TCG code generator

git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@3943 c046a42c-6fe2-441c-8c8c-71466251a162

13 files changed, 7026 insertions, 0 deletions

diff --git a/tcg/LICENSE b/tcg/LICENSE
new file mode 100644
index 0000000000..be817fa162
--- /dev/null
+++ b/tcg/LICENSE
@@ -0,0 +1,3 @@
+All the files in this directory and subdirectories are released under
+a BSD like license (see header in each file). No other license is
+accepted.
diff --git a/tcg/README b/tcg/README
new file mode 100644
index 0000000000..f5b1280913
--- /dev/null
+++ b/tcg/README
@@ -0,0 +1,420 @@
+Tiny Code Generator - Fabrice Bellard.
+
+1) Introduction
+
+TCG (Tiny Code Generator) began as a generic backend for a C
+compiler. It was simplified to be used in QEMU. It also has its roots
+in the QOP code generator written by Paul Brook. 
+
+2) Definitions
+
+The TCG "target" is the architecture for which we generate the
+code. It is of course not the same as the "target" of QEMU which is
+the emulated architecture. As TCG started as a generic C backend used
+for cross compiling, it is assumed that the TCG target is different
+from the host, although it is never the case for QEMU.
+
+A TCG "function" corresponds to a QEMU Translated Block (TB).
+
+A TCG "temporary" is a variable only live in a given
+function. Temporaries are allocated explicitely in each function.
+
+A TCG "global" is a variable which is live in all the functions. They
+are defined before the functions defined. A TCG global can be a memory
+location (e.g. a QEMU CPU register), a fixed host register (e.g. the
+QEMU CPU state pointer) or a memory location which is stored in a
+register outside QEMU TBs (not implemented yet).
+
+A TCG "basic block" corresponds to a list of instructions terminated
+by a branch instruction. 
+
+3) Intermediate representation
+
+3.1) Introduction
+
+TCG instructions operate on variables which are temporaries or
+globals. TCG instructions and variables are strongly typed. Two types
+are supported: 32 bit integers and 64 bit integers. Pointers are
+defined as an alias to 32 bit or 64 bit integers depending on the TCG
+target word size.
+
+Each instruction has a fixed number of output variable operands, input
+variable operands and always constant operands.
+
+The notable exception is the call instruction which has a variable
+number of outputs and inputs.
+
+In the textual form, output operands come first, followed by input
+operands, followed by constant operands. The output type is included
+in the instruction name. Constants are prefixed with a '$'.
+
+add_i32 t0, t1, t2  (t0 <- t1 + t2)
+
+sub_i64 t2, t3, $4  (t2 <- t3 - 4)
+
+3.2) Assumptions
+
+* Basic blocks
+
+- Basic blocks end after branches (e.g. brcond_i32 instruction),
+  goto_tb and exit_tb instructions.
+- Basic blocks end before legacy dyngen operations.
+- Basic blocks start after the end of a previous basic block, at a
+  set_label instruction or after a legacy dyngen operation.
+
+After the end of a basic block, temporaries at destroyed and globals
+are stored at their initial storage (register or memory place
+depending on their declarations).
+
+* Floating point types are not supported yet
+
+* Pointers: depending on the TCG target, pointer size is 32 bit or 64
+  bit. The type TCG_TYPE_PTR is an alias to TCG_TYPE_I32 or
+  TCG_TYPE_I64.
+
+* Helpers:
+
+Using the tcg_gen_helper_x_y it is possible to call any function
+taking i32, i64 or pointer types types. Before calling an helper, all
+globals are stored at their canonical location and it is assumed that
+the function can modify them. In the future, function modifiers will
+be allowed to tell that the helper does not read or write some globals.
+
+On some TCG targets (e.g. x86), several calling conventions are
+supported.
+
+* Branches:
+
+Use the instruction 'br' to jump to a label. Use 'jmp' to jump to an
+explicit address. Conditional branches can only jump to labels.
+
+3.3) Code Optimizations
+
+When generating instructions, you can count on at least the following
+optimizations:
+
+- Single instructions are simplified, e.g.
+
+   and_i32 t0, t0, $0xffffffff
+    
+  is suppressed.
+
+- A liveness analysis is done at the basic block level. The
+  information is used to suppress moves from a dead temporary to
+  another one. It is also used to remove instructions which compute
+  dead results. The later is especially useful for condition code
+  optimisation in QEMU.
+
+  In the following example:
+
+  add_i32 t0, t1, t2
+  add_i32 t0, t0, $1
+  mov_i32 t0, $1
+
+  only the last instruction is kept.
+
+- A macro system is supported (may get closer to function inlining
+  some day). It is useful if the liveness analysis is likely to prove
+  that some results of a computation are indeed not useful. With the
+  macro system, the user can provide several alternative
+  implementations which are used depending on the used results. It is
+  especially useful for condition code optimisation in QEMU.
+
+  Here is an example:
+
+  macro_2 t0, t1, $1
+  mov_i32 t0, $0x1234
+
+  The macro identified by the ID "$1" normally returns the values t0
+  and t1. Suppose its implementation is:
+
+  macro_start
+  brcond_i32  t2, $0, $TCG_COND_EQ, $1
+  mov_i32 t0, $2
+  br $2
+  set_label $1
+  mov_i32 t0, $3
+  set_label $2
+  add_i32 t1, t3, t4
+  macro_end
+  
+  If t0 is not used after the macro, the user can provide a simpler
+  implementation:
+
+  macro_start
+  add_i32 t1, t2, t4
+  macro_end
+
+  TCG automatically chooses the right implementation depending on
+  which macro outputs are used after it.
+
+  Note that if TCG did more expensive optimizations, macros would be
+  less useful. In the previous example a macro is useful because the
+  liveness analysis is done on each basic block separately. Hence TCG
+  cannot remove the code computing 't0' even if it is not used after
+  the first macro implementation.
+
+3.4) Instruction Reference
+
+********* Function call
+
+* call <ret> <params> ptr
+
+call function 'ptr' (pointer type)
+
+<ret> optional 32 bit or 64 bit return value
+<params> optional 32 bit or 64 bit parameters
+
+********* Jumps/Labels
+
+* jmp t0
+
+Absolute jump to address t0 (pointer type).
+
+* set_label $label
+
+Define label 'label' at the current program point.
+
+* br $label
+
+Jump to label.
+
+* brcond_i32/i64 cond, t0, t1, label
+
+Conditional jump if t0 cond t1 is true. cond can be:
+    TCG_COND_EQ
+    TCG_COND_NE
+    TCG_COND_LT /* signed */
+    TCG_COND_GE /* signed */
+    TCG_COND_LE /* signed */
+    TCG_COND_GT /* signed */
+    TCG_COND_LTU /* unsigned */
+    TCG_COND_GEU /* unsigned */
+    TCG_COND_LEU /* unsigned */
+    TCG_COND_GTU /* unsigned */
+
+********* Arithmetic
+
+* add_i32/i64 t0, t1, t2
+
+t0=t1+t2
+
+* sub_i32/i64 t0, t1, t2
+
+t0=t1-t2
+
+* mul_i32/i64 t0, t1, t2
+
+t0=t1*t2
+
+* div_i32/i64 t0, t1, t2
+
+t0=t1/t2 (signed). Undefined behavior if division by zero or overflow.
+
+* divu_i32/i64 t0, t1, t2
+
+t0=t1/t2 (unsigned). Undefined behavior if division by zero.
+
+* rem_i32/i64 t0, t1, t2
+
+t0=t1%t2 (signed). Undefined behavior if division by zero or overflow.
+
+* remu_i32/i64 t0, t1, t2
+
+t0=t1%t2 (unsigned). Undefined behavior if division by zero.
+
+* and_i32/i64 t0, t1, t2
+
+********* Logical
+
+t0=t1&t2
+
+* or_i32/i64 t0, t1, t2
+
+t0=t1|t2
+
+* xor_i32/i64 t0, t1, t2
+
+t0=t1^t2
+
+* shl_i32/i64 t0, t1, t2
+
+********* Shifts
+
+* shl_i32/i64 t0, t1, t2
+
+t0=t1 << t2. Undefined behavior if t2 < 0 or t2 >= 32 (resp 64)
+
+* shr_i32/i64 t0, t1, t2
+
+t0=t1 >> t2 (unsigned). Undefined behavior if t2 < 0 or t2 >= 32 (resp 64)
+
+* sar_i32/i64 t0, t1, t2
+
+t0=t1 >> t2 (signed). Undefined behavior if t2 < 0 or t2 >= 32 (resp 64)
+
+********* Misc
+
+* mov_i32/i64 t0, t1
+
+t0 = t1
+
+Move t1 to t0 (both operands must have the same type).
+
+* ext8s_i32/i64 t0, t1
+ext16s_i32/i64 t0, t1
+ext32s_i64 t0, t1
+
+8, 16 or 32 bit sign extension (both operands must have the same type)
+
+* bswap16_i32 t0, t1
+
+16 bit byte swap on a 32 bit value. The two high order bytes must be set
+to zero.
+
+* bswap_i32 t0, t1
+
+32 bit byte swap
+
+* bswap_i64 t0, t1
+
+64 bit byte swap
+
+********* Type conversions
+
+* ext_i32_i64 t0, t1
+Convert t1 (32 bit) to t0 (64 bit) and does sign extension
+
+* extu_i32_i64 t0, t1
+Convert t1 (32 bit) to t0 (64 bit) and does zero extension
+
+* trunc_i64_i32 t0, t1
+Truncate t1 (64 bit) to t0 (32 bit)
+
+********* Load/Store
+
+* ld_i32/i64 t0, t1, offset
+ld8s_i32/i64 t0, t1, offset
+ld8u_i32/i64 t0, t1, offset
+ld16s_i32/i64 t0, t1, offset
+ld16u_i32/i64 t0, t1, offset
+ld32s_i64 t0, t1, offset
+ld32u_i64 t0, t1, offset
+
+t0 = read(t1 + offset)
+Load 8, 16, 32 or 64 bits with or without sign extension from host memory. 
+offset must be a constant.
+
+* st_i32/i64 t0, t1, offset
+st8_i32/i64 t0, t1, offset
+st16_i32/i64 t0, t1, offset
+st32_i64 t0, t1, offset
+
+write(t0, t1 + offset)
+Write 8, 16, 32 or 64 bits to host memory.
+
+********* QEMU specific operations
+
+* tb_exit t0
+
+Exit the current TB and return the value t0 (word type).
+
+* goto_tb index
+
+Exit the current TB and jump to the TB index 'index' (constant) if the
+current TB was linked to this TB. Otherwise execute the next
+instructions.
+
+* qemu_ld_i32/i64 t0, t1, flags
+qemu_ld8u_i32/i64 t0, t1, flags
+qemu_ld8s_i32/i64 t0, t1, flags
+qemu_ld16u_i32/i64 t0, t1, flags
+qemu_ld16s_i32/i64 t0, t1, flags
+qemu_ld32u_i64 t0, t1, flags
+qemu_ld32s_i64 t0, t1, flags
+
+Load data at the QEMU CPU address t1 into t0. t1 has the QEMU CPU
+address type. 'flags' contains the QEMU memory index (selects user or
+kernel access) for example.
+
+* qemu_st_i32/i64 t0, t1, flags
+qemu_st8_i32/i64 t0, t1, flags
+qemu_st16_i32/i64 t0, t1, flags
+qemu_st32_i64 t0, t1, flags
+
+Store the data t0 at the QEMU CPU Address t1. t1 has the QEMU CPU
+address type. 'flags' contains the QEMU memory index (selects user or
+kernel access) for example.
+
+Note 1: Some shortcuts are defined when the last operand is known to be
+a constant (e.g. addi for add, movi for mov).
+
+Note 2: When using TCG, the opcodes must never be generated directly
+as some of them may not be available as "real" opcodes. Always use the
+function tcg_gen_xxx(args).
+
+4) Backend
+
+tcg-target.h contains the target specific definitions. tcg-target.c
+contains the target specific code.
+
+4.1) Assumptions
+
+The target word size (TCG_TARGET_REG_BITS) is expected to be 32 bit or
+64 bit. It is expected that the pointer has the same size as the word.
+
+On a 32 bit target, all 64 bit operations are converted to 32 bits. A
+few specific operations must be implemented to allow it (see add2_i32,
+sub2_i32, brcond2_i32).
+
+Floating point operations are not supported in this version. A
+previous incarnation of the code generator had full support of them,
+but it is better to concentrate on integer operations first.
+
+On a 64 bit target, no assumption is made in TCG about the storage of
+the 32 bit values in 64 bit registers.
+
+4.2) Constraints
+
+GCC like constraints are used to define the constraints of every
+instruction. Memory constraints are not supported in this
+version. Aliases are specified in the input operands as for GCC.
+
+A target can define specific register or constant constraints. If an
+operation uses a constant input constraint which does not allow all
+constants, it must also accept registers in order to have a fallback.
+
+The movi_i32 and movi_i64 operations must accept any constants.
+
+The mov_i32 and mov_i64 operations must accept any registers of the
+same type.
+
+The ld/st instructions must accept signed 32 bit constant offsets. It
+can be implemented by reserving a specific register to compute the
+address if the offset is too big.
+
+The ld/st instructions must accept any destination (ld) or source (st)
+register.
+
+4.3) Function call assumptions
+
+- The only supported types for parameters and return value are: 32 and
+  64 bit integers and pointer.
+- The stack grows downwards.
+- The first N parameters are passed in registers.
+- The next parameters are passed on the stack by storing them as words.
+- Some registers are clobbered during the call. 
+- The function can return 0 or 1 value in registers. On a 32 bit
+  target, functions must be able to return 2 values in registers for
+  64 bit return type.
+
+5) Migration from dyngen to TCG
+
+TCG is backward compatible with QEMU "dyngen" operations. It means
+that TCG instructions can be freely mixed with dyngen operations. It
+is expected that QEMU targets will be progressively fully converted to
+TCG. Once a target is fully converted to dyngen, it will be possible
+to apply more optimizations because more registers will be free for
+the generated code.
+
+The exception model is the same as the dyngen one.
diff --git a/tcg/TODO b/tcg/TODO
new file mode 100644
index 0000000000..9189926106
--- /dev/null
+++ b/tcg/TODO
@@ -0,0 +1,32 @@
+- test macro system
+
+- test conditional jumps
+
+- test mul, div, ext8s, ext16s, bswap
+
+- generate a global TB prologue and epilogue to save/restore registers
+  to/from the CPU state and to reserve a stack frame to optimize
+  helper calls. Modify cpu-exec.c so that it does not use global
+  register variables (except maybe for 'env').
+
+- fully convert the x86 target. The minimal amount of work includes:
+  - add cc_src, cc_dst and cc_op as globals
+  - disable its eflags optimization (the liveness analysis should
+    suffice)
+  - move complicated operations to helpers (in particular FPU, SSE, MMX).
+
+- optimize the x86 target:
+  - move some or all the registers as globals
+  - use the TB prologue and epilogue to have QEMU target registers in
+    pre assigned host registers.
+
+Ideas:
+
+- Move the slow part of the qemu_ld/st ops after the end of the TB.
+
+- Experiment: change instruction storage to simplify macro handling
+  and to handle dynamic allocation and see if the translation speed is
+  OK.
+
+- change exception syntax to get closer to QOP system (exception
+  parameters given with a specific instruction).
diff --git a/tcg/i386/tcg-target.c b/tcg/i386/tcg-target.c
new file mode 100644
index 0000000000..20822a4bc1
--- /dev/null
+++ b/tcg/i386/tcg-target.c
@@ -0,0 +1,1163 @@
+/*
+ * Tiny Code Generator for QEMU
+ *
+ * Copyright (c) 2008 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+const char *tcg_target_reg_names[TCG_TARGET_NB_REGS] = {
+    "%eax",
+    "%ecx",
+    "%edx",
+    "%ebx",
+    "%esp",
+    "%ebp",
+    "%esi",
+    "%edi",
+};
+
+int tcg_target_reg_alloc_order[TCG_TARGET_NB_REGS] = {
+    TCG_REG_EAX,
+    TCG_REG_EDX,
+    TCG_REG_ECX,
+    TCG_REG_EBX,
+    TCG_REG_ESI,
+    TCG_REG_EDI,
+    TCG_REG_EBP,
+    TCG_REG_ESP,
+};
+
+const int tcg_target_call_iarg_regs[3] = { TCG_REG_EAX, TCG_REG_EDX, TCG_REG_ECX };
+const int tcg_target_call_oarg_regs[2] = { TCG_REG_EAX, TCG_REG_EDX };
+
+static void patch_reloc(uint8_t *code_ptr, int type, 
+                        tcg_target_long value)
+{
+    switch(type) {
+    case R_386_32:
+        *(uint32_t *)code_ptr = value;
+        break;
+    case R_386_PC32:
+        *(uint32_t *)code_ptr = value - (long)code_ptr;
+        break;
+    default:
+        tcg_abort();
+    }
+}
+
+/* maximum number of register used for input function arguments */
+static inline int tcg_target_get_call_iarg_regs_count(int flags)
+{
+    flags &= TCG_CALL_TYPE_MASK;
+    switch(flags) {
+    case TCG_CALL_TYPE_STD:
+        return 0;
+    case TCG_CALL_TYPE_REGPARM_1:
+    case TCG_CALL_TYPE_REGPARM_2:
+    case TCG_CALL_TYPE_REGPARM:
+        return flags - TCG_CALL_TYPE_REGPARM_1 + 1;
+    default:
+        tcg_abort();
+    }
+}
+
+/* parse target specific constraints */
+int target_parse_constraint(TCGArgConstraint *ct, const char **pct_str)
+{
+    const char *ct_str;
+
+    ct_str = *pct_str;
+    switch(ct_str[0]) {
+    case 'a':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set_reg(ct->u.regs, TCG_REG_EAX);
+        break;
+    case 'b':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set_reg(ct->u.regs, TCG_REG_EBX);
+        break;
+    case 'c':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set_reg(ct->u.regs, TCG_REG_ECX);
+        break;
+    case 'd':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set_reg(ct->u.regs, TCG_REG_EDX);
+        break;
+    case 'S':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set_reg(ct->u.regs, TCG_REG_ESI);
+        break;
+    case 'D':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set_reg(ct->u.regs, TCG_REG_EDI);
+        break;
+    case 'q':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set32(ct->u.regs, 0, 0xf);
+        break;
+    case 'r':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set32(ct->u.regs, 0, 0xff);
+        break;
+
+        /* qemu_ld/st address constraint */
+    case 'L':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set32(ct->u.regs, 0, 0xff);
+        tcg_regset_reset_reg(ct->u.regs, TCG_REG_EAX);
+        tcg_regset_reset_reg(ct->u.regs, TCG_REG_EDX);
+        break;
+    default:
+        return -1;
+    }
+    ct_str++;
+    *pct_str = ct_str;
+    return 0;
+}
+
+/* test if a constant matches the constraint */
+static inline int tcg_target_const_match(tcg_target_long val,
+                                         const TCGArgConstraint *arg_ct)
+{
+    int ct;
+    ct = arg_ct->ct;
+    if (ct & TCG_CT_CONST)
+        return 1;
+    else
+        return 0;
+}
+
+#define ARITH_ADD 0
+#define ARITH_OR  1
+#define ARITH_ADC 2
+#define ARITH_SBB 3
+#define ARITH_AND 4
+#define ARITH_SUB 5
+#define ARITH_XOR 6
+#define ARITH_CMP 7
+
+#define SHIFT_SHL 4
+#define SHIFT_SHR 5
+#define SHIFT_SAR 7
+
+#define JCC_JMP (-1)
+#define JCC_JO  0x0
+#define JCC_JNO 0x1
+#define JCC_JB  0x2
+#define JCC_JAE 0x3
+#define JCC_JE  0x4
+#define JCC_JNE 0x5
+#define JCC_JBE 0x6
+#define JCC_JA  0x7
+#define JCC_JS  0x8
+#define JCC_JNS 0x9
+#define JCC_JP  0xa
+#define JCC_JNP 0xb
+#define JCC_JL  0xc
+#define JCC_JGE 0xd
+#define JCC_JLE 0xe
+#define JCC_JG  0xf
+
+#define P_EXT   0x100 /* 0x0f opcode prefix */
+
+static const uint8_t tcg_cond_to_jcc[10] = {
+    [TCG_COND_EQ] = JCC_JE,
+    [TCG_COND_NE] = JCC_JNE,
+    [TCG_COND_LT] = JCC_JL,
+    [TCG_COND_GE] = JCC_JGE,
+    [TCG_COND_LE] = JCC_JLE,
+    [TCG_COND_GT] = JCC_JG,
+    [TCG_COND_LTU] = JCC_JB,
+    [TCG_COND_GEU] = JCC_JAE,
+    [TCG_COND_LEU] = JCC_JBE,
+    [TCG_COND_GTU] = JCC_JA,
+};
+
+static inline void tcg_out_opc(TCGContext *s, int opc)
+{
+    if (opc & P_EXT)
+        tcg_out8(s, 0x0f);
+    tcg_out8(s, opc);
+}
+
+static inline void tcg_out_modrm(TCGContext *s, int opc, int r, int rm)
+{
+    tcg_out_opc(s, opc);
+    tcg_out8(s, 0xc0 | (r << 3) | rm);
+}
+
+/* rm == -1 means no register index */
+static inline void tcg_out_modrm_offset(TCGContext *s, int opc, int r, int rm, 
+                                        int32_t offset)
+{
+    tcg_out_opc(s, opc);
+    if (rm == -1) {
+        tcg_out8(s, 0x05 | (r << 3));
+        tcg_out32(s, offset);
+    } else if (offset == 0 && rm != TCG_REG_EBP) {
+        if (rm == TCG_REG_ESP) {
+            tcg_out8(s, 0x04 | (r << 3));
+            tcg_out8(s, 0x24);
+        } else {
+            tcg_out8(s, 0x00 | (r << 3) | rm);
+        }
+    } else if ((int8_t)offset == offset) {
+        if (rm == TCG_REG_ESP) {
+            tcg_out8(s, 0x44 | (r << 3));
+            tcg_out8(s, 0x24);
+        } else {
+            tcg_out8(s, 0x40 | (r << 3) | rm);
+        }
+        tcg_out8(s, offset);
+    } else {
+        if (rm == TCG_REG_ESP) {
+            tcg_out8(s, 0x84 | (r << 3));
+            tcg_out8(s, 0x24);
+        } else {
+            tcg_out8(s, 0x80 | (r << 3) | rm);
+        }
+        tcg_out32(s, offset);
+    }
+}
+
+static inline void tcg_out_mov(TCGContext *s, int ret, int arg)
+{
+    if (arg != ret)
+        tcg_out_modrm(s, 0x8b, ret, arg);
+}
+
+static inline void tcg_out_movi(TCGContext *s, TCGType type,
+                                int ret, int32_t arg)
+{
+    if (arg == 0) {
+        /* xor r0,r0 */
+        tcg_out_modrm(s, 0x01 | (ARITH_XOR << 3), ret, ret);
+    } else {
+        tcg_out8(s, 0xb8 + ret);
+        tcg_out32(s, arg);
+    }
+}
+
+static inline void tcg_out_ld(TCGContext *s, int ret, 
+                              int arg1, int32_t arg2)
+{
+    /* movl */
+    tcg_out_modrm_offset(s, 0x8b, ret, arg1, arg2);
+}
+
+static inline void tcg_out_st(TCGContext *s, int arg, 
+                              int arg1, int32_t arg2)
+{
+    /* movl */
+    tcg_out_modrm_offset(s, 0x89, arg, arg1, arg2);
+}
+
+static inline void tgen_arithi(TCGContext *s, int c, int r0, int32_t val)
+{
+    if (val == (int8_t)val) {
+        tcg_out_modrm(s, 0x83, c, r0);
+        tcg_out8(s, val);
+    } else {
+        tcg_out_modrm(s, 0x81, c, r0);
+        tcg_out32(s, val);
+    }
+}
+
+void tcg_out_addi(TCGContext *s, int reg, tcg_target_long val)
+{
+    if (val != 0)
+        tgen_arithi(s, ARITH_ADD, reg, val);
+}
+
+static void tcg_out_jxx(TCGContext *s, int opc, int label_index)
+{
+    int32_t val, val1;
+    TCGLabel *l = &s->labels[label_index];
+    
+    if (l->has_value) {
+        val = l->u.value - (tcg_target_long)s->code_ptr;
+        val1 = val - 2;
+        if ((int8_t)val1 == val1) {
+            if (opc == -1)
+                tcg_out8(s, 0xeb);
+            else
+                tcg_out8(s, 0x70 + opc);
+            tcg_out8(s, val1);
+        } else {
+            if (opc == -1) {
+                tcg_out8(s, 0xe9);
+                tcg_out32(s, val - 5);
+            } else {
+                tcg_out8(s, 0x0f);
+                tcg_out8(s, 0x80 + opc);
+                tcg_out32(s, val - 6);
+            }
+        }
+    } else {
+        if (opc == -1) {
+            tcg_out8(s, 0xe9);
+        } else {
+            tcg_out8(s, 0x0f);
+            tcg_out8(s, 0x80 + opc);
+        }
+        tcg_out_reloc(s, s->code_ptr, R_386_PC32, label_index, -4);
+        tcg_out32(s, -4);
+    }
+}
+
+static void tcg_out_brcond(TCGContext *s, int cond, 
+                           TCGArg arg1, TCGArg arg2, int const_arg2,
+                           int label_index)
+{
+    int c;
+    if (const_arg2) {
+        if (arg2 == 0) {
+            /* use test */
+            switch(cond) {
+            case TCG_COND_EQ:
+                c = JCC_JNE;
+                break;
+            case TCG_COND_NE:
+                c = JCC_JNE;
+                break;
+            case TCG_COND_LT:
+                c = JCC_JS;
+                break;
+            case TCG_COND_GE:
+                c = JCC_JNS;
+                break;
+            default:
+                goto do_cmpi;
+            }
+            /* test r, r */
+            tcg_out_modrm(s, 0x85, arg1, arg1);
+            tcg_out_jxx(s, c, label_index);
+        } else {
+        do_cmpi:
+            tgen_arithi(s, ARITH_CMP, arg1, arg2);
+            tcg_out_jxx(s, tcg_cond_to_jcc[cond], label_index);
+        }
+    } else {
+        tcg_out_modrm(s, 0x01 | (ARITH_CMP << 3), arg1, arg2);
+        tcg_out_jxx(s, tcg_cond_to_jcc[cond], label_index);
+    }
+}
+
+/* XXX: we implement it at the target level to avoid having to
+   handle cross basic blocks temporaries */
+static void tcg_out_brcond2(TCGContext *s,
+                            const TCGArg *args, const int *const_args)
+{
+    int label_next;
+    label_next = gen_new_label();
+    switch(args[4]) {
+    case TCG_COND_EQ:
+        tcg_out_brcond(s, TCG_COND_NE, args[0], args[2], const_args[2], label_next);
+        tcg_out_brcond(s, TCG_COND_EQ, args[1], args[3], const_args[3], args[5]);
+        break;
+    case TCG_COND_NE:
+        tcg_out_brcond(s, TCG_COND_NE, args[0], args[2], const_args[2], args[5]);
+        tcg_out_brcond(s, TCG_COND_EQ, args[1], args[3], const_args[3], label_next);
+        break;
+    case TCG_COND_LT:
+        tcg_out_brcond(s, TCG_COND_LT, args[1], args[3], const_args[3], args[5]);
