--- qemu/tcg/x86_64/tcg-target.c	2018/04/24 17:21:45	1.1.1.2
+++ qemu/tcg/x86_64/tcg-target.c	2018/04/24 17:35:27	1.1.1.3
@@ -363,6 +363,20 @@ static inline void tcg_out_movi(TCGConte
     }
 }
 
+static void tcg_out_goto(TCGContext *s, int call, uint8_t *target)
+{
+    int32_t disp;
+
+    disp = target - s->code_ptr - 5;
+    if (disp == (target - s->code_ptr - 5)) {
+        tcg_out8(s, call ? 0xe8 : 0xe9);
+        tcg_out32(s, disp);
+    } else {
+        tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R10, (tcg_target_long) target);
+        tcg_out_modrm(s, 0xff, call ? 2 : 4, TCG_REG_R10);
+    }
+}
+
 static inline void tcg_out_ld(TCGContext *s, TCGType type, int ret,
                               int arg1, tcg_target_long arg2)
 {
@@ -383,7 +397,13 @@ static inline void tcg_out_st(TCGContext
 
 static inline void tgen_arithi32(TCGContext *s, int c, int r0, int32_t val)
 {
-    if (val == (int8_t)val) {
+    if ((c == ARITH_ADD && val == 1) || (c == ARITH_SUB && val == -1)) {
+        /* inc */
+        tcg_out_modrm(s, 0xff, 0, r0);
+    } else if ((c == ARITH_ADD && val == -1) || (c == ARITH_SUB && val == 1)) {
+        /* dec */
+        tcg_out_modrm(s, 0xff, 1, r0);
+    } else if (val == (int8_t)val) {
         tcg_out_modrm(s, 0x83, c, r0);
         tcg_out8(s, val);
     } else if (c == ARITH_AND && val == 0xffu) {
@@ -400,7 +420,13 @@ static inline void tgen_arithi32(TCGCont
 
 static inline void tgen_arithi64(TCGContext *s, int c, int r0, int64_t val)
 {
-    if (val == (int8_t)val) {
+    if ((c == ARITH_ADD && val == 1) || (c == ARITH_SUB && val == -1)) {
+        /* inc */
+        tcg_out_modrm(s, 0xff | P_REXW, 0, r0);
+    } else if ((c == ARITH_ADD && val == -1) || (c == ARITH_SUB && val == 1)) {
+        /* dec */
+        tcg_out_modrm(s, 0xff | P_REXW, 1, r0);
+    } else if (val == (int8_t)val) {
         tcg_out_modrm(s, 0x83 | P_REXW, c, r0);
         tcg_out8(s, val);
     } else if (c == ARITH_AND && val == 0xffu) {
@@ -508,6 +534,7 @@ static void tcg_out_qemu_ld(TCGContext *
                             int opc)
 {
     int addr_reg, data_reg, r0, r1, mem_index, s_bits, bswap, rexw;
+    int32_t offset;
 #if defined(CONFIG_SOFTMMU)
     uint8_t *label1_ptr, *label2_ptr;
 #endif
@@ -558,9 +585,7 @@ static void tcg_out_qemu_ld(TCGContext *
 
     /* XXX: move that code at the end of the TB */
     tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_RSI, mem_index);
-    tcg_out8(s, 0xe8);
-    tcg_out32(s, (tcg_target_long)qemu_ld_helpers[s_bits] - 
-              (tcg_target_long)s->code_ptr - 4);
+    tcg_out_goto(s, 1, qemu_ld_helpers[s_bits]);
 
