XIP adaptation for xtensa platform (#1636)

Add macro WASM_ENABLE_WORD_ALING_READ to enable reading
1/2/4 and n bytes data from vram buffer, which requires 4-byte addr
alignment reading.

Eliminate XIP AOT relocations related to the below ones:
   i32_div_u, f32_min, f32_max, f32_ceil, f32_floor, f32_trunc, f32_rint

core/shared/utils/bh_common.c

@@ -5,6 +5,73 @@
 
 #include "bh_common.h"
 
+static char *
+align_ptr(char *src, unsigned int b)
+{
+    uintptr_t v = (uintptr_t)src;
+    uintptr_t m = b - 1;
+    return (char *)((v + m) & ~m);
+}
+
+/*
+Memory copy, with word alignment
+*/
+int
+b_memcpy_wa(void *s1, unsigned int s1max, const void *s2, unsigned int n)
+{
+    char *dest = (char *)s1;
+    char *src = (char *)s2;
+
+    char *pa = align_ptr(src, 4);
+    char *pb = align_ptr((src + n), 4);
+
+    unsigned int buff;
+    const char *p_byte_read;
+
+    unsigned int *p;
+    char *ps;
+
+    if (pa > src) {
+        pa -= 4;
+    }
+
+    for (p = (unsigned int *)pa; p < (unsigned int *)pb; p++) {
+        buff = *(p);
+        p_byte_read = ((char *)&buff);
+
+        /* read leading word */
+        if ((char *)p <= src) {
+            for (ps = src; ps < ((char *)p + 4); ps++) {
+                if (ps >= src + n) {
+                    break;
+                }
+                p_byte_read = ((char *)&buff) + (ps - (char *)p);
+                *dest++ = *p_byte_read;
+            }
+        }
+        /* read trailing word */
+        else if ((char *)p >= pb - 4) {
+            for (ps = (char *)p; ps < src + n; ps++) {
+                *dest++ = *p_byte_read++;
+            }
+        }
+        /* read meaning word(s) */
+        else {
+            if ((char *)p + 4 >= src + n) {
+                for (ps = (char *)p; ps < src + n; ps++) {
+                    *dest++ = *p_byte_read++;
+                }
+            }
+            else {
+                *(unsigned int *)dest = buff;
+                dest += 4;
+            }
+        }
+    }
+
+    return 0;
+}
+
 int
 b_memcpy_s(void *s1, unsigned int s1max, const void *s2, unsigned int n)
 {

core/shared/utils/bh_common.h

@@ -19,6 +19,13 @@ extern "C" {
         bh_assert(_ret == 0);                                         \
     } while (0)
 
+#define bh_memcpy_wa(dest, dlen, src, slen)                            \
+    do {                                                               \
+        int _ret = slen == 0 ? 0 : b_memcpy_wa(dest, dlen, src, slen); \
+        (void)_ret;                                                    \
+        bh_assert(_ret == 0);                                          \
+    } while (0)
+
 #define bh_memmove_s(dest, dlen, src, slen)                            \
     do {                                                               \
         int _ret = slen == 0 ? 0 : b_memmove_s(dest, dlen, src, slen); \
@@ -43,6 +50,8 @@ extern "C" {
 int
 b_memcpy_s(void *s1, unsigned int s1max, const void *s2, unsigned int n);
 int
+b_memcpy_wa(void *s1, unsigned int s1max, const void *s2, unsigned int n);
+int
 b_memmove_s(void *s1, unsigned int s1max, const void *s2, unsigned int n);
 int
 b_strcat_s(char *s1, unsigned int s1max, const char *s2);