+        tcg_out_brcond(s, TCG_COND_NE, args[1], args[3], const_args[3], label_next);
+        tcg_out_brcond(s, TCG_COND_LT, args[0], args[2], const_args[2], args[5]);
+        break;
+    case TCG_COND_LE:
+        tcg_out_brcond(s, TCG_COND_LT, args[1], args[3], const_args[3], args[5]);
+        tcg_out_brcond(s, TCG_COND_NE, args[1], args[3], const_args[3], label_next);
+        tcg_out_brcond(s, TCG_COND_LE, args[0], args[2], const_args[2], args[5]);
+        break;
+    case TCG_COND_GT:
+        tcg_out_brcond(s, TCG_COND_GT, args[1], args[3], const_args[3], args[5]);
+        tcg_out_brcond(s, TCG_COND_NE, args[1], args[3], const_args[3], label_next);
+        tcg_out_brcond(s, TCG_COND_GT, args[0], args[2], const_args[2], args[5]);
+        break;
+    case TCG_COND_GE:
+        tcg_out_brcond(s, TCG_COND_GT, args[1], args[3], const_args[3], args[5]);
+        tcg_out_brcond(s, TCG_COND_NE, args[1], args[3], const_args[3], label_next);
+        tcg_out_brcond(s, TCG_COND_GE, args[0], args[2], const_args[2], args[5]);
+        break;
+    case TCG_COND_LTU:
+        tcg_out_brcond(s, TCG_COND_LTU, args[1], args[3], const_args[3], args[5]);
+        tcg_out_brcond(s, TCG_COND_NE, args[1], args[3], const_args[3], label_next);
+        tcg_out_brcond(s, TCG_COND_LTU, args[0], args[2], const_args[2], args[5]);
+        break;
+    case TCG_COND_LEU:
+        tcg_out_brcond(s, TCG_COND_LTU, args[1], args[3], const_args[3], args[5]);
+        tcg_out_brcond(s, TCG_COND_NE, args[1], args[3], const_args[3], label_next);
+        tcg_out_brcond(s, TCG_COND_LEU, args[0], args[2], const_args[2], args[5]);
+        break;
+    case TCG_COND_GTU:
+        tcg_out_brcond(s, TCG_COND_GTU, args[1], args[3], const_args[3], args[5]);
+        tcg_out_brcond(s, TCG_COND_NE, args[1], args[3], const_args[3], label_next);
+        tcg_out_brcond(s, TCG_COND_GTU, args[0], args[2], const_args[2], args[5]);
+        break;
+    case TCG_COND_GEU:
+        tcg_out_brcond(s, TCG_COND_GTU, args[1], args[3], const_args[3], args[5]);
+        tcg_out_brcond(s, TCG_COND_NE, args[1], args[3], const_args[3], label_next);
+        tcg_out_brcond(s, TCG_COND_GEU, args[0], args[2], const_args[2], args[5]);
+        break;
+    default:
+        tcg_abort();
+    }
+    tcg_out_label(s, label_next, (tcg_target_long)s->code_ptr);
+}
+
+#if defined(CONFIG_SOFTMMU)
+extern void __ldb_mmu(void);
+extern void __ldw_mmu(void);
+extern void __ldl_mmu(void);
+extern void __ldq_mmu(void);
+
+extern void __stb_mmu(void);
+extern void __stw_mmu(void);
+extern void __stl_mmu(void);
+extern void __stq_mmu(void);
+
+static void *qemu_ld_helpers[4] = {
+    __ldb_mmu,
+    __ldw_mmu,
+    __ldl_mmu,
+    __ldq_mmu,
+};
+
+static void *qemu_st_helpers[4] = {
+    __stb_mmu,
+    __stw_mmu,
+    __stl_mmu,
+    __stq_mmu,
+};
+#endif
+
+/* XXX: qemu_ld and qemu_st could be modified to clobber only EDX and
+   EAX. It will be useful once fixed registers globals are less
+   common. */
+static void tcg_out_qemu_ld(TCGContext *s, const TCGArg *args,
+                            int opc)
+{
+    int addr_reg, data_reg, data_reg2, r0, r1, mem_index, s_bits, bswap;
+#if defined(CONFIG_SOFTMMU)
+    uint8_t *label1_ptr, *label2_ptr;
+#endif
+#if TARGET_LONG_BITS == 64
+#if defined(CONFIG_SOFTMMU)
+    uint8_t *label3_ptr;
+#endif
+    int addr_reg2;
+#endif
+
+    data_reg = *args++;
+    if (opc == 3)
+        data_reg2 = *args++;
+    else
+        data_reg2 = 0;
+    addr_reg = *args++;
+#if TARGET_LONG_BITS == 64
+    addr_reg2 = *args++;
+#endif
+    mem_index = *args;
+    s_bits = opc & 3;
+
+    r0 = TCG_REG_EAX;
+    r1 = TCG_REG_EDX;
+
+#if defined(CONFIG_SOFTMMU)
+    tcg_out_mov(s, r1, addr_reg); 
+
+    tcg_out_mov(s, r0, addr_reg); 
+ 
+    tcg_out_modrm(s, 0xc1, 5, r1); /* shr $x, r1 */
+    tcg_out8(s, TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); 
+    
+    tcg_out_modrm(s, 0x81, 4, r0); /* andl $x, r0 */
+    tcg_out32(s, TARGET_PAGE_MASK | ((1 << s_bits) - 1));
+    
+    tcg_out_modrm(s, 0x81, 4, r1); /* andl $x, r1 */
+    tcg_out32(s, (CPU_TLB_SIZE - 1) << CPU_TLB_ENTRY_BITS);
+
+    tcg_out_opc(s, 0x8d); /* lea offset(r1, %ebp), r1 */
+    tcg_out8(s, 0x80 | (r1 << 3) | 0x04);
+    tcg_out8(s, (5 << 3) | r1);
+    tcg_out32(s, offsetof(CPUState, tlb_table[mem_index][0].addr_read));
+
+    /* cmp 0(r1), r0 */
+    tcg_out_modrm_offset(s, 0x3b, r0, r1, 0);
+    
+    tcg_out_mov(s, r0, addr_reg);
+    
+#if TARGET_LONG_BITS == 32
+    /* je label1 */
+    tcg_out8(s, 0x70 + JCC_JE);
+    label1_ptr = s->code_ptr;
+    s->code_ptr++;
+#else
+    /* jne label3 */
+    tcg_out8(s, 0x70 + JCC_JNE);
+    label3_ptr = s->code_ptr;
+    s->code_ptr++;
+    
+    /* cmp 4(r1), addr_reg2 */
+    tcg_out_modrm_offset(s, 0x3b, addr_reg2, r1, 4);
+
+    /* je label1 */
+    tcg_out8(s, 0x70 + JCC_JE);
+    label1_ptr = s->code_ptr;
+    s->code_ptr++;
+    
+    /* label3: */
+    *label3_ptr = s->code_ptr - label3_ptr - 1;
+#endif
+
+    /* XXX: move that code at the end of the TB */
+#if TARGET_LONG_BITS == 32
+    tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_EDX, mem_index);
+#else
+    tcg_out_mov(s, TCG_REG_EDX, addr_reg2);
+    tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_ECX, mem_index);
+#endif
+    tcg_out8(s, 0xe8);
+    tcg_out32(s, (tcg_target_long)qemu_ld_helpers[s_bits] - 
+              (tcg_target_long)s->code_ptr - 4);
+
+    switch(opc) {
+    case 0 | 4:
+        /* movsbl */
+        tcg_out_modrm(s, 0xbe | P_EXT, data_reg, TCG_REG_EAX);
+        break;
+    case 1 | 4:
+        /* movswl */
+        tcg_out_modrm(s, 0xbf | P_EXT, data_reg, TCG_REG_EAX);
+        break;
+    case 0:
+    case 1:
+    case 2:
+    default:
+        tcg_out_mov(s, data_reg, TCG_REG_EAX);
+        break;
+    case 3:
+        if (data_reg == TCG_REG_EDX) {
+            tcg_out_opc(s, 0x90 + TCG_REG_EDX); /* xchg %edx, %eax */
+            tcg_out_mov(s, data_reg2, TCG_REG_EAX);
+        } else {
+            tcg_out_mov(s, data_reg, TCG_REG_EAX);
+            tcg_out_mov(s, data_reg2, TCG_REG_EDX);
+        }
+        break;
+    }
+
+    /* jmp label2 */
+    tcg_out8(s, 0xeb);
+    label2_ptr = s->code_ptr;
+    s->code_ptr++;
+    
+    /* label1: */
+    *label1_ptr = s->code_ptr - label1_ptr - 1;
+
+    /* add x(r1), r0 */
+    tcg_out_modrm_offset(s, 0x03, r0, r1, offsetof(CPUTLBEntry, addend) - 
+                         offsetof(CPUTLBEntry, addr_read));
+#else
+    r0 = addr_reg;
+#endif
+
+#ifdef TARGET_WORDS_BIGENDIAN
+    bswap = 1;
+#else
+    bswap = 0;
+#endif
+    switch(opc) {
+    case 0:
+        /* movzbl */
+        tcg_out_modrm_offset(s, 0xb6 | P_EXT, data_reg, r0, 0);
+        break;
+    case 0 | 4:
+        /* movsbl */
+        tcg_out_modrm_offset(s, 0xbe | P_EXT, data_reg, r0, 0);
+        break;
+    case 1:
+        /* movzwl */
+        tcg_out_modrm_offset(s, 0xb7 | P_EXT, data_reg, r0, 0);
+        if (bswap) {
+            /* rolw $8, data_reg */
+            tcg_out8(s, 0x66); 
+            tcg_out_modrm(s, 0xc1, 0, data_reg);
+            tcg_out8(s, 8);
+        }
+        break;
+    case 1 | 4:
+        /* movswl */
+        tcg_out_modrm_offset(s, 0xbf | P_EXT, data_reg, r0, 0);
+        if (bswap) {
+            /* rolw $8, data_reg */
+            tcg_out8(s, 0x66); 
+            tcg_out_modrm(s, 0xc1, 0, data_reg);
+            tcg_out8(s, 8);
+
+            /* movswl data_reg, data_reg */
+            tcg_out_modrm(s, 0xbf | P_EXT, data_reg, data_reg);
+        }
+        break;
+    case 2:
+        /* movl (r0), data_reg */
+        tcg_out_modrm_offset(s, 0x8b, data_reg, r0, 0);
+        if (bswap) {
+            /* bswap */
+            tcg_out_opc(s, (0xc8 + data_reg) | P_EXT);
+        }
+        break;
+    case 3:
+        /* XXX: could be nicer */
+        if (r0 == data_reg) {
+            r1 = TCG_REG_EDX;
+            if (r1 == data_reg)
+                r1 = TCG_REG_EAX;
+            tcg_out_mov(s, r1, r0);
+            r0 = r1;
+        }
+        if (!bswap) {
+            tcg_out_modrm_offset(s, 0x8b, data_reg, r0, 0);
+            tcg_out_modrm_offset(s, 0x8b, data_reg2, r0, 4);
+        } else {
+            tcg_out_modrm_offset(s, 0x8b, data_reg, r0, 4);
+            tcg_out_opc(s, (0xc8 + data_reg) | P_EXT);
+
+            tcg_out_modrm_offset(s, 0x8b, data_reg2, r0, 0);
+            /* bswap */
+            tcg_out_opc(s, (0xc8 + data_reg2) | P_EXT);
+        }
+        break;
+    default:
+        tcg_abort();
+    }
+
+#if defined(CONFIG_SOFTMMU)
+    /* label2: */
+    *label2_ptr = s->code_ptr - label2_ptr - 1;
+#endif
+}
+
+
+static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args,
+                            int opc)
+{
+    int addr_reg, data_reg, data_reg2, r0, r1, mem_index, s_bits, bswap;
+#if defined(CONFIG_SOFTMMU)
+    uint8_t *label1_ptr, *label2_ptr;
+#endif
+#if TARGET_LONG_BITS == 64
+#if defined(CONFIG_SOFTMMU)
+    uint8_t *label3_ptr;
+#endif
+    int addr_reg2;
+#endif
+
+    data_reg = *args++;
+    if (opc == 3)
+        data_reg2 = *args++;
+    else
+        data_reg2 = 0;
+    addr_reg = *args++;
+#if TARGET_LONG_BITS == 64
+    addr_reg2 = *args++;
+#endif
+    mem_index = *args;
+
+    s_bits = opc;
+
+    r0 = TCG_REG_EAX;
+    r1 = TCG_REG_EDX;
+
+#if defined(CONFIG_SOFTMMU)
+    tcg_out_mov(s, r1, addr_reg); 
+
+    tcg_out_mov(s, r0, addr_reg); 
+ 
+    tcg_out_modrm(s, 0xc1, 5, r1); /* shr $x, r1 */
+    tcg_out8(s, TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); 
+    
+    tcg_out_modrm(s, 0x81, 4, r0); /* andl $x, r0 */
+    tcg_out32(s, TARGET_PAGE_MASK | ((1 << s_bits) - 1));
+    
+    tcg_out_modrm(s, 0x81, 4, r1); /* andl $x, r1 */
+    tcg_out32(s, (CPU_TLB_SIZE - 1) << CPU_TLB_ENTRY_BITS);
+
+    tcg_out_opc(s, 0x8d); /* lea offset(r1, %ebp), r1 */
+    tcg_out8(s, 0x80 | (r1 << 3) | 0x04);
+    tcg_out8(s, (5 << 3) | r1);
+    tcg_out32(s, offsetof(CPUState, tlb_table[mem_index][0].addr_write));
+
+    /* cmp 0(r1), r0 */
+    tcg_out_modrm_offset(s, 0x3b, r0, r1, 0);
+    
+    tcg_out_mov(s, r0, addr_reg);
+    
+#if TARGET_LONG_BITS == 32
+    /* je label1 */
+    tcg_out8(s, 0x70 + JCC_JE);
+    label1_ptr = s->code_ptr;
+    s->code_ptr++;
+#else
+    /* jne label3 */
+    tcg_out8(s, 0x70 + JCC_JNE);
+    label3_ptr = s->code_ptr;
+    s->code_ptr++;
+    
+    /* cmp 4(r1), addr_reg2 */
+    tcg_out_modrm_offset(s, 0x3b, addr_reg2, r1, 4);
+
+    /* je label1 */
+    tcg_out8(s, 0x70 + JCC_JE);
+    label1_ptr = s->code_ptr;
+    s->code_ptr++;
+    
+    /* label3: */
+    *label3_ptr = s->code_ptr - label3_ptr - 1;
+#endif
+
+    /* XXX: move that code at the end of the TB */
+#if TARGET_LONG_BITS == 32
+    if (opc == 3) {
+        tcg_out_mov(s, TCG_REG_EDX, data_reg);
+        tcg_out_mov(s, TCG_REG_ECX, data_reg2);
+        tcg_out8(s, 0x6a); /* push Ib */
+        tcg_out8(s, 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_addi(s, TCG_REG_ESP, 4);
+    } else {
+        switch(opc) {
+        case 0:
+            /* movzbl */
+            tcg_out_modrm(s, 0xb6 | P_EXT, TCG_REG_EDX, data_reg);
+            break;
+        case 1:
+            /* movzwl */
+            tcg_out_modrm(s, 0xb7 | P_EXT, TCG_REG_EDX, data_reg);
+            break;
+        case 2:
+            tcg_out_mov(s, TCG_REG_EDX, data_reg);
+            break;
+        }
+        tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_ECX, mem_index);
+        tcg_out8(s, 0xe8);
+        tcg_out32(s, (tcg_target_long)qemu_st_helpers[s_bits] - 
+                  (tcg_target_long)s->code_ptr - 4);
+    }
+#else
+    if (opc == 3) {
+        tcg_out_mov(s, TCG_REG_EDX, addr_reg2);
+        tcg_out8(s, 0x6a); /* push Ib */
+        tcg_out8(s, mem_index);
+        tcg_out_opc(s, 0x50 + data_reg2); /* push */
+        tcg_out_opc(s, 0x50 + data_reg); /* push */
+        tcg_out8(s, 0xe8);
+        tcg_out32(s, (tcg_target_long)qemu_st_helpers[s_bits] - 
+                  (tcg_target_long)s->code_ptr - 4);
+        tcg_out_addi(s, TCG_REG_ESP, 12);
+    } else {
+        tcg_out_mov(s, TCG_REG_EDX, addr_reg2);
+        switch(opc) {
+        case 0:
+            /* movzbl */
+            tcg_out_modrm(s, 0xb6 | P_EXT, TCG_REG_ECX, data_reg);
+            break;
+        case 1:
+            /* movzwl */
+            tcg_out_modrm(s, 0xb7 | P_EXT, TCG_REG_ECX, data_reg);
+            break;
+        case 2:
+            tcg_out_mov(s, TCG_REG_ECX, data_reg);
+            break;
+        }
+        tcg_out8(s, 0x6a); /* push Ib */
+        tcg_out8(s, 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_addi(s, TCG_REG_ESP, 4);
+    }
+#endif
+    
+    /* jmp label2 */
+    tcg_out8(s, 0xeb);
+    label2_ptr = s->code_ptr;
+    s->code_ptr++;
+    
+    /* label1: */
+    *label1_ptr = s->code_ptr - label1_ptr - 1;
+
+    /* add x(r1), r0 */
+    tcg_out_modrm_offset(s, 0x03, r0, r1, offsetof(CPUTLBEntry, addend) - 
+                         offsetof(CPUTLBEntry, addr_write));
+#else
+    r0 = addr_reg;
+#endif
+
+#ifdef TARGET_WORDS_BIGENDIAN
+    bswap = 1;
+#else
+    bswap = 0;
+#endif
+    switch(opc) {
+    case 0:
+        /* movb */
+        tcg_out_modrm_offset(s, 0x88, data_reg, r0, 0);
+        break;
+    case 1:
+        if (bswap) {
+            tcg_out_mov(s, r1, data_reg);
+            tcg_out8(s, 0x66); /* rolw $8, %ecx */
+            tcg_out_modrm(s, 0xc1, 0, r1);
+            tcg_out8(s, 8);
+            data_reg = r1;
+        }
+        /* movw */
+        tcg_out8(s, 0x66);
+        tcg_out_modrm_offset(s, 0x89, data_reg, r0, 0);
+        break;
+    case 2:
+        if (bswap) {
+            tcg_out_mov(s, r1, data_reg);
+            /* bswap data_reg */
+            tcg_out_opc(s, (0xc8 + r1) | P_EXT);
+            data_reg = r1;
+        }
+        /* movl */
+        tcg_out_modrm_offset(s, 0x89, data_reg, r0, 0);
+        break;
+    case 3:
+        if (bswap) {
+            tcg_out_mov(s, r1, data_reg2);
+            /* bswap data_reg */
+            tcg_out_opc(s, (0xc8 + r1) | P_EXT);
+            tcg_out_modrm_offset(s, 0x89, r1, r0, 0);
+            tcg_out_mov(s, r1, data_reg);
+            /* bswap data_reg */
+            tcg_out_opc(s, (0xc8 + r1) | P_EXT);
+            tcg_out_modrm_offset(s, 0x89, r1, r0, 4);
+        } else {
+            tcg_out_modrm_offset(s, 0x89, data_reg, r0, 0);
+            tcg_out_modrm_offset(s, 0x89, data_reg2, r0, 4);
+        }
+        break;
+    default:
+        tcg_abort();
+    }
+
+#if defined(CONFIG_SOFTMMU)
+    /* label2: */
+    *label2_ptr = s->code_ptr - label2_ptr - 1;
+#endif
+}
+
+static inline void tcg_out_op(TCGContext *s, int opc, 
+                              const TCGArg *args, const int *const_args)
+{
+    int c;
+    
+    switch(opc) {
+    case INDEX_op_exit_tb:
+        tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_EAX, args[0]);
+        tcg_out8(s, 0xc3); /* ret */
+        break;
+    case INDEX_op_goto_tb:
+        if (s->tb_jmp_offset) {
+            /* direct jump method */
+            tcg_out8(s, 0xe9); /* jmp im */
+            s->tb_jmp_offset[args[0]] = s->code_ptr - s->code_buf;
+            tcg_out32(s, 0);
+        } else {
+            /* indirect jump method */
+            /* jmp Ev */
+            tcg_out_modrm_offset(s, 0xff, 4, -1, 
+                                 (tcg_target_long)(s->tb_next + args[0]));
+        }
+        s->tb_next_offset[args[0]] = s->code_ptr - s->code_buf;
+        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);
+        } 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);
+        } else {
+            tcg_out_modrm(s, 0xff, 4, args[0]);
+        }
+        break;
+    case INDEX_op_br:
+        tcg_out_jxx(s, JCC_JMP, args[0]);
+        break;
+    case INDEX_op_movi_i32:
+        tcg_out_movi(s, TCG_TYPE_I32, args[0], args[1]);
+        break;
+    case INDEX_op_ld8u_i32:
+        /* movzbl */
+        tcg_out_modrm_offset(s, 0xb6 | P_EXT, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_ld8s_i32:
+        /* movsbl */
+        tcg_out_modrm_offset(s, 0xbe | P_EXT, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_ld16u_i32:
+        /* movzwl */
+        tcg_out_modrm_offset(s, 0xb7 | P_EXT, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_ld16s_i32:
+        /* movswl */
+        tcg_out_modrm_offset(s, 0xbf | P_EXT, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_ld_i32:
+        /* movl */
+        tcg_out_modrm_offset(s, 0x8b, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_st8_i32:
+        /* movb */
+        tcg_out_modrm_offset(s, 0x88, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_st16_i32:
+        /* movw */
+        tcg_out8(s, 0x66);
+        tcg_out_modrm_offset(s, 0x89, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_st_i32:
+        /* movl */
+        tcg_out_modrm_offset(s, 0x89, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_sub_i32:
+        c = ARITH_SUB;
+        goto gen_arith;
+    case INDEX_op_and_i32:
+        c = ARITH_AND;
+        goto gen_arith;
+    case INDEX_op_or_i32:
+        c = ARITH_OR;
+        goto gen_arith;
+    case INDEX_op_xor_i32:
+        c = ARITH_XOR;
+        goto gen_arith;
+    case INDEX_op_add_i32:
+        c = ARITH_ADD;
+    gen_arith:
+        if (const_args[2]) {
+            tgen_arithi(s, c, args[0], args[2]);
+        } else {
+            tcg_out_modrm(s, 0x01 | (c << 3), args[2], args[0]);
+        }
+        break;
+    case INDEX_op_mul_i32:
+        if (const_args[2]) {
+            int32_t val;
+            val = args[2];
+            if (val == (int8_t)val) {
+                tcg_out_modrm(s, 0x6b, args[0], args[0]);
+                tcg_out8(s, val);
+            } else {
+                tcg_out_modrm(s, 0x69, args[0], args[0]);
+                tcg_out32(s, val);
+            }
+        } else {
+            tcg_out_modrm(s, 0xaf | P_EXT, args[0], args[2]);
+        }
+        break;
+    case INDEX_op_mulu2_i32:
+        tcg_out_modrm(s, 0xf7, 4, args[3]);
+        break;
+    case INDEX_op_div2_i32:
+        tcg_out_modrm(s, 0xf7, 7, args[4]);
+        break;
+    case INDEX_op_divu2_i32:
+        tcg_out_modrm(s, 0xf7, 6, args[4]);
+        break;
+    case INDEX_op_shl_i32:
+        c = SHIFT_SHL;
+    gen_shift32:
+        if (const_args[2]) {
+            if (args[2] == 1) {
+                tcg_out_modrm(s, 0xd1, c, args[0]);
+            } else {
+                tcg_out_modrm(s, 0xc1, c, args[0]);
+                tcg_out8(s, args[2]);
+            }
+        } else {
+            tcg_out_modrm(s, 0xd3, c, args[0]);
+        }
+        break;
+    case INDEX_op_shr_i32:
+        c = SHIFT_SHR;
+        goto gen_shift32;
+    case INDEX_op_sar_i32:
+        c = SHIFT_SAR;
+        goto gen_shift32;
+        
+    case INDEX_op_add2_i32:
+        if (const_args[4]) 
+            tgen_arithi(s, ARITH_ADD, args[0], args[4]);
+        else
+            tcg_out_modrm(s, 0x01 | (ARITH_ADD << 3), args[4], args[0]);
+        if (const_args[5]) 
+            tgen_arithi(s, ARITH_ADC, args[1], args[5]);
+        else
+            tcg_out_modrm(s, 0x01 | (ARITH_ADC << 3), args[5], args[1]);
+        break;
+    case INDEX_op_sub2_i32:
+        if (const_args[4]) 
+            tgen_arithi(s, ARITH_SUB, args[0], args[4]);
+        else
+            tcg_out_modrm(s, 0x01 | (ARITH_SUB << 3), args[4], args[0]);
+        if (const_args[5]) 
+            tgen_arithi(s, ARITH_SBB, args[1], args[5]);
+        else
+            tcg_out_modrm(s, 0x01 | (ARITH_SBB << 3), args[5], args[1]);
+        break;
+    case INDEX_op_brcond_i32:
+        tcg_out_brcond(s, args[2], args[0], args[1], const_args[1], args[3]);
+        break;
+    case INDEX_op_brcond2_i32:
+        tcg_out_brcond2(s, args, const_args);
+        break;
+
+    case INDEX_op_qemu_ld8u:
+        tcg_out_qemu_ld(s, args, 0);
+        break;
+    case INDEX_op_qemu_ld8s:
+        tcg_out_qemu_ld(s, args, 0 | 4);
+        break;
+    case INDEX_op_qemu_ld16u:
+        tcg_out_qemu_ld(s, args, 1);
+        break;
+    case INDEX_op_qemu_ld16s:
+        tcg_out_qemu_ld(s, args, 1 | 4);
+        break;
+    case INDEX_op_qemu_ld32u:
+        tcg_out_qemu_ld(s, args, 2);
+        break;
+    case INDEX_op_qemu_ld64:
+        tcg_out_qemu_ld(s, args, 3);
+        break;
+        
+    case INDEX_op_qemu_st8:
+        tcg_out_qemu_st(s, args, 0);
+        break;
+    case INDEX_op_qemu_st16:
+        tcg_out_qemu_st(s, args, 1);
+        break;
+    case INDEX_op_qemu_st32:
+        tcg_out_qemu_st(s, args, 2);
+        break;
+    case INDEX_op_qemu_st64:
+        tcg_out_qemu_st(s, args, 3);
+        break;
+
+    default:
+        tcg_abort();
+    }
+}
+
+static const TCGTargetOpDef x86_op_defs[] = {
+    { INDEX_op_exit_tb, { } },
+    { INDEX_op_goto_tb, { } },
+    { INDEX_op_call, { "ri" } },
+    { INDEX_op_jmp, { "ri" } },
+    { INDEX_op_br, { } },
+    { INDEX_op_mov_i32, { "r", "r" } },
+    { INDEX_op_movi_i32, { "r" } },
+    { INDEX_op_ld8u_i32, { "r", "r" } },
+    { INDEX_op_ld8s_i32, { "r", "r" } },
+    { INDEX_op_ld16u_i32, { "r", "r" } },
+    { INDEX_op_ld16s_i32, { "r", "r" } },
+    { INDEX_op_ld_i32, { "r", "r" } },
+    { INDEX_op_st8_i32, { "q", "r" } },
+    { INDEX_op_st16_i32, { "r", "r" } },
+    { INDEX_op_st_i32, { "r", "r" } },
+
+    { INDEX_op_add_i32, { "r", "0", "ri" } },
+    { INDEX_op_sub_i32, { "r", "0", "ri" } },
+    { INDEX_op_mul_i32, { "r", "0", "ri" } },
+    { INDEX_op_mulu2_i32, { "a", "d", "a", "r" } },
+    { INDEX_op_div2_i32, { "a", "d", "0", "1", "r" } },
+    { INDEX_op_divu2_i32, { "a", "d", "0", "1", "r" } },
+    { INDEX_op_and_i32, { "r", "0", "ri" } },
+    { INDEX_op_or_i32, { "r", "0", "ri" } },
+    { INDEX_op_xor_i32, { "r", "0", "ri" } },
+
+    { INDEX_op_shl_i32, { "r", "0", "ci" } },
+    { INDEX_op_shr_i32, { "r", "0", "ci" } },
+    { INDEX_op_sar_i32, { "r", "0", "ci" } },
+
+    { INDEX_op_brcond_i32, { "r", "ri" } },
+
+    { INDEX_op_add2_i32, { "r", "r", "0", "1", "ri", "ri" } },
+    { INDEX_op_sub2_i32, { "r", "r", "0", "1", "ri", "ri" } },
+    { INDEX_op_brcond2_i32, { "r", "r", "ri", "ri" } },
+
+#if TARGET_LONG_BITS == 32
+    { INDEX_op_qemu_ld8u, { "r", "L" } },
+    { INDEX_op_qemu_ld8s, { "r", "L" } },
+    { INDEX_op_qemu_ld16u, { "r", "L" } },
+    { INDEX_op_qemu_ld16s, { "r", "L" } },
+    { INDEX_op_qemu_ld32u, { "r", "L" } },
+    { INDEX_op_qemu_ld64, { "r", "r", "L" } },
+
+    { INDEX_op_qemu_st8, { "cb", "L" } },
+    { INDEX_op_qemu_st16, { "L", "L" } },
+    { INDEX_op_qemu_st32, { "L", "L" } },
+    { INDEX_op_qemu_st64, { "L", "L", "L" } },
+#else
+    { INDEX_op_qemu_ld8u, { "r", "L", "L" } },
+    { INDEX_op_qemu_ld8s, { "r", "L", "L" } },
+    { INDEX_op_qemu_ld16u, { "r", "L", "L" } },
+    { INDEX_op_qemu_ld16s, { "r", "L", "L" } },
+    { INDEX_op_qemu_ld32u, { "r", "L", "L" } },
+    { INDEX_op_qemu_ld64, { "r", "r", "L", "L" } },
+
+    { INDEX_op_qemu_st8, { "cb", "L", "L" } },
+    { INDEX_op_qemu_st16, { "L", "L", "L" } },
+    { INDEX_op_qemu_st32, { "L", "L", "L" } },
+    { INDEX_op_qemu_st64, { "L", "L", "L", "L" } },
+#endif
+    { -1 },
+};
+
+void tcg_target_init(TCGContext *s)
+{
+    /* fail safe */
+    if ((1 << CPU_TLB_ENTRY_BITS) != sizeof(CPUTLBEntry))
+        tcg_abort();
+
+    tcg_regset_set32(tcg_target_available_regs[TCG_TYPE_I32], 0, 0xff);
+    tcg_regset_set32(tcg_target_call_clobber_regs, 0,
+                     (1 << TCG_REG_EAX) | 
+                     (1 << TCG_REG_EDX) | 
+                     (1 << TCG_REG_ECX));
+    
+    tcg_regset_clear(s->reserved_regs);
+    tcg_regset_set_reg(s->reserved_regs, TCG_REG_ESP);
+
+    tcg_add_target_add_op_defs(x86_op_defs);
+}
diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h
new file mode 100644
index 0000000000..468e2d503e
--- /dev/null
+++ b/tcg/i386/tcg-target.h
@@ -0,0 +1,54 @@
+/*
+ * Tiny Code Generator for QEMU
+ *
+ * Copyright (c) 2008 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#define TCG_TARGET_I386 1
