user/x265: add patch for armv7. fixes #1023.

2 files changed, 366 insertions, 3 deletions

diff --git a/user/x265/APKBUILD b/user/x265/APKBUILD
index 685f75ea8..9a884cd30 100644
--- a/user/x265/APKBUILD
+++ b/user/x265/APKBUILD
@@ -2,7 +2,7 @@
 # Maintainer: A. Wilcox <awilfox@adelielinux.org>
 pkgname=x265
 pkgver=3.2.1
-pkgrel=0
+pkgrel=1
 pkgdesc="Open Source H265/HEVC video encoder"
 url="http://x265.org"
 arch="all"
@@ -12,7 +12,9 @@ depends=""
 makedepends="cmake yasm" # 2.6+ will use nasm instead of yasm
 subpackages="$pkgname-dev"
 source="$pkgname-$pkgver.tar.gz::https://download.videolan.org/pub/videolan/${pkgname}/${pkgname}_${pkgver}.tar.gz
-	x265-unbreak-my-arm.patch"
+	x265-unbreak-my-arm.patch
+	x265-arm-asm-primitives.patch
+	"
 builddir="$srcdir/${pkgname}_$pkgver"
 
 build() {
@@ -49,4 +51,5 @@ package() {
 }
 
 sha512sums="5cb29b9d4475c1f686f6e31e47ccddde3db1639d5dcf9c058513389a9fdeb35a83b14e30e59498a76dec74bb0b410b6d8f5b1b45d18927be9811bb6a40d8d568  x265-3.2.1.tar.gz
-ac3ecc613fe4da11672de8f397b5541fe2af29b7dd91a48859bfe8824934a5966c41ca30de242564c46f1a0cdd2a0da9331d9f547fc21120066fb3c45347c4c4  x265-unbreak-my-arm.patch"
+ac3ecc613fe4da11672de8f397b5541fe2af29b7dd91a48859bfe8824934a5966c41ca30de242564c46f1a0cdd2a0da9331d9f547fc21120066fb3c45347c4c4  x265-unbreak-my-arm.patch
+94c0894c1bc0ccdb21526df0ab5fbd5d3c9fc4ff7ad93ab9cb2fbe831be49f01624136c0e9d8968ea44d8c91e42bf790b2c75e7afe313532972ebe154cb76627  x265-arm-asm-primitives.patch"
diff --git a/user/x265/x265-arm-asm-primitives.patch b/user/x265/x265-arm-asm-primitives.patch
new file mode 100644
index 000000000..f49aafe57
--- /dev/null
+++ b/user/x265/x265-arm-asm-primitives.patch
@@ -0,0 +1,360 @@
+From <https://git.busybox.net/buildroot/tree/package/x265/0003-arm-asm-primitives.patch?id=57d4a27eaf1a9e59d767c321e7b7500c5060a2ac>.
+This fixes build errors like:
+
+  cd /tmp/guix-build-x265-2.8.drv-0/x265_2.8/build/encoder && /gnu/store/cd5q2pni1d95fs3cdabbclyh9hqhw2nq-gcc-5.5.0/bin/c++  -DEXPORT_C_API=1 -DHAVE_ARMV6=1 -DHAVE_INT_TYPES_H=1 -DHAVE_NEON -DHIGH_BIT_DEPTH=0 -DX265_ARCH_ARM=1 -DX265_DEPTH=8 -DX265_NS=x265 -D__STDC_LIMIT_MACROS=1 -I/tmp/guix-build-x265-2.8.drv-0/x265_2.8/source/. -I/tmp/guix-build-x265-2.8.drv-0/x265_2.8/source/common -I/tmp/guix-build-x265-2.8.drv-0/x265_2.8/source/encoder -I/tmp/guix-build-x265-2.8.drv-0/x265_2.8/build  -O2 -g -DNDEBUG   -Wall -Wextra -Wshadow -std=gnu++98 -fPIC -mcpu=native -mfloat-abi=hard -mfpu=neon -marm -fPIC -Wno-array-bounds -ffast-math -fno-exceptions -Wno-uninitialized -o CMakeFiles/encoder.dir/search.cpp.o -c /tmp/guix-build-x265-2.8.drv-0/x265_2.8/source/encoder/search.cpp
+  /tmp/guix-build-x265-2.8.drv-0/x265_2.8/source/common/arm/asm-primitives.cpp:437:38: error: incompatible types in assignment of ?void(const pixel*, intptr_t, int16_t*, intptr_t) {aka void(const unsigned char*, int, short int*, int)}? to ?void (* [2])(const pixel*, intptr_t, int16_t*, intptr_t) {aka void (* [2])(const unsigned char*, int, short int*, int)}?
