diff --git a/disas/sparc.c b/disas/sparc.c
index 8eb22e6..8e755d1 100644
--- a/disas/sparc.c
+++ b/disas/sparc.c
@@ -1175,15 +1175,11 @@
 { "subcc",      F3(2, 0x14, 0), F3(~2, ~0x14, ~0)|ASI(~0),      "1,2,d", 0, v6 },
 { "subcc",      F3(2, 0x14, 1), F3(~2, ~0x14, ~1),              "1,i,d", 0, v6 },
 
-{ "subx",       F3(2, 0x0c, 0), F3(~2, ~0x0c, ~0)|ASI(~0),      "1,2,d", 0, v6notv9 },
-{ "subx",       F3(2, 0x0c, 1), F3(~2, ~0x0c, ~1),              "1,i,d", 0, v6notv9 },
-{ "subc",       F3(2, 0x0c, 0), F3(~2, ~0x0c, ~0)|ASI(~0),      "1,2,d", 0, v9 },
-{ "subc",       F3(2, 0x0c, 1), F3(~2, ~0x0c, ~1),              "1,i,d", 0, v9 },
+{ "subc",       F3(2, 0x0c, 0), F3(~2, ~0x0c, ~0)|ASI(~0),      "1,2,d", 0, v6 },
+{ "subc",       F3(2, 0x0c, 1), F3(~2, ~0x0c, ~1),              "1,i,d", 0, v6 },
 
-{ "subxcc",     F3(2, 0x1c, 0), F3(~2, ~0x1c, ~0)|ASI(~0),      "1,2,d", 0, v6notv9 },
-{ "subxcc",     F3(2, 0x1c, 1), F3(~2, ~0x1c, ~1),              "1,i,d", 0, v6notv9 },
-{ "subccc",     F3(2, 0x1c, 0), F3(~2, ~0x1c, ~0)|ASI(~0),      "1,2,d", 0, v9 },
-{ "subccc",     F3(2, 0x1c, 1), F3(~2, ~0x1c, ~1),              "1,i,d", 0, v9 },
+{ "subccc",     F3(2, 0x1c, 0), F3(~2, ~0x1c, ~0)|ASI(~0),      "1,2,d", 0, v6 },
+{ "subccc",     F3(2, 0x1c, 1), F3(~2, ~0x1c, ~1),              "1,i,d", 0, v6 },
 
 { "and",        F3(2, 0x01, 0), F3(~2, ~0x01, ~0)|ASI(~0),      "1,2,d", 0, v6 },
 { "and",        F3(2, 0x01, 1), F3(~2, ~0x01, ~1),              "1,i,d", 0, v6 },
@@ -1215,19 +1211,13 @@
 { "addcc",      F3(2, 0x10, 1), F3(~2, ~0x10, ~1),              "1,i,d", 0, v6 },
 { "addcc",      F3(2, 0x10, 1), F3(~2, ~0x10, ~1),              "i,1,d", 0, v6 },
 
-{ "addx",       F3(2, 0x08, 0), F3(~2, ~0x08, ~0)|ASI(~0),      "1,2,d", 0, v6notv9 },
-{ "addx",       F3(2, 0x08, 1), F3(~2, ~0x08, ~1),              "1,i,d", 0, v6notv9 },
-{ "addx",       F3(2, 0x08, 1), F3(~2, ~0x08, ~1),              "i,1,d", 0, v6notv9 },
-{ "addc",       F3(2, 0x08, 0), F3(~2, ~0x08, ~0)|ASI(~0),      "1,2,d", 0, v9 },
-{ "addc",       F3(2, 0x08, 1), F3(~2, ~0x08, ~1),              "1,i,d", 0, v9 },
-{ "addc",       F3(2, 0x08, 1), F3(~2, ~0x08, ~1),              "i,1,d", 0, v9 },
+{ "addc",       F3(2, 0x08, 0), F3(~2, ~0x08, ~0)|ASI(~0),      "1,2,d", 0, v6 },
+{ "addc",       F3(2, 0x08, 1), F3(~2, ~0x08, ~1),              "1,i,d", 0, v6 },
+{ "addc",       F3(2, 0x08, 1), F3(~2, ~0x08, ~1),              "i,1,d", 0, v6 },
 
