Merge pull request #14250 from kwizart/armv7_build_neon

Armv7 build neon

4 files changed, 63 insertions, 42 deletions

diff --git a/xbmc/rendering/MatrixGL.cpp b/xbmc/rendering/MatrixGL.cpp
index 8e9d944850..fc82cb35c9 100644
--- a/xbmc/rendering/MatrixGL.cpp
+++ b/xbmc/rendering/MatrixGL.cpp
@@ -9,8 +9,9 @@
 #include "MatrixGL.h"
 #include "utils/TransformMatrix.h"
 
-#if defined(HAS_NEON)
+#if defined(HAS_NEON) && !defined(__LP64__)
 #include "utils/CPUInfo.h"
+void Matrix4Mul(float* src_mat_1, const float* src_mat_2);
 #endif
 
 #include <cmath>
@@ -128,47 +129,6 @@ void CMatrixGL::Rotatef(GLfloat angle, GLfloat x, GLfloat y, GLfloat z)
   MultMatrixf(matrix);
 }
 
-#if defined(HAS_NEON) && !defined(__LP64__)
-
-static inline void Matrix4Mul(float* src_mat_1, const float* src_mat_2)
-{
-  asm volatile (
-    // Store A & B leaving room at top of registers for result (q0-q3)
-    "vldmia %0, { q4-q7 }  \n\t"
-    "vldmia %1, { q8-q11 } \n\t"
-
-    // result = first column of B x first row of A
-    "vmul.f32 q0, q8, d8[0]\n\t"
-    "vmul.f32 q1, q8, d10[0]\n\t"
-    "vmul.f32 q2, q8, d12[0]\n\t"
-    "vmul.f32 q3, q8, d14[0]\n\t"
-
-    // result += second column of B x second row of A
-    "vmla.f32 q0, q9, d8[1]\n\t"
-    "vmla.f32 q1, q9, d10[1]\n\t"
-    "vmla.f32 q2, q9, d12[1]\n\t"
-    "vmla.f32 q3, q9, d14[1]\n\t"
-
-    // result += third column of B x third row of A
-    "vmla.f32 q0, q10, d9[0]\n\t"
-    "vmla.f32 q1, q10, d11[0]\n\t"
-    "vmla.f32 q2, q10, d13[0]\n\t"
-    "vmla.f32 q3, q10, d15[0]\n\t"
-
-    // result += last column of B x last row of A
-    "vmla.f32 q0, q11, d9[1]\n\t"
-    "vmla.f32 q1, q11, d11[1]\n\t"
-    "vmla.f32 q2, q11, d13[1]\n\t"
-    "vmla.f32 q3, q11, d15[1]\n\t"
-
-    // output = result registers
-    "vstmia %1, { q0-q3 }"
-    : //no output
-    : "r" (src_mat_2), "r" (src_mat_1)       // input - note *value* of pointer doesn't change
-    : "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7", "q8", "q9", "q10", "q11" //clobber
-    );
-}
-#endif
 void CMatrixGL::MultMatrixf(const CMatrixGL &matrix) noexcept
 {
 #if defined(HAS_NEON) && !defined(__LP64__)
diff --git a/xbmc/rendering/MatrixGL.neon.cpp b/xbmc/rendering/MatrixGL.neon.cpp
new file mode 100644
index 0000000000..3f3f33ce36
--- /dev/null
+++ b/xbmc/rendering/MatrixGL.neon.cpp
@@ -0,0 +1,47 @@
+/*
+ *  Copyright (C) 2005-2018 Team Kodi
+ *  This file is part of Kodi - https://kodi.tv
+ *
+ *  SPDX-License-Identifier: GPL-2.0-or-later
+ *  See LICENSES/README.md for more information.
+ */
+
+
+void Matrix4Mul(float* src_mat_1, const float* src_mat_2)
+{
+  asm volatile (
+    // Store A & B leaving room at top of registers for result (q0-q3)
+    "vldmia %0, { q4-q7 }  \n\t"
+    "vldmia %1, { q8-q11 } \n\t"
+
+    // result = first column of B x first row of A
+    "vmul.f32 q0, q8, d8[0]\n\t"
+    "vmul.f32 q1, q8, d10[0]\n\t"
+    "vmul.f32 q2, q8, d12[0]\n\t"
+    "vmul.f32 q3, q8, d14[0]\n\t"
+
+    // result += second column of B x second row of A
+    "vmla.f32 q0, q9, d8[1]\n\t"
+    "vmla.f32 q1, q9, d10[1]\n\t"
+    "vmla.f32 q2, q9, d12[1]\n\t"
+    "vmla.f32 q3, q9, d14[1]\n\t"
+
+    // result += third column of B x third row of A
+    "vmla.f32 q0, q10, d9[0]\n\t"
+    "vmla.f32 q1, q10, d11[0]\n\t"
+    "vmla.f32 q2, q10, d13[0]\n\t"
+    "vmla.f32 q3, q10, d15[0]\n\t"
+
+    // result += last column of B x last row of A
+    "vmla.f32 q0, q11, d9[1]\n\t"
+    "vmla.f32 q1, q11, d11[1]\n\t"
+    "vmla.f32 q2, q11, d13[1]\n\t"
+    "vmla.f32 q3, q11, d15[1]\n\t"
+
+    // output = result registers
+    "vstmia %1, { q0-q3 }"
+    : //no output
+    : "r" (src_mat_2), "r" (src_mat_1)       // input - note *value* of pointer doesn't change
+    : "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7", "q8", "q9", "q10", "q11" //clobber
+    );
+}
diff --git a/xbmc/rendering/gl/CMakeLists.txt b/xbmc/rendering/gl/CMakeLists.txt
index 4c4a245fbe..00fff7668f 100644
--- a/xbmc/rendering/gl/CMakeLists.txt
+++ b/xbmc/rendering/gl/CMakeLists.txt
@@ -8,4 +8,11 @@ set(HEADERS GUIWindowTestPatternGL.h
             ../MatrixGL.h
             GLShader.h)
 
+if(ARCH MATCHES arm AND ENABLE_NEON)
+  list(APPEND SOURCES ../MatrixGL.neon.cpp)
+  if(NOT DEFINED NEON_FLAGS)
+    set_source_files_properties(../MatrixGL.neon.cpp PROPERTIES COMPILE_OPTIONS -mfpu=neon)
+  endif()
+endif()
+
 core_add_library(rendering_gl)
diff --git a/xbmc/rendering/gles/CMakeLists.txt b/xbmc/rendering/gles/CMakeLists.txt
index 4df58e3c68..fc39514128 100644
--- a/xbmc/rendering/gles/CMakeLists.txt
+++ b/xbmc/rendering/gles/CMakeLists.txt
@@ -7,5 +7,12 @@ if(OPENGLES_FOUND)
               ../MatrixGL.h
               GLESShader.h)
 
+  if(ARCH MATCHES arm AND ENABLE_NEON)
+    list(APPEND SOURCES ../MatrixGL.neon.cpp)
+    if(NOT DEFINED NEON_FLAGS)
+      set_source_files_properties(../MatrixGL.neon.cpp PROPERTIES COMPILE_OPTIONS -mfpu=neon)
+    endif()
+  endif()
+
   core_add_library(rendering_gles)
 endif()