+	   p.pu[LUMA_64x48].convert_p2s = PFX(filterPixelToShort_64x48_neon);
+					^
+
+Downloaded from upstream bug report:
+https://bitbucket.org/multicoreware/x265/issues/406
+
+Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
+
+--- ./source/common/arm/asm-primitives.cpp.orig	2018-05-21 02:33:10.000000000 -0600
++++ ./source/common/arm/asm-primitives.cpp	2018-05-28 20:38:37.302378303 -0600
+@@ -48,77 +48,77 @@ void setupAssemblyPrimitives(EncoderPrim
+         p.ssim_4x4x2_core = PFX(ssim_4x4x2_core_neon);
+ 
+         // addAvg
+-         p.pu[LUMA_4x4].addAvg   = PFX(addAvg_4x4_neon);
+-         p.pu[LUMA_4x8].addAvg   = PFX(addAvg_4x8_neon);
+-         p.pu[LUMA_4x16].addAvg  = PFX(addAvg_4x16_neon);
+-         p.pu[LUMA_8x4].addAvg   = PFX(addAvg_8x4_neon);
+-         p.pu[LUMA_8x8].addAvg   = PFX(addAvg_8x8_neon);
+-         p.pu[LUMA_8x16].addAvg  = PFX(addAvg_8x16_neon);
+-         p.pu[LUMA_8x32].addAvg  = PFX(addAvg_8x32_neon);
+-         p.pu[LUMA_12x16].addAvg = PFX(addAvg_12x16_neon);
+-         p.pu[LUMA_16x4].addAvg  = PFX(addAvg_16x4_neon);
+-         p.pu[LUMA_16x8].addAvg  = PFX(addAvg_16x8_neon);
+-         p.pu[LUMA_16x12].addAvg = PFX(addAvg_16x12_neon);
+-         p.pu[LUMA_16x16].addAvg = PFX(addAvg_16x16_neon);
+-         p.pu[LUMA_16x32].addAvg = PFX(addAvg_16x32_neon);
+-         p.pu[LUMA_16x64].addAvg = PFX(addAvg_16x64_neon);
+-         p.pu[LUMA_24x32].addAvg = PFX(addAvg_24x32_neon);
+-         p.pu[LUMA_32x8].addAvg  = PFX(addAvg_32x8_neon);
+-         p.pu[LUMA_32x16].addAvg = PFX(addAvg_32x16_neon);
+-         p.pu[LUMA_32x24].addAvg = PFX(addAvg_32x24_neon);
+-         p.pu[LUMA_32x32].addAvg = PFX(addAvg_32x32_neon);
+-         p.pu[LUMA_32x64].addAvg = PFX(addAvg_32x64_neon);
+-         p.pu[LUMA_48x64].addAvg = PFX(addAvg_48x64_neon);
+-         p.pu[LUMA_64x16].addAvg = PFX(addAvg_64x16_neon);
+-         p.pu[LUMA_64x32].addAvg = PFX(addAvg_64x32_neon);
+-         p.pu[LUMA_64x48].addAvg = PFX(addAvg_64x48_neon);
+-         p.pu[LUMA_64x64].addAvg = PFX(addAvg_64x64_neon);
++         p.pu[LUMA_4x4].addAvg[ALIGNED]   = PFX(addAvg_4x4_neon);
++         p.pu[LUMA_4x8].addAvg[ALIGNED]   = PFX(addAvg_4x8_neon);
++         p.pu[LUMA_4x16].addAvg[ALIGNED]  = PFX(addAvg_4x16_neon);
++         p.pu[LUMA_8x4].addAvg[ALIGNED]   = PFX(addAvg_8x4_neon);
++         p.pu[LUMA_8x8].addAvg[ALIGNED]   = PFX(addAvg_8x8_neon);
++         p.pu[LUMA_8x16].addAvg[ALIGNED]  = PFX(addAvg_8x16_neon);
++         p.pu[LUMA_8x32].addAvg[ALIGNED]  = PFX(addAvg_8x32_neon);
++         p.pu[LUMA_12x16].addAvg[ALIGNED] = PFX(addAvg_12x16_neon);
++         p.pu[LUMA_16x4].addAvg[ALIGNED]  = PFX(addAvg_16x4_neon);
++         p.pu[LUMA_16x8].addAvg[ALIGNED]  = PFX(addAvg_16x8_neon);
++         p.pu[LUMA_16x12].addAvg[ALIGNED] = PFX(addAvg_16x12_neon);
++         p.pu[LUMA_16x16].addAvg[ALIGNED] = PFX(addAvg_16x16_neon);
++         p.pu[LUMA_16x32].addAvg[ALIGNED] = PFX(addAvg_16x32_neon);
++         p.pu[LUMA_16x64].addAvg[ALIGNED] = PFX(addAvg_16x64_neon);
++         p.pu[LUMA_24x32].addAvg[ALIGNED] = PFX(addAvg_24x32_neon);
++         p.pu[LUMA_32x8].addAvg[ALIGNED]  = PFX(addAvg_32x8_neon);
++         p.pu[LUMA_32x16].addAvg[ALIGNED] = PFX(addAvg_32x16_neon);
++         p.pu[LUMA_32x24].addAvg[ALIGNED] = PFX(addAvg_32x24_neon);
++         p.pu[LUMA_32x32].addAvg[ALIGNED] = PFX(addAvg_32x32_neon);
++         p.pu[LUMA_32x64].addAvg[ALIGNED] = PFX(addAvg_32x64_neon);
++         p.pu[LUMA_48x64].addAvg[ALIGNED] = PFX(addAvg_48x64_neon);
++         p.pu[LUMA_64x16].addAvg[ALIGNED] = PFX(addAvg_64x16_neon);
++         p.pu[LUMA_64x32].addAvg[ALIGNED] = PFX(addAvg_64x32_neon);