-{ "addxcc",     F3(2, 0x18, 0), F3(~2, ~0x18, ~0)|ASI(~0),      "1,2,d", 0, v6notv9 },
-{ "addxcc",     F3(2, 0x18, 1), F3(~2, ~0x18, ~1),              "1,i,d", 0, v6notv9 },
-{ "addxcc",     F3(2, 0x18, 1), F3(~2, ~0x18, ~1),              "i,1,d", 0, v6notv9 },
-{ "addccc",     F3(2, 0x18, 0), F3(~2, ~0x18, ~0)|ASI(~0),      "1,2,d", 0, v9 },
-{ "addccc",     F3(2, 0x18, 1), F3(~2, ~0x18, ~1),              "1,i,d", 0, v9 },
-{ "addccc",     F3(2, 0x18, 1), F3(~2, ~0x18, ~1),              "i,1,d", 0, v9 },
+{ "addccc",     F3(2, 0x18, 0), F3(~2, ~0x18, ~0)|ASI(~0),      "1,2,d", 0, v6 },
+{ "addccc",     F3(2, 0x18, 1), F3(~2, ~0x18, ~1),              "1,i,d", 0, v6 },
+{ "addccc",     F3(2, 0x18, 1), F3(~2, ~0x18, ~1),              "i,1,d", 0, v6 },
 
 { "smul",       F3(2, 0x0b, 0), F3(~2, ~0x0b, ~0)|ASI(~0),      "1,2,d", 0, v8 },
 { "smul",       F3(2, 0x0b, 1), F3(~2, ~0x0b, ~1),              "1,i,d", 0, v8 },
@@ -2042,6 +2032,10 @@
 
 #undef IMPDEP
 
+{ "addxc", F3F(2, 0x36, 0x011), F3F(~2, ~0x36, ~0x011), "1,2,d", 0, v9b },
+{ "addxccc", F3F(2, 0x36, 0x013), F3F(~2, ~0x36, ~0x013), "1,2,d", 0, v9b },
+{ "umulxhi", F3F(2, 0x36, 0x016), F3F(~2, ~0x36, ~0x016), "1,2,d", 0, v9b },
+
 };
 
 static const int sparc_num_opcodes = ((sizeof sparc_opcodes)/(sizeof sparc_opcodes[0]));
diff --git a/include/elf.h b/include/elf.h
index 70107f0..a516584 100644
--- a/include/elf.h
+++ b/include/elf.h
@@ -473,14 +473,35 @@
 #define PPC_FEATURE_TRUE_LE             0x00000002
 #define PPC_FEATURE_PPC_LE              0x00000001
 
-/* Bits present in AT_HWCAP, primarily for Sparc32.  */
+/* Bits present in AT_HWCAP for Sparc.  */
 
-#define HWCAP_SPARC_FLUSH       1    /* CPU supports flush instruction. */
-#define HWCAP_SPARC_STBAR       2
-#define HWCAP_SPARC_SWAP        4
-#define HWCAP_SPARC_MULDIV      8
-#define HWCAP_SPARC_V9		16
-#define HWCAP_SPARC_ULTRA3	32
+#define HWCAP_SPARC_FLUSH               0x00000001
+#define HWCAP_SPARC_STBAR               0x00000002
+#define HWCAP_SPARC_SWAP                0x00000004
+#define HWCAP_SPARC_MULDIV              0x00000008
+#define HWCAP_SPARC_V9                  0x00000010
+#define HWCAP_SPARC_ULTRA3              0x00000020
+#define HWCAP_SPARC_BLKINIT             0x00000040
+#define HWCAP_SPARC_N2                  0x00000080
+#define HWCAP_SPARC_MUL32               0x00000100
+#define HWCAP_SPARC_DIV32               0x00000200
+#define HWCAP_SPARC_FSMULD              0x00000400
+#define HWCAP_SPARC_V8PLUS              0x00000800
+#define HWCAP_SPARC_POPC                0x00001000
+#define HWCAP_SPARC_VIS                 0x00002000
+#define HWCAP_SPARC_VIS2                0x00004000
+#define HWCAP_SPARC_ASI_BLK_INIT        0x00008000
+#define HWCAP_SPARC_FMAF                0x00010000
+#define HWCAP_SPARC_VIS3                0x00020000
+#define HWCAP_SPARC_HPC                 0x00040000
+#define HWCAP_SPARC_RANDOM              0x00080000
+#define HWCAP_SPARC_TRANS               0x00100000
+#define HWCAP_SPARC_FJFMAU              0x00200000
+#define HWCAP_SPARC_IMA                 0x00400000
+#define HWCAP_SPARC_ASI_CACHE_SPARING   0x00800000
+#define HWCAP_SPARC_PAUSE               0x01000000
+#define HWCAP_SPARC_CBCOND              0x02000000
+#define HWCAP_SPARC_CRYPTO              0x04000000
 
