Fix compilation, still needs more work.

,gitignore

@@ -1,2 +0,0 @@
-__PYCACHE__
-.*

src/cha4block.h

@@ -1,3 +1,10 @@
+#if defined(__x86_64__)
+#include <immintrin.h> // AVX2
+#include <tmmintrin.h> // SSSE3
+#elif defined(__aarch64__)
+#include "sse2neon.h"
+#endif
+
 #define VEC4_ROT(A, IMM)                                                       \
     _mm_or_si128(_mm_slli_epi32(A, IMM), _mm_srli_epi32(A, (32 - IMM)))
 

src/cha8block.h

@@ -1,3 +1,6 @@
+#include <immintrin.h> // AVX2
+#include <tmmintrin.h> // SSSE3
+
 #define VEC8_ROT(A, IMM)                                                       \
     _mm256_or_si256(_mm256_slli_epi32(A, IMM), _mm256_srli_epi32(A, (32 - IMM)))
 

src/chacha20.h

@@ -1,7 +1,7 @@
 #include <stdbool.h>
 #include <stdint.h>
 
-static const uint64_t CHA_BLOCK_SIZE = 64;
+#define CHA_BLOCK_SIZE 64
 
 typedef struct cha_ctx {
     uint32_t state[16];
@@ -15,20 +15,11 @@ typedef struct cha_ctx {
 #pragma GCC target("ssse3")
 #pragma GCC target("avx2")
 #endif
-
-#include <emmintrin.h>  // SSE2
-#include "cha4block.h"
-
-#include <immintrin.h>  // AVX2
-#include <tmmintrin.h>  // SSSE3
 #include "cha8block.h"
-
-#elif defined(__aarch64__)
-#include "sse2neon.h"
-#include "cha4block.h"
 #endif
 
 #include "cha1block.h"
+#include "cha4block.h"
 
 #include <assert.h>
 #include <stdbool.h>
@@ -40,7 +31,6 @@ typedef struct cha_ctx {
 #include <time.h>
 #include <unistd.h>
 
-
 /// @brief Initialize cha_ctx
 /// @param ctx holds ChaCha20 state
 /// @param key 32 byte key
@@ -79,7 +69,8 @@ void cha_update(cha_ctx* ctx, uint8_t* out, uint64_t outlen) {
         if (ctx->uncount) {
             memmove(ctx->unconsumed, ctx->unconsumed + N, ctx->uncount);
         }
-        if (c == out + outlen) return;
+        if (c == out + outlen)
+            return;
     }
 #if defined(__x86_64__)
     // TODO: Handle resume if we are not at block boundary