++         p.pu[LUMA_64x48].addAvg[ALIGNED] = PFX(addAvg_64x48_neon);
++         p.pu[LUMA_64x64].addAvg[ALIGNED] = PFX(addAvg_64x64_neon);
+ 
+         // chroma addAvg
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_4x2].addAvg   = PFX(addAvg_4x2_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_4x4].addAvg   = PFX(addAvg_4x4_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_4x8].addAvg   = PFX(addAvg_4x8_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_4x16].addAvg  = PFX(addAvg_4x16_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_6x8].addAvg   = PFX(addAvg_6x8_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_8x2].addAvg   = PFX(addAvg_8x2_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_8x4].addAvg   = PFX(addAvg_8x4_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_8x6].addAvg   = PFX(addAvg_8x6_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_8x8].addAvg   = PFX(addAvg_8x8_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_8x16].addAvg  = PFX(addAvg_8x16_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_8x32].addAvg  = PFX(addAvg_8x32_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_12x16].addAvg = PFX(addAvg_12x16_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_16x4].addAvg  = PFX(addAvg_16x4_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_16x8].addAvg  = PFX(addAvg_16x8_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_16x12].addAvg = PFX(addAvg_16x12_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_16x16].addAvg = PFX(addAvg_16x16_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_16x32].addAvg = PFX(addAvg_16x32_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_24x32].addAvg = PFX(addAvg_24x32_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_32x8].addAvg  = PFX(addAvg_32x8_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_32x16].addAvg = PFX(addAvg_32x16_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_32x24].addAvg = PFX(addAvg_32x24_neon);
+-        p.chroma[X265_CSP_I420].pu[CHROMA_420_32x32].addAvg = PFX(addAvg_32x32_neon);
+-
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_4x8].addAvg   = PFX(addAvg_4x8_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_4x16].addAvg  = PFX(addAvg_4x16_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_4x32].addAvg  = PFX(addAvg_4x32_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].addAvg  = PFX(addAvg_6x16_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x4].addAvg   = PFX(addAvg_8x4_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x8].addAvg   = PFX(addAvg_8x8_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x12].addAvg  = PFX(addAvg_8x12_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x16].addAvg  = PFX(addAvg_8x16_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x32].addAvg  = PFX(addAvg_8x32_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x64].addAvg  = PFX(addAvg_8x64_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_12x32].addAvg = PFX(addAvg_12x32_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_16x8].addAvg  = PFX(addAvg_16x8_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_16x16].addAvg = PFX(addAvg_16x16_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_16x24].addAvg = PFX(addAvg_16x24_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_16x32].addAvg = PFX(addAvg_16x32_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_16x64].addAvg = PFX(addAvg_16x64_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_24x64].addAvg = PFX(addAvg_24x64_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_32x16].addAvg = PFX(addAvg_32x16_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_32x32].addAvg = PFX(addAvg_32x32_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_32x48].addAvg = PFX(addAvg_32x48_neon);