 /* Bits present in AT_HWCAP for s390.  */
 
diff --git a/include/qemu/compiler.h b/include/qemu/compiler.h
index 155b358..ac7c4c4 100644
--- a/include/qemu/compiler.h
+++ b/include/qemu/compiler.h
@@ -24,6 +24,12 @@
 #define QEMU_WARN_UNUSED_RESULT
 #endif
 
+#if QEMU_GNUC_PREREQ(4, 3)
+#define QEMU_ARTIFICIAL __attribute__((always_inline, artificial))
+#else
+#define QEMU_ARTIFICIAL
+#endif
+
 #if defined(_WIN32)
 # define QEMU_PACKED __attribute__((gcc_struct, packed))
 #else
diff --git a/tcg/aarch64/tcg-target.c b/tcg/aarch64/tcg-target.c
index 56dae66..987c0bd 100644
--- a/tcg/aarch64/tcg-target.c
+++ b/tcg/aarch64/tcg-target.c
@@ -1007,7 +1007,7 @@
     tcg_out_adr(s, TCG_REG_X3, lb->raddr);
     tcg_out_call(s, qemu_ld_helpers[opc & ~MO_SIGN]);
     if (opc & MO_SIGN) {
-        tcg_out_sxt(s, TCG_TYPE_I64, size, lb->datalo_reg, TCG_REG_X0);
+        tcg_out_sxt(s, lb->type, size, lb->datalo_reg, TCG_REG_X0);
     } else {
         tcg_out_mov(s, size == MO_64, lb->datalo_reg, TCG_REG_X0);
     }
@@ -1032,7 +1032,7 @@
 }
 
 static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOp opc,
-                                TCGReg data_reg, TCGReg addr_reg,
+                                TCGType ext, TCGReg data_reg, TCGReg addr_reg,
                                 int mem_index, tcg_insn_unit *raddr,
                                 tcg_insn_unit *label_ptr)
 {
@@ -1040,6 +1040,7 @@
 
     label->is_ld = is_ld;
     label->opc = opc;
+    label->type = ext;
     label->datalo_reg = data_reg;
     label->addrlo_reg = addr_reg;
     label->mem_index = mem_index;
@@ -1108,7 +1109,7 @@
 
 #endif /* CONFIG_SOFTMMU */
 
-static void tcg_out_qemu_ld_direct(TCGContext *s, TCGMemOp memop,
+static void tcg_out_qemu_ld_direct(TCGContext *s, TCGMemOp memop, TCGType ext,
                                    TCGReg data_r, TCGReg addr_r, TCGReg off_r)
 {
     const TCGMemOp bswap = memop & MO_BSWAP;
@@ -1118,7 +1119,8 @@
         tcg_out_ldst_r(s, I3312_LDRB, data_r, addr_r, off_r);
         break;
     case MO_SB:
-        tcg_out_ldst_r(s, I3312_LDRSBX, data_r, addr_r, off_r);
+        tcg_out_ldst_r(s, ext ? I3312_LDRSBX : I3312_LDRSBW,
+                       data_r, addr_r, off_r);
         break;
     case MO_UW:
         tcg_out_ldst_r(s, I3312_LDRH, data_r, addr_r, off_r);
@@ -1130,9 +1132,10 @@
         if (bswap) {
             tcg_out_ldst_r(s, I3312_LDRH, data_r, addr_r, off_r);
             tcg_out_rev16(s, data_r, data_r);
-            tcg_out_sxt(s, TCG_TYPE_I64, MO_16, data_r, data_r);
+            tcg_out_sxt(s, ext, MO_16, data_r, data_r);
         } else {
-            tcg_out_ldst_r(s, I3312_LDRSHX, data_r, addr_r, off_r);
+            tcg_out_ldst_r(s, ext ? I3312_LDRSHX : I3312_LDRSHW,
+                           data_r, addr_r, off_r);
         }
         break;
     case MO_UL:
@@ -1197,18 +1200,18 @@
 }
 
