From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Misty De Meo <mistydemeo@gmail.com>
Date: Tue, 15 Sep 2020 16:57:26 -0700
Subject: [PATCH] Check for __arm64__, not just __arm__

On at least one 64-bit ARM processor I've tested (Apple Silicon on macOS),
__arm__ isn't defined but __arm64__ is. As a result, some of the
ARM-specific macros are missing and calls to them fail.
---
 src/cr-core.c    | 2 +-
 src/dev32s.h     | 2 +-
 src/pffft-wrap.c | 2 +-
 src/pffft.c      | 4 ++--
 4 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/src/cr-core.c b/src/cr-core.c
index 159a5d9..fe5ea8a 100644
--- a/src/cr-core.c
+++ b/src/cr-core.c
@@ -80,7 +80,7 @@ static void cubic_stage_fn(stage_t * p, fifo_t * output_fifo)
   #define DEFINED_X86 0
 #endif
 
-#if defined __arm__
+#if defined(__arm__) || defined(__arm64__)
   #define DEFINED_ARM 1
 #else
   #define DEFINED_ARM 0
diff --git a/src/dev32s.h b/src/dev32s.h
index 7edae86..a14d7ad 100644
--- a/src/dev32s.h
+++ b/src/dev32s.h
@@ -31,7 +31,7 @@ SIMD_INLINE(void) vStorSum(float * a, v4_t b) {
   v4_t t = vAdd(_mm_movehl_ps(b, b), b);
   _mm_store_ss(a, vAdd(t, _mm_shuffle_ps(t,t,1)));}
 
-#elif defined __arm__
+#elif defined(__arm__) || defined(__arm64__)
 
 #include <arm_neon.h>
 
diff --git a/src/pffft-wrap.c b/src/pffft-wrap.c
index c920f06..1641fc4 100644
--- a/src/pffft-wrap.c
+++ b/src/pffft-wrap.c
@@ -40,7 +40,7 @@ static void pffft_zconvolve(PFFFT_Setup *s, const float *a, const float *b, floa
 
   float ar, ai, br, bi;
 
-#ifdef __arm__
+#if defined(__arm__) || defined(__arm64__)
   __builtin_prefetch(va);
   __builtin_prefetch(vb);
   __builtin_prefetch(va+2);
diff --git a/src/pffft.c b/src/pffft.c
index 46c841e..8c775a9 100644
--- a/src/pffft.c
+++ b/src/pffft.c
@@ -157,7 +157,7 @@ typedef __m128 v4sf;
 /*
   ARM NEON support macros
 */
-#elif !defined(PFFFT_SIMD_DISABLE) && defined(__arm__)
+#elif !defined(PFFFT_SIMD_DISABLE) && (defined(__arm__) || defined(__arm64__))
 #  include <arm_neon.h>
 typedef float32x4_t v4sf;
 #  define SIMD_SZ 4
@@ -1732,7 +1732,7 @@ void pffft_zconvolve_accumulate(PFFFT_Setup *s, const float *a, const float *b,
   const v4sf * RESTRICT vb = (const v4sf*)b;
   v4sf * RESTRICT vab = (v4sf*)ab;
 
-#ifdef __arm__
+#if defined(__arm__) || defined(__arm64__)
   __builtin_prefetch(va);
   __builtin_prefetch(vb);
   __builtin_prefetch(vab);
-- 
2.30.1