     switch(opc) {
     case 0 | 4:
@@ -604,8 +629,20 @@ static void tcg_out_qemu_ld(TCGContext *
     /* add x(r1), r0 */
     tcg_out_modrm_offset(s, 0x03 | P_REXW, r0, r1, offsetof(CPUTLBEntry, addend) - 
                          offsetof(CPUTLBEntry, addr_read));
+    offset = 0;
 #else
-    r0 = addr_reg;
+    if (GUEST_BASE == (int32_t)GUEST_BASE) {
+        r0 = addr_reg;
+        offset = GUEST_BASE;
+    } else {
+        offset = 0;
+        /* movq $GUEST_BASE, r0 */
+        tcg_out_opc(s, (0xb8 + (r0 & 7)) | P_REXW, 0, r0, 0);
+        tcg_out32(s, GUEST_BASE);
+        tcg_out32(s, GUEST_BASE >> 32);
+        /* addq addr_reg, r0 */
+        tcg_out_modrm(s, 0x01 | P_REXW, addr_reg, r0);
+    }
 #endif    
 
 #ifdef TARGET_WORDS_BIGENDIAN
@@ -616,15 +653,15 @@ static void tcg_out_qemu_ld(TCGContext *
     switch(opc) {
     case 0:
         /* movzbl */
-        tcg_out_modrm_offset(s, 0xb6 | P_EXT, data_reg, r0, 0);
+        tcg_out_modrm_offset(s, 0xb6 | P_EXT, data_reg, r0, offset);
         break;
     case 0 | 4:
         /* movsbX */
-        tcg_out_modrm_offset(s, 0xbe | P_EXT | rexw, data_reg, r0, 0);
+        tcg_out_modrm_offset(s, 0xbe | P_EXT | rexw, data_reg, r0, offset);
         break;
     case 1:
         /* movzwl */
-        tcg_out_modrm_offset(s, 0xb7 | P_EXT, data_reg, r0, 0);
+        tcg_out_modrm_offset(s, 0xb7 | P_EXT, data_reg, r0, offset);
         if (bswap) {
             /* rolw $8, data_reg */
             tcg_out8(s, 0x66); 
@@ -635,7 +672,7 @@ static void tcg_out_qemu_ld(TCGContext *
     case 1 | 4:
         if (bswap) {
             /* movzwl */
-            tcg_out_modrm_offset(s, 0xb7 | P_EXT, data_reg, r0, 0);
+            tcg_out_modrm_offset(s, 0xb7 | P_EXT, data_reg, r0, offset);
             /* rolw $8, data_reg */
             tcg_out8(s, 0x66); 
             tcg_out_modrm(s, 0xc1, 0, data_reg);
@@ -645,12 +682,12 @@ static void tcg_out_qemu_ld(TCGContext *
             tcg_out_modrm(s, 0xbf | P_EXT | rexw, data_reg, data_reg);
         } else {
             /* movswX */
-            tcg_out_modrm_offset(s, 0xbf | P_EXT | rexw, data_reg, r0, 0);
+            tcg_out_modrm_offset(s, 0xbf | P_EXT | rexw, data_reg, r0, offset);
         }
         break;
     case 2:
         /* movl (r0), data_reg */
-        tcg_out_modrm_offset(s, 0x8b, data_reg, r0, 0);
+        tcg_out_modrm_offset(s, 0x8b, data_reg, r0, offset);
         if (bswap) {
             /* bswap */
             tcg_out_opc(s, (0xc8 + (data_reg & 7)) | P_EXT, 0, data_reg, 0);
@@ -659,19 +696,19 @@ static void tcg_out_qemu_ld(TCGContext *
     case 2 | 4:
         if (bswap) {
             /* movl (r0), data_reg */
-            tcg_out_modrm_offset(s, 0x8b, data_reg, r0, 0);
+            tcg_out_modrm_offset(s, 0x8b, data_reg, r0, offset);
             /* bswap */
             tcg_out_opc(s, (0xc8 + (data_reg & 7)) | P_EXT, 0, data_reg, 0);
             /* movslq */
             tcg_out_modrm(s, 0x63 | P_REXW, data_reg, data_reg);
         } else {
             /* movslq */
-            tcg_out_modrm_offset(s, 0x63 | P_REXW, data_reg, r0, 0);
+            tcg_out_modrm_offset(s, 0x63 | P_REXW, data_reg, r0, offset);
         }
         break;
     case 3:
         /* movq (r0), data_reg */
-        tcg_out_modrm_offset(s, 0x8b | P_REXW, data_reg, r0, 0);
+        tcg_out_modrm_offset(s, 0x8b | P_REXW, data_reg, r0, offset);
         if (bswap) {
             /* bswap */
             tcg_out_opc(s, (0xc8 + (data_reg & 7)) | P_EXT | P_REXW, 0, data_reg, 0);
@@ -691,6 +728,7 @@ static void tcg_out_qemu_st(TCGContext *
                             int opc)
 {
     int addr_reg, data_reg, r0, r1, mem_index, s_bits, bswap, rexw;
+    int32_t offset;
 #if defined(CONFIG_SOFTMMU)
     uint8_t *label1_ptr, *label2_ptr;
 #endif
@@ -760,9 +798,7 @@ static void tcg_out_qemu_st(TCGContext *
         break;
     }
     tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_RDX, mem_index);
-    tcg_out8(s, 0xe8);
-    tcg_out32(s, (tcg_target_long)qemu_st_helpers[s_bits] - 
-              (tcg_target_long)s->code_ptr - 4);
+    tcg_out_goto(s, 1, qemu_st_helpers[s_bits]);
 