 static void tcg_out_qemu_ld(TCGContext *s, TCGReg data_reg, TCGReg addr_reg,
-                            TCGMemOp memop, int mem_index)
+                            TCGMemOp memop, TCGType ext, int mem_index)
 {
 #ifdef CONFIG_SOFTMMU
     TCGMemOp s_bits = memop & MO_SIZE;
     tcg_insn_unit *label_ptr;
 
     tcg_out_tlb_read(s, addr_reg, s_bits, &label_ptr, mem_index, 1);
-    tcg_out_qemu_ld_direct(s, memop, data_reg, addr_reg, TCG_REG_X1);
-    add_qemu_ldst_label(s, true, memop, data_reg, addr_reg,
+    tcg_out_qemu_ld_direct(s, memop, ext, data_reg, addr_reg, TCG_REG_X1);
+    add_qemu_ldst_label(s, true, memop, ext, data_reg, addr_reg,
                         mem_index, s->code_ptr, label_ptr);
 #else /* !CONFIG_SOFTMMU */
-    tcg_out_qemu_ld_direct(s, memop, data_reg, addr_reg,
+    tcg_out_qemu_ld_direct(s, memop, ext, data_reg, addr_reg,
                            GUEST_BASE ? TCG_REG_GUEST_BASE : TCG_REG_XZR);
 #endif /* CONFIG_SOFTMMU */
 }
@@ -1222,7 +1225,7 @@
 
     tcg_out_tlb_read(s, addr_reg, s_bits, &label_ptr, mem_index, 0);
     tcg_out_qemu_st_direct(s, memop, data_reg, addr_reg, TCG_REG_X1);
-    add_qemu_ldst_label(s, false, memop, data_reg, addr_reg,
+    add_qemu_ldst_label(s, false, memop, s_bits == MO_64, data_reg, addr_reg,
                         mem_index, s->code_ptr, label_ptr);
 #else /* !CONFIG_SOFTMMU */
     tcg_out_qemu_st_direct(s, memop, data_reg, addr_reg,
@@ -1515,7 +1518,7 @@
 
     case INDEX_op_qemu_ld_i32:
     case INDEX_op_qemu_ld_i64:
-        tcg_out_qemu_ld(s, a0, a1, a2, args[3]);
+        tcg_out_qemu_ld(s, a0, a1, a2, ext, args[3]);
         break;
     case INDEX_op_qemu_st_i32:
     case INDEX_op_qemu_st_i64:
diff --git a/tcg/sparc/tcg-target.c b/tcg/sparc/tcg-target.c
index 40f2ec1..0c4b028 100644
--- a/tcg/sparc/tcg-target.c
+++ b/tcg/sparc/tcg-target.c
@@ -197,8 +197,8 @@
 #define ARITH_XOR  (INSN_OP(2) | INSN_OP3(0x03))
 #define ARITH_SUB  (INSN_OP(2) | INSN_OP3(0x04))
 #define ARITH_SUBCC (INSN_OP(2) | INSN_OP3(0x14))
-#define ARITH_ADDX (INSN_OP(2) | INSN_OP3(0x08))
-#define ARITH_SUBX (INSN_OP(2) | INSN_OP3(0x0c))
+#define ARITH_ADDC (INSN_OP(2) | INSN_OP3(0x08))
+#define ARITH_SUBC (INSN_OP(2) | INSN_OP3(0x0c))
 #define ARITH_UMUL (INSN_OP(2) | INSN_OP3(0x0a))
 #define ARITH_SMUL (INSN_OP(2) | INSN_OP3(0x0b))
 #define ARITH_UDIV (INSN_OP(2) | INSN_OP3(0x0e))
@@ -209,6 +209,9 @@
 #define ARITH_MOVCC (INSN_OP(2) | INSN_OP3(0x2c))
 #define ARITH_MOVR (INSN_OP(2) | INSN_OP3(0x2f))
 
+#define ARITH_ADDXC (INSN_OP(2) | INSN_OP3(0x36) | INSN_OPF(0x11))
+#define ARITH_UMULXHI (INSN_OP(2) | INSN_OP3(0x36) | INSN_OPF(0x16))
+
 #define SHIFT_SLL  (INSN_OP(2) | INSN_OP3(0x25))
 #define SHIFT_SRL  (INSN_OP(2) | INSN_OP3(0x26))
 #define SHIFT_SRA  (INSN_OP(2) | INSN_OP3(0x27))
@@ -262,6 +265,10 @@
 #define STW_LE     (STWA  | INSN_ASI(ASI_PRIMARY_LITTLE))
 #define STX_LE     (STXA  | INSN_ASI(ASI_PRIMARY_LITTLE))
 
