From bd6d015e6ff7ccc68fa50662e5b714aa4d5e9d45 Mon Sep 17 00:00:00 2001
From: Berkus Decker <berkus+github@metta.systems>
Date: Sun, 10 Jan 2021 23:08:03 +0200
Subject: [PATCH] Fix arm simd shuffling arguments

According to packed_simd docs, the syntax for simd_shuffle4()
is as follows:

"The indices must be in range [0, M * N) where M is the number of input
vectors (1 or 2)and N is the number of lanes of the input vectors. The
indices i in range [0, N) refer to the i-th element of vec0, while the
indices in range [N, 2*N) refer to the i - N-th element of vec1."

I did not find implementation or documentation for simd_shuffle4() but
I believe packed_simd implements exactly the same interface.

Plus, implementing this change has fixed font-kit glyph output on
an Apple M1 mac (64-bit arm).
---
 simd/src/arm/mod.rs | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/simd/src/arm/mod.rs b/simd/src/arm/mod.rs
index 606914e4..931a64a5 100644
--- a/simd/src/arm/mod.rs
+++ b/simd/src/arm/mod.rs
@@ -129,7 +129,7 @@ impl F32x2 {
 
     #[inline]
     pub fn concat_xy_xy(self, other: F32x2) -> F32x4 {
-        unsafe { F32x4(simd_shuffle4(self.0, other.0, [0, 1, 0, 1])) }
+        unsafe { F32x4(simd_shuffle4(self.0, other.0, [0, 1, 2, 3])) }
     }
 }
 
@@ -314,17 +314,17 @@ impl F32x4 {
 
     #[inline]
     pub fn concat_xy_xy(self, other: F32x4) -> F32x4 {
-        unsafe { F32x4(simd_shuffle4(self.0, other.0, [0, 1, 0, 1])) }
-    }
-
-    #[inline]
-    pub fn concat_xy_zw(self, other: F32x4) -> F32x4 {
         unsafe { F32x4(simd_shuffle4(self.0, other.0, [0, 1, 2, 3])) }
     }
 
+    #[inline]
+    pub fn concat_xy_zw(self, other: F32x4) -> F32x4 {
+        unsafe { F32x4(simd_shuffle4(self.0, other.0, [0, 1, 6, 7])) }
+    }
+
     #[inline]
     pub fn concat_zw_zw(self, other: F32x4) -> F32x4 {
-        unsafe { F32x4(simd_shuffle4(self.0, other.0, [2, 3, 2, 3])) }
+        unsafe { F32x4(simd_shuffle4(self.0, other.0, [2, 3, 6, 7])) }
     }
 
     // Conversions
@@ -461,7 +461,7 @@ impl I32x2 {
 
     #[inline]
     pub fn concat_xy_xy(self, other: I32x2) -> I32x4 {
-        unsafe { I32x4(simd_shuffle4(self.0, other.0, [0, 1, 0, 1])) }
+        unsafe { I32x4(simd_shuffle4(self.0, other.0, [0, 1, 2, 3])) }
     }
 
     // Conversions
@@ -471,7 +471,7 @@ impl I32x2 {
     pub fn to_f32x2(self) -> F32x2 {
         unsafe { F32x2(simd_cast(self.0)) }
     }
-    
+
     #[inline]
     pub fn to_i32x4(self) -> I32x4 {
         self.concat_xy_xy(I32x2::default())