_luma_horz_bslt4_av8
+
+.global _ih264_deblk_luma_vert_bs4_av8
+_ih264_deblk_luma_vert_bs4_av8 = ih264_deblk_luma_vert_bs4_av8
+
+.global _ih264_deblk_luma_vert_bslt4_av8
+_ih264_deblk_luma_vert_bslt4_av8 = ih264_deblk_luma_vert_bslt4_av8
+
+.global _ih264_default_weighted_pred_chroma_av8
+_ih264_default_weighted_pred_chroma_av8 = ih264_default_weighted_pred_chroma_av8
+
+.global _ih264_default_weighted_pred_luma_av8
+_ih264_default_weighted_pred_luma_av8 = ih264_default_weighted_pred_luma_av8
+
+.global _ih264_ihadamard_scaling_4x4_av8
+_ih264_ihadamard_scaling_4x4_av8 = ih264_ihadamard_scaling_4x4_av8
+
+.global _ih264_inter_pred_chroma_av8
+_ih264_inter_pred_chroma_av8 = ih264_inter_pred_chroma_av8
+
+.global _ih264_inter_pred_luma_copy_av8
+_ih264_inter_pred_luma_copy_av8 = ih264_inter_pred_luma_copy_av8
+
+.global _ih264_inter_pred_luma_horz_av8
+_ih264_inter_pred_luma_horz_av8 = ih264_inter_pred_luma_horz_av8
+
+.global _ih264_inter_pred_luma_horz_hpel_vert_hpel_av8
+_ih264_inter_pred_luma_horz_hpel_vert_hpel_av8 = ih264_inter_pred_luma_horz_hpel_vert_hpel_av8
+
+.global _ih264_inter_pred_luma_horz_hpel_vert_qpel_av8
+_ih264_inter_pred_luma_horz_hpel_vert_qpel_av8 = ih264_inter_pred_luma_horz_hpel_vert_qpel_av8
+
+.global _ih264_inter_pred_luma_horz_qpel_av8
+_ih264_inter_pred_luma_horz_qpel_av8 = ih264_inter_pred_luma_horz_qpel_av8
+
+.global _ih264_inter_pred_luma_horz_qpel_vert_hpel_av8
+_ih264_inter_pred_luma_horz_qpel_vert_hpel_av8 = ih264_inter_pred_luma_horz_qpel_vert_hpel_av8
+
+.global _ih264_inter_pred_luma_horz_qpel_vert_qpel_av8
+_ih264_inter_pred_luma_horz_qpel_vert_qpel_av8 = ih264_inter_pred_luma_horz_qpel_vert_qpel_av8
+
+.global _ih264_inter_pred_luma_vert_av8
+_ih264_inter_pred_luma_vert_av8 = ih264_inter_pred_luma_vert_av8
+
+.global _ih264_inter_pred_luma_vert_qpel_av8
+_ih264_inter_pred_luma_vert_qpel_av8 = ih264_inter_pred_luma_vert_qpel_av8
+
+.global _ih264_intra_pred_chroma_8x8_mode_horz_av8
+_ih264_intra_pred_chroma_8x8_mode_horz_av8 = ih264_intra_pred_chroma_8x8_mode_horz_av8
+
+.global _ih264_intra_pred_chroma_8x8_mode_plane_av8
+_ih264_intra_pred_chroma_8x8_mode_plane_av8 = ih264_intra_pred_chroma_8x8_mode_plane_av8
+
+.global _ih264_intra_pred_chroma_8x8_mode_vert_av8
+_ih264_intra_pred_chroma_8x8_mode_vert_av8 = ih264_intra_pred_chroma_8x8_mode_vert_av8
+
+.global _ih264_intra_pred_luma_16x16_mode_dc_av8
+_ih264_intra_pred_luma_16x16_mode_dc_av8 = ih264_intra_pred_luma_16x16_mode_dc_av8
+
+.global _ih264_intra_pred_luma_16x16_mode_horz_av8
+_ih264_intra_pred_luma_16x16_mode_horz_av8 = ih264_intra_pred_luma_16x16_mode_horz_av8
+
+.global _ih264_intra_pred_luma_16x16_mode_plane_av8
+_ih264_intra_pred_luma_16x16_mode_plane_av8 = ih264_intra_pred_luma_16x16_mode_plane_av8
+
+.global _ih264_intra_pred_luma_16x16_mode_vert_av8
+_ih264_intra_pred_luma_16x16_mode_vert_av8 = ih264_intra_pred_luma_16x16_mode_vert_av8
+
+.global _ih264_intra_pred_luma_4x4_mode_dc_av8
+_ih264_intra_pred_luma_4x4_mode_dc_av8 = ih264_intra_pred_luma_4x4_mode_dc_av8
+
+.global _ih264_intra_pred_luma_4x4_mode_diag_dl_av8
+_ih264_intra_pred_luma_4x4_mode_diag_dl_av8 = ih264_intra_pred_luma_4x4_mode_diag_dl_av8
+
+.global _ih264_intra_pred_luma_4x4_mode_diag_dr_av8
+_ih264_intra_pred_luma_4x4_mode_diag_dr_av8 = ih264_intra_pred_luma_4x4_mode_diag_dr_av8
+
+.global _ih264_intra_pred_luma_4x4_mode_horz_av8
+_ih264_intra_pred_luma_4x4_mode_horz_av8 = ih264_intra_pred_luma_4x4_mode_horz_av8
+
+.global _ih264_intra_pred_luma_4x4_mode_horz_d_av8