     /* jmp label2 */
     tcg_out8(s, 0xeb);
@@ -775,8 +811,20 @@ static void tcg_out_qemu_st(TCGContext *
     /* add x(r1), r0 */
     tcg_out_modrm_offset(s, 0x03 | P_REXW, r0, r1, offsetof(CPUTLBEntry, addend) - 
                          offsetof(CPUTLBEntry, addr_write));
+    offset = 0;
 #else
-    r0 = addr_reg;
+    if (GUEST_BASE == (int32_t)GUEST_BASE) {
+        r0 = addr_reg;
+        offset = GUEST_BASE;
+    } else {
+        offset = 0;
+        /* movq $GUEST_BASE, r0 */
+        tcg_out_opc(s, (0xb8 + (r0 & 7)) | P_REXW, 0, r0, 0);
+        tcg_out32(s, GUEST_BASE);
+        tcg_out32(s, GUEST_BASE >> 32);
+        /* addq addr_reg, r0 */
+        tcg_out_modrm(s, 0x01 | P_REXW, addr_reg, r0);
+    }
 #endif
 
 #ifdef TARGET_WORDS_BIGENDIAN
@@ -787,7 +835,7 @@ static void tcg_out_qemu_st(TCGContext *
     switch(opc) {
     case 0:
         /* movb */
-        tcg_out_modrm_offset(s, 0x88 | P_REXB, data_reg, r0, 0);
+        tcg_out_modrm_offset(s, 0x88 | P_REXB, data_reg, r0, offset);
         break;
     case 1:
         if (bswap) {
@@ -799,7 +847,7 @@ static void tcg_out_qemu_st(TCGContext *
         }
         /* movw */
         tcg_out8(s, 0x66);
-        tcg_out_modrm_offset(s, 0x89, data_reg, r0, 0);
+        tcg_out_modrm_offset(s, 0x89, data_reg, r0, offset);
         break;
     case 2:
         if (bswap) {
@@ -809,7 +857,7 @@ static void tcg_out_qemu_st(TCGContext *
             data_reg = r1;
         }
         /* movl */
-        tcg_out_modrm_offset(s, 0x89, data_reg, r0, 0);
+        tcg_out_modrm_offset(s, 0x89, data_reg, r0, offset);
         break;
     case 3:
         if (bswap) {
@@ -819,7 +867,7 @@ static void tcg_out_qemu_st(TCGContext *
             data_reg = r1;
         }
         /* movq */
-        tcg_out_modrm_offset(s, 0x89 | P_REXW, data_reg, r0, 0);
+        tcg_out_modrm_offset(s, 0x89 | P_REXW, data_reg, r0, offset);
         break;
     default:
         tcg_abort();
@@ -839,8 +887,7 @@ static inline void tcg_out_op(TCGContext
     switch(opc) {
     case INDEX_op_exit_tb:
         tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_RAX, args[0]);
-        tcg_out8(s, 0xe9); /* jmp tb_ret_addr */
-        tcg_out32(s, tb_ret_addr - s->code_ptr - 4);
+        tcg_out_goto(s, 0, tb_ret_addr);
         break;