+
+#define TCG_TARGET_REG_BITS 32
+//#define TCG_TARGET_WORDS_BIGENDIAN
+
+#define TCG_TARGET_NB_REGS 8
+
+enum {
+    TCG_REG_EAX = 0,
+    TCG_REG_ECX,
+    TCG_REG_EDX,
+    TCG_REG_EBX,
+    TCG_REG_ESP,
+    TCG_REG_EBP,
+    TCG_REG_ESI,
+    TCG_REG_EDI,
+};
+
+/* used for function call generation */
+#define TCG_REG_CALL_STACK TCG_REG_ESP 
+#define TCG_TARGET_STACK_ALIGN 16
+
+/* Note: must be synced with dyngen-exec.h */
+#define TCG_AREG0 TCG_REG_EBP
+#define TCG_AREG1 TCG_REG_EBX
+#define TCG_AREG2 TCG_REG_ESI
+#define TCG_AREG3 TCG_REG_EDI
+
+static inline void flush_icache_range(unsigned long start, unsigned long stop)
+{
+}
diff --git a/tcg/tcg-dyngen.c b/tcg/tcg-dyngen.c
new file mode 100644
index 0000000000..1cc084e6ef
--- /dev/null
+++ b/tcg/tcg-dyngen.c
@@ -0,0 +1,482 @@
+/*
+ * Tiny Code Generator for QEMU
+ *
+ * Copyright (c) 2008 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#include <assert.h>
+#include <stdarg.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <inttypes.h>
+
+#include "config.h"
+#include "osdep.h"
+
+#include "tcg.h"
+
+int __op_param1, __op_param2, __op_param3;
+#if defined(__sparc__) || defined(__arm__)
+  void __op_gen_label1(){}
+  void __op_gen_label2(){}
+  void __op_gen_label3(){}
+#else
+  int __op_gen_label1, __op_gen_label2, __op_gen_label3;
+#endif
+int __op_jmp0, __op_jmp1, __op_jmp2, __op_jmp3;
+
+#if 0
+#if defined(__s390__)
+static inline void flush_icache_range(unsigned long start, unsigned long stop)
+{
+}
+#elif defined(__ia64__)
+static inline void flush_icache_range(unsigned long start, unsigned long stop)
+{
+    while (start < stop) {
+	asm volatile ("fc %0" :: "r"(start));
+	start += 32;
+    }
+    asm volatile (";;sync.i;;srlz.i;;");
+}
+#elif defined(__powerpc__)
+
+#define MIN_CACHE_LINE_SIZE 8 /* conservative value */
+
+static inline void flush_icache_range(unsigned long start, unsigned long stop)
+{
+    unsigned long p;
+
+    start &= ~(MIN_CACHE_LINE_SIZE - 1);
+    stop = (stop + MIN_CACHE_LINE_SIZE - 1) & ~(MIN_CACHE_LINE_SIZE - 1);
+
+    for (p = start; p < stop; p += MIN_CACHE_LINE_SIZE) {
+        asm volatile ("dcbst 0,%0" : : "r"(p) : "memory");
+    }
+    asm volatile ("sync" : : : "memory");
+    for (p = start; p < stop; p += MIN_CACHE_LINE_SIZE) {
+        asm volatile ("icbi 0,%0" : : "r"(p) : "memory");
+    }
+    asm volatile ("sync" : : : "memory");
+    asm volatile ("isync" : : : "memory");
+}
+#elif defined(__alpha__)
+static inline void flush_icache_range(unsigned long start, unsigned long stop)
+{
+    asm ("imb");
+}
+#elif defined(__sparc__)
+static inline void flush_icache_range(unsigned long start, unsigned long stop)
+{
+	unsigned long p;
+
+	p = start & ~(8UL - 1UL);
+	stop = (stop + (8UL - 1UL)) & ~(8UL - 1UL);
+
+	for (; p < stop; p += 8)
+		__asm__ __volatile__("flush\t%0" : : "r" (p));
+}
+#elif defined(__arm__)
+static inline void flush_icache_range(unsigned long start, unsigned long stop)
+{
+    register unsigned long _beg __asm ("a1") = start;
+    register unsigned long _end __asm ("a2") = stop;
+    register unsigned long _flg __asm ("a3") = 0;
+    __asm __volatile__ ("swi 0x9f0002" : : "r" (_beg), "r" (_end), "r" (_flg));
+}
+#elif defined(__mc68000)
+
+# include <asm/cachectl.h>
+static inline void flush_icache_range(unsigned long start, unsigned long stop)
+{
+    cacheflush(start,FLUSH_SCOPE_LINE,FLUSH_CACHE_BOTH,stop-start+16);
+}
+#elif defined(__mips__)
+
+#include <sys/cachectl.h>
+static inline void flush_icache_range(unsigned long start, unsigned long stop)
+{
+    _flush_cache ((void *)start, stop - start, BCACHE);
+}
+#else
+#error unsupported CPU
+#endif
+
+#ifdef __alpha__
+
+register int gp asm("$29");
+
+static inline void immediate_ldah(void *p, int val) {
+    uint32_t *dest = p;
+    long high = ((val >> 16) + ((val >> 15) & 1)) & 0xffff;
+
+    *dest &= ~0xffff;
+    *dest |= high;
+    *dest |= 31 << 16;
+}
+static inline void immediate_lda(void *dest, int val) {
+    *(uint16_t *) dest = val;
+}
+void fix_bsr(void *p, int offset) {
+    uint32_t *dest = p;
+    *dest &= ~((1 << 21) - 1);
+    *dest |= (offset >> 2) & ((1 << 21) - 1);
+}
+
+#endif /* __alpha__ */
+
+#ifdef __arm__
+
+#define ARM_LDR_TABLE_SIZE 1024
+
+typedef struct LDREntry {
+    uint8_t *ptr;
+    uint32_t *data_ptr;
+    unsigned type:2;
+} LDREntry;
+
+static LDREntry arm_ldr_table[1024];
+static uint32_t arm_data_table[ARM_LDR_TABLE_SIZE];
+
+extern char exec_loop;
+
+static inline void arm_reloc_pc24(uint32_t *ptr, uint32_t insn, int val)
+{
+    *ptr = (insn & ~0xffffff) | ((insn + ((val - (int)ptr) >> 2)) & 0xffffff);
+}
+
+static uint8_t *arm_flush_ldr(uint8_t *gen_code_ptr,
+                              LDREntry *ldr_start, LDREntry *ldr_end,
+                              uint32_t *data_start, uint32_t *data_end,
+                              int gen_jmp)
+{
+    LDREntry *le;
+    uint32_t *ptr;
+    int offset, data_size, target;
+    uint8_t *data_ptr;
+    uint32_t insn;
+    uint32_t mask;
+
+    data_size = (data_end - data_start) << 2;
+
+    if (gen_jmp) {
+        /* generate branch to skip the data */
+        if (data_size == 0)
+            return gen_code_ptr;
+        target = (long)gen_code_ptr + data_size + 4;
+        arm_reloc_pc24((uint32_t *)gen_code_ptr, 0xeafffffe, target);
+        gen_code_ptr += 4;
+    }
+
+    /* copy the data */
+    data_ptr = gen_code_ptr;
+    memcpy(gen_code_ptr, data_start, data_size);
+    gen_code_ptr += data_size;
+
+    /* patch the ldr to point to the data */
+    for(le = ldr_start; le < ldr_end; le++) {
+        ptr = (uint32_t *)le->ptr;
+        offset = ((unsigned long)(le->data_ptr) - (unsigned long)data_start) +
+            (unsigned long)data_ptr -
+            (unsigned long)ptr - 8;
+        if (offset < 0) {
+            fprintf(stderr, "Negative constant pool offset\n");
+            tcg_abort();
+        }
+        switch (le->type) {
+          case 0: /* ldr */
+            mask = ~0x00800fff;
+            if (offset >= 4096) {
+                fprintf(stderr, "Bad ldr offset\n");
+                tcg_abort();
+            }
+            break;
+          case 1: /* ldc */
+            mask = ~0x008000ff;
+            if (offset >= 1024 ) {
+                fprintf(stderr, "Bad ldc offset\n");
+                tcg_abort();
+            }
+            break;
+          case 2: /* add */
+            mask = ~0xfff;
+            if (offset >= 1024 ) {
+                fprintf(stderr, "Bad add offset\n");
+                tcg_abort();
+            }
+            break;
+          default:
+            fprintf(stderr, "Bad pc relative fixup\n");
+            tcg_abort();
+          }
+        insn = *ptr & mask;
+        switch (le->type) {
+          case 0: /* ldr */
+            insn |= offset | 0x00800000;
+            break;
+          case 1: /* ldc */
+            insn |= (offset >> 2) | 0x00800000;
+            break;
+          case 2: /* add */
+            insn |= (offset >> 2) | 0xf00;
+            break;
+          }
+        *ptr = insn;
+    }
+    return gen_code_ptr;
+}
+
+#endif /* __arm__ */
+
+#ifdef __ia64
+
+/* Patch instruction with "val" where "mask" has 1 bits. */
+static inline void ia64_patch (uint64_t insn_addr, uint64_t mask, uint64_t val)
+{
+    uint64_t m0, m1, v0, v1, b0, b1, *b = (uint64_t *) (insn_addr & -16);
+#   define insn_mask ((1UL << 41) - 1)
+    unsigned long shift;
+
+    b0 = b[0]; b1 = b[1];
+    shift = 5 + 41 * (insn_addr % 16); /* 5 template, 3 x 41-bit insns */
+    if (shift >= 64) {
+	m1 = mask << (shift - 64);
+	v1 = val << (shift - 64);
+    } else {
+	m0 = mask << shift; m1 = mask >> (64 - shift);
+	v0 = val  << shift; v1 = val >> (64 - shift);
+	b[0] = (b0 & ~m0) | (v0 & m0);
+    }
+    b[1] = (b1 & ~m1) | (v1 & m1);
+}
+
+static inline void ia64_patch_imm60 (uint64_t insn_addr, uint64_t val)
+{
+	ia64_patch(insn_addr,
+		   0x011ffffe000UL,
+		   (  ((val & 0x0800000000000000UL) >> 23) /* bit 59 -> 36 */
+		    | ((val & 0x00000000000fffffUL) << 13) /* bit 0 -> 13 */));
+	ia64_patch(insn_addr - 1, 0x1fffffffffcUL, val >> 18);
+}
+
+static inline void ia64_imm64 (void *insn, uint64_t val)
+{
+    /* Ignore the slot number of the relocation; GCC and Intel
+       toolchains differed for some time on whether IMM64 relocs are
+       against slot 1 (Intel) or slot 2 (GCC).  */
+    uint64_t insn_addr = (uint64_t) insn & ~3UL;
+
+    ia64_patch(insn_addr + 2,
+	       0x01fffefe000UL,
+	       (  ((val & 0x8000000000000000UL) >> 27) /* bit 63 -> 36 */
+		| ((val & 0x0000000000200000UL) <<  0) /* bit 21 -> 21 */
+		| ((val & 0x00000000001f0000UL) <<  6) /* bit 16 -> 22 */
+		| ((val & 0x000000000000ff80UL) << 20) /* bit  7 -> 27 */
+		| ((val & 0x000000000000007fUL) << 13) /* bit  0 -> 13 */)
+	    );
+    ia64_patch(insn_addr + 1, 0x1ffffffffffUL, val >> 22);
+}
+
+static inline void ia64_imm60b (void *insn, uint64_t val)
+{
+    /* Ignore the slot number of the relocation; GCC and Intel
+       toolchains differed for some time on whether IMM64 relocs are
+       against slot 1 (Intel) or slot 2 (GCC).  */
+    uint64_t insn_addr = (uint64_t) insn & ~3UL;
+
+    if (val + ((uint64_t) 1 << 59) >= (1UL << 60))
+	fprintf(stderr, "%s: value %ld out of IMM60 range\n",
+		__FUNCTION__, (int64_t) val);
+    ia64_patch_imm60(insn_addr + 2, val);
+}
+
+static inline void ia64_imm22 (void *insn, uint64_t val)
+{
+    if (val + (1 << 21) >= (1 << 22))
+	fprintf(stderr, "%s: value %li out of IMM22 range\n",
+		__FUNCTION__, (int64_t)val);
+    ia64_patch((uint64_t) insn, 0x01fffcfe000UL,
+	       (  ((val & 0x200000UL) << 15) /* bit 21 -> 36 */
+		| ((val & 0x1f0000UL) <<  6) /* bit 16 -> 22 */
+		| ((val & 0x00ff80UL) << 20) /* bit  7 -> 27 */
+		| ((val & 0x00007fUL) << 13) /* bit  0 -> 13 */));
+}
+
+/* Like ia64_imm22(), but also clear bits 20-21.  For addl, this has
+   the effect of turning "addl rX=imm22,rY" into "addl
+   rX=imm22,r0".  */
+static inline void ia64_imm22_r0 (void *insn, uint64_t val)
+{
+    if (val + (1 << 21) >= (1 << 22))
+	fprintf(stderr, "%s: value %li out of IMM22 range\n",
+		__FUNCTION__, (int64_t)val);
+    ia64_patch((uint64_t) insn, 0x01fffcfe000UL | (0x3UL << 20),
+	       (  ((val & 0x200000UL) << 15) /* bit 21 -> 36 */
+		| ((val & 0x1f0000UL) <<  6) /* bit 16 -> 22 */
+		| ((val & 0x00ff80UL) << 20) /* bit  7 -> 27 */
+		| ((val & 0x00007fUL) << 13) /* bit  0 -> 13 */));
+}
+
+static inline void ia64_imm21b (void *insn, uint64_t val)
+{
+    if (val + (1 << 20) >= (1 << 21))
+	fprintf(stderr, "%s: value %li out of IMM21b range\n",
+		__FUNCTION__, (int64_t)val);
+    ia64_patch((uint64_t) insn, 0x11ffffe000UL,
+	       (  ((val & 0x100000UL) << 16) /* bit 20 -> 36 */
+		| ((val & 0x0fffffUL) << 13) /* bit  0 -> 13 */));
+}
+
+static inline void ia64_nop_b (void *insn)
+{
+    ia64_patch((uint64_t) insn, (1UL << 41) - 1, 2UL << 37);
+}
+
+static inline void ia64_ldxmov(void *insn, uint64_t val)
+{
+    if (val + (1 << 21) < (1 << 22))
+	ia64_patch((uint64_t) insn, 0x1fff80fe000UL, 8UL << 37);
+}
+
+static inline int ia64_patch_ltoff(void *insn, uint64_t val,
+				   int relaxable)
+{
+    if (relaxable && (val + (1 << 21) < (1 << 22))) {
+	ia64_imm22_r0(insn, val);
+	return 0;
+    }
+    return 1;
+}
+
+struct ia64_fixup {
+    struct ia64_fixup *next;
+    void *addr;			/* address that needs to be patched */
+    long value;
+};
+
+#define IA64_PLT(insn, plt_index)			\
+do {							\
+    struct ia64_fixup *fixup = alloca(sizeof(*fixup));	\
+    fixup->next = plt_fixes;				\
+    plt_fixes = fixup;					\
+    fixup->addr = (insn);				\
+    fixup->value = (plt_index);				\
+    plt_offset[(plt_index)] = 1;			\
+} while (0)
+
+#define IA64_LTOFF(insn, val, relaxable)			\
+do {								\
+    if (ia64_patch_ltoff(insn, val, relaxable)) {		\
+	struct ia64_fixup *fixup = alloca(sizeof(*fixup));	\
+	fixup->next = ltoff_fixes;				\
+	ltoff_fixes = fixup;					\
+	fixup->addr = (insn);					\
+	fixup->value = (val);					\
+    }								\
+} while (0)
+
+static inline void ia64_apply_fixes (uint8_t **gen_code_pp,
+				     struct ia64_fixup *ltoff_fixes,
+				     uint64_t gp,
+				     struct ia64_fixup *plt_fixes,
+				     int num_plts,
+				     unsigned long *plt_target,
+				     unsigned int *plt_offset)
+{
+    static const uint8_t plt_bundle[] = {
+	0x04, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,	/* nop 0; movl r1=GP */
+	0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x00, 0x60,
+
+	0x05, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,	/* nop 0; brl IP */
+	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc0
+    };
+    uint8_t *gen_code_ptr = *gen_code_pp, *plt_start, *got_start;
+    uint64_t *vp;
+    struct ia64_fixup *fixup;
+    unsigned int offset = 0;
+    struct fdesc {
+	long ip;
+	long gp;
+    } *fdesc;
+    int i;
+
+    if (plt_fixes) {
+	plt_start = gen_code_ptr;
+
+	for (i = 0; i < num_plts; ++i) {
+	    if (plt_offset[i]) {
+		plt_offset[i] = offset;
+		offset += sizeof(plt_bundle);
+
+		fdesc = (struct fdesc *) plt_target[i];
+		memcpy(gen_code_ptr, plt_bundle, sizeof(plt_bundle));
+		ia64_imm64 (gen_code_ptr + 0x02, fdesc->gp);
+		ia64_imm60b(gen_code_ptr + 0x12,
+			    (fdesc->ip - (long) (gen_code_ptr + 0x10)) >> 4);
+		gen_code_ptr += sizeof(plt_bundle);
+	    }
+	}
+
+	for (fixup = plt_fixes; fixup; fixup = fixup->next)
+	    ia64_imm21b(fixup->addr,
+			((long) plt_start + plt_offset[fixup->value]
+			 - ((long) fixup->addr & ~0xf)) >> 4);
+    }
+
+    got_start = gen_code_ptr;
+
+    /* First, create the GOT: */
+    for (fixup = ltoff_fixes; fixup; fixup = fixup->next) {
+	/* first check if we already have this value in the GOT: */
+	for (vp = (uint64_t *) got_start; vp < (uint64_t *) gen_code_ptr; ++vp)
+	    if (*vp == fixup->value)
+		break;
+	if (vp == (uint64_t *) gen_code_ptr) {
+	    /* Nope, we need to put the value in the GOT: */
+	    *vp = fixup->value;
+	    gen_code_ptr += 8;
+	}
+	ia64_imm22(fixup->addr, (long) vp - gp);
+    }
+    /* Keep code ptr aligned. */
+    if ((long) gen_code_ptr & 15)
+	gen_code_ptr += 8;
+    *gen_code_pp = gen_code_ptr;
+}
+#endif
+#endif
+
+const TCGArg *dyngen_op(TCGContext *s, int opc, const TCGArg *opparam_ptr)
+{
+    uint8_t *gen_code_ptr;
+
+    gen_code_ptr = s->code_ptr;
+    switch(opc) {
+
+/* op.h is dynamically generated by dyngen.c from op.c */
+#include "op.h"
+
+    default:
+        tcg_abort();
+    }
+    s->code_ptr = gen_code_ptr;
+    return opparam_ptr;
+}
diff --git a/tcg/tcg-op.h b/tcg/tcg-op.h
new file mode 100644
index 0000000000..705e0df971
--- /dev/null
+++ b/tcg/tcg-op.h
@@ -0,0 +1,1175 @@
+/*
+ * Tiny Code Generator for QEMU
+ *
+ * Copyright (c) 2008 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#include "tcg.h"
+
+/* legacy dyngen operations */
+#include "gen-op.h"
+
+int gen_new_label(void);
+
+static inline void tcg_gen_op1(int opc, TCGArg arg1)
+{
+    *gen_opc_ptr++ = opc;
+    *gen_opparam_ptr++ = arg1;
+}
+
+static inline void tcg_gen_op2(int opc, TCGArg arg1, TCGArg arg2)
+{
+    *gen_opc_ptr++ = opc;
+    *gen_opparam_ptr++ = arg1;
+    *gen_opparam_ptr++ = arg2;
+}
+
+static inline void tcg_gen_op3(int opc, TCGArg arg1, TCGArg arg2, TCGArg arg3)
+{
+    *gen_opc_ptr++ = opc;
+    *gen_opparam_ptr++ = arg1;
+    *gen_opparam_ptr++ = arg2;
+    *gen_opparam_ptr++ = arg3;
+}
+
+static inline void tcg_gen_op4(int opc, TCGArg arg1, TCGArg arg2, TCGArg arg3, 
+                               TCGArg arg4)
+{
+    *gen_opc_ptr++ = opc;
+    *gen_opparam_ptr++ = arg1;
+    *gen_opparam_ptr++ = arg2;
+    *gen_opparam_ptr++ = arg3;
+    *gen_opparam_ptr++ = arg4;
+}
+
+static inline void tcg_gen_op5(int opc, TCGArg arg1, TCGArg arg2, 
+                               TCGArg arg3, TCGArg arg4,
+                               TCGArg arg5)
+{
+    *gen_opc_ptr++ = opc;
+    *gen_opparam_ptr++ = arg1;
+    *gen_opparam_ptr++ = arg2;
+    *gen_opparam_ptr++ = arg3;
+    *gen_opparam_ptr++ = arg4;
+    *gen_opparam_ptr++ = arg5;
+}
+
+static inline void tcg_gen_op6(int opc, TCGArg arg1, TCGArg arg2, 
+                               TCGArg arg3, TCGArg arg4,
+                               TCGArg arg5, TCGArg arg6)
+{
+    *gen_opc_ptr++ = opc;
+    *gen_opparam_ptr++ = arg1;
+    *gen_opparam_ptr++ = arg2;
+    *gen_opparam_ptr++ = arg3;
+    *gen_opparam_ptr++ = arg4;
+    *gen_opparam_ptr++ = arg5;
+    *gen_opparam_ptr++ = arg6;
+}
+
+static inline void gen_set_label(int n)
+{
+    tcg_gen_op1(INDEX_op_set_label, n);
+}
+
+static inline void tcg_gen_mov_i32(int ret, int arg)
+{
+    tcg_gen_op2(INDEX_op_mov_i32, ret, arg);
+}
+
+static inline void tcg_gen_movi_i32(int ret, int32_t arg)
+{
+    tcg_gen_op2(INDEX_op_movi_i32, ret, arg);
+}
+
+/* helper calls */
+#define TCG_HELPER_CALL_FLAGS 0
+
+static inline void tcg_gen_helper_0_0(void *func)
+{
+    tcg_gen_call(&tcg_ctx, 
+                 tcg_const_ptr((tcg_target_long)func), TCG_HELPER_CALL_FLAGS,
+                 0, NULL, 0, NULL);
+}
+
+static inline void tcg_gen_helper_0_1(void *func, TCGArg arg)
+{
+    tcg_gen_call(&tcg_ctx,
+                 tcg_const_ptr((tcg_target_long)func), TCG_HELPER_CALL_FLAGS,
+                 0, NULL, 1, &arg);
+}
+
+static inline void tcg_gen_helper_0_2(void *func, TCGArg arg1, TCGArg arg2)
+{
+    TCGArg args[2];
+    args[0] = arg1;
+    args[1] = arg2;
+    tcg_gen_call(&tcg_ctx, 
+                 tcg_const_ptr((tcg_target_long)func), TCG_HELPER_CALL_FLAGS,
+                 0, NULL, 2, args);
+}
+
+static inline void tcg_gen_helper_1_2(void *func, TCGArg ret, 
+                                      TCGArg arg1, TCGArg arg2)
+{
+    TCGArg args[2];
+    args[0] = arg1;
+    args[1] = arg2;
+    tcg_gen_call(&tcg_ctx, 
+                 tcg_const_ptr((tcg_target_long)func), TCG_HELPER_CALL_FLAGS,
+                 1, &ret, 2, args);
+}
+
+/* 32 bit ops */
+
+static inline void tcg_gen_ld8u_i32(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_ld8u_i32, ret, arg2, offset);
+}
+
+static inline void tcg_gen_ld8s_i32(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_ld8s_i32, ret, arg2, offset);
+}
+
+static inline void tcg_gen_ld16u_i32(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_ld16u_i32, ret, arg2, offset);
+}
+
+static inline void tcg_gen_ld16s_i32(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_ld16s_i32, ret, arg2, offset);
+}
+
+static inline void tcg_gen_ld_i32(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_ld_i32, ret, arg2, offset);
+}
+
+static inline void tcg_gen_st8_i32(int arg1, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_st8_i32, arg1, arg2, offset);
+}
+
+static inline void tcg_gen_st16_i32(int arg1, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_st16_i32, arg1, arg2, offset);
+}
+
+static inline void tcg_gen_st_i32(int arg1, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_st_i32, arg1, arg2, offset);
+}
+
+static inline void tcg_gen_add_i32(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_add_i32, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_addi_i32(int ret, int arg1, int32_t arg2)
+{
+    tcg_gen_add_i32(ret, arg1, tcg_const_i32(arg2));
+}
+
+static inline void tcg_gen_sub_i32(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_sub_i32, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_subi_i32(int ret, int arg1, int32_t arg2)
+{
+    tcg_gen_sub_i32(ret, arg1, tcg_const_i32(arg2));
+}
+
+static inline void tcg_gen_and_i32(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_and_i32, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_andi_i32(int ret, int arg1, int32_t arg2)
+{
+    /* some cases can be optimized here */
+    if (arg2 == 0) {
+        tcg_gen_movi_i32(ret, 0);
+    } else if (arg2 == 0xffffffff) {
+        tcg_gen_mov_i32(ret, arg1);
+    } else {
+        tcg_gen_and_i32(ret, arg1, tcg_const_i32(arg2));
+    }
+}
+
+static inline void tcg_gen_or_i32(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_or_i32, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_ori_i32(int ret, int arg1, int32_t arg2)
+{
+    /* some cases can be optimized here */
+    if (arg2 == 0xffffffff) {
+        tcg_gen_movi_i32(ret, 0);
+    } else if (arg2 == 0) {
+        tcg_gen_mov_i32(ret, arg1);
+    } else {
+        tcg_gen_or_i32(ret, arg1, tcg_const_i32(arg2));
+    }
+}
+
+static inline void tcg_gen_xor_i32(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_xor_i32, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_xori_i32(int ret, int arg1, int32_t arg2)
+{
+    /* some cases can be optimized here */
+    if (arg2 == 0) {
+        tcg_gen_mov_i32(ret, arg1);
+    } else {
+        tcg_gen_xor_i32(ret, arg1, tcg_const_i32(arg2));
+    }
+}
+
+static inline void tcg_gen_shl_i32(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_shl_i32, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_shli_i32(int ret, int arg1, int32_t arg2)
+{
+    tcg_gen_shl_i32(ret, arg1, tcg_const_i32(arg2));
+}
+
+static inline void tcg_gen_shr_i32(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_shr_i32, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_shri_i32(int ret, int arg1, int32_t arg2)
+{
+    tcg_gen_shr_i32(ret, arg1, tcg_const_i32(arg2));
+}
+
+static inline void tcg_gen_sar_i32(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_sar_i32, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_sari_i32(int ret, int arg1, int32_t arg2)
+{
+    tcg_gen_sar_i32(ret, arg1, tcg_const_i32(arg2));
+}
+
+static inline void tcg_gen_brcond_i32(int cond, TCGArg arg1, TCGArg arg2, 
+                                      int label_index)
+{
+    tcg_gen_op4(INDEX_op_brcond_i32, arg1, arg2, cond, label_index);
+}
+
+static inline void tcg_gen_mul_i32(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_mul_i32, ret, arg1, arg2);
+}
+
+#ifdef TCG_TARGET_HAS_div_i32
+static inline void tcg_gen_div_i32(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_div_i32, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_rem_i32(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_rem_i32, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_divu_i32(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_divu_i32, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_remu_i32(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_remu_i32, ret, arg1, arg2);
+}
+#else
+static inline void tcg_gen_div_i32(int ret, int arg1, int arg2)
+{
+    int t0;
+    t0 = tcg_temp_new(TCG_TYPE_I32);
+    tcg_gen_sari_i32(t0, arg1, 31);
+    tcg_gen_op5(INDEX_op_div2_i32, ret, t0, arg1, t0, arg2);
+}
+
+static inline void tcg_gen_rem_i32(int ret, int arg1, int arg2)
+{
+    int t0;
+    t0 = tcg_temp_new(TCG_TYPE_I32);
+    tcg_gen_sari_i32(t0, arg1, 31);
+    tcg_gen_op5(INDEX_op_div2_i32, t0, ret, arg1, t0, arg2);
+}
+
+static inline void tcg_gen_divu_i32(int ret, int arg1, int arg2)
+{
+    int t0;
+    t0 = tcg_temp_new(TCG_TYPE_I32);
+    tcg_gen_movi_i32(t0, 0);
+    tcg_gen_op5(INDEX_op_divu2_i32, ret, t0, arg1, t0, arg2);
+}
+
+static inline void tcg_gen_remu_i32(int ret, int arg1, int arg2)
+{
+    int t0;
+    t0 = tcg_temp_new(TCG_TYPE_I32);
+    tcg_gen_movi_i32(t0, 0);
+    tcg_gen_op5(INDEX_op_divu2_i32, t0, ret, arg1, t0, arg2);
+}
+#endif
+
+#if TCG_TARGET_REG_BITS == 32
+
+static inline void tcg_gen_mov_i64(int ret, int arg)
+{
+    tcg_gen_mov_i32(ret, arg);
+    tcg_gen_mov_i32(ret + 1, arg + 1);
+}
+
+static inline void tcg_gen_movi_i64(int ret, int64_t arg)
+{
+    tcg_gen_movi_i32(ret, arg);
+    tcg_gen_movi_i32(ret + 1, arg >> 32);
+}
+
+static inline void tcg_gen_ld8u_i64(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_ld8u_i32(ret, arg2, offset);
+    tcg_gen_movi_i32(ret + 1, 0);
+}
+
+static inline void tcg_gen_ld8s_i64(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_ld8s_i32(ret, arg2, offset);
+    tcg_gen_sari_i32(ret + 1, ret, 31);
+}
+
+static inline void tcg_gen_ld16u_i64(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_ld16u_i32(ret, arg2, offset);
+    tcg_gen_movi_i32(ret + 1, 0);