+_ih264_intra_pred_luma_4x4_mode_horz_d_av8 = ih264_intra_pred_luma_4x4_mode_horz_d_av8
+
+.global _ih264_intra_pred_luma_4x4_mode_horz_u_av8
+_ih264_intra_pred_luma_4x4_mode_horz_u_av8 = ih264_intra_pred_luma_4x4_mode_horz_u_av8
+
+.global _ih264_intra_pred_luma_4x4_mode_vert_av8
+_ih264_intra_pred_luma_4x4_mode_vert_av8 = ih264_intra_pred_luma_4x4_mode_vert_av8
+
+.global _ih264_intra_pred_luma_4x4_mode_vert_l_av8
+_ih264_intra_pred_luma_4x4_mode_vert_l_av8 = ih264_intra_pred_luma_4x4_mode_vert_l_av8
+
+.global _ih264_intra_pred_luma_4x4_mode_vert_r_av8
+_ih264_intra_pred_luma_4x4_mode_vert_r_av8 = ih264_intra_pred_luma_4x4_mode_vert_r_av8
+
+.global _ih264_intra_pred_luma_8x8_mode_dc_av8
+_ih264_intra_pred_luma_8x8_mode_dc_av8 = ih264_intra_pred_luma_8x8_mode_dc_av8
+
+.global _ih264_intra_pred_luma_8x8_mode_diag_dl_av8
+_ih264_intra_pred_luma_8x8_mode_diag_dl_av8 = ih264_intra_pred_luma_8x8_mode_diag_dl_av8
+
+.global _ih264_intra_pred_luma_8x8_mode_diag_dr_av8
+_ih264_intra_pred_luma_8x8_mode_diag_dr_av8 = ih264_intra_pred_luma_8x8_mode_diag_dr_av8
+
+.global _ih264_intra_pred_luma_8x8_mode_horz_av8
+_ih264_intra_pred_luma_8x8_mode_horz_av8 = ih264_intra_pred_luma_8x8_mode_horz_av8
+
+.global _ih264_intra_pred_luma_8x8_mode_horz_d_av8
+_ih264_intra_pred_luma_8x8_mode_horz_d_av8 = ih264_intra_pred_luma_8x8_mode_horz_d_av8
+
+.global _ih264_intra_pred_luma_8x8_mode_horz_u_av8
+_ih264_intra_pred_luma_8x8_mode_horz_u_av8 = ih264_intra_pred_luma_8x8_mode_horz_u_av8
+
+.global _ih264_intra_pred_luma_8x8_mode_vert_av8
+_ih264_intra_pred_luma_8x8_mode_vert_av8 = ih264_intra_pred_luma_8x8_mode_vert_av8
+
+.global _ih264_intra_pred_luma_8x8_mode_vert_l_av8
+_ih264_intra_pred_luma_8x8_mode_vert_l_av8 = ih264_intra_pred_luma_8x8_mode_vert_l_av8
+
+.global _ih264_intra_pred_luma_8x8_mode_vert_r_av8
+_ih264_intra_pred_luma_8x8_mode_vert_r_av8 = ih264_intra_pred_luma_8x8_mode_vert_r_av8
+
+.global _ih264_iquant_itrans_recon_4x4_av8
+_ih264_iquant_itrans_recon_4x4_av8 = ih264_iquant_itrans_recon_4x4_av8
+
+.global _ih264_iquant_itrans_recon_4x4_dc_av8
+_ih264_iquant_itrans_recon_4x4_dc_av8 = ih264_iquant_itrans_recon_4x4_dc_av8
+
+.global _ih264_iquant_itrans_recon_8x8_av8
+_ih264_iquant_itrans_recon_8x8_av8 = ih264_iquant_itrans_recon_8x8_av8
+
+.global _ih264_iquant_itrans_recon_8x8_dc_av8
+_ih264_iquant_itrans_recon_8x8_dc_av8 = ih264_iquant_itrans_recon_8x8_dc_av8
+
+.global _ih264_iquant_itrans_recon_chroma_4x4_av8
+_ih264_iquant_itrans_recon_chroma_4x4_av8 = ih264_iquant_itrans_recon_chroma_4x4_av8
+
+.global _ih264_iquant_itrans_recon_chroma_4x4_dc_av8
+_ih264_iquant_itrans_recon_chroma_4x4_dc_av8 = ih264_iquant_itrans_recon_chroma_4x4_dc_av8
+
+.global _ih264_pad_left_chroma_av8
+_ih264_pad_left_chroma_av8 = ih264_pad_left_chroma_av8
+
+.global _ih264_pad_left_luma_av8
+_ih264_pad_left_luma_av8 = ih264_pad_left_luma_av8
+
+.global _ih264_pad_right_chroma_av8
+_ih264_pad_right_chroma_av8 = ih264_pad_right_chroma_av8
+
+.global _ih264_pad_right_luma_av8
+_ih264_pad_right_luma_av8 = ih264_pad_right_luma_av8
+
+.global _ih264_pad_top_av8
+_ih264_pad_top_av8 = ih264_pad_top_av8
+
+.global _ih264_weighted_bi_pred_chroma_av8
+_ih264_weighted_bi_pred_chroma_av8 = ih264_weighted_bi_pred_chroma_av8
+
+.global _ih264_weighted_bi_pred_luma_av8
+_ih264_weighted_bi_pred_luma_av8 = ih264_weighted_bi_pred_luma_av8
+
+.global _ih264_weighted_pred_chroma_av8
+_ih264_weighted_pred_chroma_av8 = ih264_weighted_pred_chroma_av8
+
+.global _ih264_weighted_pred_luma_av8
+_ih264_weighted_pred_luma_av8 = ih264_weighted_pred_luma_av8