+-        p.chroma[X265_CSP_I422].pu[CHROMA_422_32x64].addAvg = PFX(addAvg_32x64_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_4x2].addAvg[ALIGNED]   = PFX(addAvg_4x2_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_4x4].addAvg[ALIGNED]   = PFX(addAvg_4x4_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_4x8].addAvg[ALIGNED]   = PFX(addAvg_4x8_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_4x16].addAvg[ALIGNED]  = PFX(addAvg_4x16_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_6x8].addAvg[ALIGNED]   = PFX(addAvg_6x8_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_8x2].addAvg[ALIGNED]   = PFX(addAvg_8x2_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_8x4].addAvg[ALIGNED]   = PFX(addAvg_8x4_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_8x6].addAvg[ALIGNED]   = PFX(addAvg_8x6_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_8x8].addAvg[ALIGNED]   = PFX(addAvg_8x8_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_8x16].addAvg[ALIGNED]  = PFX(addAvg_8x16_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_8x32].addAvg[ALIGNED]  = PFX(addAvg_8x32_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_12x16].addAvg[ALIGNED] = PFX(addAvg_12x16_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_16x4].addAvg[ALIGNED]  = PFX(addAvg_16x4_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_16x8].addAvg[ALIGNED]  = PFX(addAvg_16x8_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_16x12].addAvg[ALIGNED] = PFX(addAvg_16x12_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_16x16].addAvg[ALIGNED] = PFX(addAvg_16x16_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_16x32].addAvg[ALIGNED] = PFX(addAvg_16x32_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_24x32].addAvg[ALIGNED] = PFX(addAvg_24x32_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_32x8].addAvg[ALIGNED]  = PFX(addAvg_32x8_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_32x16].addAvg[ALIGNED] = PFX(addAvg_32x16_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_32x24].addAvg[ALIGNED] = PFX(addAvg_32x24_neon);
++        p.chroma[X265_CSP_I420].pu[CHROMA_420_32x32].addAvg[ALIGNED] = PFX(addAvg_32x32_neon);
++
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_4x8].addAvg[ALIGNED]   = PFX(addAvg_4x8_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_4x16].addAvg[ALIGNED]  = PFX(addAvg_4x16_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_4x32].addAvg[ALIGNED]  = PFX(addAvg_4x32_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].addAvg[ALIGNED]  = PFX(addAvg_6x16_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x4].addAvg[ALIGNED]   = PFX(addAvg_8x4_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x8].addAvg[ALIGNED]   = PFX(addAvg_8x8_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x12].addAvg[ALIGNED]  = PFX(addAvg_8x12_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x16].addAvg[ALIGNED]  = PFX(addAvg_8x16_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x32].addAvg[ALIGNED]  = PFX(addAvg_8x32_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x64].addAvg[ALIGNED]  = PFX(addAvg_8x64_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_12x32].addAvg[ALIGNED] = PFX(addAvg_12x32_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_16x8].addAvg[ALIGNED]  = PFX(addAvg_16x8_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_16x16].addAvg[ALIGNED] = PFX(addAvg_16x16_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_16x24].addAvg[ALIGNED] = PFX(addAvg_16x24_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_16x32].addAvg[ALIGNED] = PFX(addAvg_16x32_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_16x64].addAvg[ALIGNED] = PFX(addAvg_16x64_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_24x64].addAvg[ALIGNED] = PFX(addAvg_24x64_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_32x16].addAvg[ALIGNED] = PFX(addAvg_32x16_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_32x32].addAvg[ALIGNED] = PFX(addAvg_32x32_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_32x48].addAvg[ALIGNED] = PFX(addAvg_32x48_neon);
++        p.chroma[X265_CSP_I422].pu[CHROMA_422_32x64].addAvg[ALIGNED] = PFX(addAvg_32x64_neon);
+ 
+         // quant
+          p.quant = PFX(quant_neon);
+@@ -402,7 +402,7 @@ void setupAssemblyPrimitives(EncoderPrim
+         p.scale2D_64to32  = PFX(scale2D_64to32_neon);
+ 
+         // scale1D_128to64
+-        p.scale1D_128to64 = PFX(scale1D_128to64_neon);
++        p.scale1D_128to64[ALIGNED] = PFX(scale1D_128to64_neon);
+ 
+         // copy_count
+         p.cu[BLOCK_4x4].copy_cnt     = PFX(copy_cnt_4_neon);
+@@ -411,37 +411,37 @@ void setupAssemblyPrimitives(EncoderPrim
+         p.cu[BLOCK_32x32].copy_cnt   = PFX(copy_cnt_32_neon);
+ 
+         // filterPixelToShort
+-        p.pu[LUMA_4x4].convert_p2s   = PFX(filterPixelToShort_4x4_neon);
+-        p.pu[LUMA_4x8].convert_p2s   = PFX(filterPixelToShort_4x8_neon);
+-        p.pu[LUMA_4x16].convert_p2s  = PFX(filterPixelToShort_4x16_neon);
+-        p.pu[LUMA_8x4].convert_p2s   = PFX(filterPixelToShort_8x4_neon);
+-        p.pu[LUMA_8x8].convert_p2s   = PFX(filterPixelToShort_8x8_neon);
+-        p.pu[LUMA_8x16].convert_p2s  = PFX(filterPixelToShort_8x16_neon);
+-        p.pu[LUMA_8x32].convert_p2s  = PFX(filterPixelToShort_8x32_neon);
+-        p.pu[LUMA_12x16].convert_p2s = PFX(filterPixelToShort_12x16_neon);
+-        p.pu[LUMA_16x4].convert_p2s  = PFX(filterPixelToShort_16x4_neon);
+-        p.pu[LUMA_16x8].convert_p2s  = PFX(filterPixelToShort_16x8_neon);
+-        p.pu[LUMA_16x12].convert_p2s = PFX(filterPixelToShort_16x12_neon);
+-        p.pu[LUMA_16x16].convert_p2s = PFX(filterPixelToShort_16x16_neon);
+-        p.pu[LUMA_16x32].convert_p2s = PFX(filterPixelToShort_16x32_neon);
+-        p.pu[LUMA_16x64].convert_p2s = PFX(filterPixelToShort_16x64_neon);
+-        p.pu[LUMA_24x32].convert_p2s = PFX(filterPixelToShort_24x32_neon);
+-        p.pu[LUMA_32x8].convert_p2s  = PFX(filterPixelToShort_32x8_neon);
+-        p.pu[LUMA_32x16].convert_p2s = PFX(filterPixelToShort_32x16_neon);
+-        p.pu[LUMA_32x24].convert_p2s = PFX(filterPixelToShort_32x24_neon);
+-        p.pu[LUMA_32x32].convert_p2s = PFX(filterPixelToShort_32x32_neon);
+-        p.pu[LUMA_32x64].convert_p2s = PFX(filterPixelToShort_32x64_neon);
+-        p.pu[LUMA_48x64].convert_p2s = PFX(filterPixelToShort_48x64_neon);
+-        p.pu[LUMA_64x16].convert_p2s = PFX(filterPixelToShort_64x16_neon);
+-        p.pu[LUMA_64x32].convert_p2s = PFX(filterPixelToShort_64x32_neon);
+-        p.pu[LUMA_64x48].convert_p2s = PFX(filterPixelToShort_64x48_neon);
+-        p.pu[LUMA_64x64].convert_p2s = PFX(filterPixelToShort_64x64_neon);
++        p.pu[LUMA_4x4].convert_p2s[ALIGNED]   = PFX(filterPixelToShort_4x4_neon);
++        p.pu[LUMA_4x8].convert_p2s[ALIGNED]   = PFX(filterPixelToShort_4x8_neon);
++        p.pu[LUMA_4x16].convert_p2s[ALIGNED]  = PFX(filterPixelToShort_4x16_neon);
++        p.pu[LUMA_8x4].convert_p2s[ALIGNED]   = PFX(filterPixelToShort_8x4_neon);
++        p.pu[LUMA_8x8].convert_p2s[ALIGNED]   = PFX(filterPixelToShort_8x8_neon);