+}
+
+static inline void tcg_gen_ld16s_i64(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_ld16s_i32(ret, arg2, offset);
+    tcg_gen_sari_i32(ret + 1, ret, 31);
+}
+
+static inline void tcg_gen_ld32u_i64(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_ld_i32(ret, arg2, offset);
+    tcg_gen_movi_i32(ret + 1, 0);
+}
+
+static inline void tcg_gen_ld32s_i64(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_ld_i32(ret, arg2, offset);
+    tcg_gen_sari_i32(ret + 1, ret, 31);
+}
+
+static inline void tcg_gen_ld_i64(int ret, int arg2, tcg_target_long offset)
+{
+    /* since arg2 and ret have different types, they cannot be the
+       same temporary */
+#ifdef TCG_TARGET_WORDS_BIGENDIAN
+    tcg_gen_ld_i32(ret + 1, arg2, offset);
+    tcg_gen_ld_i32(ret, arg2, offset + 4);
+#else
+    tcg_gen_ld_i32(ret, arg2, offset);
+    tcg_gen_ld_i32(ret + 1, arg2, offset + 4);
+#endif
+}
+
+static inline void tcg_gen_st8_i64(int arg1, int arg2, tcg_target_long offset)
+{
+    tcg_gen_st8_i32(arg1, arg2, offset);
+}
+
+static inline void tcg_gen_st16_i64(int arg1, int arg2, tcg_target_long offset)
+{
+    tcg_gen_st16_i32(arg1, arg2, offset);
+}
+
+static inline void tcg_gen_st32_i64(int arg1, int arg2, tcg_target_long offset)
+{
+    tcg_gen_st_i32(arg1, arg2, offset);
+}
+
+static inline void tcg_gen_st_i64(int arg1, int arg2, tcg_target_long offset)
+{
+#ifdef TCG_TARGET_WORDS_BIGENDIAN
+    tcg_gen_st_i32(arg1 + 1, arg2, offset);
+    tcg_gen_st_i32(arg1, arg2, offset + 4);
+#else
+    tcg_gen_st_i32(arg1, arg2, offset);
+    tcg_gen_st_i32(arg1 + 1, arg2, offset + 4);
+#endif
+}
+
+static inline void tcg_gen_add_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op6(INDEX_op_add2_i32, ret, ret + 1, 
+                arg1, arg1 + 1, arg2, arg2 + 1);
+}
+
+static inline void tcg_gen_addi_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_add_i64(ret, arg1, tcg_const_i64(arg2));
+}
+
+static inline void tcg_gen_sub_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op6(INDEX_op_sub2_i32, ret, ret + 1, 
+                arg1, arg1 + 1, arg2, arg2 + 1);
+}
+
+static inline void tcg_gen_subi_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_sub_i64(ret, arg1, tcg_const_i64(arg2));
+}
+
+static inline void tcg_gen_and_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_and_i32(ret, arg1, arg2);
+    tcg_gen_and_i32(ret + 1, arg1 + 1, arg2 + 1);
+}
+
+static inline void tcg_gen_andi_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_andi_i32(ret, arg1, arg2);
+    tcg_gen_andi_i32(ret + 1, arg1 + 1, arg2 >> 32);
+}
+
+static inline void tcg_gen_or_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_or_i32(ret, arg1, arg2);
+    tcg_gen_or_i32(ret + 1, arg1 + 1, arg2 + 1);
+}
+
+static inline void tcg_gen_ori_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_ori_i32(ret, arg1, arg2);
+    tcg_gen_ori_i32(ret + 1, arg1 + 1, arg2 >> 32);
+}
+
+static inline void tcg_gen_xor_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_xor_i32(ret, arg1, arg2);
+    tcg_gen_xor_i32(ret + 1, arg1 + 1, arg2 + 1);
+}
+
+static inline void tcg_gen_xori_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_xori_i32(ret, arg1, arg2);
+    tcg_gen_xori_i32(ret + 1, arg1 + 1, arg2 >> 32);
+}
+
+/* XXX: use generic code when basic block handling is OK or CPU
+   specific code (x86) */
+static inline void tcg_gen_shl_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_helper_1_2(tcg_helper_shl_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_shli_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_shifti_i64(ret, arg1, arg2, 0, 0);
+}
+
+static inline void tcg_gen_shr_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_helper_1_2(tcg_helper_shr_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_shri_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_shifti_i64(ret, arg1, arg2, 1, 0);
+}
+
+static inline void tcg_gen_sar_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_helper_1_2(tcg_helper_sar_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_sari_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_shifti_i64(ret, arg1, arg2, 1, 1);
+}
+
+static inline void tcg_gen_brcond_i64(int cond, TCGArg arg1, TCGArg arg2, 
+                                      int label_index)
+{
+    tcg_gen_op6(INDEX_op_brcond2_i32, 
+                arg1, arg1 + 1, arg2, arg2 + 1, cond, label_index);
+}
+
+static inline void tcg_gen_mul_i64(int ret, int arg1, int arg2)
+{
+    int t0, t1;
+    
+    t0 = tcg_temp_new(TCG_TYPE_I64);
+    t1 = tcg_temp_new(TCG_TYPE_I32);
+
+    tcg_gen_op4(INDEX_op_mulu2_i32, t0, t0 + 1, arg1, arg2);
+    
+    tcg_gen_mul_i32(t1, arg1, arg2 + 1);
+    tcg_gen_add_i32(t0 + 1, t0 + 1, t1);
+    tcg_gen_mul_i32(t1, arg1 + 1, arg2);
+    tcg_gen_add_i32(t0 + 1, t0 + 1, t1);
+    
+    tcg_gen_mov_i64(ret, t0);
+}
+
+static inline void tcg_gen_div_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_helper_1_2(tcg_helper_div_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_rem_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_helper_1_2(tcg_helper_rem_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_divu_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_helper_1_2(tcg_helper_divu_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_remu_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_helper_1_2(tcg_helper_remu_i64, ret, arg1, arg2);
+}
+
+#else
+
+static inline void tcg_gen_mov_i64(int ret, int arg)
+{
+    tcg_gen_op2(INDEX_op_mov_i64, ret, arg);
+}
+
+static inline void tcg_gen_movi_i64(int ret, int64_t arg)
+{
+    tcg_gen_op2(INDEX_op_movi_i64, ret, arg);
+}
+
+static inline void tcg_gen_ld8u_i64(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_ld8u_i64, ret, arg2, offset);
+}
+
+static inline void tcg_gen_ld8s_i64(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_ld8s_i64, ret, arg2, offset);
+}
+
+static inline void tcg_gen_ld16u_i64(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_ld16u_i64, ret, arg2, offset);
+}
+
+static inline void tcg_gen_ld16s_i64(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_ld16s_i64, ret, arg2, offset);
+}
+
+static inline void tcg_gen_ld32u_i64(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_ld32u_i64, ret, arg2, offset);
+}
+
+static inline void tcg_gen_ld32s_i64(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_ld32s_i64, ret, arg2, offset);
+}
+
+static inline void tcg_gen_ld_i64(int ret, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_ld_i64, ret, arg2, offset);
+}
+
+static inline void tcg_gen_st8_i64(int arg1, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_st8_i64, arg1, arg2, offset);
+}
+
+static inline void tcg_gen_st16_i64(int arg1, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_st16_i64, arg1, arg2, offset);
+}
+
+static inline void tcg_gen_st32_i64(int arg1, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_st32_i64, arg1, arg2, offset);
+}
+
+static inline void tcg_gen_st_i64(int arg1, int arg2, tcg_target_long offset)
+{
+    tcg_gen_op3(INDEX_op_st_i64, arg1, arg2, offset);
+}
+
+static inline void tcg_gen_add_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_add_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_addi_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_add_i64(ret, arg1, tcg_const_i64(arg2));
+}
+
+static inline void tcg_gen_sub_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_sub_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_subi_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_sub_i64(ret, arg1, tcg_const_i64(arg2));
+}
+
+static inline void tcg_gen_and_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_and_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_andi_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_and_i64(ret, arg1, tcg_const_i64(arg2));
+}
+
+static inline void tcg_gen_or_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_or_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_ori_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_or_i64(ret, arg1, tcg_const_i64(arg2));
+}
+
+static inline void tcg_gen_xor_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_xor_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_xori_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_xor_i64(ret, arg1, tcg_const_i64(arg2));
+}
+
+static inline void tcg_gen_shl_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_shl_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_shli_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_shl_i64(ret, arg1, tcg_const_i64(arg2));
+}
+
+static inline void tcg_gen_shr_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_shr_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_shri_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_shr_i64(ret, arg1, tcg_const_i64(arg2));
+}
+
+static inline void tcg_gen_sar_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_sar_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_sari_i64(int ret, int arg1, int64_t arg2)
+{
+    tcg_gen_sar_i64(ret, arg1, tcg_const_i64(arg2));
+}
+
+static inline void tcg_gen_brcond_i64(int cond, TCGArg arg1, TCGArg arg2, 
+                                      int label_index)
+{
+    tcg_gen_op4(INDEX_op_brcond_i64, arg1, arg2, cond, label_index);
+}
+
+static inline void tcg_gen_mul_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_mul_i64, ret, arg1, arg2);
+}
+
+#ifdef TCG_TARGET_HAS_div_i64
+static inline void tcg_gen_div_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_div_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_rem_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_rem_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_divu_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_divu_i64, ret, arg1, arg2);
+}
+
+static inline void tcg_gen_remu_i64(int ret, int arg1, int arg2)
+{
+    tcg_gen_op3(INDEX_op_remu_i64, ret, arg1, arg2);
+}
+#else
+static inline void tcg_gen_div_i64(int ret, int arg1, int arg2)
+{
+    int t0;
+    t0 = tcg_temp_new(TCG_TYPE_I64);
+    tcg_gen_sari_i64(t0, arg1, 63);
+    tcg_gen_op5(INDEX_op_div2_i64, ret, t0, arg1, t0, arg2);
+}
+
+static inline void tcg_gen_rem_i64(int ret, int arg1, int arg2)
+{
+    int t0;
+    t0 = tcg_temp_new(TCG_TYPE_I64);
+    tcg_gen_sari_i64(t0, arg1, 63);
+    tcg_gen_op5(INDEX_op_div2_i64, t0, ret, arg1, t0, arg2);
+}
+
+static inline void tcg_gen_divu_i64(int ret, int arg1, int arg2)
+{
+    int t0;
+    t0 = tcg_temp_new(TCG_TYPE_I64);
+    tcg_gen_movi_i64(t0, 0);
+    tcg_gen_op5(INDEX_op_divu2_i64, ret, t0, arg1, t0, arg2);
+}
+
+static inline void tcg_gen_remu_i64(int ret, int arg1, int arg2)
+{
+    int t0;
+    t0 = tcg_temp_new(TCG_TYPE_I64);
+    tcg_gen_movi_i64(t0, 0);
+    tcg_gen_op5(INDEX_op_divu2_i64, t0, ret, arg1, t0, arg2);
+}
+#endif
+
+#endif
+
+/***************************************/
+/* optional operations */
+
+static inline void tcg_gen_ext8s_i32(int ret, int arg)
+{
+#ifdef TCG_TARGET_HAS_ext8s_i32
+    tcg_gen_op2(INDEX_op_ext8s_i32, ret, arg);
+#else
+    tcg_gen_shli_i32(ret, arg, 24);
+    tcg_gen_sari_i32(ret, arg, 24);
+#endif
+}
+
+static inline void tcg_gen_ext16s_i32(int ret, int arg)
+{
+#ifdef TCG_TARGET_HAS_ext16s_i32
+    tcg_gen_op2(INDEX_op_ext16s_i32, ret, arg);
+#else
+    tcg_gen_shli_i32(ret, arg, 16);
+    tcg_gen_sari_i32(ret, arg, 16);
+#endif
+}
+
+/* Note: we assume the two high bytes are set to zero */
+static inline void tcg_gen_bswap16_i32(TCGArg ret, TCGArg arg)
+{
+#ifdef TCG_TARGET_HAS_bswap16_i32
+    tcg_gen_op2(INDEX_op_bswap16_i32, ret, arg);
+#else
+    TCGArg t0, t1;
+    t0 = tcg_temp_new(TCG_TYPE_I32);
+    t1 = tcg_temp_new(TCG_TYPE_I32);
+    
+    tcg_gen_shri_i32(t0, arg, 8);
+    tcg_gen_andi_i32(t1, arg, 0x000000ff);
+    tcg_gen_shli_i32(t1, t1, 8);
+    tcg_gen_or_i32(ret, t0, t1);
+#endif
+}
+
+static inline void tcg_gen_bswap_i32(TCGArg ret, TCGArg arg)
+{
+#ifdef TCG_TARGET_HAS_bswap_i32
+    tcg_gen_op2(INDEX_op_bswap_i32, ret, arg);
+#else
+    TCGArg t0, t1;
+    t0 = tcg_temp_new(TCG_TYPE_I32);
+    t1 = tcg_temp_new(TCG_TYPE_I32);
+    
+    tcg_gen_shli_i32(t0, arg, 24);
+    
+    tcg_gen_andi_i32(t1, arg, 0x0000ff00);
+    tcg_gen_shli_i32(t1, t1, 8);
+    tcg_gen_or_i32(t0, t0, t1);
+    
+    tcg_gen_shri_i32(t1, arg, 8);
+    tcg_gen_andi_i32(t1, t1, 0x0000ff00);
+    tcg_gen_or_i32(t0, t0, t1);
+    
+    tcg_gen_shri_i32(t1, arg, 24);
+    tcg_gen_or_i32(ret, t0, t1);
+#endif
+}
+
+#if TCG_TARGET_REG_BITS == 32
+static inline void tcg_gen_ext8s_i64(int ret, int arg)
+{
+    tcg_gen_ext8s_i32(ret, arg);
+    tcg_gen_sari_i32(ret + 1, ret, 31);
+}
+
+static inline void tcg_gen_ext16s_i64(int ret, int arg)
+{
+    tcg_gen_ext16s_i32(ret, arg);
+    tcg_gen_sari_i32(ret + 1, ret, 31);
+}
+
+static inline void tcg_gen_ext32s_i64(int ret, int arg)
+{
+    tcg_gen_mov_i32(ret, arg);
+    tcg_gen_sari_i32(ret + 1, ret, 31);
+}
+
+static inline void tcg_gen_trunc_i64_i32(int ret, int arg)
+{
+    tcg_gen_mov_i32(ret, arg);
+}
+
+static inline void tcg_gen_extu_i32_i64(int ret, int arg)
+{
+    tcg_gen_mov_i32(ret, arg);
+    tcg_gen_movi_i32(ret + 1, 0);
+}
+
+static inline void tcg_gen_ext_i32_i64(int ret, int arg)
+{
+    tcg_gen_mov_i32(ret, arg);
+    tcg_gen_sari_i32(ret + 1, ret, 31);
+}
+
+static inline void tcg_gen_bswap_i64(int ret, int arg)
+{
+    int t0, t1;
+    t0 = tcg_temp_new(TCG_TYPE_I32);
+    t1 = tcg_temp_new(TCG_TYPE_I32);
+
+    tcg_gen_bswap_i32(t0, arg);
+    tcg_gen_bswap_i32(t1, arg + 1);
+    tcg_gen_mov_i32(ret, t1);
+    tcg_gen_mov_i32(ret + 1, t0);
+}
+#else
+
+static inline void tcg_gen_ext8s_i64(int ret, int arg)
+{
+#ifdef TCG_TARGET_HAS_ext8s_i64
+    tcg_gen_op2(INDEX_op_ext8s_i64, ret, arg);
+#else
+    tcg_gen_shli_i64(ret, arg, 56);
+    tcg_gen_sari_i64(ret, arg, 56);
+#endif
+}
+
+static inline void tcg_gen_ext16s_i64(int ret, int arg)
+{
+#ifdef TCG_TARGET_HAS_ext16s_i64
+    tcg_gen_op2(INDEX_op_ext16s_i64, ret, arg);
+#else
+    tcg_gen_shli_i64(ret, arg, 48);
+    tcg_gen_sari_i64(ret, arg, 48);
+#endif
+}
+
+static inline void tcg_gen_ext32s_i64(int ret, int arg)
+{
+#ifdef TCG_TARGET_HAS_ext32s_i64
+    tcg_gen_op2(INDEX_op_ext32s_i64, ret, arg);
+#else
+    tcg_gen_shli_i64(ret, arg, 32);
+    tcg_gen_sari_i64(ret, arg, 32);
+#endif
+}
+
+/* Note: we assume the target supports move between 32 and 64 bit
+   registers */
+static inline void tcg_gen_trunc_i64_i32(int ret, int arg)
+{
+    tcg_gen_mov_i32(ret, arg);
+}
+
+/* Note: we assume the target supports move between 32 and 64 bit
+   registers */
+static inline void tcg_gen_extu_i32_i64(int ret, int arg)
+{
+    tcg_gen_andi_i64(ret, arg, 0xffffffff);
+}
+
+/* Note: we assume the target supports move between 32 and 64 bit
+   registers */
+static inline void tcg_gen_ext_i32_i64(int ret, int arg)
+{
+    tcg_gen_ext32s_i64(ret, arg);
+}
+
+static inline void tcg_gen_bswap_i64(TCGArg ret, TCGArg arg)
+{
+#ifdef TCG_TARGET_HAS_bswap_i64
+    tcg_gen_op2(INDEX_op_bswap_i64, ret, arg);
+#else
+    TCGArg t0, t1;
+    t0 = tcg_temp_new(TCG_TYPE_I32);
+    t1 = tcg_temp_new(TCG_TYPE_I32);
+    
+    tcg_gen_shli_i64(t0, arg, 56);
+    
+    tcg_gen_andi_i64(t1, arg, 0x0000ff00);
+    tcg_gen_shli_i64(t1, t1, 40);
+    tcg_gen_or_i64(t0, t0, t1);
+    
+    tcg_gen_andi_i64(t1, arg, 0x00ff0000);
+    tcg_gen_shli_i64(t1, t1, 24);
+    tcg_gen_or_i64(t0, t0, t1);
+
+    tcg_gen_andi_i64(t1, arg, 0xff000000);
+    tcg_gen_shli_i64(t1, t1, 8);
+    tcg_gen_or_i64(t0, t0, t1);
+
+    tcg_gen_shri_i64(t1, arg, 8);
+    tcg_gen_andi_i64(t1, t1, 0xff000000);
+    tcg_gen_or_i64(t0, t0, t1);
+    
+    tcg_gen_shri_i64(t1, arg, 24);
+    tcg_gen_andi_i64(t1, t1, 0x00ff0000);
+    tcg_gen_or_i64(t0, t0, t1);
+
+    tcg_gen_shri_i64(t1, arg, 40);
+    tcg_gen_andi_i64(t1, t1, 0x0000ff00);
+    tcg_gen_or_i64(t0, t0, t1);
+
+    tcg_gen_shri_i64(t1, arg, 56);
+    tcg_gen_or_i64(ret, t0, t1);
+#endif
+}
+
+#endif
+
+/***************************************/
+static inline void tcg_gen_macro_2(int ret0, int ret1, int macro_id)
+{
+    tcg_gen_op3(INDEX_op_macro_2, ret0, ret1, macro_id);
+}
+
+/***************************************/
+/* QEMU specific operations. Their type depend on the QEMU CPU
+   type. */
+#ifndef TARGET_LONG_BITS
+#error must include QEMU headers
+#endif
+
+static inline void tcg_gen_exit_tb(tcg_target_long val)
+{
+    tcg_gen_op1(INDEX_op_exit_tb, val);
+}
+
+static inline void tcg_gen_goto_tb(int idx)
+{
+    tcg_gen_op1(INDEX_op_goto_tb, idx);
+}
+
+#if TCG_TARGET_REG_BITS == 32
+static inline void tcg_gen_qemu_ld8u(int ret, int addr, int mem_index)
+{
+#if TARGET_LONG_BITS == 32
+    tcg_gen_op3(INDEX_op_qemu_ld8u, ret, addr, mem_index);
+#else
+    tcg_gen_op4(INDEX_op_qemu_ld8u, ret, addr, addr + 1, mem_index);
+    tcg_gen_movi_i32(ret + 1, 0);
+#endif
+}
+
+static inline void tcg_gen_qemu_ld8s(int ret, int addr, int mem_index)
+{
+#if TARGET_LONG_BITS == 32
+    tcg_gen_op3(INDEX_op_qemu_ld8s, ret, addr, mem_index);
+#else
+    tcg_gen_op4(INDEX_op_qemu_ld8s, ret, addr, addr + 1, mem_index);
+    tcg_gen_ext8s_i32(ret + 1, ret);
+#endif
+}
+
+static inline void tcg_gen_qemu_ld16u(int ret, int addr, int mem_index)
+{
+#if TARGET_LONG_BITS == 32
+    tcg_gen_op3(INDEX_op_qemu_ld16u, ret, addr, mem_index);
+#else
+    tcg_gen_op4(INDEX_op_qemu_ld16u, ret, addr, addr + 1, mem_index);
+    tcg_gen_movi_i32(ret + 1, 0);
+#endif
+}
+
+static inline void tcg_gen_qemu_ld16s(int ret, int addr, int mem_index)
+{
+#if TARGET_LONG_BITS == 32
+    tcg_gen_op3(INDEX_op_qemu_ld16s, ret, addr, mem_index);
+#else
+    tcg_gen_op4(INDEX_op_qemu_ld16s, ret, addr, addr + 1, mem_index);
+    tcg_gen_ext16s_i32(ret + 1, ret);
+#endif
+}
+
+static inline void tcg_gen_qemu_ld32u(int ret, int addr, int mem_index)
+{
+#if TARGET_LONG_BITS == 32
+    tcg_gen_op3(INDEX_op_qemu_ld32u, ret, addr, mem_index);
+#else
+    tcg_gen_op4(INDEX_op_qemu_ld32u, ret, addr, addr + 1, mem_index);
+    tcg_gen_movi_i32(ret + 1, 0);
+#endif
+}
+
+static inline void tcg_gen_qemu_ld32s(int ret, int addr, int mem_index)
+{
+#if TARGET_LONG_BITS == 32
+    tcg_gen_op3(INDEX_op_qemu_ld32u, ret, addr, mem_index);
+#else
+    tcg_gen_op4(INDEX_op_qemu_ld32u, ret, addr, addr + 1, mem_index);
+    tcg_gen_sari_i32(ret + 1, ret, 31);
+#endif
+}
+
+static inline void tcg_gen_qemu_ld64(int ret, int addr, int mem_index)
+{
+#if TARGET_LONG_BITS == 32
+    tcg_gen_op4(INDEX_op_qemu_ld64, ret, ret + 1, addr, mem_index);
+#else
+    tcg_gen_op5(INDEX_op_qemu_ld64, ret, ret + 1, addr, addr + 1, mem_index);
+#endif
+}
+
+static inline void tcg_gen_qemu_st8(int arg, int addr, int mem_index)
+{
+#if TARGET_LONG_BITS == 32
+    tcg_gen_op3(INDEX_op_qemu_st8, arg, addr, mem_index);
+#else
+    tcg_gen_op4(INDEX_op_qemu_st8, arg, addr, addr + 1, mem_index);
+#endif
+}
+
+static inline void tcg_gen_qemu_st16(int arg, int addr, int mem_index)
+{
+#if TARGET_LONG_BITS == 32
+    tcg_gen_op3(INDEX_op_qemu_st16, arg, addr, mem_index);
+#else
+    tcg_gen_op4(INDEX_op_qemu_st16, arg, addr, addr + 1, mem_index);
+#endif
+}
+
+static inline void tcg_gen_qemu_st32(int arg, int addr, int mem_index)
+{
+#if TARGET_LONG_BITS == 32
+    tcg_gen_op3(INDEX_op_qemu_st32, arg, addr, mem_index);
+#else
+    tcg_gen_op4(INDEX_op_qemu_st32, arg, addr, addr + 1, mem_index);
+#endif
+}
+
+static inline void tcg_gen_qemu_st64(int arg, int addr, int mem_index)
+{
+#if TARGET_LONG_BITS == 32
+    tcg_gen_op4(INDEX_op_qemu_st64, arg, arg + 1, addr, mem_index);
+#else
+    tcg_gen_op5(INDEX_op_qemu_st64, arg, arg + 1, addr, addr + 1, mem_index);
+#endif
+}
+
+#else /* TCG_TARGET_REG_BITS == 32 */
+
+static inline void tcg_gen_qemu_ld8u(int ret, int addr, int mem_index)
+{
+    tcg_gen_op3(INDEX_op_qemu_ld8u, ret, addr, mem_index);
+}
+
+static inline void tcg_gen_qemu_ld8s(int ret, int addr, int mem_index)
+{
+    tcg_gen_op3(INDEX_op_qemu_ld8s, ret, addr, mem_index);
+}
+
+static inline void tcg_gen_qemu_ld16u(int ret, int addr, int mem_index)
+{
+    tcg_gen_op3(INDEX_op_qemu_ld16u, ret, addr, mem_index);
+}
+
+static inline void tcg_gen_qemu_ld16s(int ret, int addr, int mem_index)
+{
+    tcg_gen_op3(INDEX_op_qemu_ld16s, ret, addr, mem_index);
+}
+
+static inline void tcg_gen_qemu_ld32u(int ret, int addr, int mem_index)
+{
+    tcg_gen_op3(INDEX_op_qemu_ld32u, ret, addr, mem_index);
+}
+
+static inline void tcg_gen_qemu_ld32s(int ret, int addr, int mem_index)
+{
+    tcg_gen_op3(INDEX_op_qemu_ld32s, ret, addr, mem_index);
+}
+
+static inline void tcg_gen_qemu_ld64(int ret, int addr, int mem_index)
+{
+    tcg_gen_op3(INDEX_op_qemu_ld64, ret, addr, mem_index);
+}
+
+static inline void tcg_gen_qemu_st8(int arg, int addr, int mem_index)
+{
+    tcg_gen_op3(INDEX_op_qemu_st8, arg, addr, mem_index);
+}
+
+static inline void tcg_gen_qemu_st16(int arg, int addr, int mem_index)
+{
+    tcg_gen_op3(INDEX_op_qemu_st16, arg, addr, mem_index);
+}
+
+static inline void tcg_gen_qemu_st32(int arg, int addr, int mem_index)
+{
+    tcg_gen_op3(INDEX_op_qemu_st32, arg, addr, mem_index);
+}
+
+static inline void tcg_gen_qemu_st64(int arg, int addr, int mem_index)
+{
+    tcg_gen_op3(INDEX_op_qemu_st64, arg, addr, mem_index);
+}
+
+#endif /* TCG_TARGET_REG_BITS != 32 */
diff --git a/tcg/tcg-opc.h b/tcg/tcg-opc.h
new file mode 100644
index 0000000000..ebf4f82fb8
--- /dev/null
+++ b/tcg/tcg-opc.h
@@ -0,0 +1,228 @@
+/*
+ * Tiny Code Generator for QEMU
+ *
+ * Copyright (c) 2008 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#include "dyngen-opc.h"
+
+#ifndef DEF2
+#define DEF2(name, oargs, iargs, cargs, flags) DEF(name, oargs + iargs + cargs, 0)
+#endif
+
+/* predefined ops */
+DEF2(end, 0, 0, 0, 0) /* must be kept first */
+DEF2(nop, 0, 0, 0, 0)
+DEF2(nop1, 0, 0, 1, 0)
+DEF2(nop2, 0, 0, 2, 0)
+DEF2(nop3, 0, 0, 3, 0)
+DEF2(nopn, 0, 0, 1, 0) /* variable number of parameters */
+/* macro handling */
+DEF2(macro_2, 2, 0, 1, 0)
+DEF2(macro_start, 0, 0, 2, 0)
+DEF2(macro_end, 0, 0, 2, 0)
+DEF2(macro_goto, 0, 0, 3, 0)
+
+DEF2(set_label, 0, 0, 1, 0)
+DEF2(call, 0, 1, 2, 0) /* variable number of parameters */
+DEF2(jmp, 0, 1, 0, TCG_OPF_BB_END)
+DEF2(br, 0, 0, 1, TCG_OPF_BB_END)
+
+DEF2(mov_i32, 1, 1, 0, 0)
+DEF2(movi_i32, 1, 0, 1, 0)
+/* load/store */
+DEF2(ld8u_i32, 1, 1, 1, 0)
+DEF2(ld8s_i32, 1, 1, 1, 0)
+DEF2(ld16u_i32, 1, 1, 1, 0)
+DEF2(ld16s_i32, 1, 1, 1, 0)
+DEF2(ld_i32, 1, 1, 1, 0)
+DEF2(st8_i32, 0, 2, 1, 0)
+DEF2(st16_i32, 0, 2, 1, 0)
+DEF2(st_i32, 0, 2, 1, 0)
+/* arith */
+DEF2(add_i32, 1, 2, 0, 0)
+DEF2(sub_i32, 1, 2, 0, 0)
+DEF2(mul_i32, 1, 2, 0, 0)
+#ifdef TCG_TARGET_HAS_div_i32
+DEF2(div_i32, 1, 2, 0, 0)
+DEF2(divu_i32, 1, 2, 0, 0)
+DEF2(rem_i32, 1, 2, 0, 0)
+DEF2(remu_i32, 1, 2, 0, 0)
+#else
+DEF2(div2_i32, 2, 3, 0, 0)
+DEF2(divu2_i32, 2, 3, 0, 0)
+#endif
+DEF2(and_i32, 1, 2, 0, 0)
+DEF2(or_i32, 1, 2, 0, 0)
+DEF2(xor_i32, 1, 2, 0, 0)
+/* shifts */
+DEF2(shl_i32, 1, 2, 0, 0)
+DEF2(shr_i32, 1, 2, 0, 0)
+DEF2(sar_i32, 1, 2, 0, 0)
+
+DEF2(brcond_i32, 0, 2, 2, TCG_OPF_BB_END)
+#if TCG_TARGET_REG_BITS == 32
+DEF2(add2_i32, 2, 4, 0, 0)
+DEF2(sub2_i32, 2, 4, 0, 0)
+DEF2(brcond2_i32, 0, 4, 2, TCG_OPF_BB_END)
+DEF2(mulu2_i32, 2, 2, 0, 0)
+#endif
+#ifdef TCG_TARGET_HAS_ext8s_i32
+DEF2(ext8s_i32, 1, 1, 0, 0)
+#endif
+#ifdef TCG_TARGET_HAS_ext16s_i32
+DEF2(ext16s_i32, 1, 1, 0, 0)
+#endif
+#ifdef TCG_TARGET_HAS_bswap_i32
+DEF2(bswap_i32, 1, 1, 0, 0)
+#endif
+
+#if TCG_TARGET_REG_BITS == 64
+DEF2(mov_i64, 1, 1, 0, 0)
+DEF2(movi_i64, 1, 0, 1, 0)
+/* load/store */
+DEF2(ld8u_i64, 1, 1, 1, 0)
+DEF2(ld8s_i64, 1, 1, 1, 0)
+DEF2(ld16u_i64, 1, 1, 1, 0)
+DEF2(ld16s_i64, 1, 1, 1, 0)
+DEF2(ld32u_i64, 1, 1, 1, 0)
+DEF2(ld32s_i64, 1, 1, 1, 0)
+DEF2(ld_i64, 1, 1, 1, 0)
+DEF2(st8_i64, 0, 2, 1, 0)
+DEF2(st16_i64, 0, 2, 1, 0)
+DEF2(st32_i64, 0, 2, 1, 0)
+DEF2(st_i64, 0, 2, 1, 0)
+/* arith */
+DEF2(add_i64, 1, 2, 0, 0)
+DEF2(sub_i64, 1, 2, 0, 0)
+DEF2(mul_i64, 1, 2, 0, 0)
+#ifdef TCG_TARGET_HAS_div_i64
+DEF2(div_i64, 1, 2, 0, 0)
+DEF2(divu_i64, 1, 2, 0, 0)
+DEF2(rem_i64, 1, 2, 0, 0)
+DEF2(remu_i64, 1, 2, 0, 0)
+#else
+DEF2(div2_i64, 2, 3, 0, 0)
+DEF2(divu2_i64, 2, 3, 0, 0)
+#endif
+DEF2(and_i64, 1, 2, 0, 0)