src/CMakeLists.txt

@@ -101,9 +101,9 @@ if (MACOS_BUNDLE)
 	endforeach(folder)
 
 	if(CMAKE_BUILD_TYPE STREQUAL "Debug")
-	    set(LIBUSB_PATH "${CMAKE_BINARY_DIR}/vcpkg_installed/x64-osx/debug/lib/libusb-1.0.0.dylib")
+	    set(LIBUSB_PATH "${CMAKE_BINARY_DIR}/vcpkg_installed/${VCPKG_TARGET_TRIPLET}/debug/lib/libusb-1.0.0.dylib")
 	else()
-	    set(LIBUSB_PATH "${CMAKE_BINARY_DIR}/vcpkg_installed/x64-osx/lib/libusb-1.0.0.dylib")
+	    set(LIBUSB_PATH "${CMAKE_BINARY_DIR}/vcpkg_installed/${VCPKG_TARGET_TRIPLET}/lib/libusb-1.0.0.dylib")
 	endif()
 
 	add_custom_command (TARGET CemuBin POST_BUILD

src/Cafe/CMakeLists.txt

@@ -537,7 +537,7 @@ if(APPLE)
 	target_sources(CemuCafe PRIVATE "HW/Latte/Renderer/Vulkan/CocoaSurface.mm")
 endif()
 
-if(CMAKE_SYSTEM_PROCESSOR MATCHES "(aarch64)|(AARCH64)")
+if(CEMU_ARCHITECTURE MATCHES "(aarch64)|(AARCH64)|(arm64)|(ARM64)")
   target_sources(CemuCafe PRIVATE
     HW/Espresso/Recompiler/BackendAArch64/BackendAArch64.cpp
     HW/Espresso/Recompiler/BackendAArch64/BackendAArch64.h

src/Cafe/HW/Espresso/Recompiler/BackendAArch64/BackendAArch64.cpp

@@ -169,8 +169,10 @@ struct AArch64GenContext_t : CodeGenerator
 
 	bool processAllJumps()
 	{
-		for (auto&& [jumpStart, jumpInfo] : jumps)
+		for (auto jump : jumps)
 		{
+			auto jumpStart = jump.first;
+			auto jumpInfo = jump.second;
 			bool success = std::visit(
 				[&, this](const auto& jump) {
 					setSize(jumpStart);