++        p.pu[LUMA_8x16].convert_p2s[ALIGNED]  = PFX(filterPixelToShort_8x16_neon);
++        p.pu[LUMA_8x32].convert_p2s[ALIGNED]  = PFX(filterPixelToShort_8x32_neon);
++        p.pu[LUMA_12x16].convert_p2s[ALIGNED] = PFX(filterPixelToShort_12x16_neon);
++        p.pu[LUMA_16x4].convert_p2s[ALIGNED]  = PFX(filterPixelToShort_16x4_neon);
++        p.pu[LUMA_16x8].convert_p2s[ALIGNED]  = PFX(filterPixelToShort_16x8_neon);
++        p.pu[LUMA_16x12].convert_p2s[ALIGNED] = PFX(filterPixelToShort_16x12_neon);
++        p.pu[LUMA_16x16].convert_p2s[ALIGNED] = PFX(filterPixelToShort_16x16_neon);
++        p.pu[LUMA_16x32].convert_p2s[ALIGNED] = PFX(filterPixelToShort_16x32_neon);
++        p.pu[LUMA_16x64].convert_p2s[ALIGNED] = PFX(filterPixelToShort_16x64_neon);
++        p.pu[LUMA_24x32].convert_p2s[ALIGNED] = PFX(filterPixelToShort_24x32_neon);
++        p.pu[LUMA_32x8].convert_p2s[ALIGNED]  = PFX(filterPixelToShort_32x8_neon);
++        p.pu[LUMA_32x16].convert_p2s[ALIGNED] = PFX(filterPixelToShort_32x16_neon);
++        p.pu[LUMA_32x24].convert_p2s[ALIGNED] = PFX(filterPixelToShort_32x24_neon);
++        p.pu[LUMA_32x32].convert_p2s[ALIGNED] = PFX(filterPixelToShort_32x32_neon);
++        p.pu[LUMA_32x64].convert_p2s[ALIGNED] = PFX(filterPixelToShort_32x64_neon);
++        p.pu[LUMA_48x64].convert_p2s[ALIGNED] = PFX(filterPixelToShort_48x64_neon);
++        p.pu[LUMA_64x16].convert_p2s[ALIGNED] = PFX(filterPixelToShort_64x16_neon);
++        p.pu[LUMA_64x32].convert_p2s[ALIGNED] = PFX(filterPixelToShort_64x32_neon);
++        p.pu[LUMA_64x48].convert_p2s[ALIGNED] = PFX(filterPixelToShort_64x48_neon);
++        p.pu[LUMA_64x64].convert_p2s[ALIGNED] = PFX(filterPixelToShort_64x64_neon);
+ 
+         // Block_fill
+-        p.cu[BLOCK_4x4].blockfill_s   = PFX(blockfill_s_4x4_neon);
+-        p.cu[BLOCK_8x8].blockfill_s   = PFX(blockfill_s_8x8_neon);
+-        p.cu[BLOCK_16x16].blockfill_s = PFX(blockfill_s_16x16_neon);
+-        p.cu[BLOCK_32x32].blockfill_s = PFX(blockfill_s_32x32_neon);
++        p.cu[BLOCK_4x4].blockfill_s[ALIGNED]   = PFX(blockfill_s_4x4_neon);
++        p.cu[BLOCK_8x8].blockfill_s[ALIGNED]   = PFX(blockfill_s_8x8_neon);
++        p.cu[BLOCK_16x16].blockfill_s[ALIGNED] = PFX(blockfill_s_16x16_neon);
++        p.cu[BLOCK_32x32].blockfill_s[ALIGNED] = PFX(blockfill_s_32x32_neon);
+ 
+         // Blockcopy_ss
+         p.cu[BLOCK_4x4].copy_ss   = PFX(blockcopy_ss_4x4_neon);
+@@ -495,21 +495,21 @@ void setupAssemblyPrimitives(EncoderPrim
+         p.chroma[X265_CSP_I422].cu[BLOCK_422_32x64].copy_sp = PFX(blockcopy_sp_32x64_neon);
+ 
+         // pixel_add_ps
+-        p.cu[BLOCK_4x4].add_ps   = PFX(pixel_add_ps_4x4_neon);
+-        p.cu[BLOCK_8x8].add_ps   = PFX(pixel_add_ps_8x8_neon);
+-        p.cu[BLOCK_16x16].add_ps = PFX(pixel_add_ps_16x16_neon);
+-        p.cu[BLOCK_32x32].add_ps = PFX(pixel_add_ps_32x32_neon);
+-        p.cu[BLOCK_64x64].add_ps = PFX(pixel_add_ps_64x64_neon);
++        p.cu[BLOCK_4x4].add_ps[ALIGNED]   = PFX(pixel_add_ps_4x4_neon);
++        p.cu[BLOCK_8x8].add_ps[ALIGNED]   = PFX(pixel_add_ps_8x8_neon);
++        p.cu[BLOCK_16x16].add_ps[ALIGNED] = PFX(pixel_add_ps_16x16_neon);
++        p.cu[BLOCK_32x32].add_ps[ALIGNED] = PFX(pixel_add_ps_32x32_neon);
++        p.cu[BLOCK_64x64].add_ps[ALIGNED] = PFX(pixel_add_ps_64x64_neon);
+ 
+         // chroma add_ps
+-        p.chroma[X265_CSP_I420].cu[BLOCK_420_4x4].add_ps   = PFX(pixel_add_ps_4x4_neon);
+-        p.chroma[X265_CSP_I420].cu[BLOCK_420_8x8].add_ps   = PFX(pixel_add_ps_8x8_neon);
+-        p.chroma[X265_CSP_I420].cu[BLOCK_420_16x16].add_ps = PFX(pixel_add_ps_16x16_neon);
