From cce3259b5694bce0d5193f388e80fe7248ca951f Mon Sep 17 00:00:00 2001
From: Aryadev Chavali <aryadev@aryadevchavali.com>
Date: Sun, 28 Apr 2024 17:37:18 +0530
Subject: [PATCH] Fixed bugs in base.c

In particular, __LITTLE_ENDIAN__ was not a functioning macro.
Instead, I implemented a version by hand (copied from IBM) that
actually figures out if the machine is little endian or not.

Thank you unit testing!
---
 lib/base.c | 46 +++++++++++++++++++---------------------------
 lib/base.h | 15 +++++++++++----
 2 files changed, 30 insertions(+), 31 deletions(-)

diff --git a/lib/base.c b/lib/base.c
index 99186e6..f34307b 100644
--- a/lib/base.c
+++ b/lib/base.c
@@ -28,61 +28,61 @@ union word_pun
 
 hword_t hword_htobc(hword_t w)
 {
-#if __LITTLE_ENDIAN__
-  return w;
-#else
+  if (LITTLE_ENDIAN)
+    return w;
   union hword_pun x = {w};
   union hword_pun y = {0};
   for (size_t i = 0, j = HWORD_SIZE; i < HWORD_SIZE; ++i, --j)
     y.bytes[j - 1] = x.bytes[i];
   return y.h;
-#endif
 }
 
 hword_t hword_bctoh(hword_t w)
 {
-#if __LITTLE_ENDIAN__
-  return w;
-#else
+  if (LITTLE_ENDIAN)
+    return w;
   union hword_pun x = {w};
   union hword_pun y = {0};
   for (size_t i = 0, j = HWORD_SIZE; i < HWORD_SIZE; ++i, --j)
     y.bytes[j - 1] = x.bytes[i];
   return y.h;
-#endif
 }
 
 word_t word_htobc(word_t w)
 {
-#if __LITTLE_ENDIAN__
-  return w;
-#else
+  if (LITTLE_ENDIAN)
+    return w;
   union word_pun x = {w};
   union word_pun y = {0};
   for (size_t i = 0, j = WORD_SIZE; i < WORD_SIZE; ++i, --j)
     y.bytes[j - 1] = x.bytes[i];
   return y.h;
-#endif
 }
 
 word_t word_bctoh(word_t w)
 {
-#if __LITTLE_ENDIAN__
-  return w;
-#else
+  if (LITTLE_ENDIAN)
+    return w;
   union word_pun x = {w};
   union word_pun y = {0};
   for (size_t i = 0, j = WORD_SIZE; i < WORD_SIZE; ++i, --j)
     y.bytes[j - 1] = x.bytes[i];
   return y.h;
-#endif
 }
 
 hword_t convert_bytes_to_hword(byte_t *bytes)
 {
-  hword_t be_h = 0;
-  memcpy(&be_h, bytes, HWORD_SIZE);
-  hword_t h = hword_bctoh(be_h);
+  hword_t h = 0;
+  for (size_t i = 0; i < HWORD_SIZE; ++i)
+    h |= ((hword_t)(bytes[i]) << (8 * i));
+  return h;
+}
+
+word_t convert_bytes_to_word(byte_t *bytes)
+{
+  word_t h = 0;
+  for (size_t i = 0; i < WORD_SIZE; ++i)
+    h |= ((word_t)(bytes[i]) << (8 * i));
   return h;
 }
 
@@ -97,11 +97,3 @@ void convert_word_to_bytes(word_t w, byte_t *bytes)
   word_t be_w = word_htobc(w);
   memcpy(bytes, &be_w, WORD_SIZE);
 }
-
-word_t convert_bytes_to_word(byte_t *bytes)
-{
-  word_t be_w = 0;
-  memcpy(&be_w, bytes, WORD_SIZE);
-  word_t w = word_bctoh(be_w);
-  return w;
-}
diff --git a/lib/base.h b/lib/base.h
index 97bfeea..66043f1 100644
--- a/lib/base.h
+++ b/lib/base.h
@@ -97,6 +97,12 @@ typedef enum
 #define DHWORD(HWORD) ((data_t){.as_hword = (HWORD)})
 #define DWORD(WORD)   ((data_t){.as_word = (WORD)})
 
+// Macro to determine little endian
+#ifndef LITTLE_ENDIAN
+static const int __i = 1;
+#define LITTLE_ENDIAN ((*((byte_t *)&__i)) == 0)
+#endif
+
 /**
    @brief Safely subtract SUB from W, where both are words (64 bit integers).
 
@@ -124,8 +130,8 @@ typedef enum
    @brief Convert a buffer of bytes to a half word.
 
    @details We assume the buffer of bytes are in virtual machine byte
-   code format (big endian) and that they are at least HWORD_SIZE in
-   size.
+   code format (little endian) and that they are at least HWORD_SIZE
+   in size.
 */
 hword_t convert_bytes_to_hword(byte_t *buffer);
 
@@ -143,13 +149,14 @@ void convert_hword_to_bytes(hword_t h, byte_t *buffer);
    @brief Convert a buffer of bytes to a word.
 
    @details We assume the buffer of bytes are in virtual machine byte
-   code format (big endian) and that they are at least WORD_SIZE in
+   code format (little endian) and that they are at least WORD_SIZE in
    size.
 */
 word_t convert_bytes_to_word(byte_t *);
 
 /**
-   @brief Convert a word into a VM byte code format bytes (big endian)
+   @brief Convert a word into a VM byte code format bytes (little
+   endian)
 
    @param w: Word to convert