+DEF2(or_i64, 1, 2, 0, 0)
+DEF2(xor_i64, 1, 2, 0, 0)
+/* shifts */
+DEF2(shl_i64, 1, 2, 0, 0)
+DEF2(shr_i64, 1, 2, 0, 0)
+DEF2(sar_i64, 1, 2, 0, 0)
+
+DEF2(brcond_i64, 0, 2, 2, TCG_OPF_BB_END)
+#ifdef TCG_TARGET_HAS_ext8s_i64
+DEF2(ext8s_i64, 1, 1, 0, 0)
+#endif
+#ifdef TCG_TARGET_HAS_ext16s_i64
+DEF2(ext16s_i64, 1, 1, 0, 0)
+#endif
+#ifdef TCG_TARGET_HAS_ext32s_i64
+DEF2(ext32s_i64, 1, 1, 0, 0)
+#endif
+#ifdef TCG_TARGET_HAS_bswap_i64
+DEF2(bswap_i64, 1, 1, 0, 0)
+#endif
+#endif
+
+/* QEMU specific */
+DEF2(exit_tb, 0, 0, 1, TCG_OPF_BB_END)
+DEF2(goto_tb, 0, 0, 1, TCG_OPF_BB_END)
+/* Note: even if TARGET_LONG_BITS is not defined, the INDEX_op
+   constants must be defined */
+#if TCG_TARGET_REG_BITS == 32
+#if TARGET_LONG_BITS == 32
+DEF2(qemu_ld8u, 1, 1, 1, TCG_OPF_CALL_CLOBBER)
+#else
+DEF2(qemu_ld8u, 1, 2, 1, TCG_OPF_CALL_CLOBBER)
+#endif
+#if TARGET_LONG_BITS == 32
+DEF2(qemu_ld8s, 1, 1, 1, TCG_OPF_CALL_CLOBBER)
+#else
+DEF2(qemu_ld8s, 1, 2, 1, TCG_OPF_CALL_CLOBBER)
+#endif
+#if TARGET_LONG_BITS == 32
+DEF2(qemu_ld16u, 1, 1, 1, TCG_OPF_CALL_CLOBBER)
+#else
+DEF2(qemu_ld16u, 1, 2, 1, TCG_OPF_CALL_CLOBBER)
+#endif
+#if TARGET_LONG_BITS == 32
+DEF2(qemu_ld16s, 1, 1, 1, TCG_OPF_CALL_CLOBBER)
+#else
+DEF2(qemu_ld16s, 1, 2, 1, TCG_OPF_CALL_CLOBBER)
+#endif
+#if TARGET_LONG_BITS == 32
+DEF2(qemu_ld32u, 1, 1, 1, TCG_OPF_CALL_CLOBBER)
+#else
+DEF2(qemu_ld32u, 1, 2, 1, TCG_OPF_CALL_CLOBBER)
+#endif
+#if TARGET_LONG_BITS == 32
+DEF2(qemu_ld32s, 1, 1, 1, TCG_OPF_CALL_CLOBBER)
+#else
+DEF2(qemu_ld32s, 1, 2, 1, TCG_OPF_CALL_CLOBBER)
+#endif
+#if TARGET_LONG_BITS == 32
+DEF2(qemu_ld64, 2, 1, 1, TCG_OPF_CALL_CLOBBER)
+#else
+DEF2(qemu_ld64, 2, 2, 1, TCG_OPF_CALL_CLOBBER)
+#endif
+
+#if TARGET_LONG_BITS == 32
+DEF2(qemu_st8, 0, 2, 1, TCG_OPF_CALL_CLOBBER)
+#else
+DEF2(qemu_st8, 0, 3, 1, TCG_OPF_CALL_CLOBBER)
+#endif
+#if TARGET_LONG_BITS == 32
+DEF2(qemu_st16, 0, 2, 1, TCG_OPF_CALL_CLOBBER)
+#else
+DEF2(qemu_st16, 0, 3, 1, TCG_OPF_CALL_CLOBBER)
+#endif
+#if TARGET_LONG_BITS == 32
+DEF2(qemu_st32, 0, 2, 1, TCG_OPF_CALL_CLOBBER)
+#else
+DEF2(qemu_st32, 0, 3, 1, TCG_OPF_CALL_CLOBBER)
+#endif
+#if TARGET_LONG_BITS == 32
+DEF2(qemu_st64, 0, 3, 1, TCG_OPF_CALL_CLOBBER)
+#else
+DEF2(qemu_st64, 0, 4, 1, TCG_OPF_CALL_CLOBBER)
+#endif
+
+#else /* TCG_TARGET_REG_BITS == 32 */
+
+DEF2(qemu_ld8u, 1, 1, 1, TCG_OPF_CALL_CLOBBER)
+DEF2(qemu_ld8s, 1, 1, 1, TCG_OPF_CALL_CLOBBER)
+DEF2(qemu_ld16u, 1, 1, 1, TCG_OPF_CALL_CLOBBER)
+DEF2(qemu_ld16s, 1, 1, 1, TCG_OPF_CALL_CLOBBER)
+DEF2(qemu_ld32u, 1, 1, 1, TCG_OPF_CALL_CLOBBER)
+DEF2(qemu_ld32s, 1, 1, 1, TCG_OPF_CALL_CLOBBER)
+DEF2(qemu_ld64, 1, 1, 1, TCG_OPF_CALL_CLOBBER)
+
+DEF2(qemu_st8, 0, 2, 1, TCG_OPF_CALL_CLOBBER)
+DEF2(qemu_st16, 0, 2, 1, TCG_OPF_CALL_CLOBBER)
+DEF2(qemu_st32, 0, 2, 1, TCG_OPF_CALL_CLOBBER)
+DEF2(qemu_st64, 0, 2, 1, TCG_OPF_CALL_CLOBBER)
+
+#endif /* TCG_TARGET_REG_BITS != 32 */
+
+#undef DEF2
diff --git a/tcg/tcg-runtime.c b/tcg/tcg-runtime.c
new file mode 100644
index 0000000000..f41703ec8b
--- /dev/null
+++ b/tcg/tcg-runtime.c
@@ -0,0 +1,68 @@
+/*
+ * Tiny Code Generator for QEMU
+ *
+ * Copyright (c) 2008 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#include <stdarg.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <inttypes.h>
+
+#include "config.h"
+#include "osdep.h"
+#include "tcg.h"
+
+int64_t tcg_helper_shl_i64(int64_t arg1, int64_t arg2)
+{
+    return arg1 << arg2;
+}
+
+int64_t tcg_helper_shr_i64(int64_t arg1, int64_t arg2)
+{
+    return (uint64_t)arg1 >> arg2;
+}
+
+int64_t tcg_helper_sar_i64(int64_t arg1, int64_t arg2)
+{
+    return arg1 >> arg2;
+}
+
+int64_t tcg_helper_div_i64(int64_t arg1, int64_t arg2)
+{
+    return arg1 / arg2;
+}
+
+int64_t tcg_helper_rem_i64(int64_t arg1, int64_t arg2)
+{
+    return arg1 / arg2;
+}
+
+uint64_t tcg_helper_divu_i64(uint64_t arg1, uint64_t arg2)
+{
+    return arg1 / arg2;
+}
+
+uint64_t tcg_helper_remu_i64(uint64_t arg1, uint64_t arg2)
+{
+    return arg1 / arg2;
+}
+
diff --git a/tcg/tcg.c b/tcg/tcg.c
new file mode 100644
index 0000000000..211e7fc987
--- /dev/null
+++ b/tcg/tcg.c
@@ -0,0 +1,1781 @@
+/*
+ * Tiny Code Generator for QEMU
+ *
+ * Copyright (c) 2008 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+/* define it to suppress various consistency checks (faster) */
+#define NDEBUG
+
+/* define it to use liveness analysis (better code) */
+#define USE_LIVENESS_ANALYSIS
+
+#include <assert.h>
+#include <stdarg.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <inttypes.h>
+
+#include "config.h"
+#include "osdep.h"
+
+/* Note: the long term plan is to reduce the dependancies on the QEMU
+   CPU definitions. Currently they are used for qemu_ld/st
+   instructions */
+#define NO_CPU_IO_DEFS
+#include "cpu.h"
+#include "exec-all.h"
+
+#include "tcg-op.h"
+#include "elf.h"
+
+
+static void patch_reloc(uint8_t *code_ptr, int type, 
+                        tcg_target_long value);
+
+TCGOpDef tcg_op_defs[] = {
+#define DEF(s, n, copy_size) { #s, 0, 0, n, n, 0, copy_size },
+#define DEF2(s, iargs, oargs, cargs, flags) { #s, iargs, oargs, cargs, iargs + oargs + cargs, flags, 0 },
+#include "tcg-opc.h"
+#undef DEF
+#undef DEF2
+};
+
+TCGRegSet tcg_target_available_regs[2];
+TCGRegSet tcg_target_call_clobber_regs;
+
+/* XXX: move that inside the context */
+uint16_t *gen_opc_ptr;
+TCGArg *gen_opparam_ptr;
+
+static inline void tcg_out8(TCGContext *s, uint8_t v)
+{
+    *s->code_ptr++ = v;
+}
+
+static inline void tcg_out16(TCGContext *s, uint16_t v)
+{
+    *(uint16_t *)s->code_ptr = v;
+    s->code_ptr += 2;
+}
+
+static inline void tcg_out32(TCGContext *s, uint32_t v)
+{
+    *(uint32_t *)s->code_ptr = v;
+    s->code_ptr += 4;
+}
+
+/* label relocation processing */
+
+void tcg_out_reloc(TCGContext *s, uint8_t *code_ptr, int type, 
+                   int label_index, long addend)
+{
+    TCGLabel *l;
+    TCGRelocation *r;
+
+    l = &s->labels[label_index];
+    if (l->has_value) {
+        patch_reloc(code_ptr, type, l->u.value + addend);
+    } else {
+        /* add a new relocation entry */
+        r = tcg_malloc(sizeof(TCGRelocation));
+        r->type = type;
+        r->ptr = code_ptr;
+        r->addend = addend;
+        r->next = l->u.first_reloc;
+        l->u.first_reloc = r;
+    }
+}
+
+static void tcg_out_label(TCGContext *s, int label_index, 
+                          tcg_target_long value)
+{
+    TCGLabel *l;
+    TCGRelocation *r;
+
+    l = &s->labels[label_index];
+    if (l->has_value)
+        tcg_abort();
+    r = l->u.first_reloc;
+    while (r != NULL) {
+        patch_reloc(r->ptr, r->type, value + r->addend);
+        r = r->next;
+    }
+    l->has_value = 1;
+    l->u.value = value;
+}
+
+int gen_new_label(void)
+{
+    TCGContext *s = &tcg_ctx;
+    int idx;
+    TCGLabel *l;
+
+    if (s->nb_labels >= TCG_MAX_LABELS)
+        tcg_abort();
+    idx = s->nb_labels++;
+    l = &s->labels[idx];
+    l->has_value = 0;
+    l->u.first_reloc = NULL;
+    return idx;
+}
+
+#include "tcg-target.c"
+
+/* XXX: factorize */
+static void pstrcpy(char *buf, int buf_size, const char *str)
+{
+    int c;
+    char *q = buf;
+
+    if (buf_size <= 0)
+        return;
+
+    for(;;) {
+        c = *str++;
+        if (c == 0 || q >= buf + buf_size - 1)
+            break;
+        *q++ = c;
+    }
+    *q = '\0';
+}
+
+#if TCG_TARGET_REG_BITS == 32
+/* strcat and truncate. */
+static char *pstrcat(char *buf, int buf_size, const char *s)
+{
+    int len;
+    len = strlen(buf);
+    if (len < buf_size)
+        pstrcpy(buf + len, buf_size - len, s);
+    return buf;
+}
+#endif
+
+/* pool based memory allocation */
+void *tcg_malloc_internal(TCGContext *s, int size)
+{
+    TCGPool *p;
+    int pool_size;
+    
+    if (size > TCG_POOL_CHUNK_SIZE) {
+        /* big malloc: insert a new pool (XXX: could optimize) */
+        p = qemu_malloc(sizeof(TCGPool) + size);
+        p->size = size;
+        if (s->pool_current)
+            s->pool_current->next = p;
+        else
+            s->pool_first = p;
+        p->next = s->pool_current;
+    } else {
+        p = s->pool_current;
+        if (!p) {
+            p = s->pool_first;
+            if (!p)
+                goto new_pool;
+        } else {
+            if (!p->next) {
+            new_pool:
+                pool_size = TCG_POOL_CHUNK_SIZE;
+                p = qemu_malloc(sizeof(TCGPool) + pool_size);
+                p->size = pool_size;
+                p->next = NULL;
+                if (s->pool_current) 
+                    s->pool_current->next = p;
+                else
+                    s->pool_first = p;
+            } else {
+                p = p->next;
+            }
+        }
+    }
+    s->pool_current = p;
+    s->pool_cur = p->data + size;
+    s->pool_end = p->data + p->size;
+    return p->data;
+}
+
+void tcg_pool_reset(TCGContext *s)
+{
+    s->pool_cur = s->pool_end = NULL;
+    s->pool_current = NULL;
+}
+
+/* free all the pool */
+void tcg_pool_free(TCGContext *s)
+{
+    TCGPool *p, *p1;
+
+    for(p = s->pool_first; p != NULL; p = p1) {
+        p1 = p->next;
+        qemu_free(p);
+    }
+    s->pool_first = NULL;
+    s->pool_cur = s->pool_end = NULL;
+}
+
+void tcg_context_init(TCGContext *s)
+{
+    int op, total_args, n;
+    TCGOpDef *def;
+    TCGArgConstraint *args_ct;
+    int *sorted_args;
+
+    memset(s, 0, sizeof(*s));
+    s->temps = s->static_temps;
+    s->nb_globals = 0;
+    
+    /* Count total number of arguments and allocate the corresponding
+       space */
+    total_args = 0;
+    for(op = 0; op < NB_OPS; op++) {
+        def = &tcg_op_defs[op];
+        n = def->nb_iargs + def->nb_oargs;
+        total_args += n;
+    }
+
+    args_ct = qemu_malloc(sizeof(TCGArgConstraint) * total_args);
+    sorted_args = qemu_malloc(sizeof(int) * total_args);
+
+    for(op = 0; op < NB_OPS; op++) {
+        def = &tcg_op_defs[op];
+        def->args_ct = args_ct;
+        def->sorted_args = sorted_args;
+        n = def->nb_iargs + def->nb_oargs;
+        sorted_args += n;
+        args_ct += n;
+    }
+    
+    tcg_target_init(s);
+}
+
+void tcg_set_frame(TCGContext *s, int reg,
+                   tcg_target_long start, tcg_target_long size)
+{
+    s->frame_start = start;
+    s->frame_end = start + size;
+    s->frame_reg = reg;
+}
+
+void tcg_set_macro_func(TCGContext *s, TCGMacroFunc *func)
+{
+    s->macro_func = func;
+}
+
+void tcg_func_start(TCGContext *s)
+{
+    tcg_pool_reset(s);
+    s->nb_temps = s->nb_globals;
+    s->labels = tcg_malloc(sizeof(TCGLabel) * TCG_MAX_LABELS);
+    s->nb_labels = 0;
+    s->current_frame_offset = s->frame_start;
+
+    gen_opc_ptr = gen_opc_buf;
+    gen_opparam_ptr = gen_opparam_buf;
+}
+
+static inline void tcg_temp_alloc(TCGContext *s, int n)
+{
+    if (n > TCG_MAX_TEMPS)
+        tcg_abort();
+}
+
+int tcg_global_reg_new(TCGType type, int reg, const char *name)
+{
+    TCGContext *s = &tcg_ctx;
+    TCGTemp *ts;
+    int idx;
+
+#if TCG_TARGET_REG_BITS == 32
+    if (type != TCG_TYPE_I32)
+        tcg_abort();
+#endif
+    if (tcg_regset_test_reg(s->reserved_regs, reg))
+        tcg_abort();
+    idx = s->nb_globals;
+    tcg_temp_alloc(s, s->nb_globals + 1);
+    ts = &s->temps[s->nb_globals];
+    ts->base_type = type;
+    ts->type = type;
+    ts->fixed_reg = 1;
+    ts->reg = reg;
+    ts->val_type = TEMP_VAL_REG;
+    ts->name = name;
+    s->nb_globals++;
+    tcg_regset_set_reg(s->reserved_regs, reg);
+    return idx;
+}
+
+int tcg_global_mem_new(TCGType type, int reg, tcg_target_long offset,
+                       const char *name)
+{
+    TCGContext *s = &tcg_ctx;
+    TCGTemp *ts;
+    int idx;
+
+    idx = s->nb_globals;
+#if TCG_TARGET_REG_BITS == 32
+    if (type == TCG_TYPE_I64) {
+        char buf[64];
+        tcg_temp_alloc(s, s->nb_globals + 1);
+        ts = &s->temps[s->nb_globals];
+        ts->base_type = type;
+        ts->type = TCG_TYPE_I32;
+        ts->fixed_reg = 0;
+        ts->mem_allocated = 1;
+        ts->mem_reg = reg;
+#ifdef TCG_TARGET_WORDS_BIGENDIAN
+        ts->mem_offset = offset + 4;
+#else
+        ts->mem_offset = offset;
+#endif
+        ts->val_type = TEMP_VAL_MEM;
+        pstrcpy(buf, sizeof(buf), name);
+        pstrcat(buf, sizeof(buf), "_0");
+        ts->name = strdup(buf);
+        ts++;
+
+        ts->base_type = type;
+        ts->type = TCG_TYPE_I32;
+        ts->fixed_reg = 0;
+        ts->mem_allocated = 1;
+        ts->mem_reg = reg;
+#ifdef TCG_TARGET_WORDS_BIGENDIAN
+        ts->mem_offset = offset;
+#else
+        ts->mem_offset = offset + 4;
+#endif
+        ts->val_type = TEMP_VAL_MEM;
+        pstrcpy(buf, sizeof(buf), name);
+        pstrcat(buf, sizeof(buf), "_1");
+        ts->name = strdup(buf);
+
+        s->nb_globals += 2;
+    } else
+#endif
+    {
+        tcg_temp_alloc(s, s->nb_globals + 1);
+        ts = &s->temps[s->nb_globals];
+        ts->base_type = type;
+        ts->type = type;
+        ts->fixed_reg = 0;
+        ts->mem_allocated = 1;
+        ts->mem_reg = reg;
+        ts->mem_offset = offset;
+        ts->val_type = TEMP_VAL_MEM;
+        ts->name = name;
+        s->nb_globals++;
+    }
+    return idx;
+}
+
+int tcg_temp_new(TCGType type)
+{
+    TCGContext *s = &tcg_ctx;
+    TCGTemp *ts;
+    int idx;
+
+    idx = s->nb_temps;
+#if TCG_TARGET_REG_BITS == 32
+    if (type == TCG_TYPE_I64) {
+        tcg_temp_alloc(s, s->nb_temps + 1);
+        ts = &s->temps[s->nb_temps];
+        ts->base_type = type;
+        ts->type = TCG_TYPE_I32;
+        ts->val_type = TEMP_VAL_DEAD;
+        ts->mem_allocated = 0;
+        ts->name = NULL;
+        ts++;
+        ts->base_type = TCG_TYPE_I32;
+        ts->type = TCG_TYPE_I32;
+        ts->val_type = TEMP_VAL_DEAD;
+        ts->mem_allocated = 0;
+        ts->name = NULL;
+        s->nb_temps += 2;
+    } else
+#endif
+    {
+        tcg_temp_alloc(s, s->nb_temps + 1);
+        ts = &s->temps[s->nb_temps];
+        ts->base_type = type;
+        ts->type = type;
+        ts->val_type = TEMP_VAL_DEAD;
+        ts->mem_allocated = 0;
+        ts->name = NULL;
+        s->nb_temps++;
+    }
+    return idx;
+}
+
+int tcg_const_i32(int32_t val)
+{
+    TCGContext *s = &tcg_ctx;
+    TCGTemp *ts;
+    int idx;
+
+    idx = s->nb_temps;
+    tcg_temp_alloc(s, idx + 1);
+    ts = &s->temps[idx];
+    ts->base_type = ts->type = TCG_TYPE_I32;
+    ts->val_type = TEMP_VAL_CONST;
+    ts->name = NULL;
+    ts->val = val;
+    s->nb_temps++;
+    return idx;
+}
+
+int tcg_const_i64(int64_t val)
+{
+    TCGContext *s = &tcg_ctx;
+    TCGTemp *ts;
+    int idx;
+
+    idx = s->nb_temps;
+#if TCG_TARGET_REG_BITS == 32
+    tcg_temp_alloc(s, idx + 2);
+    ts = &s->temps[idx];
+    ts->base_type = TCG_TYPE_I64;
+    ts->type = TCG_TYPE_I32;
+    ts->val_type = TEMP_VAL_CONST;
+    ts->name = NULL;
+    ts->val = val;
+    ts++;
+    ts->base_type = TCG_TYPE_I32;
+    ts->type = TCG_TYPE_I32;
+    ts->val_type = TEMP_VAL_CONST;
+    ts->name = NULL;
+    ts->val = val >> 32;
+    s->nb_temps += 2;
+#else
+    tcg_temp_alloc(s, idx + 1);
+    ts = &s->temps[idx];
+    ts->base_type = ts->type = TCG_TYPE_I64;
+    ts->val_type = TEMP_VAL_CONST;
+    ts->name = NULL;
+    ts->val = val;
+    s->nb_temps++;
+#endif    
+    return idx;
+}
+
+void tcg_register_helper(void *func, const char *name)
+{
+    TCGContext *s = &tcg_ctx;
+    int n;
+    if ((s->nb_helpers + 1) > s->allocated_helpers) {
+        n = s->allocated_helpers;
+        if (n == 0) {
+            n = 4;
+        } else {
+            n *= 2;
+        }
+        s->helpers = realloc(s->helpers, n * sizeof(TCGHelperInfo));
+        s->allocated_helpers = n;
+    }
+    s->helpers[s->nb_helpers].func = func;
+    s->helpers[s->nb_helpers].name = name;
+    s->nb_helpers++;
+}
+
+const char *tcg_helper_get_name(TCGContext *s, void *func)
+{
+    int i;
+
+    for(i = 0; i < s->nb_helpers; i++) {
+        if (s->helpers[i].func == func)
+            return s->helpers[i].name;
+    }
+    return NULL;
+}
+
+static inline TCGType tcg_get_base_type(TCGContext *s, TCGArg arg)
+{
+    return s->temps[arg].base_type;
+}
+
+static void tcg_gen_call_internal(TCGContext *s, TCGArg func, 
+                                  unsigned int flags,
+                                  unsigned int nb_rets, const TCGArg *rets,
+                                  unsigned int nb_params, const TCGArg *params)
+{
+    int i;
+    *gen_opc_ptr++ = INDEX_op_call;
+    *gen_opparam_ptr++ = (nb_rets << 16) | (nb_params + 1);
+    for(i = 0; i < nb_rets; i++) {
+        *gen_opparam_ptr++ = rets[i];
+    }
+    for(i = 0; i < nb_params; i++) {
+        *gen_opparam_ptr++ = params[i];
+    }
+    *gen_opparam_ptr++ = func;
+
+    *gen_opparam_ptr++ = flags;
+    /* total parameters, needed to go backward in the instruction stream */
+    *gen_opparam_ptr++ = 1 + nb_rets + nb_params + 3;
+}
+
+
+#if TCG_TARGET_REG_BITS < 64
+/* Note: we convert the 64 bit args to 32 bit */
+void tcg_gen_call(TCGContext *s, TCGArg func, unsigned int flags,
+                  unsigned int nb_rets, const TCGArg *rets,
+                  unsigned int nb_params, const TCGArg *args1)
+{
+    TCGArg ret, *args2, rets_2[2], arg;
+    int j, i, call_type;
+
+    if (nb_rets == 1) {
+        ret = rets[0];
+        if (tcg_get_base_type(s, ret) == TCG_TYPE_I64) {
+            nb_rets = 2;
+            rets_2[0] = ret;
+            rets_2[1] = ret + 1;
+            rets = rets_2;
+        }
+    }
+    args2 = alloca((nb_params * 2) * sizeof(TCGArg));
+    j = 0;
+    call_type = (flags & TCG_CALL_TYPE_MASK);
+    for(i = 0; i < nb_params; i++) {
+        arg = args1[i];
+        if (tcg_get_base_type(s, arg) == TCG_TYPE_I64) {
+#ifdef TCG_TARGET_I386
+            /* REGPARM case: if the third parameter is 64 bit, it is
+               allocated on the stack */
+            if (j == 2 && call_type == TCG_CALL_TYPE_REGPARM) {
+                call_type = TCG_CALL_TYPE_REGPARM_2;
+                flags = (flags & ~TCG_CALL_TYPE_MASK) | call_type;
+            }
+            args2[j++] = arg;
+            args2[j++] = arg + 1;
+#else
+#ifdef TCG_TARGET_WORDS_BIGENDIAN
+            args2[j++] = arg + 1;
+            args2[j++] = arg;
+#else
+            args2[j++] = arg;
+            args2[j++] = arg + 1;
+#endif
+#endif
+        } else {
+            args2[j++] = arg;
+        }
+    }
+    tcg_gen_call_internal(s, func, flags, 
+                          nb_rets, rets, j, args2);
+}
+#else
+void tcg_gen_call(TCGContext *s, TCGArg func, unsigned int flags,
+                  unsigned int nb_rets, const TCGArg *rets,
+                  unsigned int nb_params, const TCGArg *args1)
+{
+    tcg_gen_call_internal(s, func, flags, 
+                          nb_rets, rets, nb_params, args1);
+}
+#endif
+
+void tcg_gen_shifti_i64(TCGArg ret, TCGArg arg1, 
+                        int c, int right, int arith)
+{
+    if (c == 0)
+        return;
+    if (c >= 32) {
+        c -= 32;
+        if (right) {
+            if (arith) {
+                tcg_gen_sari_i32(ret, arg1 + 1, c);
+                tcg_gen_sari_i32(ret + 1, arg1 + 1, 31);
+            } else {
+                tcg_gen_shri_i32(ret, arg1 + 1, c);
+                tcg_gen_movi_i32(ret + 1, 0);
+            }
+        } else {
+            tcg_gen_shli_i32(ret + 1, arg1, c);
+            tcg_gen_movi_i32(ret, 0);
+        }
+    } else {
+        int t0, t1;
+
+        t0 = tcg_temp_new(TCG_TYPE_I32);
+        t1 = tcg_temp_new(TCG_TYPE_I32);
+        if (right) {
+            tcg_gen_shli_i32(t0, arg1 + 1, 32 - c);
+            if (arith)
+                tcg_gen_sari_i32(t1, arg1 + 1, c);
+            else 
+                tcg_gen_shri_i32(t1, arg1 + 1, c);
+            tcg_gen_shri_i32(ret, arg1, c); 
+            tcg_gen_or_i32(ret, ret, t0);
+            tcg_gen_mov_i32(ret + 1, t1);
+        } else {
+            tcg_gen_shri_i32(t0, arg1, 32 - c);
+            /* Note: ret can be the same as arg1, so we use t1 */
+            tcg_gen_shli_i32(t1, arg1, c); 
+            tcg_gen_shli_i32(ret + 1, arg1 + 1, c);
+            tcg_gen_or_i32(ret + 1, ret + 1, t0);
+            tcg_gen_mov_i32(ret, t1);
+        }
+    }
+}
+
+void tcg_reg_alloc_start(TCGContext *s)
+{
+    int i;
+    TCGTemp *ts;
+    for(i = 0; i < s->nb_globals; i++) {
+        ts = &s->temps[i];
+        if (ts->fixed_reg) {
+            ts->val_type = TEMP_VAL_REG;
+        } else {
+            ts->val_type = TEMP_VAL_MEM;
+        }
+    }
+    for(i = 0; i < TCG_TARGET_NB_REGS; i++) {
+        s->reg_to_temp[i] = -1;
+    }
+}
+
+char *tcg_get_arg_str(TCGContext *s, char *buf, int buf_size, TCGArg arg)
+{
+    TCGTemp *ts;
+    if (arg < s->nb_globals) {
+        pstrcpy(buf, buf_size, s->temps[arg].name);
+    } else {
+        ts = &s->temps[arg];
+        if (ts->val_type == TEMP_VAL_CONST) {
+            snprintf(buf, buf_size, "$0x%" TCG_PRIlx , ts->val);
+        } else {
+            snprintf(buf, buf_size, "tmp%d", (int)arg - s->nb_globals);
+        }
+    }
+    return buf;
+}
+
+void tcg_dump_ops(TCGContext *s, FILE *outfile)
+{
+    const uint16_t *opc_ptr;
+    const TCGArg *args;
+    TCGArg arg;
+    int c, i, k, nb_oargs, nb_iargs, nb_cargs;
+    const TCGOpDef *def;
+    char buf[128];
+
+    opc_ptr = gen_opc_buf;
+    args = gen_opparam_buf;
+    while (opc_ptr < gen_opc_ptr) {
+        c = *opc_ptr++;
+        def = &tcg_op_defs[c];
+        fprintf(outfile, " %s ", def->name);
+        if (c == INDEX_op_call) {
+            TCGArg arg;
+            /* variable number of arguments */
+            arg = *args++;
+            nb_oargs = arg >> 16;
+            nb_iargs = arg & 0xffff;
+            nb_cargs = def->nb_cargs;
+        } else if (c == INDEX_op_nopn) {
+            /* variable number of arguments */
+            nb_cargs = *args;
+            nb_oargs = 0;
+            nb_iargs = 0;
+        } else {
+            nb_oargs = def->nb_oargs;
+            nb_iargs = def->nb_iargs;
+            nb_cargs = def->nb_cargs;
+        }
+
+        k = 0;
+        for(i = 0; i < nb_oargs; i++) {
+            if (k != 0)
+                fprintf(outfile, ",");
+            fprintf(outfile, "%s", tcg_get_arg_str(s, buf, sizeof(buf), args[k++]));
+        }
+        for(i = 0; i < nb_iargs; i++) {
+            if (k != 0)
+                fprintf(outfile, ",");
+            /* XXX: dump helper name for call */
+            fprintf(outfile, "%s", tcg_get_arg_str(s, buf, sizeof(buf), args[k++]));
+        }
+        for(i = 0; i < nb_cargs; i++) {
+            if (k != 0)
+                fprintf(outfile, ",");
+            arg = args[k++];
+            fprintf(outfile, "$0x%" TCG_PRIlx, arg);
+        }
+        fprintf(outfile, "\n");
+        args += nb_iargs + nb_oargs + nb_cargs;
+    }
+}
+
+/* we give more priority to constraints with less registers */
+static int get_constraint_priority(const TCGOpDef *def, int k)
+{
+    const TCGArgConstraint *arg_ct;
+
+    int i, n;
+    arg_ct = &def->args_ct[k];
+    if (arg_ct->ct & TCG_CT_ALIAS) {
+        /* an alias is equivalent to a single register */
+        n = 1;
+    } else {
+        if (!(arg_ct->ct & TCG_CT_REG))
+            return 0;
+        n = 0;
+        for(i = 0; i < TCG_TARGET_NB_REGS; i++) {
+            if (tcg_regset_test_reg(arg_ct->u.regs, i))
+                n++;
+        }
+    }
+    return TCG_TARGET_NB_REGS - n + 1;
+}
+
+/* sort from highest priority to lowest */
+static void sort_constraints(TCGOpDef *def, int start, int n)
+{
+    int i, j, p1, p2, tmp;
+
+    for(i = 0; i < n; i++)
+        def->sorted_args[start + i] = start + i;
+    if (n <= 1)
+        return;
+    for(i = 0; i < n - 1; i++) {
+        for(j = i + 1; j < n; j++) {
+            p1 = get_constraint_priority(def, def->sorted_args[start + i]);
+            p2 = get_constraint_priority(def, def->sorted_args[start + j]);
+            if (p1 < p2) {
+                tmp = def->sorted_args[start + i];
+                def->sorted_args[start + i] = def->sorted_args[start + j];