+#ifndef use_vis3_instructions
+bool use_vis3_instructions;
+#endif
+
 static inline int check_fit_i64(int64_t val, unsigned int bits)
 {
     return val == sextract64(val, 0, bits);
@@ -657,7 +664,7 @@
 static void tcg_out_setcond_i32(TCGContext *s, TCGCond cond, TCGReg ret,
                                 TCGReg c1, int32_t c2, int c2const)
 {
-    /* For 32-bit comparisons, we can play games with ADDX/SUBX.  */
+    /* For 32-bit comparisons, we can play games with ADDC/SUBC.  */
     switch (cond) {
     case TCG_COND_LTU:
     case TCG_COND_GEU:
@@ -668,9 +675,12 @@
     case TCG_COND_NE:
         /* For equality, we can transform to inequality vs zero.  */
         if (c2 != 0) {
-            tcg_out_arithc(s, ret, c1, c2, c2const, ARITH_XOR);
+            tcg_out_arithc(s, TCG_REG_T1, c1, c2, c2const, ARITH_XOR);
+            c2 = TCG_REG_T1;
+        } else {
+            c2 = c1;
         }
-        c1 = TCG_REG_G0, c2 = ret, c2const = 0;
+        c1 = TCG_REG_G0, c2const = 0;
         cond = (cond == TCG_COND_EQ ? TCG_COND_GEU : TCG_COND_LTU);
 	break;
 
@@ -698,15 +708,32 @@
 
     tcg_out_cmp(s, c1, c2, c2const);
     if (cond == TCG_COND_LTU) {
-        tcg_out_arithi(s, ret, TCG_REG_G0, 0, ARITH_ADDX);
+        tcg_out_arithi(s, ret, TCG_REG_G0, 0, ARITH_ADDC);
     } else {
-        tcg_out_arithi(s, ret, TCG_REG_G0, -1, ARITH_SUBX);
+        tcg_out_arithi(s, ret, TCG_REG_G0, -1, ARITH_SUBC);
     }
 }
 
 static void tcg_out_setcond_i64(TCGContext *s, TCGCond cond, TCGReg ret,
                                 TCGReg c1, int32_t c2, int c2const)
 {
+    if (use_vis3_instructions) {
+        switch (cond) {
+        case TCG_COND_NE:
+            if (c2 != 0) {
+                break;
+            }
+            c2 = c1, c2const = 0, c1 = TCG_REG_G0;
+            /* FALLTHRU */
+        case TCG_COND_LTU:
+            tcg_out_cmp(s, c1, c2, c2const);
+            tcg_out_arith(s, ret, TCG_REG_G0, TCG_REG_G0, ARITH_ADDXC);
+            return;
+        default:
+            break;
+        }
+    }
+
     /* For 64-bit signed comparisons vs zero, we can avoid the compare
        if the input does not overlap the output.  */
     if (c2 == 0 && !is_unsigned_cond(cond) && c1 != ret) {
@@ -719,9 +746,9 @@
     }
 }
 
-static void tcg_out_addsub2(TCGContext *s, TCGReg rl, TCGReg rh,
-                            TCGReg al, TCGReg ah, int32_t bl, int blconst,
-                            int32_t bh, int bhconst, int opl, int oph)
+static void tcg_out_addsub2_i32(TCGContext *s, TCGReg rl, TCGReg rh,
+                                TCGReg al, TCGReg ah, int32_t bl, int blconst,
+                                int32_t bh, int bhconst, int opl, int oph)
 {
     TCGReg tmp = TCG_REG_T1;
 