+-        p.chroma[X265_CSP_I420].cu[BLOCK_420_32x32].add_ps = PFX(pixel_add_ps_32x32_neon);
+-        p.chroma[X265_CSP_I422].cu[BLOCK_422_4x8].add_ps   = PFX(pixel_add_ps_4x8_neon);
+-        p.chroma[X265_CSP_I422].cu[BLOCK_422_8x16].add_ps  = PFX(pixel_add_ps_8x16_neon);
+-        p.chroma[X265_CSP_I422].cu[BLOCK_422_16x32].add_ps = PFX(pixel_add_ps_16x32_neon);
+-        p.chroma[X265_CSP_I422].cu[BLOCK_422_32x64].add_ps = PFX(pixel_add_ps_32x64_neon);
++        p.chroma[X265_CSP_I420].cu[BLOCK_420_4x4].add_ps[ALIGNED]   = PFX(pixel_add_ps_4x4_neon);
++        p.chroma[X265_CSP_I420].cu[BLOCK_420_8x8].add_ps[ALIGNED]   = PFX(pixel_add_ps_8x8_neon);
++        p.chroma[X265_CSP_I420].cu[BLOCK_420_16x16].add_ps[ALIGNED] = PFX(pixel_add_ps_16x16_neon);
++        p.chroma[X265_CSP_I420].cu[BLOCK_420_32x32].add_ps[ALIGNED] = PFX(pixel_add_ps_32x32_neon);
++        p.chroma[X265_CSP_I422].cu[BLOCK_422_4x8].add_ps[ALIGNED]   = PFX(pixel_add_ps_4x8_neon);
++        p.chroma[X265_CSP_I422].cu[BLOCK_422_8x16].add_ps[ALIGNED]  = PFX(pixel_add_ps_8x16_neon);
++        p.chroma[X265_CSP_I422].cu[BLOCK_422_16x32].add_ps[ALIGNED] = PFX(pixel_add_ps_16x32_neon);
++        p.chroma[X265_CSP_I422].cu[BLOCK_422_32x64].add_ps[ALIGNED] = PFX(pixel_add_ps_32x64_neon);
+ 
+         // cpy2Dto1D_shr
+         p.cu[BLOCK_4x4].cpy2Dto1D_shr   = PFX(cpy2Dto1D_shr_4x4_neon);
+@@ -518,10 +518,10 @@ void setupAssemblyPrimitives(EncoderPrim
+         p.cu[BLOCK_32x32].cpy2Dto1D_shr = PFX(cpy2Dto1D_shr_32x32_neon);
+ 
+         // ssd_s
+-        p.cu[BLOCK_4x4].ssd_s   = PFX(pixel_ssd_s_4x4_neon);
+-        p.cu[BLOCK_8x8].ssd_s   = PFX(pixel_ssd_s_8x8_neon);
+-        p.cu[BLOCK_16x16].ssd_s = PFX(pixel_ssd_s_16x16_neon);
+-        p.cu[BLOCK_32x32].ssd_s = PFX(pixel_ssd_s_32x32_neon);
++        p.cu[BLOCK_4x4].ssd_s[ALIGNED]   = PFX(pixel_ssd_s_4x4_neon);
++        p.cu[BLOCK_8x8].ssd_s[ALIGNED]   = PFX(pixel_ssd_s_8x8_neon);
++        p.cu[BLOCK_16x16].ssd_s[ALIGNED] = PFX(pixel_ssd_s_16x16_neon);
++        p.cu[BLOCK_32x32].ssd_s[ALIGNED] = PFX(pixel_ssd_s_32x32_neon);
+ 
+         // sse_ss
+         p.cu[BLOCK_4x4].sse_ss   = PFX(pixel_sse_ss_4x4_neon);
+@@ -548,10 +548,10 @@ void setupAssemblyPrimitives(EncoderPrim
+         p.chroma[X265_CSP_I422].cu[BLOCK_422_32x64].sub_ps = PFX(pixel_sub_ps_32x64_neon);
+ 
+         // calc_Residual
+-        p.cu[BLOCK_4x4].calcresidual   = PFX(getResidual4_neon);
+-        p.cu[BLOCK_8x8].calcresidual   = PFX(getResidual8_neon);
+-        p.cu[BLOCK_16x16].calcresidual = PFX(getResidual16_neon);
+-        p.cu[BLOCK_32x32].calcresidual = PFX(getResidual32_neon);
++        p.cu[BLOCK_4x4].calcresidual[ALIGNED]   = PFX(getResidual4_neon);
++        p.cu[BLOCK_8x8].calcresidual[ALIGNED]   = PFX(getResidual8_neon);
++        p.cu[BLOCK_16x16].calcresidual[ALIGNED] = PFX(getResidual16_neon);
++        p.cu[BLOCK_32x32].calcresidual[ALIGNED] = PFX(getResidual32_neon);
+ 
+         // sse_pp
+         p.cu[BLOCK_4x4].sse_pp   = PFX(pixel_sse_pp_4x4_neon);
+@@ -722,31 +722,31 @@ void setupAssemblyPrimitives(EncoderPrim
+         p.pu[LUMA_64x64].sad_x4 = PFX(sad_x4_64x64_neon);
+ 
+         // pixel_avg_pp
+-        p.pu[LUMA_4x4].pixelavg_pp   = PFX(pixel_avg_pp_4x4_neon);
+-        p.pu[LUMA_4x8].pixelavg_pp   = PFX(pixel_avg_pp_4x8_neon);
+-        p.pu[LUMA_4x16].pixelavg_pp  = PFX(pixel_avg_pp_4x16_neon);
+-        p.pu[LUMA_8x4].pixelavg_pp   = PFX(pixel_avg_pp_8x4_neon);
+-        p.pu[LUMA_8x8].pixelavg_pp   = PFX(pixel_avg_pp_8x8_neon);
+-        p.pu[LUMA_8x16].pixelavg_pp  = PFX(pixel_avg_pp_8x16_neon);