+                def->sorted_args[start + j] = tmp;
+            }
+        }
+    }
+}
+
+void tcg_add_target_add_op_defs(const TCGTargetOpDef *tdefs)
+{
+    int op;
+    TCGOpDef *def;
+    const char *ct_str;
+    int i, nb_args;
+
+    for(;;) {
+        if (tdefs->op < 0)
+            break;
+        op = tdefs->op;
+        assert(op >= 0 && op < NB_OPS);
+        def = &tcg_op_defs[op];
+        nb_args = def->nb_iargs + def->nb_oargs;
+        for(i = 0; i < nb_args; i++) {
+            ct_str = tdefs->args_ct_str[i];
+            tcg_regset_clear(def->args_ct[i].u.regs);
+            def->args_ct[i].ct = 0;
+            if (ct_str[0] >= '0' && ct_str[0] <= '9') {
+                int oarg;
+                oarg = ct_str[0] - '0';
+                assert(oarg < def->nb_oargs);
+                assert(def->args_ct[oarg].ct & TCG_CT_REG);
+                /* TCG_CT_ALIAS is for the output arguments. The input
+                   argument is tagged with TCG_CT_IALIAS for
+                   informative purposes. */
+                def->args_ct[i] = def->args_ct[oarg];
+                def->args_ct[oarg].ct = i | TCG_CT_ALIAS;
+                def->args_ct[i].ct |= TCG_CT_IALIAS;
+            } else {
+                for(;;) {
+                    if (*ct_str == '\0')
+                        break;
+                    switch(*ct_str) {
+                    case 'i':
+                        def->args_ct[i].ct |= TCG_CT_CONST;
+                        ct_str++;
+                        break;
+                    default:
+                        if (target_parse_constraint(&def->args_ct[i], &ct_str) < 0) {
+                            fprintf(stderr, "Invalid constraint '%s' for arg %d of operation '%s'\n",
+                                    ct_str, i, def->name);
+                            exit(1);
+                        }
+                    }
+                }
+            }
+        }
+
+        /* sort the constraints (XXX: this is just an heuristic) */
+        sort_constraints(def, 0, def->nb_oargs);
+        sort_constraints(def, def->nb_oargs, def->nb_iargs);
+
+#if 0
+        {
+            int i;
+
+            printf("%s: sorted=", def->name);
+            for(i = 0; i < def->nb_oargs + def->nb_iargs; i++)
+                printf(" %d", def->sorted_args[i]);
+            printf("\n");
+        }
+#endif
+        tdefs++;
+    }
+
+}
+
+#ifdef USE_LIVENESS_ANALYSIS
+
+/* set a nop for an operation using 'nb_args' */
+static inline void tcg_set_nop(TCGContext *s, uint16_t *opc_ptr, 
+                               TCGArg *args, int nb_args)
+{
+    if (nb_args == 0) {
+        *opc_ptr = INDEX_op_nop;
+    } else {
+        *opc_ptr = INDEX_op_nopn;
+        args[0] = nb_args;
+        args[nb_args - 1] = nb_args;
+    }
+}
+
+/* liveness analysis: end of basic block: globals are live, temps are dead */
+static inline void tcg_la_bb_end(TCGContext *s, uint8_t *dead_temps)
+{
+    memset(dead_temps, 0, s->nb_globals);
+    memset(dead_temps + s->nb_globals, 1, s->nb_temps - s->nb_globals);
+}
+
+/* Liveness analysis : update the opc_dead_iargs array to tell if a
+   given input arguments is dead. Instructions updating dead
+   temporaries are removed. */
+void tcg_liveness_analysis(TCGContext *s)
+{
+    int i, op_index, op, nb_args, nb_iargs, nb_oargs, arg, nb_ops;
+    TCGArg *args;
+    const TCGOpDef *def;
+    uint8_t *dead_temps;
+    unsigned int dead_iargs;
+    
+    gen_opc_ptr++; /* skip end */
+
+    nb_ops = gen_opc_ptr - gen_opc_buf;
+
+    /* XXX: make it really dynamic */
+    s->op_dead_iargs = tcg_malloc(OPC_BUF_SIZE * sizeof(uint16_t));
+    
+    dead_temps = tcg_malloc(s->nb_temps);
+    memset(dead_temps, 1, s->nb_temps);
+
+    args = gen_opparam_ptr;
+    op_index = nb_ops - 1;
+    while (op_index >= 0) {
+        op = gen_opc_buf[op_index];
+        def = &tcg_op_defs[op];
+        switch(op) {
+        case INDEX_op_call:
+            nb_args = args[-1];
+            args -= nb_args;
+            nb_iargs = args[0] & 0xffff;
+            nb_oargs = args[0] >> 16;
+            args++;
+
+            /* output args are dead */
+            for(i = 0; i < nb_oargs; i++) {
+                arg = args[i];
+                dead_temps[arg] = 1;
+            }
+            
+            /* globals are live (they may be used by the call) */
+            memset(dead_temps, 0, s->nb_globals);
+
+            /* input args are live */
+            dead_iargs = 0;
+            for(i = 0; i < nb_iargs; i++) {
+                arg = args[i + nb_oargs];
+                if (dead_temps[arg]) {
+                    dead_iargs |= (1 << i);
+                }
+                dead_temps[arg] = 0;
+            }
+            s->op_dead_iargs[op_index] = dead_iargs;
+            args--;
+            break;
+        case INDEX_op_set_label:
+            args--;
+            /* mark end of basic block */
+            tcg_la_bb_end(s, dead_temps);
+            break;
+        case INDEX_op_nopn:
+            nb_args = args[-1];
+            args -= nb_args;
+            break;
+        case INDEX_op_macro_2:
+            {
+                int dead_args[2], macro_id;
+                int saved_op_index, saved_arg_index;
+                int macro_op_index, macro_arg_index;
+                int macro_end_op_index, macro_end_arg_index;
+                int last_nb_temps;
+                
+                nb_args = 3;
+                args -= nb_args;
+                dead_args[0] = dead_temps[args[0]];
+                dead_args[1] = dead_temps[args[1]];
+                macro_id = args[2];
+
+                /* call the macro function which generate code
+                   depending on the live outputs */
+                saved_op_index = op_index;
+                saved_arg_index = args - gen_opparam_buf;
+
+                /* add a macro start instruction */
+                *gen_opc_ptr++ = INDEX_op_macro_start;
+                *gen_opparam_ptr++ = saved_op_index;
+                *gen_opparam_ptr++ = saved_arg_index;
+
+                macro_op_index = gen_opc_ptr - gen_opc_buf;
+                macro_arg_index = gen_opparam_ptr -  gen_opparam_buf;
+
+                last_nb_temps = s->nb_temps;
+
+                s->macro_func(s, macro_id, dead_args);
+
+                /* realloc temp info (XXX: make it faster) */
+                if (s->nb_temps > last_nb_temps) {
+                    uint8_t *new_dead_temps;
+
+                    new_dead_temps = tcg_malloc(s->nb_temps);
+                    memcpy(new_dead_temps, dead_temps, last_nb_temps);
+                    memset(new_dead_temps + last_nb_temps, 1, 
+                           s->nb_temps - last_nb_temps);
+                    dead_temps = new_dead_temps;
+                }
+
+                macro_end_op_index = gen_opc_ptr - gen_opc_buf;
+                macro_end_arg_index = gen_opparam_ptr - gen_opparam_buf;
+
+                /* end of macro: add a goto to the next instruction */
+                *gen_opc_ptr++ = INDEX_op_macro_end;
+                *gen_opparam_ptr++ = op_index + 1;
+                *gen_opparam_ptr++ = saved_arg_index + nb_args;
+
+                /* modify the macro operation to be a macro_goto */
+                gen_opc_buf[op_index] = INDEX_op_macro_goto;
+                args[0] = macro_op_index;
+                args[1] = macro_arg_index;
+                args[2] = 0; /* dummy third arg to match the 
+                                macro parameters */
+
+                /* set the next instruction to the end of the macro */
+                op_index = macro_end_op_index;
+                args = macro_end_arg_index + gen_opparam_buf;
+            }
+            break;
+        case INDEX_op_macro_start:
+            args -= 2;
+            op_index = args[0];
+            args = gen_opparam_buf + args[1];
+            break;
+        case INDEX_op_macro_goto:
+        case INDEX_op_macro_end:
+            tcg_abort(); /* should never happen in liveness analysis */
+        case INDEX_op_end:
+            break;
+            /* XXX: optimize by hardcoding common cases (e.g. triadic ops) */
+        default:
+            if (op > INDEX_op_end) {
+                args -= def->nb_args;
+                nb_iargs = def->nb_iargs;
+                nb_oargs = def->nb_oargs;
+
+                /* Test if the operation can be removed because all
+                   its outputs are dead. We may add a flag to
+                   explicitely tell if the op has side
+                   effects. Currently we assume that if nb_oargs == 0
+                   or OPF_BB_END is set, the operation has side
+                   effects and cannot be removed */
+                if (nb_oargs != 0 && !(def->flags & TCG_OPF_BB_END)) {
+                    for(i = 0; i < nb_oargs; i++) {
+                        arg = args[i];
+                        if (!dead_temps[arg])
+                            goto do_not_remove;
+                    }
+                    tcg_set_nop(s, gen_opc_buf + op_index, args, def->nb_args);
+#ifdef CONFIG_PROFILER
+                    {
+                        extern int64_t dyngen_tcg_del_op_count;
+                        dyngen_tcg_del_op_count++;
+                    }
+#endif
+                } else {
+                do_not_remove:
+
+                    /* output args are dead */
+                    for(i = 0; i < nb_oargs; i++) {
+                        arg = args[i];
+                        dead_temps[arg] = 1;
+                    }
+                    
+                    /* if end of basic block, update */
+                    if (def->flags & TCG_OPF_BB_END) {
+                        tcg_la_bb_end(s, dead_temps);
+                    }
+                    
+                    /* input args are live */
+                    dead_iargs = 0;
+                    for(i = 0; i < nb_iargs; i++) {
+                        arg = args[i + nb_oargs];
+                        if (dead_temps[arg]) {
+                            dead_iargs |= (1 << i);
+                        }
+                        dead_temps[arg] = 0;
+                    }
+                    s->op_dead_iargs[op_index] = dead_iargs;
+                }
+            } else {
+                /* legacy dyngen operations */
+                args -= def->nb_args;
+                /* mark end of basic block */
+                tcg_la_bb_end(s, dead_temps);
+            }
+            break;
+        }
+        op_index--;
+    }
+
+    if (args != gen_opparam_buf)
+        tcg_abort();
+}
+#else
+/* dummy liveness analysis */
+void tcg_liveness_analysis(TCGContext *s)
+{
+    int nb_ops;
+    nb_ops = gen_opc_ptr - gen_opc_buf;
+
+    s->op_dead_iargs = tcg_malloc(nb_ops * sizeof(uint16_t));
+    memset(s->op_dead_iargs, 0, nb_ops * sizeof(uint16_t));
+}
+#endif
+
+#ifndef NDEBUG
+static void dump_regs(TCGContext *s)
+{
+    TCGTemp *ts;
+    int i;
+    char buf[64];
+
+    for(i = 0; i < s->nb_temps; i++) {
+        ts = &s->temps[i];
+        printf("  %10s: ", tcg_get_arg_str(s, buf, sizeof(buf), i));
+        switch(ts->val_type) {
+        case TEMP_VAL_REG:
+            printf("%s", tcg_target_reg_names[ts->reg]);
+            break;
+        case TEMP_VAL_MEM:
+            printf("%d(%s)", (int)ts->mem_offset, tcg_target_reg_names[ts->mem_reg]);
+            break;
+        case TEMP_VAL_CONST:
+            printf("$0x%" TCG_PRIlx, ts->val);
+            break;
+        case TEMP_VAL_DEAD:
+            printf("D");
+            break;
+        default:
+            printf("???");
+            break;
+        }
+        printf("\n");
+    }
+
+    for(i = 0; i < TCG_TARGET_NB_REGS; i++) {
+        if (s->reg_to_temp[i] >= 0) {
+            printf("%s: %s\n", 
+                   tcg_target_reg_names[i], 
+                   tcg_get_arg_str(s, buf, sizeof(buf), s->reg_to_temp[i]));
+        }
+    }
+}
+
+static void check_regs(TCGContext *s)
+{
+    int reg, k;
+    TCGTemp *ts;
+    char buf[64];
+
+    for(reg = 0; reg < TCG_TARGET_NB_REGS; reg++) {
+        k = s->reg_to_temp[reg];
+        if (k >= 0) {
+            ts = &s->temps[k];
+            if (ts->val_type != TEMP_VAL_REG ||
+                ts->reg != reg) {
+                printf("Inconsistency for register %s:\n", 
+                       tcg_target_reg_names[reg]);
+                printf("reg state:\n");
+                dump_regs(s);
+                tcg_abort();
+            }
+        }
+    }
+    for(k = 0; k < s->nb_temps; k++) {
+        ts = &s->temps[k];
+        if (ts->val_type == TEMP_VAL_REG &&
+            !ts->fixed_reg &&
+            s->reg_to_temp[ts->reg] != k) {
+                printf("Inconsistency for temp %s:\n", 
+                       tcg_get_arg_str(s, buf, sizeof(buf), k));
+                printf("reg state:\n");
+                dump_regs(s);
+                tcg_abort();
+        }
+    }
+}
+#endif
+
+static void temp_allocate_frame(TCGContext *s, int temp)
+{
+    TCGTemp *ts;
+    ts = &s->temps[temp];
+    s->current_frame_offset = (s->current_frame_offset + sizeof(tcg_target_long) - 1) & ~(sizeof(tcg_target_long) - 1);
+    if (s->current_frame_offset + sizeof(tcg_target_long) > s->frame_end)
+        abort();
+    ts->mem_offset = s->current_frame_offset;
+    ts->mem_reg = s->frame_reg;
+    ts->mem_allocated = 1;
+    s->current_frame_offset += sizeof(tcg_target_long);
+}
+
+/* free register 'reg' by spilling the corresponding temporary if necessary */
+static void tcg_reg_free(TCGContext *s, int reg)
+{
+    TCGTemp *ts;
+    int temp;
+
+    temp = s->reg_to_temp[reg];
+    if (temp != -1) {
+        ts = &s->temps[temp];
+        assert(ts->val_type == TEMP_VAL_REG);
+        if (!ts->mem_coherent) {
+            if (!ts->mem_allocated) 
+                temp_allocate_frame(s, temp);
+            tcg_out_st(s, reg, ts->mem_reg, ts->mem_offset);
+        }
+        ts->val_type = TEMP_VAL_MEM;
+        s->reg_to_temp[reg] = -1;
+    }
+}
+
+/* Allocate a register belonging to reg1 & ~reg2 */
+static int tcg_reg_alloc(TCGContext *s, TCGRegSet reg1, TCGRegSet reg2)
+{
+    int i, reg;
+    TCGRegSet reg_ct;
+
+    tcg_regset_andnot(reg_ct, reg1, reg2);
+
+    /* first try free registers */
+    for(i = 0; i < TCG_TARGET_NB_REGS; i++) {
+        reg = tcg_target_reg_alloc_order[i];
+        if (tcg_regset_test_reg(reg_ct, reg) && s->reg_to_temp[reg] == -1)
+            return reg;
+    }
+
+    /* XXX: do better spill choice */
+    for(i = 0; i < TCG_TARGET_NB_REGS; i++) {
+        reg = tcg_target_reg_alloc_order[i];
+        if (tcg_regset_test_reg(reg_ct, reg)) {
+            tcg_reg_free(s, reg);
+            return reg;
+        }
+    }
+
+    tcg_abort();
+}
+
+/* at the end of a basic block, we assume all temporaries are dead and
+   all globals are stored at their canonical location */
+/* XXX: optimize by handling constants in another array ? */
+void tcg_reg_alloc_bb_end(TCGContext *s)
+{
+    TCGTemp *ts;
+    int i;
+
+    for(i = 0; i < s->nb_globals; i++) {
+        ts = &s->temps[i];
+        if (!ts->fixed_reg) {
+            if (ts->val_type == TEMP_VAL_REG) {
+                tcg_reg_free(s, ts->reg);
+            }
+        }
+    }
+
+    for(i = s->nb_globals; i < s->nb_temps; i++) {
+        ts = &s->temps[i];
+        if (ts->val_type != TEMP_VAL_CONST) {
+            if (ts->val_type == TEMP_VAL_REG) {
+                s->reg_to_temp[ts->reg] = -1;
+            }
+            ts->val_type = TEMP_VAL_DEAD;
+        }
+    }
+}
+
+#define IS_DEAD_IARG(n) ((dead_iargs >> (n)) & 1)
+
+static void tcg_reg_alloc_mov(TCGContext *s, const TCGOpDef *def,
+                              const TCGArg *args,
+                              unsigned int dead_iargs)
+{
+    TCGTemp *ts, *ots;
+    int reg;
+    const TCGArgConstraint *arg_ct;
+
+    ots = &s->temps[args[0]];
+    ts = &s->temps[args[1]];
+    arg_ct = &def->args_ct[0];
+
+    if (ts->val_type == TEMP_VAL_REG) {
+        if (IS_DEAD_IARG(0) && !ts->fixed_reg && !ots->fixed_reg) {
+            /* the mov can be suppressed */
+            if (ots->val_type == TEMP_VAL_REG)
+                s->reg_to_temp[ots->reg] = -1;
+            reg = ts->reg;
+            s->reg_to_temp[reg] = -1;
+            ts->val_type = TEMP_VAL_DEAD;
+        } else {
+            if (ots->val_type == TEMP_VAL_REG) {
+                reg = ots->reg;
+            } else {
+                reg = tcg_reg_alloc(s, arg_ct->u.regs, s->reserved_regs);
+            }
+            if (ts->reg != reg) {
+                tcg_out_mov(s, reg, ts->reg);
+            }
+        }
+    } else if (ts->val_type == TEMP_VAL_MEM) {
+        if (ots->val_type == TEMP_VAL_REG) {
+            reg = ots->reg;
+        } else {
+            reg = tcg_reg_alloc(s, arg_ct->u.regs, s->reserved_regs);
+        }
+        tcg_out_ld(s, reg, ts->mem_reg, ts->mem_offset);
+    } else if (ts->val_type == TEMP_VAL_CONST) {
+        if (ots->val_type == TEMP_VAL_REG) {
+            reg = ots->reg;
+        } else {
+            reg = tcg_reg_alloc(s, arg_ct->u.regs, s->reserved_regs);
+        }
+        tcg_out_movi(s, ots->type, reg, ts->val);
+    } else {
+        tcg_abort();
+    }
+    s->reg_to_temp[reg] = args[0];
+    ots->reg = reg;
+    ots->val_type = TEMP_VAL_REG;
+    ots->mem_coherent = 0;
+}
+
+static void tcg_reg_alloc_op(TCGContext *s, 
+                             const TCGOpDef *def, int opc,
+                             const TCGArg *args,
+                             unsigned int dead_iargs)
+{
+    TCGRegSet allocated_regs;
+    int i, k, nb_iargs, nb_oargs, reg;
+    TCGArg arg;
+    const TCGArgConstraint *arg_ct;
+    TCGTemp *ts;
+    TCGArg new_args[TCG_MAX_OP_ARGS];
+    int const_args[TCG_MAX_OP_ARGS];
+
+    nb_oargs = def->nb_oargs;
+    nb_iargs = def->nb_iargs;
+
+    /* copy constants */
+    memcpy(new_args + nb_oargs + nb_iargs, 
+           args + nb_oargs + nb_iargs, 
+           sizeof(TCGArg) * def->nb_cargs);
+
+    /* satisfy input constraints */ 
+    tcg_regset_set(allocated_regs, s->reserved_regs);
+    for(k = 0; k < nb_iargs; k++) {
+        i = def->sorted_args[nb_oargs + k];
+        arg = args[i];
+        arg_ct = &def->args_ct[i];
+        ts = &s->temps[arg];
+        if (ts->val_type == TEMP_VAL_MEM) {
+            reg = tcg_reg_alloc(s, arg_ct->u.regs, allocated_regs);
+            tcg_out_ld(s, reg, ts->mem_reg, ts->mem_offset);
+            ts->val_type = TEMP_VAL_REG;
+            ts->reg = reg;
+            ts->mem_coherent = 1;
+            s->reg_to_temp[reg] = arg;
+        } else if (ts->val_type == TEMP_VAL_CONST) {
+            if (tcg_target_const_match(ts->val, arg_ct)) {
+                /* constant is OK for instruction */
+                const_args[i] = 1;
+                new_args[i] = ts->val;
+                goto iarg_end;
+            } else {
+                /* need to move to a register*/
+                reg = tcg_reg_alloc(s, arg_ct->u.regs, allocated_regs);
+                tcg_out_movi(s, ts->type, reg, ts->val);
+                goto iarg_end1;
+            }
+        }
+        assert(ts->val_type == TEMP_VAL_REG);
+        if ((arg_ct->ct & TCG_CT_IALIAS) &&
+            !IS_DEAD_IARG(i - nb_oargs)) {
+            /* if the input is aliased to an output and if it is
+               not dead after the instruction, we must allocate
+               a new register and move it */
+            goto allocate_in_reg;
+        }
+        reg = ts->reg;
+        if (tcg_regset_test_reg(arg_ct->u.regs, reg)) {
+            /* nothing to do : the constraint is satisfied */
+        } else {
+        allocate_in_reg:
+            /* allocate a new register matching the constraint 
+               and move the temporary register into it */
+            reg = tcg_reg_alloc(s, arg_ct->u.regs, allocated_regs);
+            tcg_out_mov(s, reg, ts->reg);
+        }
+    iarg_end1:
+        new_args[i] = reg;
+        const_args[i] = 0;
+        tcg_regset_set_reg(allocated_regs, reg);
+    iarg_end: ;
+    }
+    
+    /* mark dead temporaries and free the associated registers */
+    for(i = 0; i < nb_iargs; i++) {
+        arg = args[nb_oargs + i];
+        if (IS_DEAD_IARG(i)) {
+            ts = &s->temps[arg];
+            if (ts->val_type != TEMP_VAL_CONST && !ts->fixed_reg) {
+                if (ts->val_type == TEMP_VAL_REG)
+                    s->reg_to_temp[ts->reg] = -1;
+                ts->val_type = TEMP_VAL_DEAD;
+            }
+        }
+    }
+
+    /* XXX: permit generic clobber register list ? */ 
+    if (def->flags & TCG_OPF_CALL_CLOBBER) {
+        for(reg = 0; reg < TCG_TARGET_NB_REGS; reg++) {
+            if (tcg_regset_test_reg(tcg_target_call_clobber_regs, reg)) {
+                tcg_reg_free(s, reg);
+            }
+        }
+    }
+
+    /* satisfy the output constraints */
+    tcg_regset_set(allocated_regs, s->reserved_regs);
+    for(k = 0; k < nb_oargs; k++) {
+        i = def->sorted_args[k];
+        arg = args[i];
+        arg_ct = &def->args_ct[i];
+        ts = &s->temps[arg];
+        if (arg_ct->ct & TCG_CT_ALIAS) {
+            reg = new_args[arg_ct->ct & ~TCG_CT_ALIAS];
+        } else {
+            /* if fixed register, we try to use it */
+            reg = ts->reg;
+            if (ts->fixed_reg &&
+                tcg_regset_test_reg(arg_ct->u.regs, reg)) {
+                goto oarg_end;
+            }
+            reg = tcg_reg_alloc(s, arg_ct->u.regs, allocated_regs);
+        }
+        tcg_regset_set_reg(allocated_regs, reg);
+        /* if a fixed register is used, then a move will be done afterwards */
+        if (!ts->fixed_reg) {
+            if (ts->val_type == TEMP_VAL_REG)
+                s->reg_to_temp[ts->reg] = -1;
+            ts->val_type = TEMP_VAL_REG;
+            ts->reg = reg;
+            /* temp value is modified, so the value kept in memory is
+               potentially not the same */
+            ts->mem_coherent = 0; 
+            s->reg_to_temp[reg] = arg;
+        }
+    oarg_end:
+        new_args[i] = reg;
+    }
+
+    if (def->flags & TCG_OPF_BB_END)
+        tcg_reg_alloc_bb_end(s);
+
+    /* emit instruction */
+    tcg_out_op(s, opc, new_args, const_args);
+    
+    /* move the outputs in the correct register if needed */
+    for(i = 0; i < nb_oargs; i++) {
+        ts = &s->temps[args[i]];
+        reg = new_args[i];
+        if (ts->fixed_reg && ts->reg != reg) {
+            tcg_out_mov(s, ts->reg, reg);
+        }
+    }
+}
+
+static int tcg_reg_alloc_call(TCGContext *s, const TCGOpDef *def,
+                              int opc, const TCGArg *args,
+                              unsigned int dead_iargs)
+{
+    int nb_iargs, nb_oargs, flags, nb_regs, i, reg, nb_params;
+    TCGArg arg, func_arg;
+    TCGTemp *ts;
+    tcg_target_long stack_offset, call_stack_size;
+    int const_func_arg;
+    TCGRegSet allocated_regs;
+    const TCGArgConstraint *arg_ct;
+
+    arg = *args++;
+
+    nb_oargs = arg >> 16;
+    nb_iargs = arg & 0xffff;
+    nb_params = nb_iargs - 1;
+
+    flags = args[nb_oargs + nb_iargs];
+
+    nb_regs = tcg_target_get_call_iarg_regs_count(flags);
+    if (nb_regs > nb_params)
+        nb_regs = nb_params;
+
+    /* assign stack slots first */
+    /* XXX: preallocate call stack */
+    call_stack_size = (nb_params - nb_regs) * sizeof(tcg_target_long);
+    call_stack_size = (call_stack_size + TCG_TARGET_STACK_ALIGN - 1) & 
+        ~(TCG_TARGET_STACK_ALIGN - 1);
+    tcg_out_addi(s, TCG_REG_CALL_STACK, -call_stack_size);
+
+    stack_offset = 0;
+    for(i = nb_regs; i < nb_params; i++) {
+        arg = args[nb_oargs + i];
+        ts = &s->temps[arg];
+        if (ts->val_type == TEMP_VAL_REG) {
+            tcg_out_st(s, ts->reg, TCG_REG_CALL_STACK, stack_offset);
+        } else if (ts->val_type == TEMP_VAL_MEM) {
+            reg = tcg_reg_alloc(s, tcg_target_available_regs[ts->type], 
+                                s->reserved_regs);
+            /* XXX: not correct if reading values from the stack */
