esp32: xtensa_user: Implement a few more instructions

You can find them used in the ROM version of memcpy.
While it might be controversial if it's a good idea to use the ROM version
of these functions, it's nicer to support more instructions here anyway.

arch/xtensa/src/esp32/esp32_user.c

@@ -158,6 +158,36 @@ static int decode_s8i(const uint8_t *p, uint8_t *imm8, uint8_t *s,
   return 0;
 }
 
+/****************************************************************************
+ * Name: decode_s16i
+ *
+ * Description:
+ *   Decode S16I instruction using 32-bit aligned access.
+ *   Return non-zero on successful decoding.
+ *
+ ****************************************************************************/
+
+static int decode_s16i(const uint8_t *p, uint8_t *imm8, uint8_t *s,
+                       uint8_t *t)
+{
+  /*  23           16 15   12 11    8 7     4 3     0
+   * | imm8          |0 1 0 1| s     | t     |0 0 1 0|
+   */
+
+  uint8_t b0 = load_uint8(p);
+  uint8_t b1 = load_uint8(p + 1);
+
+  if ((b0 & 0xf) == 2 && (b1 & 0xf0) == 0x50)
+    {
+      *t = b0 >> 4;
+      *s = b1 & 0xf;
+      *imm8 = load_uint8(p + 2);
+      return 1;
+    }
+
+  return 0;
+}
+
 /****************************************************************************
  * Name: decode_l8ui
  *
@@ -188,6 +218,36 @@ static int decode_l8ui(const uint8_t *p, uint8_t *imm8, uint8_t *s,
   return 0;
 }
 
+/****************************************************************************
+ * Name: decode_l16ui
+ *
+ * Description:
+ *   Decode L16UI instruction using 32-bit aligned access.
+ *   Return non-zero on successful decoding.
+ *
+ ****************************************************************************/
+
+static int decode_l16ui(const uint8_t *p, uint8_t *imm8, uint8_t *s,
+                       uint8_t *t)
+{
+  /*  23           16 15   12 11    8 7     4 3     0
+   * | imm8          |0 0 0 1| s     | t     |0 0 1 0|
+   */
+
+  uint8_t b0 = load_uint8(p);
+  uint8_t b1 = load_uint8(p + 1);
+
+  if ((b0 & 0xf) == 2 && (b1 & 0xf0) == 0x10)
+    {
+      *t = b0 >> 4;
+      *s = b1 & 0xf;
+      *imm8 = load_uint8(p + 2);
+      return 1;
+    }
+
+  return 0;
+}
+
 /****************************************************************************
  * Name: advance_pc
  *
@@ -274,6 +334,23 @@ uint32_t *xtensa_user(int exccause, uint32_t *regs)
           advance_pc(regs, 3);
           return regs;
         }
+      else if (decode_s16i(pc, &imm8, &s, &t))
+        {
+          binfo("Emulating S16I imm8=%u, s=%u (%p), t=%u (%p)\n",
+                (unsigned int)imm8,
+                (unsigned int)s,
+                (void *)regs[REG_A0 + s],
+                (unsigned int)t,
+                (void *)regs[REG_A0 + t]);
+
+          DEBUGASSERT(regs[REG_A0 + s] + imm8 == regs[REG_EXCVADDR]);
+          store_uint8(((uint8_t *)regs[REG_A0 + s]) + imm8,
+                      regs[REG_A0 + t]);
+          store_uint8(((uint8_t *)regs[REG_A0 + s]) + imm8 + 1,
+                      regs[REG_A0 + t] >> 8);
+          advance_pc(regs, 3);
+          return regs;
+        }
       else if (decode_l8ui(pc, &imm8, &s, &t))
         {
           binfo("Emulating L8UI imm8=%u, s=%u (%p), t=%u (%p)\n",
@@ -289,6 +366,22 @@ uint32_t *xtensa_user(int exccause, uint32_t *regs)
           advance_pc(regs, 3);
           return regs;
         }
+      else if (decode_l16ui(pc, &imm8, &s, &t))
+        {
+          binfo("Emulating L16UI imm8=%u, s=%u (%p), t=%u (%p)\n",
+                (unsigned int)imm8,
+                (unsigned int)s,
+                (void *)regs[REG_A0 + s],
+                (unsigned int)t,
+                (void *)regs[REG_A0 + t]);
+
+          DEBUGASSERT(regs[REG_A0 + s] + imm8 == regs[REG_EXCVADDR]);
+          uint8_t lo = load_uint8(((uint8_t *)regs[REG_A0 + s]) + imm8);
+          uint8_t hi = load_uint8(((uint8_t *)regs[REG_A0 + s]) + imm8 + 1);
+          regs[REG_A0 + t] = (hi << 8) | lo;
+          advance_pc(regs, 3);
+          return regs;
+        }
     }
 
 #endif