+-        p.pu[LUMA_8x32].pixelavg_pp  = PFX(pixel_avg_pp_8x32_neon);
+-        p.pu[LUMA_12x16].pixelavg_pp = PFX(pixel_avg_pp_12x16_neon);
+-        p.pu[LUMA_16x4].pixelavg_pp  = PFX(pixel_avg_pp_16x4_neon);
+-        p.pu[LUMA_16x8].pixelavg_pp  = PFX(pixel_avg_pp_16x8_neon);
+-        p.pu[LUMA_16x12].pixelavg_pp = PFX(pixel_avg_pp_16x12_neon);
+-        p.pu[LUMA_16x16].pixelavg_pp = PFX(pixel_avg_pp_16x16_neon);
+-        p.pu[LUMA_16x32].pixelavg_pp = PFX(pixel_avg_pp_16x32_neon);
+-        p.pu[LUMA_16x64].pixelavg_pp = PFX(pixel_avg_pp_16x64_neon);
+-        p.pu[LUMA_24x32].pixelavg_pp = PFX(pixel_avg_pp_24x32_neon);
+-        p.pu[LUMA_32x8].pixelavg_pp  = PFX(pixel_avg_pp_32x8_neon);
+-        p.pu[LUMA_32x16].pixelavg_pp = PFX(pixel_avg_pp_32x16_neon);
+-        p.pu[LUMA_32x24].pixelavg_pp = PFX(pixel_avg_pp_32x24_neon);
+-        p.pu[LUMA_32x32].pixelavg_pp = PFX(pixel_avg_pp_32x32_neon);
+-        p.pu[LUMA_32x64].pixelavg_pp = PFX(pixel_avg_pp_32x64_neon);
+-        p.pu[LUMA_48x64].pixelavg_pp = PFX(pixel_avg_pp_48x64_neon);
+-        p.pu[LUMA_64x16].pixelavg_pp = PFX(pixel_avg_pp_64x16_neon);
+-        p.pu[LUMA_64x32].pixelavg_pp = PFX(pixel_avg_pp_64x32_neon);
+-        p.pu[LUMA_64x48].pixelavg_pp = PFX(pixel_avg_pp_64x48_neon);
+-        p.pu[LUMA_64x64].pixelavg_pp = PFX(pixel_avg_pp_64x64_neon);
++        p.pu[LUMA_4x4].pixelavg_pp[ALIGNED]   = PFX(pixel_avg_pp_4x4_neon);
++        p.pu[LUMA_4x8].pixelavg_pp[ALIGNED]   = PFX(pixel_avg_pp_4x8_neon);
++        p.pu[LUMA_4x16].pixelavg_pp[ALIGNED]  = PFX(pixel_avg_pp_4x16_neon);
++        p.pu[LUMA_8x4].pixelavg_pp[ALIGNED]   = PFX(pixel_avg_pp_8x4_neon);
++        p.pu[LUMA_8x8].pixelavg_pp[ALIGNED]   = PFX(pixel_avg_pp_8x8_neon);
++        p.pu[LUMA_8x16].pixelavg_pp[ALIGNED]  = PFX(pixel_avg_pp_8x16_neon);
++        p.pu[LUMA_8x32].pixelavg_pp[ALIGNED]  = PFX(pixel_avg_pp_8x32_neon);
++        p.pu[LUMA_12x16].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_12x16_neon);
++        p.pu[LUMA_16x4].pixelavg_pp[ALIGNED]  = PFX(pixel_avg_pp_16x4_neon);
++        p.pu[LUMA_16x8].pixelavg_pp[ALIGNED]  = PFX(pixel_avg_pp_16x8_neon);
++        p.pu[LUMA_16x12].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_16x12_neon);
++        p.pu[LUMA_16x16].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_16x16_neon);
++        p.pu[LUMA_16x32].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_16x32_neon);
++        p.pu[LUMA_16x64].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_16x64_neon);
++        p.pu[LUMA_24x32].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_24x32_neon);
++        p.pu[LUMA_32x8].pixelavg_pp[ALIGNED]  = PFX(pixel_avg_pp_32x8_neon);
++        p.pu[LUMA_32x16].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_32x16_neon);
++        p.pu[LUMA_32x24].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_32x24_neon);
++        p.pu[LUMA_32x32].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_32x32_neon);
++        p.pu[LUMA_32x64].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_32x64_neon);
++        p.pu[LUMA_48x64].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_48x64_neon);
++        p.pu[LUMA_64x16].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_64x16_neon);
++        p.pu[LUMA_64x32].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_64x32_neon);
++        p.pu[LUMA_64x48].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_64x48_neon);
++        p.pu[LUMA_64x64].pixelavg_pp[ALIGNED] = PFX(pixel_avg_pp_64x64_neon);
+ 
+         // planecopy
+         p.planecopy_cp = PFX(pixel_planecopy_cp_neon);