+            tcg_out_ld(s, reg, ts->mem_reg, ts->mem_offset);
+            tcg_out_st(s, reg, TCG_REG_CALL_STACK, stack_offset);
+        } else if (ts->val_type == TEMP_VAL_CONST) {
+            reg = tcg_reg_alloc(s, tcg_target_available_regs[ts->type], 
+                                s->reserved_regs);
+            /* XXX: sign extend may be needed on some targets */
+            tcg_out_movi(s, ts->type, reg, ts->val);
+            tcg_out_st(s, reg, TCG_REG_CALL_STACK, stack_offset);
+        } else {
+            tcg_abort();
+        }
+        stack_offset += sizeof(tcg_target_long);
+    }
+    
+    /* assign input registers */
+    tcg_regset_set(allocated_regs, s->reserved_regs);
+    for(i = 0; i < nb_regs; i++) {
+        arg = args[nb_oargs + i];
+        ts = &s->temps[arg];
+        reg = tcg_target_call_iarg_regs[i];
+        tcg_reg_free(s, reg);
+        if (ts->val_type == TEMP_VAL_REG) {
+            if (ts->reg != reg) {
+                tcg_out_mov(s, reg, ts->reg);
+            }
+        } else if (ts->val_type == TEMP_VAL_MEM) {
+            tcg_out_ld(s, reg, ts->mem_reg, ts->mem_offset);
+        } else if (ts->val_type == TEMP_VAL_CONST) {
+            /* XXX: sign extend ? */
+            tcg_out_movi(s, ts->type, reg, ts->val);
+        } else {
+            tcg_abort();
+        }
+        tcg_regset_set_reg(allocated_regs, reg);
+    }
+    
+    /* assign function address */
+    func_arg = args[nb_oargs + nb_iargs - 1];
+    arg_ct = &def->args_ct[0];
+    ts = &s->temps[func_arg];
+    const_func_arg = 0;
+    if (ts->val_type == TEMP_VAL_MEM) {
+        reg = tcg_reg_alloc(s, arg_ct->u.regs, allocated_regs);
+        tcg_out_ld(s, reg, ts->mem_reg, ts->mem_offset);
+        func_arg = reg;
+    } else if (ts->val_type == TEMP_VAL_REG) {
+        reg = ts->reg;
+        if (!tcg_regset_test_reg(arg_ct->u.regs, reg)) {
+            reg = tcg_reg_alloc(s, arg_ct->u.regs, allocated_regs);
+            tcg_out_mov(s, reg, ts->reg);
+        }
+        func_arg = reg;
+    } else if (ts->val_type == TEMP_VAL_CONST) {
+        if (tcg_target_const_match(ts->val, arg_ct)) {
+            const_func_arg = 1;
+            func_arg = ts->val;
+        } else {
+            reg = tcg_reg_alloc(s, arg_ct->u.regs, allocated_regs);
+            tcg_out_movi(s, ts->type, reg, ts->val);
+            func_arg = reg;
+        }
+    } else {
+        tcg_abort();
+    }
+    
+    /* mark dead temporaries and free the associated registers */
+    for(i = 0; i < nb_params; i++) {
+        arg = args[nb_oargs + i];
+        if (IS_DEAD_IARG(i)) {
+            ts = &s->temps[arg];
+            if (ts->val_type != TEMP_VAL_CONST && !ts->fixed_reg) {
+                if (ts->val_type == TEMP_VAL_REG)
+                    s->reg_to_temp[ts->reg] = -1;
+                ts->val_type = TEMP_VAL_DEAD;
+            }
+        }
+    }
+    
+    /* clobber call registers */
+    for(reg = 0; reg < TCG_TARGET_NB_REGS; reg++) {
+        if (tcg_regset_test_reg(tcg_target_call_clobber_regs, reg)) {
+            tcg_reg_free(s, reg);
+        }
+    }
+    
+    /* store globals and free associated registers (we assume the call
+       can modify any global. */
+    for(i = 0; i < s->nb_globals; i++) {
+        ts = &s->temps[i];
+        if (!ts->fixed_reg) {
+            if (ts->val_type == TEMP_VAL_REG) {
+                tcg_reg_free(s, ts->reg);
+            }
+        }
+    }
+
+    tcg_out_op(s, opc, &func_arg, &const_func_arg);
+    
+    tcg_out_addi(s, TCG_REG_CALL_STACK, call_stack_size);
+
+    /* assign output registers and emit moves if needed */
+    for(i = 0; i < nb_oargs; i++) {
+        arg = args[i];
+        ts = &s->temps[arg];
+        reg = tcg_target_call_oarg_regs[i];
+        tcg_reg_free(s, reg);
+        if (ts->fixed_reg) {
+            if (ts->reg != reg) {
+                tcg_out_mov(s, ts->reg, reg);
+            }
+        } else {
+            if (ts->val_type == TEMP_VAL_REG)
+                s->reg_to_temp[ts->reg] = -1;
+            ts->val_type = TEMP_VAL_REG;
+            ts->reg = reg;
+            ts->mem_coherent = 0; 
+            s->reg_to_temp[reg] = arg;
+        }
+    }
+    
+    return nb_iargs + nb_oargs + def->nb_cargs + 1;
+}
+
+#ifdef CONFIG_PROFILER
+
+static int64_t dyngen_table_op_count[NB_OPS];
+
+void dump_op_count(void)
+{
+    int i;
+    FILE *f;
+    f = fopen("/tmp/op1.log", "w");
+    for(i = 0; i < INDEX_op_end; i++) {
+        fprintf(f, "%s %" PRId64 "\n", tcg_op_defs[i].name, dyngen_table_op_count[i]);
+    }
+    fclose(f);
+    f = fopen("/tmp/op2.log", "w");
+    for(i = INDEX_op_end; i < NB_OPS; i++) {
+        fprintf(f, "%s %" PRId64 "\n", tcg_op_defs[i].name, dyngen_table_op_count[i]);
+    }
+    fclose(f);
+}
+#endif
+
+
+static inline int tcg_gen_code_common(TCGContext *s, uint8_t *gen_code_buf,
+                                      int do_search_pc,
+                                      const uint8_t *searched_pc)
+{
+    int opc, op_index, macro_op_index;
+    const TCGOpDef *def;
+    unsigned int dead_iargs;
+    const TCGArg *args;
+
+#ifdef DEBUG_DISAS
+    if (unlikely(loglevel & CPU_LOG_TB_OP)) {
+        fprintf(logfile, "OP:\n");
+        tcg_dump_ops(s, logfile);
+        fprintf(logfile, "\n");
+    }
+#endif
+
+    tcg_liveness_analysis(s);
+
+#ifdef DEBUG_DISAS
+    if (unlikely(loglevel & CPU_LOG_TB_OP_OPT)) {
+        fprintf(logfile, "OP after la:\n");
+        tcg_dump_ops(s, logfile);
+        fprintf(logfile, "\n");
+    }
+#endif
+
+    tcg_reg_alloc_start(s);
+
+    s->code_buf = gen_code_buf;
+    s->code_ptr = gen_code_buf;
+
+    macro_op_index = -1;
+    args = gen_opparam_buf;
+    op_index = 0;
+    for(;;) {
+        opc = gen_opc_buf[op_index];
+#ifdef CONFIG_PROFILER
+        dyngen_table_op_count[opc]++;
+#endif
+        def = &tcg_op_defs[opc];
+#if 0
+        printf("%s: %d %d %d\n", def->name,
+               def->nb_oargs, def->nb_iargs, def->nb_cargs);
+        //        dump_regs(s);
+#endif
+        switch(opc) {
+        case INDEX_op_mov_i32:
+#if TCG_TARGET_REG_BITS == 64
+        case INDEX_op_mov_i64:
+#endif
+            dead_iargs = s->op_dead_iargs[op_index];
+            tcg_reg_alloc_mov(s, def, args, dead_iargs);
+            break;
+        case INDEX_op_nop:
+        case INDEX_op_nop1:
+        case INDEX_op_nop2:
+        case INDEX_op_nop3:
+            break;
+        case INDEX_op_nopn:
+            args += args[0];
+            goto next;
+        case INDEX_op_macro_goto:
+            macro_op_index = op_index; /* only used for exceptions */
+            op_index = args[0] - 1;
+            args = gen_opparam_buf + args[1];
+            goto next;
+        case INDEX_op_macro_end:
+            macro_op_index = -1; /* only used for exceptions */
+            op_index = args[0] - 1;
+            args = gen_opparam_buf + args[1];
+            goto next;
+        case INDEX_op_macro_start:
+            /* must never happen here */
+            tcg_abort();
+        case INDEX_op_set_label:
+            tcg_reg_alloc_bb_end(s);
+            tcg_out_label(s, args[0], (long)s->code_ptr);
+            break;
+        case INDEX_op_call:
+            dead_iargs = s->op_dead_iargs[op_index];
+            args += tcg_reg_alloc_call(s, def, opc, args, dead_iargs);
+            goto next;
+        case INDEX_op_end:
+            goto the_end;
+        case 0 ... INDEX_op_end - 1:
+            /* legacy dyngen ops */
+#ifdef CONFIG_PROFILER
+            {
+                extern int64_t dyngen_old_op_count;
+                dyngen_old_op_count++;
+            }
+#endif
+            tcg_reg_alloc_bb_end(s);
+            if (do_search_pc) {
+                s->code_ptr += def->copy_size;
+                args += def->nb_args;
+            } else {
+                args = dyngen_op(s, opc, args);
+            }
+            goto next;
+        default:
+            /* Note: in order to speed up the code, it would be much
+               faster to have specialized register allocator functions for
+               some common argument patterns */
+            dead_iargs = s->op_dead_iargs[op_index];
+            tcg_reg_alloc_op(s, def, opc, args, dead_iargs);
+            break;
+        }
+        args += def->nb_args;
+    next: ;
+        if (do_search_pc) {
+            if (searched_pc < s->code_ptr) {
+                if (macro_op_index >= 0)
+                    return macro_op_index;
+                else
+                    return op_index;
+            }
+        }
+        op_index++;
+#ifndef NDEBUG
+        check_regs(s);
+#endif
+    }
+ the_end:
+    return -1;
+}
+
+int dyngen_code(TCGContext *s, uint8_t *gen_code_buf)
+{
+#ifdef CONFIG_PROFILER
+    {
+        extern int64_t dyngen_op_count;
+        extern int dyngen_op_count_max;
+        int n;
+        n = (gen_opc_ptr - gen_opc_buf);
+        dyngen_op_count += n;
+        if (n > dyngen_op_count_max)
+            dyngen_op_count_max = n;
+    }
+#endif
+
+    tcg_gen_code_common(s, gen_code_buf, 0, NULL);
+
+    /* flush instruction cache */
+    flush_icache_range((unsigned long)gen_code_buf, 
+                       (unsigned long)s->code_ptr);
+    return s->code_ptr -  gen_code_buf;
+}
+
+/* return the index of the micro operation such as the pc after is <
+   search_pc. Note: gen_code_buf is accessed during the operation, but
+   its content should not be modified. Return -1 if not found. */
+int dyngen_code_search_pc(TCGContext *s, uint8_t *gen_code_buf,
+                          const uint8_t *searched_pc)
+{
+    return tcg_gen_code_common(s, gen_code_buf, 1, searched_pc);
+}
diff --git a/tcg/tcg.h b/tcg/tcg.h
new file mode 100644
index 0000000000..e57e5deb01
--- /dev/null
+++ b/tcg/tcg.h
@@ -0,0 +1,324 @@
+/*
+ * Tiny Code Generator for QEMU
+ *
+ * Copyright (c) 2008 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#include "tcg-target.h"
+
+#if TCG_TARGET_REG_BITS == 32
+typedef int32_t tcg_target_long;
+typedef uint32_t tcg_target_ulong;
+#define TCG_PRIlx PRIx32
+#define TCG_PRIld PRId32
+#elif TCG_TARGET_REG_BITS == 64
+typedef int64_t tcg_target_long;
+typedef uint64_t tcg_target_ulong;
+#define TCG_PRIlx PRIx64
+#define TCG_PRIld PRId64
+#else
+#error unsupported
+#endif
+
+#if TCG_TARGET_NB_REGS <= 32
+typedef uint32_t TCGRegSet;
+#elif TCG_TARGET_NB_REGS <= 64
+typedef uint64_t TCGRegSet;
+#else
+#error unsupported
+#endif
+
+enum {
+#define DEF(s, n, copy_size) INDEX_op_ ## s,
+#include "tcg-opc.h"
+#undef DEF
+    NB_OPS,
+};
+
+#define tcg_regset_clear(d) (d) = 0
+#define tcg_regset_set(d, s) (d) = (s)
+#define tcg_regset_set32(d, reg, val32) (d) |= (val32) << (reg)
+#define tcg_regset_set_reg(d, r) (d) |= 1 << (r)
+#define tcg_regset_reset_reg(d, r) (d) &= ~(1 << (r))
+#define tcg_regset_test_reg(d, r) (((d) >> (r)) & 1)
+#define tcg_regset_or(d, a, b) (d) = (a) | (b)
+#define tcg_regset_and(d, a, b) (d) = (a) & (b)
+#define tcg_regset_andnot(d, a, b) (d) = (a) & ~(b)
+#define tcg_regset_not(d, a) (d) = ~(a)
+
+typedef struct TCGRelocation {
+    struct TCGRelocation *next;
+    int type;
+    uint8_t *ptr;
+    tcg_target_long addend;
+} TCGRelocation; 
+
+typedef struct TCGLabel {
+    int has_value;
+    union {
+        tcg_target_ulong value;
+        TCGRelocation *first_reloc;
+    } u;
+} TCGLabel;
+
+typedef struct TCGPool {
+    struct TCGPool *next;
+    int size;
+    uint8_t data[0];
+} TCGPool;
+
+#define TCG_POOL_CHUNK_SIZE 32768
+
+#define TCG_MAX_LABELS 512
+
+#define TCG_MAX_TEMPS 256
+
+typedef int TCGType;
+
+#define TCG_TYPE_I32 0
+#define TCG_TYPE_I64 1
+
+#if TCG_TARGET_REG_BITS == 32
+#define TCG_TYPE_PTR TCG_TYPE_I32
+#else
+#define TCG_TYPE_PTR TCG_TYPE_I64
+#endif
+
+typedef tcg_target_ulong TCGArg;
+
+/* call flags */
+#define TCG_CALL_TYPE_MASK      0x000f
+#define TCG_CALL_TYPE_STD       0x0000 /* standard C call */
+#define TCG_CALL_TYPE_REGPARM_1 0x0001 /* i386 style regparm call (1 reg) */
+#define TCG_CALL_TYPE_REGPARM_2 0x0002 /* i386 style regparm call (2 regs) */
+#define TCG_CALL_TYPE_REGPARM   0x0003 /* i386 style regparm call (3 regs) */
+
+typedef enum {
+    TCG_COND_EQ,
+    TCG_COND_NE,
+    TCG_COND_LT,
+    TCG_COND_GE,
+    TCG_COND_LE,
+    TCG_COND_GT,
+    /* unsigned */
+    TCG_COND_LTU,
+    TCG_COND_GEU,
+    TCG_COND_LEU,
+    TCG_COND_GTU,
+} TCGCond;
+
+#define TEMP_VAL_DEAD  0
+#define TEMP_VAL_REG   1
+#define TEMP_VAL_MEM   2
+#define TEMP_VAL_CONST 3
+
+/* XXX: optimize memory layout */
+typedef struct TCGTemp {
+    TCGType base_type;
+    TCGType type;
+    int val_type;
+    int reg;
+    tcg_target_long val;
+    int mem_reg;
+    tcg_target_long mem_offset;
+    unsigned int fixed_reg:1;
+    unsigned int mem_coherent:1;
+    unsigned int mem_allocated:1;
+    const char *name;
+} TCGTemp;
+
+typedef struct TCGHelperInfo {
+    void *func;
+    const char *name;
+} TCGHelperInfo;
+
+typedef struct TCGContext TCGContext;
+
+typedef void TCGMacroFunc(TCGContext *s, int macro_id, const int *dead_args);
+
+struct TCGContext {
+    uint8_t *pool_cur, *pool_end;
+    TCGPool *pool_first, *pool_current;
+    TCGLabel *labels;
+    int nb_labels;
+    TCGTemp *temps; /* globals first, temps after */
+    int nb_globals;
+    int nb_temps;
+    /* constant indexes (end of temp array) */
+    int const_start;
+    int const_end;
+
+    /* goto_tb support */
+    uint8_t *code_buf;
+    unsigned long *tb_next;
+    uint16_t *tb_next_offset;
+    uint16_t *tb_jmp_offset; /* != NULL if USE_DIRECT_JUMP */
+
+    uint16_t *op_dead_iargs; /* for each operation, each bit tells if the
+                                corresponding input argument is dead */
+    /* tells in which temporary a given register is. It does not take
+       into account fixed registers */
+    int reg_to_temp[TCG_TARGET_NB_REGS];
+    TCGRegSet reserved_regs;
+    tcg_target_long current_frame_offset;
+    tcg_target_long frame_start;
+    tcg_target_long frame_end;
+    int frame_reg;
+
+    uint8_t *code_ptr;
+    TCGTemp static_temps[TCG_MAX_TEMPS];
+
+    TCGMacroFunc *macro_func;
+    TCGHelperInfo *helpers;
+    int nb_helpers;
+    int allocated_helpers;
+};
+
+extern TCGContext tcg_ctx;
+extern uint16_t *gen_opc_ptr;
+extern TCGArg *gen_opparam_ptr;
+extern uint16_t gen_opc_buf[];
+extern TCGArg gen_opparam_buf[];
+
+/* pool based memory allocation */
+
+void *tcg_malloc_internal(TCGContext *s, int size);
+void tcg_pool_reset(TCGContext *s);
+void tcg_pool_delete(TCGContext *s);
+
+static inline void *tcg_malloc(int size)
+{
+    TCGContext *s = &tcg_ctx;
+    uint8_t *ptr, *ptr_end;
+    size = (size + sizeof(long) - 1) & ~(sizeof(long) - 1);
+    ptr = s->pool_cur;
+    ptr_end = ptr + size;
+    if (unlikely(ptr_end > s->pool_end)) {
+        return tcg_malloc_internal(&tcg_ctx, size);
+    } else {
+        s->pool_cur = ptr_end;
+        return ptr;
+    }
+}
+
+void tcg_context_init(TCGContext *s);
+void tcg_func_start(TCGContext *s);
+
+int dyngen_code(TCGContext *s, uint8_t *gen_code_buf);
+int dyngen_code_search_pc(TCGContext *s, uint8_t *gen_code_buf,
+                          const uint8_t *searched_pc);
+
+void tcg_set_frame(TCGContext *s, int reg,
+                   tcg_target_long start, tcg_target_long size);
+void tcg_set_macro_func(TCGContext *s, TCGMacroFunc *func);
+int tcg_global_reg_new(TCGType type, int reg, const char *name);
+int tcg_global_mem_new(TCGType type, int reg, tcg_target_long offset,
+                       const char *name);
+int tcg_temp_new(TCGType type);
+char *tcg_get_arg_str(TCGContext *s, char *buf, int buf_size, TCGArg arg);
+
+#define TCG_CT_ALIAS  0x80
+#define TCG_CT_IALIAS 0x40
+#define TCG_CT_REG    0x01
+#define TCG_CT_CONST  0x02 /* any constant of register size */
+
+typedef struct TCGArgConstraint {
+    uint32_t ct;
+    union {
+        TCGRegSet regs;
+    } u;
+} TCGArgConstraint;
+
+#define TCG_MAX_OP_ARGS 16
+
+#define TCG_OPF_BB_END     0x01 /* instruction defines the end of a basic
+                                   block */
+#define TCG_OPF_CALL_CLOBBER 0x02 /* instruction clobbers call registers */
+
+typedef struct TCGOpDef {
+    const char *name;
+    uint8_t nb_oargs, nb_iargs, nb_cargs, nb_args;
+    uint8_t flags;
+    uint16_t copy_size;
+    TCGArgConstraint *args_ct;
+    int *sorted_args;
+} TCGOpDef;
+        
+typedef struct TCGTargetOpDef {
+    int op;
+    const char *args_ct_str[TCG_MAX_OP_ARGS];
+} TCGTargetOpDef;
+
+extern TCGOpDef tcg_op_defs[];
+
+void tcg_target_init(TCGContext *s);
+
+#define tcg_abort() \
+do {\
+    fprintf(stderr, "%s:%d: tcg fatal error\n", __FILE__, __LINE__);\
+    abort();\
+} while (0)
+
+void tcg_add_target_add_op_defs(const TCGTargetOpDef *tdefs);
+
+void tcg_gen_call(TCGContext *s, TCGArg func, unsigned int flags,
+                  unsigned int nb_rets, const TCGArg *rets,
+                  unsigned int nb_params, const TCGArg *args1);
+void tcg_gen_shifti_i64(TCGArg ret, TCGArg arg1, 
+                        int c, int right, int arith);
+
+/* only used for debugging purposes */
+void tcg_register_helper(void *func, const char *name);
+#define TCG_HELPER(func) tcg_register_helper(func, #func)
+const char *tcg_helper_get_name(TCGContext *s, void *func);
+void tcg_dump_ops(TCGContext *s, FILE *outfile);
+
+void dump_ops(const uint16_t *opc_buf, const TCGArg *opparam_buf);
+int tcg_const_i32(int32_t val);
+int tcg_const_i64(int64_t val);
+
+#if TCG_TARGET_REG_BITS == 32
+#define tcg_const_ptr tcg_const_i32
+#define tcg_add_ptr tcg_add_i32
+#define tcg_sub_ptr tcg_sub_i32
+#else
+#define tcg_const_ptr tcg_const_i64
+#define tcg_add_ptr tcg_add_i64
+#define tcg_sub_ptr tcg_sub_i64
+#endif
+
+void tcg_out_reloc(TCGContext *s, uint8_t *code_ptr, int type, 
+                   int label_index, long addend);
+void tcg_reg_alloc_start(TCGContext *s);
+void tcg_reg_alloc_bb_end(TCGContext *s);
+void tcg_liveness_analysis(TCGContext *s);
+const TCGArg *tcg_gen_code_op(TCGContext *s, int opc, const TCGArg *args1,
+                              unsigned int dead_iargs);
+
+const TCGArg *dyngen_op(TCGContext *s, int opc, const TCGArg *opparam_ptr);
+
+/* tcg-runtime.c */
+int64_t tcg_helper_shl_i64(int64_t arg1, int64_t arg2);
+int64_t tcg_helper_shr_i64(int64_t arg1, int64_t arg2);
+int64_t tcg_helper_sar_i64(int64_t arg1, int64_t arg2);
+int64_t tcg_helper_div_i64(int64_t arg1, int64_t arg2);
+int64_t tcg_helper_rem_i64(int64_t arg1, int64_t arg2);
+uint64_t tcg_helper_divu_i64(uint64_t arg1, uint64_t arg2);
+uint64_t tcg_helper_remu_i64(uint64_t arg1, uint64_t arg2);
diff --git a/tcg/x86_64/tcg-target.c b/tcg/x86_64/tcg-target.c
new file mode 100644
index 0000000000..a2f0e4c743
--- /dev/null
+++ b/tcg/x86_64/tcg-target.c
@@ -0,0 +1,1227 @@
+/*
+ * Tiny Code Generator for QEMU
+ *
+ * Copyright (c) 2008 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+const char *tcg_target_reg_names[TCG_TARGET_NB_REGS] = {
+    "%rax",
+    "%rcx",
+    "%rdx",
+    "%rbx",
+    "%rsp",
+    "%rbp",
+    "%rsi",
+    "%rdi",
+    "%r8",
+    "%r9",
+    "%r10",
+    "%r11",
+    "%r12",
+    "%r13",
+    "%r14",
+    "%r15",
+};
+
+int tcg_target_reg_alloc_order[TCG_TARGET_NB_REGS] = {
+    TCG_REG_RDI,
+    TCG_REG_RSI,
+    TCG_REG_RDX,
+    TCG_REG_RCX,
+    TCG_REG_R8,
+    TCG_REG_R9,
+    TCG_REG_RAX,
+    TCG_REG_R10,
+    TCG_REG_R11,
+
+    TCG_REG_RBP,
+    TCG_REG_RBX,
+    TCG_REG_R12,
+    TCG_REG_R13,
+    TCG_REG_R14,
+    TCG_REG_R15,
+};
+
+const int tcg_target_call_iarg_regs[6] = { 
+    TCG_REG_RDI,
+    TCG_REG_RSI,
+    TCG_REG_RDX,
+    TCG_REG_RCX,
+    TCG_REG_R8,
+    TCG_REG_R9,
+};
+
+const int tcg_target_call_oarg_regs[2] = { 
+    TCG_REG_RAX, 
+    TCG_REG_RDX 
+};
+
+static void patch_reloc(uint8_t *code_ptr, int type, 
+                        tcg_target_long value)
+{
+    switch(type) {
+    case R_X86_64_32:
+        if (value != (uint32_t)value)
+            tcg_abort();
+        *(uint32_t *)code_ptr = value;
+        break;
+    case R_X86_64_32S:
+        if (value != (int32_t)value)
+            tcg_abort();
+        *(uint32_t *)code_ptr = value;
+        break;
+    case R_386_PC32:
+        value -= (long)code_ptr;
+        if (value != (int32_t)value)
+            tcg_abort();
+        *(uint32_t *)code_ptr = value;
+        break;
+    default:
+        tcg_abort();
+    }
+}
+
+/* maximum number of register used for input function arguments */
+static inline int tcg_target_get_call_iarg_regs_count(int flags)
+{
+    return 6;
+}
+
+/* parse target specific constraints */
+int target_parse_constraint(TCGArgConstraint *ct, const char **pct_str)
+{
+    const char *ct_str;
+
+    ct_str = *pct_str;
+    switch(ct_str[0]) {
+    case 'a':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set_reg(ct->u.regs, TCG_REG_RAX);
+        break;
+    case 'b':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set_reg(ct->u.regs, TCG_REG_RBX);
+        break;
+    case 'c':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set_reg(ct->u.regs, TCG_REG_RCX);
+        break;
+    case 'd':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set_reg(ct->u.regs, TCG_REG_RDX);
+        break;
+    case 'S':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set_reg(ct->u.regs, TCG_REG_RSI);
+        break;
+    case 'D':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set_reg(ct->u.regs, TCG_REG_RDI);
+        break;
+    case 'q':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set32(ct->u.regs, 0, 0xf);
+        break;
+    case 'r':
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set32(ct->u.regs, 0, 0xffff);
+        break;
+    case 'L': /* qemu_ld/st constraint */
+        ct->ct |= TCG_CT_REG;
+        tcg_regset_set32(ct->u.regs, 0, 0xffff);
+        tcg_regset_reset_reg(ct->u.regs, TCG_REG_RSI);
+        tcg_regset_reset_reg(ct->u.regs, TCG_REG_RDI);
+        break;
+    case 'e':
+        ct->ct |= TCG_CT_CONST_S32;
+        break;
+    case 'Z':
+        ct->ct |= TCG_CT_CONST_U32;
+        break;
+    default:
+        return -1;
+    }
+    ct_str++;
+    *pct_str = ct_str;
+    return 0;
+}
+
+/* test if a constant matches the constraint */
+static inline int tcg_target_const_match(tcg_target_long val,
+                                         const TCGArgConstraint *arg_ct)
+{
+    int ct;
+    ct = arg_ct->ct;
+    if (ct & TCG_CT_CONST)
+        return 1;
+    else if ((ct & TCG_CT_CONST_S32) && val == (int32_t)val)
+        return 1;
+    else if ((ct & TCG_CT_CONST_U32) && val == (uint32_t)val)
+        return 1;
+    else
+        return 0;
+}
+
+#define ARITH_ADD 0
+#define ARITH_OR  1
+#define ARITH_ADC 2
+#define ARITH_SBB 3
+#define ARITH_AND 4
+#define ARITH_SUB 5
+#define ARITH_XOR 6
+#define ARITH_CMP 7
+
+#define SHIFT_SHL 4
+#define SHIFT_SHR 5
+#define SHIFT_SAR 7
+
+#define JCC_JMP (-1)
+#define JCC_JO  0x0
+#define JCC_JNO 0x1
+#define JCC_JB  0x2
+#define JCC_JAE 0x3
+#define JCC_JE  0x4
+#define JCC_JNE 0x5
+#define JCC_JBE 0x6
+#define JCC_JA  0x7
+#define JCC_JS  0x8
+#define JCC_JNS 0x9
+#define JCC_JP  0xa
+#define JCC_JNP 0xb
+#define JCC_JL  0xc
+#define JCC_JGE 0xd
+#define JCC_JLE 0xe
+#define JCC_JG  0xf
+
+#define P_EXT   0x100 /* 0x0f opcode prefix */
+#define P_REXW  0x200 /* set rex.w = 1 */
+#define P_REX   0x400 /* force rex usage */
+                                  
+static const uint8_t tcg_cond_to_jcc[10] = {
+    [TCG_COND_EQ] = JCC_JE,
+    [TCG_COND_NE] = JCC_JNE,
+    [TCG_COND_LT] = JCC_JL,
+    [TCG_COND_GE] = JCC_JGE,
+    [TCG_COND_LE] = JCC_JLE,
+    [TCG_COND_GT] = JCC_JG,
+    [TCG_COND_LTU] = JCC_JB,
+    [TCG_COND_GEU] = JCC_JAE,
+    [TCG_COND_LEU] = JCC_JBE,
+    [TCG_COND_GTU] = JCC_JA,
+};
+
+static inline void tcg_out_opc(TCGContext *s, int opc, int r, int rm, int x)
+{
+    int rex;
+    rex = ((opc >> 6) & 0x8) | ((r >> 1) & 0x4) | 
+        ((x >> 2) & 2) | ((rm >> 3) & 1);
+    if (rex || (opc & P_REX)) {
+        tcg_out8(s, rex | 0x40);
+    }
+    if (opc & P_EXT)
+        tcg_out8(s, 0x0f);
+    tcg_out8(s, opc);