@@ -735,6 +762,54 @@
     tcg_out_mov(s, TCG_TYPE_I32, rl, tmp);
 }
 
+static void tcg_out_addsub2_i64(TCGContext *s, TCGReg rl, TCGReg rh,
+                                TCGReg al, TCGReg ah, int32_t bl, int blconst,
+                                int32_t bh, int bhconst, bool is_sub)
+{
+    TCGReg tmp = TCG_REG_T1;
+
+    /* Note that the low parts are fully consumed before tmp is set.  */
+    if (rl != ah && (bhconst || rl != bh)) {
+        tmp = rl;
+    }
+
+    tcg_out_arithc(s, tmp, al, bl, blconst, is_sub ? ARITH_SUBCC : ARITH_ADDCC);
+
+    if (use_vis3_instructions && !is_sub) {
+        /* Note that ADDXC doesn't accept immediates.  */
+        if (bhconst && bh != 0) {
+           tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_T2, bh);
+           bh = TCG_REG_T2;
+        }
+        tcg_out_arith(s, rh, ah, bh, ARITH_ADDXC);
+    } else if (bh == TCG_REG_G0) {
+	/* If we have a zero, we can perform the operation in two insns,
+           with the arithmetic first, and a conditional move into place.  */
+	if (rh == ah) {
+            tcg_out_arithi(s, TCG_REG_T2, ah, 1,
+			   is_sub ? ARITH_SUB : ARITH_ADD);
+            tcg_out_movcc(s, TCG_COND_LTU, MOVCC_XCC, rh, TCG_REG_T2, 0);
+	} else {
+            tcg_out_arithi(s, rh, ah, 1, is_sub ? ARITH_SUB : ARITH_ADD);
+	    tcg_out_movcc(s, TCG_COND_GEU, MOVCC_XCC, rh, ah, 0);
+	}
+    } else {
+        /* Otherwise adjust BH as if there is carry into T2 ... */
+        if (bhconst) {
+            tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_T2, bh + (is_sub ? -1 : 1));
+        } else {
+            tcg_out_arithi(s, TCG_REG_T2, bh, 1,
+                           is_sub ? ARITH_SUB : ARITH_ADD);
+        }
+        /* ... smoosh T2 back to original BH if carry is clear ... */
+        tcg_out_movcc(s, TCG_COND_GEU, MOVCC_XCC, TCG_REG_T2, bh, bhconst);
+	/* ... and finally perform the arithmetic with the new operand.  */
+        tcg_out_arith(s, rh, ah, TCG_REG_T2, is_sub ? ARITH_SUB : ARITH_ADD);
+    }
+
+    tcg_out_mov(s, TCG_TYPE_I64, rl, tmp);
+}
+
 static void tcg_out_call_nodelay(TCGContext *s, tcg_insn_unit *dest)
 {
     ptrdiff_t disp = tcg_pcrel_diff(s, dest);
@@ -1264,12 +1339,14 @@
         break;
 
     case INDEX_op_add2_i32:
-        tcg_out_addsub2(s, a0, a1, a2, args[3], args[4], const_args[4],
-                        args[5], const_args[5], ARITH_ADDCC, ARITH_ADDX);
+        tcg_out_addsub2_i32(s, args[0], args[1], args[2], args[3],
+                            args[4], const_args[4], args[5], const_args[5],
+                            ARITH_ADDCC, ARITH_ADDC);
         break;
     case INDEX_op_sub2_i32:
-        tcg_out_addsub2(s, a0, a1, a2, args[3], args[4], const_args[4],
-                        args[5], const_args[5], ARITH_SUBCC, ARITH_SUBX);
+        tcg_out_addsub2_i32(s, args[0], args[1], args[2], args[3],
+                            args[4], const_args[4], args[5], const_args[5],
+                            ARITH_SUBCC, ARITH_SUBC);
         break;
     case INDEX_op_mulu2_i32:
         c = ARITH_UMUL;
@@ -1351,6 +1428,17 @@
     case INDEX_op_movcond_i64:
         tcg_out_movcond_i64(s, args[5], a0, a1, a2, c2, args[3], const_args[3]);
         break;
+    case INDEX_op_add2_i64:
+        tcg_out_addsub2_i64(s, args[0], args[1], args[2], args[3], args[4],
+                            const_args[4], args[5], const_args[5], false);
+        break;
+    case INDEX_op_sub2_i64:
+        tcg_out_addsub2_i64(s, args[0], args[1], args[2], args[3], args[4],
+                            const_args[4], args[5], const_args[5], true);
+        break;
+    case INDEX_op_muluh_i64:
+        tcg_out_arith(s, args[0], args[1], args[2], ARITH_UMULXHI);
+        break;
 
     gen_arith:
         tcg_out_arithc(s, a0, a1, a2, c2, c);
@@ -1449,6 +1537,10 @@
     { INDEX_op_setcond_i64, { "R", "RZ", "RJ" } },
     { INDEX_op_movcond_i64, { "R", "RZ", "RJ", "RI", "0" } },
 
