From a780997f9674ef5b2085efc5b7df1507f36f8168 Mon Sep 17 00:00:00 2001
From: Konstantin Gindemit <konstantin.gindemit@gmail.com>
Date: Mon, 10 Jul 2023 21:54:38 +0200
Subject: [PATCH] ~ Fixes in pixman-arm-intrisics.cpp

---
 dependency/pixman/pixman-arm-intrisics.cpp | 20 ++++++++++----------
 projects/CMake/CMakeLists.txt              |  2 +-
 2 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/dependency/pixman/pixman-arm-intrisics.cpp b/dependency/pixman/pixman-arm-intrisics.cpp
index 44eee7b..02a496d 100644
--- a/dependency/pixman/pixman-arm-intrisics.cpp
+++ b/dependency/pixman/pixman-arm-intrisics.cpp
@@ -22,25 +22,25 @@ extern "C" void pixman_composite_over_n_8888_asm_neon(int32_t w, int32_t h,
                                                       int32_t   dst_stride,
                                                       uint32_t  src)
 {
-    // Extract the source alpha and replicate it to all 8 lanes of a NEON vector
-    uint8x8_t v_src_alpha = vdup_n_u8(src >> 24);
-    // Extract the source color and replicate it to all 8 lanes of a NEON vector
-    uint8x8_t v_src_color = vdup_n_u8(src);
+    // Extract the source alpha and replicate it to all 16 lanes of a NEON vector
+    uint8x16_t v_src_alpha = vdupq_n_u8(src >> 24);
+    // Extract the source color and replicate it to all 16 lanes of a NEON vector
+    uint8x16_t v_src_color = vdupq_n_u8(src);
 
     for (int32_t y = 0; y < h; y++)
     {
-        for (int32_t x = 0; x < w; x += 8)
+        for (int32_t x = 0; x < w; x += 16)  // Changed to 16 to match uint8x16_t
         {
-            // Load 8 destination pixels
-            uint8x8_t v_dst_color = vld1_u8((uint8_t *)(dst + x));
+            // Load 16 destination pixels
+            uint8x16_t v_dst_color = vld1q_u8((uint8_t *)(dst + x));
 
             // Calculate the result color = source color * source alpha + destination color * (1 - source alpha)
             // Note that we need to shift right by 8 because the alpha blending operation can result in values greater than 255
-            uint8x8_t v_res_color = vshrq_n_u8(vmlaq_u8(vmlsq_u8(v_dst_color, v_dst_color, v_src_alpha), v_src_color, v_src_alpha), 8);
+            uint8x16_t v_res_color = vshrq_n_u8(vmlaq_u8(vmlsq_u8(v_dst_color, v_dst_color, v_src_alpha), v_src_color, v_src_alpha), 8);
 
             // Store the result to memory
-            vst1_u8((uint8_t *)(dst + x), v_res_color);
+            vst1q_u8((uint8_t *)(dst + x), v_res_color);
         }
         dst += dst_stride;
     }
-}
+}
\ No newline at end of file
diff --git a/projects/CMake/CMakeLists.txt b/projects/CMake/CMakeLists.txt
index 5fd4f68..c2e855c 100644
--- a/projects/CMake/CMakeLists.txt
+++ b/projects/CMake/CMakeLists.txt
@@ -61,7 +61,7 @@ if (RLOTTIE_IOS)
   message("Compile asm source for iOS")
   target_compile_options(rlottie PUBLIC -fno-integrated-as)
   target_compile_definitions(rlottie PUBLIC USE_ARM_NEON __ARM64_NEON__ __ARM_NEON__)
-  target_sources(rlottie PRIVATE ${PIXMAN_ROOT}/pixman-arma64-neon-asm.S)
+  target_sources(rlottie PRIVATE ${PIXMAN_ROOT}/pixman-arm-intrisics.cpp)
 endif()
 
 if (WIN32)