     case INDEX_op_goto_tb:
         if (s->tb_jmp_offset) {
@@ -859,16 +906,14 @@ static inline void tcg_out_op(TCGContext
         break;
     case INDEX_op_call:
         if (const_args[0]) {
-            tcg_out8(s, 0xe8);
-            tcg_out32(s, args[0] - (tcg_target_long)s->code_ptr - 4);
+            tcg_out_goto(s, 1, (void *) args[0]);
         } else {
             tcg_out_modrm(s, 0xff, 2, args[0]);
         }
         break;
     case INDEX_op_jmp:
         if (const_args[0]) {
-            tcg_out8(s, 0xe9);
-            tcg_out32(s, args[0] - (tcg_target_long)s->code_ptr - 4);
+            tcg_out_goto(s, 0, (void *) args[0]);
         } else {
             tcg_out_modrm(s, 0xff, 4, args[0]);
         }
@@ -1136,6 +1181,21 @@ static inline void tcg_out_op(TCGContext
     case INDEX_op_ext32s_i64:
         tcg_out_modrm(s, 0x63 | P_REXW, args[0], args[1]);
         break;
+    case INDEX_op_ext8u_i32:
+        tcg_out_modrm(s, 0xb6 | P_EXT | P_REXB, args[0], args[1]);
+        break;
+    case INDEX_op_ext16u_i32:
+        tcg_out_modrm(s, 0xb7 | P_EXT, args[0], args[1]);
+        break;
+    case INDEX_op_ext8u_i64:
+        tcg_out_modrm(s, 0xb6 | P_EXT | P_REXW, args[0], args[1]);
+        break;
+    case INDEX_op_ext16u_i64:
+        tcg_out_modrm(s, 0xb7 | P_EXT | P_REXW, args[0], args[1]);
+        break;
+    case INDEX_op_ext32u_i64:
+        tcg_out_modrm(s, 0x8b, args[0], args[1]);
+        break;
 
     case INDEX_op_qemu_ld8u:
         tcg_out_qemu_ld(s, args, 0);
@@ -1310,6 +1370,11 @@ static const TCGTargetOpDef x86_64_op_de
     { INDEX_op_ext8s_i64, { "r", "r"} },
     { INDEX_op_ext16s_i64, { "r", "r"} },
     { INDEX_op_ext32s_i64, { "r", "r"} },
+    { INDEX_op_ext8u_i32, { "r", "r"} },
+    { INDEX_op_ext16u_i32, { "r", "r"} },
+    { INDEX_op_ext8u_i64, { "r", "r"} },
+    { INDEX_op_ext16u_i64, { "r", "r"} },
+    { INDEX_op_ext32u_i64, { "r", "r"} },
 
     { INDEX_op_qemu_ld8u, { "r", "L" } },
     { INDEX_op_qemu_ld8s, { "r", "L" } },
@@ -1322,7 +1387,7 @@ static const TCGTargetOpDef x86_64_op_de
     { INDEX_op_qemu_st8, { "L", "L" } },
     { INDEX_op_qemu_st16, { "L", "L" } },
     { INDEX_op_qemu_st32, { "L", "L" } },
-    { INDEX_op_qemu_st64, { "L", "L", "L" } },
+    { INDEX_op_qemu_st64, { "L", "L" } },
 
     { -1 },
 };