+    { INDEX_op_add2_i64, { "R", "R", "RZ", "RZ", "RJ", "RI" } },
+    { INDEX_op_sub2_i64, { "R", "R", "RZ", "RZ", "RJ", "RI" } },
+    { INDEX_op_muluh_i64, { "R", "RZ", "RZ" } },
+
     { INDEX_op_qemu_ld_i32, { "r", "A" } },
     { INDEX_op_qemu_ld_i64, { "R", "A" } },
     { INDEX_op_qemu_st_i32, { "sZ", "A" } },
@@ -1459,6 +1551,15 @@
 
 static void tcg_target_init(TCGContext *s)
 {
+    /* Only probe for the platform and capabilities if we havn't already
+       determined maximum values at compile time.  */
+#ifndef use_vis3_instructions
+    {
+        unsigned long hwcap = qemu_getauxval(AT_HWCAP);
+        use_vis3_instructions = (hwcap & HWCAP_SPARC_VIS3) != 0;
+    }
+#endif
+
     tcg_regset_set32(tcg_target_available_regs[TCG_TYPE_I32], 0, 0xffffffff);
     tcg_regset_set32(tcg_target_available_regs[TCG_TYPE_I64], 0, ALL_64);
 
diff --git a/tcg/sparc/tcg-target.h b/tcg/sparc/tcg-target.h
index 089f976..0c4c8af 100644
--- a/tcg/sparc/tcg-target.h
+++ b/tcg/sparc/tcg-target.h
@@ -85,6 +85,12 @@
 #define TCG_TARGET_EXTEND_ARGS 1
 #endif
 
+#if defined(__VIS__) && __VIS__ >= 0x300
+#define use_vis3_instructions  1
+#else
+extern bool use_vis3_instructions;
+#endif
+
 /* optional instructions */
 #define TCG_TARGET_HAS_div_i32		1
 #define TCG_TARGET_HAS_rem_i32		0
@@ -133,11 +139,11 @@
 #define TCG_TARGET_HAS_nor_i64          0
 #define TCG_TARGET_HAS_deposit_i64      0
 #define TCG_TARGET_HAS_movcond_i64      1
-#define TCG_TARGET_HAS_add2_i64         0
-#define TCG_TARGET_HAS_sub2_i64         0
+#define TCG_TARGET_HAS_add2_i64         1
+#define TCG_TARGET_HAS_sub2_i64         1
 #define TCG_TARGET_HAS_mulu2_i64        0
 #define TCG_TARGET_HAS_muls2_i64        0
-#define TCG_TARGET_HAS_muluh_i64        0
+#define TCG_TARGET_HAS_muluh_i64        use_vis3_instructions
 #define TCG_TARGET_HAS_mulsh_i64        0
 
 #define TCG_AREG0 TCG_REG_I0
diff --git a/tcg/tcg-be-ldst.h b/tcg/tcg-be-ldst.h
index 49b3de6..429cba2 100644
--- a/tcg/tcg-be-ldst.h
+++ b/tcg/tcg-be-ldst.h
@@ -24,8 +24,9 @@
 #define TCG_MAX_QEMU_LDST       640
 
 typedef struct TCGLabelQemuLdst {
-    bool is_ld:1;           /* qemu_ld: true, qemu_st: false */
-    TCGMemOp opc:4;
+    bool is_ld;             /* qemu_ld: true, qemu_st: false */
+    TCGMemOp opc;
+    TCGType type;           /* result type of a load */
     TCGReg addrlo_reg;      /* reg index for low word of guest virtual addr */
     TCGReg addrhi_reg;      /* reg index for high word of guest virtual addr */
     TCGReg datalo_reg;      /* reg index for low word to be loaded or stored */
diff --git a/tcg/tcg.h b/tcg/tcg.h
index 997a704..7285f71 100644
--- a/tcg/tcg.h
+++ b/tcg/tcg.h
@@ -274,75 +274,54 @@
 
 typedef tcg_target_ulong TCGArg;
 
