sys/pgmspace.h: Refactor inline asm (#7510)

* Refactor asm tabs for `.S` cosmetics

* Avoid fixed regname (`a14`) allocation

* Remove unnecessary explicit cast and word masking

Author: jjsuwa

Diff for: tools/sdk/libc/xtensa-lx106-elf/include/sys/pgmspace.h

@@ -61,42 +61,43 @@ extern "C" {
 //     w1,     w0
 
 #define pgm_read_with_offset(addr, res) \
-  __asm__("extui    %0, %1, 0, 2\n"     /* Extract offset within word (in bytes) */ \
-      "sub      %1, %1, %0\n"       /* Subtract offset from addr, yielding an aligned address */ \
-      "l32i.n   %1, %1, 0x0\n"      /* Load word from aligned address */ \
-      "ssa8l    %0\n"               /* Prepare to shift by offset (in bits) */ \
-      "src      %0, %1, %1\n"       /* Shift right; now the requested byte is the first one */ \
-      :"=r"(res), "=r"(addr) \
-      :"1"(addr) \
-      :);
+  __asm__( \
+    "extui\t%0, %1, 0, 2\n\t"  /* Extract offset within word (in bytes) */ \
+    "sub\t%1, %1, %0\n\t"      /* Subtract offset from addr, yielding an aligned address */ \
+    "l32i.n\t%1, %1, 0\n\t"    /* Load word from aligned address */ \
+    "ssa8l\t%0\n\t"            /* Prepare to shift by offset (in bits) */ \
+    "src\t%0, %1, %1"          /* Shift right; now the requested byte is the first one */ \
+    : "=r"(res), "+r"(addr))
 
 #define pgm_read_dword_with_offset(addr, res) \
-  __asm__("extui    %0, %1, 0, 2\n"     /* Extract offset within word (in bytes) */ \
-      "sub      %1, %1, %0\n"       /* Subtract offset from addr, yielding an aligned address */ \
-      "l32i     a14, %1, 0\n" \
-      "l32i     %1, %1, 4\n" \
-      "ssa8l    %0\n" \
-      "src      %0, %1, a14\n" \
-      :"=r"(res), "=r"(addr) \
-      :"1"(addr) \
-      :"a14");
+  do { \
+    uint32_t temp; \
+    __asm__( \
+      "extui\t%0, %1, 0, 2\n\t"  /* Extract offset within word (in bytes) */ \
+      "sub\t%1, %1, %0\n\t"      /* Subtract offset from addr, yielding an aligned address */ \
+      "l32i.n\t%2, %1, 0\n\t"    /* Load 2 words */ \
+      "l32i.n\t%1, %1, 4\n\t"    /*     from aligned address */ \
+      "ssa8l\t%0\n\t"            /* Prepare to shift by offset (in bits) */ \
+      "src\t%0, %1, %2"          /* Shift right in order to extract the requested dword */ \
+      : "=r"(res), "+r"(addr), "=r"(temp)); \
+  } while(0)
 
 static inline uint8_t pgm_read_byte_inlined(const void* addr) {
   uint32_t res;
   pgm_read_with_offset(addr, res);
-  return (uint8_t) res;     /* This masks the lower byte from the returned word */
+  return res;  /* Implicit cast to uint8_t masks the lower byte from the returned word */
 }
 
 /* Although this says "word", it's actually 16 bit, i.e. half word on Xtensa */
 static inline uint16_t pgm_read_word_inlined(const void* addr) {
   uint32_t res;
   pgm_read_with_offset(addr, res);
-  return (uint16_t) res;    /* This masks the lower half-word from the returned word */
+  return res;  /* Implicit cast to uint16_t masks the lower half-word from the returned word */
 }
 
 /* Can't legally cast bits of uint32_t to a float w/o conversion or std::memcpy, which is inefficient. */
 /* The ASM block doesn't care the type, so just pass in what C thinks is a float and return in custom fcn. */
-static inline float pgm_read_float_unaligned(const void *addr) {
+static inline float pgm_read_float_unaligned(const void* addr) {
   float res;
   pgm_read_with_offset(addr, res);
   return res;
@@ -121,7 +122,7 @@ static inline uint32_t pgm_read_dword_unaligned(const void *addr) {
 }
 
 #define pgm_read_ptr_unaligned(addr)   ((void*)pgm_read_dword_unaligned(addr))
-#define pgm_read_word_unaligned(addr)  ((uint16_t)(pgm_read_dword_unaligned(addr) & 0xffff))
+#define pgm_read_word_unaligned(addr)  ((uint16_t)pgm_read_dword_unaligned(addr))
 
 
 // Allow selection of _aligned or _unaligned, but default to _unaligned for Arduino compatibility