-/* Define a type and accessor macros for variables.  Using a struct is
-   nice because it gives some level of type safely.  Ideally the compiler
-   be able to see through all this.  However in practice this is not true,
-   especially on targets with braindamaged ABIs (e.g. i386).
-   We use plain int by default to avoid this runtime overhead.
-   Users of tcg_gen_* don't need to know about any of this, and should
-   treat TCGv as an opaque type.
+/* Define a type and accessor macros for variables.  Using pointer types
+   is nice because it gives some level of type safely.  Converting to and
+   from intptr_t rather than int reduces the number of sign-extension
+   instructions that get implied on 64-bit hosts.  Users of tcg_gen_* don't
+   need to know about any of this, and should treat TCGv as an opaque type.
    In addition we do typechecking for different types of variables.  TCGv_i32
    and TCGv_i64 are 32/64-bit variables respectively.  TCGv and TCGv_ptr
-   are aliases for target_ulong and host pointer sized values respectively.
- */
+   are aliases for target_ulong and host pointer sized values respectively.  */
 
-#ifdef CONFIG_DEBUG_TCG
-#define DEBUG_TCGV 1
-#endif
+typedef struct TCGv_i32_d *TCGv_i32;
+typedef struct TCGv_i64_d *TCGv_i64;
+typedef struct TCGv_ptr_d *TCGv_ptr;
 
-#ifdef DEBUG_TCGV
-
-typedef struct
+static inline TCGv_i32 QEMU_ARTIFICIAL MAKE_TCGV_I32(intptr_t i)
 {
-    int i32;
-} TCGv_i32;
+    return (TCGv_i32)i;
+}
 
-typedef struct
+static inline TCGv_i64 QEMU_ARTIFICIAL MAKE_TCGV_I64(intptr_t i)
 {
-    int i64;
-} TCGv_i64;
+    return (TCGv_i64)i;
+}
 
-typedef struct {
-    int iptr;
-} TCGv_ptr;
+static inline TCGv_ptr QEMU_ARTIFICIAL MAKE_TCGV_PTR(intptr_t i)
+{
+    return (TCGv_ptr)i;
+}
 
-#define MAKE_TCGV_I32(i) __extension__                  \
-    ({ TCGv_i32 make_tcgv_tmp = {i}; make_tcgv_tmp;})
-#define MAKE_TCGV_I64(i) __extension__                  \
-    ({ TCGv_i64 make_tcgv_tmp = {i}; make_tcgv_tmp;})
-#define MAKE_TCGV_PTR(i) __extension__                  \
-    ({ TCGv_ptr make_tcgv_tmp = {i}; make_tcgv_tmp; })
-#define GET_TCGV_I32(t) ((t).i32)
-#define GET_TCGV_I64(t) ((t).i64)
-#define GET_TCGV_PTR(t) ((t).iptr)
+static inline intptr_t QEMU_ARTIFICIAL GET_TCGV_I32(TCGv_i32 t)
+{
+    return (intptr_t)t;
+}
+
+static inline intptr_t QEMU_ARTIFICIAL GET_TCGV_I64(TCGv_i64 t)
+{
+    return (intptr_t)t;
+}
+
+static inline intptr_t QEMU_ARTIFICIAL GET_TCGV_PTR(TCGv_ptr t)
+{
+    return (intptr_t)t;
+}
+
 #if TCG_TARGET_REG_BITS == 32
 #define TCGV_LOW(t) MAKE_TCGV_I32(GET_TCGV_I64(t))
 #define TCGV_HIGH(t) MAKE_TCGV_I32(GET_TCGV_I64(t) + 1)
 #endif
 
-#else /* !DEBUG_TCGV */
-
-typedef int TCGv_i32;
-typedef int TCGv_i64;
-#if TCG_TARGET_REG_BITS == 32
-#define TCGv_ptr TCGv_i32
-#else
-#define TCGv_ptr TCGv_i64
-#endif
-#define MAKE_TCGV_I32(x) (x)
-#define MAKE_TCGV_I64(x) (x)
-#define MAKE_TCGV_PTR(x) (x)
-#define GET_TCGV_I32(t) (t)
-#define GET_TCGV_I64(t) (t)
-#define GET_TCGV_PTR(t) (t)
-
-#if TCG_TARGET_REG_BITS == 32
-#define TCGV_LOW(t) (t)
-#define TCGV_HIGH(t) ((t) + 1)
-#endif
-
-#endif /* DEBUG_TCGV */
-
 #define TCGV_EQUAL_I32(a, b) (GET_TCGV_I32(a) == GET_TCGV_I32(b))
 #define TCGV_EQUAL_I64(a, b) (GET_TCGV_I64(a) == GET_TCGV_I64(b))
 #define TCGV_EQUAL_PTR(a, b) (GET_TCGV_PTR(a) == GET_TCGV_PTR(b))