+}
+
+static inline void tcg_out_modrm(TCGContext *s, int opc, int r, int rm)
+{
+    tcg_out_opc(s, opc, r, rm, 0);
+    tcg_out8(s, 0xc0 | ((r & 7) << 3) | (rm & 7));
+}
+
+/* rm < 0 means no register index plus (-rm - 1 immediate bytes) */
+static inline void tcg_out_modrm_offset(TCGContext *s, int opc, int r, int rm, 
+                                        tcg_target_long offset)
+{
+    if (rm < 0) {
+        tcg_target_long val;
+        tcg_out_opc(s, opc, r, 0, 0);
+        val = offset - ((tcg_target_long)s->code_ptr + 5 + (-rm - 1));
+        if (val == (int32_t)val) {
+            /* eip relative */
+            tcg_out8(s, 0x05 | ((r & 7) << 3));
+            tcg_out32(s, val);
+        } else if (offset == (int32_t)offset) {
+            tcg_out8(s, 0x04 | ((r & 7) << 3));
+            tcg_out8(s, 0x25); /* sib */
+            tcg_out32(s, offset);
+        } else {
+            tcg_abort();
+        }
+    } else if (offset == 0 && (rm & 7) != TCG_REG_RBP) {
+        tcg_out_opc(s, opc, r, rm, 0);
+        if ((rm & 7) == TCG_REG_RSP) {
+            tcg_out8(s, 0x04 | ((r & 7) << 3));
+            tcg_out8(s, 0x24);
+        } else {
+            tcg_out8(s, 0x00 | ((r & 7) << 3) | (rm & 7));
+        }
+    } else if ((int8_t)offset == offset) {
+        tcg_out_opc(s, opc, r, rm, 0);
+        if ((rm & 7) == TCG_REG_RSP) {
+            tcg_out8(s, 0x44 | ((r & 7) << 3));
+            tcg_out8(s, 0x24);
+        } else {
+            tcg_out8(s, 0x40 | ((r & 7) << 3) | (rm & 7));
+        }
+        tcg_out8(s, offset);
+    } else {
+        tcg_out_opc(s, opc, r, rm, 0);
+        if ((rm & 7) == TCG_REG_RSP) {
+            tcg_out8(s, 0x84 | ((r & 7) << 3));
+            tcg_out8(s, 0x24);
+        } else {
+            tcg_out8(s, 0x80 | ((r & 7) << 3) | (rm & 7));
+        }
+        tcg_out32(s, offset);
+    }
+}
+
+/* XXX: incomplete. index must be different from ESP */
+static void tcg_out_modrm_offset2(TCGContext *s, int opc, int r, int rm, 
+                                  int index, int shift,
+                                  tcg_target_long offset)
+{
+    int mod;
+    if (rm == -1)
+        tcg_abort();
+    if (offset == 0 && (rm & 7) != TCG_REG_RBP) {
+        mod = 0;
+    } else if (offset == (int8_t)offset) {
+        mod = 0x40;
+    } else if (offset == (int32_t)offset) {
+        mod = 0x80;
+    } else {
+        tcg_abort();
+    }
+    if (index == -1) {
+        tcg_out_opc(s, opc, r, rm, 0);
+        if ((rm & 7) == TCG_REG_RSP) {
+            tcg_out8(s, mod | ((r & 7) << 3) | 0x04);
+            tcg_out8(s, 0x04 | (rm & 7));
+        } else {
+            tcg_out8(s, mod | ((r & 7) << 3) | (rm & 7));
+        }
+    } else {
+        tcg_out_opc(s, opc, r, rm, index);
+        tcg_out8(s, mod | ((r & 7) << 3) | 0x04);
+        tcg_out8(s, (shift << 6) | ((index & 7) << 3) | (rm & 7));
+    }
+    if (mod == 0x40) {
+        tcg_out8(s, offset);
+    } else if (mod == 0x80) {
+        tcg_out32(s, offset);
+    }
+}
+
+static inline void tcg_out_mov(TCGContext *s, int ret, int arg)
+{
+    tcg_out_modrm(s, 0x8b | P_REXW, ret, arg);
+}
+
+static inline void tcg_out_movi(TCGContext *s, TCGType type, 
+                                int ret, tcg_target_long arg)
+{
+    if (arg == 0) {
+        tcg_out_modrm(s, 0x01 | (ARITH_XOR << 3), ret, ret); /* xor r0,r0 */
+    } else if (arg == (uint32_t)arg || type == TCG_TYPE_I32) {
+        tcg_out_opc(s, 0xb8 + (ret & 7), 0, ret, 0);
+        tcg_out32(s, arg);
+    } else if (arg == (int32_t)arg) {
+        tcg_out_modrm(s, 0xc7 | P_REXW, 0, ret);
+        tcg_out32(s, arg);
+    } else {
+        tcg_out_opc(s, (0xb8 + (ret & 7)) | P_REXW, 0, ret, 0);
+        tcg_out32(s, arg);
+        tcg_out32(s, arg >> 32);
+    }
+}
+
+static inline void tcg_out_ld(TCGContext *s, int ret, 
+                              int arg1, tcg_target_long arg2)
+{
+    tcg_out_modrm_offset(s, 0x8b | P_REXW, ret, arg1, arg2); /* movq */
+}
+
+static inline void tcg_out_st(TCGContext *s, int arg, 
+                              int arg1, tcg_target_long arg2)
+{
+    tcg_out_modrm_offset(s, 0x89 | P_REXW, arg, arg1, arg2); /* movq */
+}
+
+static inline void tgen_arithi32(TCGContext *s, int c, int r0, int32_t val)
+{
+    if (val == (int8_t)val) {
+        tcg_out_modrm(s, 0x83, c, r0);
+        tcg_out8(s, val);
+    } else {
+        tcg_out_modrm(s, 0x81, c, r0);
+        tcg_out32(s, val);
+    }
+}
+
+static inline void tgen_arithi64(TCGContext *s, int c, int r0, int64_t val)
+{
+    if (val == (int8_t)val) {
+        tcg_out_modrm(s, 0x83 | P_REXW, c, r0);
+        tcg_out8(s, val);
+    } else if (val == (int32_t)val) {
+        tcg_out_modrm(s, 0x81 | P_REXW, c, r0);
+        tcg_out32(s, val);
+    } else if (c == ARITH_AND && val == (uint32_t)val) {
+        tcg_out_modrm(s, 0x81, c, r0);
+        tcg_out32(s, val);
+    } else {
+        tcg_abort();
+    }
+}
+
+void tcg_out_addi(TCGContext *s, int reg, tcg_target_long val)
+{
+    if (val != 0)
+        tgen_arithi64(s, ARITH_ADD, reg, val);
+}
+
+static void tcg_out_jxx(TCGContext *s, int opc, int label_index)
+{
+    int32_t val, val1;
+    TCGLabel *l = &s->labels[label_index];
+    
+    if (l->has_value) {
+        val = l->u.value - (tcg_target_long)s->code_ptr;
+        val1 = val - 2;
+        if ((int8_t)val1 == val1) {
+            if (opc == -1)
+                tcg_out8(s, 0xeb);
+            else
+                tcg_out8(s, 0x70 + opc);
+            tcg_out8(s, val1);
+        } else {
+            if (opc == -1) {
+                tcg_out8(s, 0xe9);
+                tcg_out32(s, val - 5);
+            } else {
+                tcg_out8(s, 0x0f);
+                tcg_out8(s, 0x80 + opc);
+                tcg_out32(s, val - 6);
+            }
+        }
+    } else {
+        if (opc == -1) {
+            tcg_out8(s, 0xe9);
+        } else {
+            tcg_out8(s, 0x0f);
+            tcg_out8(s, 0x80 + opc);
+        }
+        tcg_out_reloc(s, s->code_ptr, R_386_PC32, label_index, -4);
+        tcg_out32(s, -4);
+    }
+}
+
+static void tcg_out_brcond(TCGContext *s, int cond, 
+                           TCGArg arg1, TCGArg arg2, int const_arg2,
+                           int label_index, int rexw)
+{
+    int c;
+    if (const_arg2) {
+        if (arg2 == 0) {
+            /* use test */
+            switch(cond) {
+            case TCG_COND_EQ:
+                c = JCC_JNE;
+                break;
+            case TCG_COND_NE:
+                c = JCC_JNE;
+                break;
+            case TCG_COND_LT:
+                c = JCC_JS;
+                break;
+            case TCG_COND_GE:
+                c = JCC_JNS;
+                break;
+            default:
+                goto do_cmpi;
+            }
+            /* test r, r */
+            tcg_out_modrm(s, 0x85 | rexw, arg1, arg1);
+            tcg_out_jxx(s, c, label_index);
+        } else {
+        do_cmpi:
+            if (rexw)
+                tgen_arithi64(s, ARITH_CMP, arg1, arg2);
+            else
+                tgen_arithi32(s, ARITH_CMP, arg1, arg2);
+            tcg_out_jxx(s, tcg_cond_to_jcc[cond], label_index);
+        }
+    } else {
+        tcg_out_modrm(s, 0x01 | (ARITH_CMP << 3) | rexw, arg1, arg2);
+        tcg_out_jxx(s, tcg_cond_to_jcc[cond], label_index);
+    }
+}
+
+#if defined(CONFIG_SOFTMMU)
+extern void __ldb_mmu(void);
+extern void __ldw_mmu(void);
+extern void __ldl_mmu(void);
+extern void __ldq_mmu(void);
+
+extern void __stb_mmu(void);
+extern void __stw_mmu(void);
+extern void __stl_mmu(void);
+extern void __stq_mmu(void);
+
+
+static void *qemu_ld_helpers[4] = {
+    __ldb_mmu,
+    __ldw_mmu,
+    __ldl_mmu,
+    __ldq_mmu,
+};
+
+static void *qemu_st_helpers[4] = {
+    __stb_mmu,
+    __stw_mmu,
+    __stl_mmu,
+    __stq_mmu,
+};
+#endif
+
+static void tcg_out_qemu_ld(TCGContext *s, const TCGArg *args,
+                            int opc)
+{
+    int addr_reg, data_reg, r0, r1, mem_index, s_bits, bswap, rexw;
+#if defined(CONFIG_SOFTMMU)
+    uint8_t *label1_ptr, *label2_ptr;
+#endif
+
+    data_reg = *args++;
+    addr_reg = *args++;
+    mem_index = *args;
+    s_bits = opc & 3;
+
+    r0 = TCG_REG_RDI;
+    r1 = TCG_REG_RSI;
+
+#if TARGET_LONG_BITS == 32
+    rexw = 0;
+#else
+    rexw = P_REXW;
+#endif
+#if defined(CONFIG_SOFTMMU)
+    /* mov */
+    tcg_out_modrm(s, 0x8b | rexw, r1, addr_reg);
+
+    /* mov */
+    tcg_out_modrm(s, 0x8b | rexw, r0, addr_reg);
+ 
+    tcg_out_modrm(s, 0xc1 | rexw, 5, r1); /* shr $x, r1 */
+    tcg_out8(s, TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); 
+    
+    tcg_out_modrm(s, 0x81 | rexw, 4, r0); /* andl $x, r0 */
+    tcg_out32(s, TARGET_PAGE_MASK | ((1 << s_bits) - 1));
+    
+    tcg_out_modrm(s, 0x81, 4, r1); /* andl $x, r1 */
+    tcg_out32(s, (CPU_TLB_SIZE - 1) << CPU_TLB_ENTRY_BITS);
+
+    /* lea offset(r1, env), r1 */
+    tcg_out_modrm_offset2(s, 0x8d | P_REXW, r1, r1, TCG_AREG0, 0,
+                          offsetof(CPUState, tlb_table[mem_index][0].addr_read));
+
+    /* cmp 0(r1), r0 */
+    tcg_out_modrm_offset(s, 0x3b | rexw, r0, r1, 0);
+    
+    /* mov */
+    tcg_out_modrm(s, 0x8b | rexw, r0, addr_reg);
+    
+    /* je label1 */
+    tcg_out8(s, 0x70 + JCC_JE);
+    label1_ptr = s->code_ptr;
+    s->code_ptr++;
+
+    /* 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);
+
+    switch(opc) {
+    case 0 | 4:
+        /* movsbq */
+        tcg_out_modrm(s, 0xbe | P_EXT | P_REXW, data_reg, TCG_REG_RAX);
+        break;
+    case 1 | 4:
+        /* movswq */
+        tcg_out_modrm(s, 0xbf | P_EXT | P_REXW, data_reg, TCG_REG_RAX);
+        break;
+    case 2 | 4:
+        /* movslq */
+        tcg_out_modrm(s, 0x63 | P_REXW, data_reg, TCG_REG_RAX);
+        break;
+    case 0:
+    case 1:
+    case 2:
+    default:
+        /* movl */
+        tcg_out_modrm(s, 0x8b, data_reg, TCG_REG_RAX);
+        break;
+    case 3:
+        tcg_out_mov(s, data_reg, TCG_REG_RAX);
+        break;
+    }
+
+    /* jmp label2 */
+    tcg_out8(s, 0xeb);
+    label2_ptr = s->code_ptr;
+    s->code_ptr++;
+    
+    /* label1: */
+    *label1_ptr = s->code_ptr - label1_ptr - 1;
+
+    /* add x(r1), r0 */
+    tcg_out_modrm_offset(s, 0x03 | P_REXW, r0, r1, offsetof(CPUTLBEntry, addend) - 
+                         offsetof(CPUTLBEntry, addr_read));
+#else
+    r0 = addr_reg;
+#endif    
+
+#ifdef TARGET_WORDS_BIGENDIAN
+    bswap = 1;
+#else
+    bswap = 0;
+#endif
+    switch(opc) {
+    case 0:
+        /* movzbl */
+        tcg_out_modrm_offset(s, 0xb6 | P_EXT, data_reg, r0, 0);
+        break;
+    case 0 | 4:
+        /* movsbX */
+        tcg_out_modrm_offset(s, 0xbe | P_EXT | rexw, data_reg, r0, 0);
+        break;
+    case 1:
+        /* movzwl */
+        tcg_out_modrm_offset(s, 0xb7 | P_EXT, data_reg, r0, 0);
+        if (bswap) {
+            /* rolw $8, data_reg */
+            tcg_out8(s, 0x66); 
+            tcg_out_modrm(s, 0xc1, 0, data_reg);
+            tcg_out8(s, 8);
+        }
+        break;
+    case 1 | 4:
+        if (bswap) {
+            /* movzwl */
+            tcg_out_modrm_offset(s, 0xb7 | P_EXT, data_reg, r0, 0);
+            /* rolw $8, data_reg */
+            tcg_out8(s, 0x66); 
+            tcg_out_modrm(s, 0xc1, 0, data_reg);
+            tcg_out8(s, 8);
+
+            /* movswX data_reg, data_reg */
+            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);
+        }
+        break;
+    case 2:
+        /* movl (r0), data_reg */
+        tcg_out_modrm_offset(s, 0x8b, data_reg, r0, 0);
+        if (bswap) {
+            /* bswap */
+            tcg_out_opc(s, (0xc8 + (data_reg & 7)) | P_EXT, 0, data_reg, 0);
+        }
+        break;
+    case 2 | 4:
+        if (bswap) {
+            /* movl (r0), data_reg */
+            tcg_out_modrm_offset(s, 0x8b, data_reg, r0, 0);
+            /* 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);
+        }
+        break;
+    case 3:
+        /* movq (r0), data_reg */
+        tcg_out_modrm_offset(s, 0x8b | P_REXW, data_reg, r0, 0);
+        if (bswap) {
+            /* bswap */
+            tcg_out_opc(s, (0xc8 + (data_reg & 7)) | P_EXT | P_REXW, 0, data_reg, 0);
+        }
+        break;
+    default:
+        tcg_abort();
+    }
+
+#if defined(CONFIG_SOFTMMU)
+    /* label2: */
+    *label2_ptr = s->code_ptr - label2_ptr - 1;
+#endif
+}
+
+static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args,
+                            int opc)
+{
+    int addr_reg, data_reg, r0, r1, mem_index, s_bits, bswap, rexw;
+#if defined(CONFIG_SOFTMMU)
+    uint8_t *label1_ptr, *label2_ptr;
+#endif
+
+    data_reg = *args++;
+    addr_reg = *args++;
+    mem_index = *args;
+
+    s_bits = opc;
+
+    r0 = TCG_REG_RDI;
+    r1 = TCG_REG_RSI;
+
+#if TARGET_LONG_BITS == 32
+    rexw = 0;
+#else
+    rexw = P_REXW;
+#endif
+#if defined(CONFIG_SOFTMMU)
+    /* mov */
+    tcg_out_modrm(s, 0x8b | rexw, r1, addr_reg);
+
+    /* mov */
+    tcg_out_modrm(s, 0x8b | rexw, r0, addr_reg);
+ 
+    tcg_out_modrm(s, 0xc1 | rexw, 5, r1); /* shr $x, r1 */
+    tcg_out8(s, TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); 
+    
+    tcg_out_modrm(s, 0x81 | rexw, 4, r0); /* andl $x, r0 */
+    tcg_out32(s, TARGET_PAGE_MASK | ((1 << s_bits) - 1));
+    
+    tcg_out_modrm(s, 0x81, 4, r1); /* andl $x, r1 */
+    tcg_out32(s, (CPU_TLB_SIZE - 1) << CPU_TLB_ENTRY_BITS);
+
+    /* lea offset(r1, env), r1 */
+    tcg_out_modrm_offset2(s, 0x8d | P_REXW, r1, r1, TCG_AREG0, 0,
+                          offsetof(CPUState, tlb_table[mem_index][0].addr_write));
+
+    /* cmp 0(r1), r0 */
+    tcg_out_modrm_offset(s, 0x3b | rexw, r0, r1, 0);
+    
+    /* mov */
+    tcg_out_modrm(s, 0x8b | rexw, r0, addr_reg);
+    
+    /* je label1 */
+    tcg_out8(s, 0x70 + JCC_JE);
+    label1_ptr = s->code_ptr;
+    s->code_ptr++;
+
+    /* XXX: move that code at the end of the TB */
+    switch(opc) {
+    case 0:
+        /* movzbl */
+        tcg_out_modrm(s, 0xb6 | P_EXT, TCG_REG_RSI, data_reg);
+        break;
+    case 1:
+        /* movzwl */
+        tcg_out_modrm(s, 0xb7 | P_EXT, TCG_REG_RSI, data_reg);
+        break;
+    case 2:
+        /* movl */
+        tcg_out_modrm(s, 0x8b, TCG_REG_RSI, data_reg);
+        break;
+    default:
+    case 3:
+        tcg_out_mov(s, TCG_REG_RSI, data_reg);
+        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);
+
+    /* jmp label2 */
+    tcg_out8(s, 0xeb);
+    label2_ptr = s->code_ptr;
+    s->code_ptr++;
+    
+    /* label1: */
+    *label1_ptr = s->code_ptr - label1_ptr - 1;
+
+    /* add x(r1), r0 */
+    tcg_out_modrm_offset(s, 0x03 | P_REXW, r0, r1, offsetof(CPUTLBEntry, addend) - 
+                         offsetof(CPUTLBEntry, addr_write));
+#else
+    r0 = addr_reg;
+#endif
+
+#ifdef TARGET_WORDS_BIGENDIAN
+    bswap = 1;
+#else
+    bswap = 0;
+#endif
+    switch(opc) {
+    case 0:
+        /* movb */
+        tcg_out_modrm_offset(s, 0x88 | P_REX, data_reg, r0, 0);
+        break;
+    case 1:
+        if (bswap) {
+            tcg_out_modrm(s, 0x8b, r1, data_reg); /* movl */
+            tcg_out8(s, 0x66); /* rolw $8, %ecx */
+            tcg_out_modrm(s, 0xc1, 0, r1);
+            tcg_out8(s, 8);
+            data_reg = r1;
+        }
+        /* movw */
+        tcg_out8(s, 0x66);
+        tcg_out_modrm_offset(s, 0x89, data_reg, r0, 0);
+        break;
+    case 2:
+        if (bswap) {
+            tcg_out_modrm(s, 0x8b, r1, data_reg); /* movl */
+            /* bswap data_reg */
+            tcg_out_opc(s, (0xc8 + r1) | P_EXT, 0, r1, 0);
+            data_reg = r1;
+        }
+        /* movl */
+        tcg_out_modrm_offset(s, 0x89, data_reg, r0, 0);
+        break;
+    case 3:
+        if (bswap) {
+            tcg_out_mov(s, r1, data_reg);
+            /* bswap data_reg */
+            tcg_out_opc(s, (0xc8 + r1) | P_EXT | P_REXW, 0, r1, 0);
+            data_reg = r1;
+        }
+        /* movq */
+        tcg_out_modrm_offset(s, 0x89 | P_REXW, data_reg, r0, 0);
+        break;
+    default:
+        tcg_abort();
+    }
+
+#if defined(CONFIG_SOFTMMU)
+    /* label2: */
+    *label2_ptr = s->code_ptr - label2_ptr - 1;
+#endif
+}
+
+static inline void tcg_out_op(TCGContext *s, int opc, const TCGArg *args,
+                              const int *const_args)
+{
+    int c;
+    
+    switch(opc) {
+    case INDEX_op_exit_tb:
+        tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_RAX, args[0]);
+        tcg_out8(s, 0xc3); /* ret */
+        break;
+    case INDEX_op_goto_tb:
+        if (s->tb_jmp_offset) {
+            /* direct jump method */
+            tcg_out8(s, 0xe9); /* jmp im */
+            s->tb_jmp_offset[args[0]] = s->code_ptr - s->code_buf;
+            tcg_out32(s, 0);
+        } else {
+            /* indirect jump method */
+            /* jmp Ev */
+            tcg_out_modrm_offset(s, 0xff, 4, -1, 
+                                 (tcg_target_long)(s->tb_next + 
+                                                   args[0]));
+        }
+        s->tb_next_offset[args[0]] = s->code_ptr - s->code_buf;
+        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);
+        } 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);
+        } else {
+            tcg_out_modrm(s, 0xff, 4, args[0]);
+        }
+        break;
+    case INDEX_op_br:
+        tcg_out_jxx(s, JCC_JMP, args[0]);
+        break;
+    case INDEX_op_movi_i32:
+        tcg_out_movi(s, TCG_TYPE_I32, args[0], (uint32_t)args[1]);
+        break;
+    case INDEX_op_movi_i64:
+        tcg_out_movi(s, TCG_TYPE_I64, args[0], args[1]);
+        break;
+    case INDEX_op_ld8u_i32:
+    case INDEX_op_ld8u_i64:
+        /* movzbl */
+        tcg_out_modrm_offset(s, 0xb6 | P_EXT, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_ld8s_i32:
+        /* movsbl */
+        tcg_out_modrm_offset(s, 0xbe | P_EXT, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_ld8s_i64:
+        /* movsbq */
+        tcg_out_modrm_offset(s, 0xbe | P_EXT | P_REXW, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_ld16u_i32:
+    case INDEX_op_ld16u_i64:
+        /* movzwl */
+        tcg_out_modrm_offset(s, 0xb7 | P_EXT, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_ld16s_i32:
+        /* movswl */
+        tcg_out_modrm_offset(s, 0xbf | P_EXT, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_ld16s_i64:
+        /* movswq */
+        tcg_out_modrm_offset(s, 0xbf | P_EXT | P_REXW, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_ld_i32:
+    case INDEX_op_ld32u_i64:
+        /* movl */
+        tcg_out_modrm_offset(s, 0x8b, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_ld32s_i64:
+        /* movslq */
+        tcg_out_modrm_offset(s, 0x63 | P_REXW, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_ld_i64:
+        /* movq */
+        tcg_out_modrm_offset(s, 0x8b | P_REXW, args[0], args[1], args[2]);
+        break;
+        
+    case INDEX_op_st8_i32:
+    case INDEX_op_st8_i64:
+        /* movb */
+        tcg_out_modrm_offset(s, 0x88 | P_REX, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_st16_i32:
+    case INDEX_op_st16_i64:
+        /* movw */
+        tcg_out8(s, 0x66);
+        tcg_out_modrm_offset(s, 0x89, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_st_i32:
+    case INDEX_op_st32_i64:
+        /* movl */
+        tcg_out_modrm_offset(s, 0x89, args[0], args[1], args[2]);
+        break;
+    case INDEX_op_st_i64:
+        /* movq */
+        tcg_out_modrm_offset(s, 0x89 | P_REXW, args[0], args[1], args[2]);
+        break;
+
+    case INDEX_op_sub_i32:
+        c = ARITH_SUB;
+        goto gen_arith32;
+    case INDEX_op_and_i32:
+        c = ARITH_AND;
+        goto gen_arith32;
+    case INDEX_op_or_i32:
+        c = ARITH_OR;
+        goto gen_arith32;
+    case INDEX_op_xor_i32:
+        c = ARITH_XOR;
+        goto gen_arith32;
+    case INDEX_op_add_i32:
+        c = ARITH_ADD;
+    gen_arith32:
+        if (const_args[2]) {
+            tgen_arithi32(s, c, args[0], args[2]);
+        } else {
+            tcg_out_modrm(s, 0x01 | (c << 3), args[2], args[0]);
+        }
+        break;
+
+    case INDEX_op_sub_i64:
+        c = ARITH_SUB;
+        goto gen_arith64;
+    case INDEX_op_and_i64:
+        c = ARITH_AND;
+        goto gen_arith64;
+    case INDEX_op_or_i64:
+        c = ARITH_OR;
+        goto gen_arith64;
+    case INDEX_op_xor_i64:
+        c = ARITH_XOR;
+        goto gen_arith64;
+    case INDEX_op_add_i64:
+        c = ARITH_ADD;
+    gen_arith64:
+        if (const_args[2]) {
+            tgen_arithi64(s, c, args[0], args[2]);
+        } else {
+            tcg_out_modrm(s, 0x01 | (c << 3) | P_REXW, args[2], args[0]);
+        }
+        break;
+
+    case INDEX_op_mul_i32:
+        if (const_args[2]) {
+            int32_t val;
+            val = args[2];
+            if (val == (int8_t)val) {
+                tcg_out_modrm(s, 0x6b, args[0], args[0]);
+                tcg_out8(s, val);
+            } else {
+                tcg_out_modrm(s, 0x69, args[0], args[0]);
+                tcg_out32(s, val);
+            }
+        } else {
+            tcg_out_modrm(s, 0xaf | P_EXT, args[0], args[2]);
+        }
+        break;
+    case INDEX_op_mul_i64:
+        if (const_args[2]) {
+            int32_t val;
+            val = args[2];
+            if (val == (int8_t)val) {
+                tcg_out_modrm(s, 0x6b | P_REXW, args[0], args[0]);
+                tcg_out8(s, val);
+            } else {
+                tcg_out_modrm(s, 0x69 | P_REXW, args[0], args[0]);
+                tcg_out32(s, val);
+            }
+        } else {
+            tcg_out_modrm(s, 0xaf | P_EXT | P_REXW, args[0], args[2]);
+        }
+        break;
+    case INDEX_op_div2_i32:
+        tcg_out_modrm(s, 0xf7, 7, args[4]);
+        break;
+    case INDEX_op_divu2_i32:
+        tcg_out_modrm(s, 0xf7, 6, args[4]);
+        break;
+    case INDEX_op_div2_i64:
+        tcg_out_modrm(s, 0xf7 | P_REXW, 7, args[4]);
+        break;
+    case INDEX_op_divu2_i64:
+        tcg_out_modrm(s, 0xf7 | P_REXW, 6, args[4]);
+        break;
+
+    case INDEX_op_shl_i32:
+        c = SHIFT_SHL;
+    gen_shift32:
+        if (const_args[2]) {
+            if (args[2] == 1) {
+                tcg_out_modrm(s, 0xd1, c, args[0]);
+            } else {
+                tcg_out_modrm(s, 0xc1, c, args[0]);
+                tcg_out8(s, args[2]);
+            }
+        } else {
+            tcg_out_modrm(s, 0xd3, c, args[0]);
+        }
+        break;
+    case INDEX_op_shr_i32:
+        c = SHIFT_SHR;
+        goto gen_shift32;
+    case INDEX_op_sar_i32:
+        c = SHIFT_SAR;
+        goto gen_shift32;
+        
+    case INDEX_op_shl_i64:
+        c = SHIFT_SHL;
+    gen_shift64:
+        if (const_args[2]) {
+            if (args[2] == 1) {
+                tcg_out_modrm(s, 0xd1 | P_REXW, c, args[0]);
+            } else {
+                tcg_out_modrm(s, 0xc1 | P_REXW, c, args[0]);
+                tcg_out8(s, args[2]);
+            }
+        } else {
+            tcg_out_modrm(s, 0xd3 | P_REXW, c, args[0]);
+        }
+        break;
+    case INDEX_op_shr_i64:
+        c = SHIFT_SHR;
+        goto gen_shift64;
+    case INDEX_op_sar_i64:
+        c = SHIFT_SAR;
+        goto gen_shift64;
+        
+    case INDEX_op_brcond_i32:
+        tcg_out_brcond(s, args[2], args[0], args[1], const_args[1], 
+                       args[3], 0);
+        break;
+    case INDEX_op_brcond_i64:
+        tcg_out_brcond(s, args[2], args[0], args[1], const_args[1], 
+                       args[3], P_REXW);
+        break;
+
+    case INDEX_op_bswap_i32:
+        tcg_out_opc(s, (0xc8 + (args[0] & 7)) | P_EXT, 0, args[0], 0);
+        break;
+    case INDEX_op_bswap_i64:
+        tcg_out_opc(s, (0xc8 + (args[0] & 7)) | P_EXT | P_REXW, 0, args[0], 0);
+        break;
+
+    case INDEX_op_qemu_ld8u:
+        tcg_out_qemu_ld(s, args, 0);
+        break;
+    case INDEX_op_qemu_ld8s:
+        tcg_out_qemu_ld(s, args, 0 | 4);
+        break;
+    case INDEX_op_qemu_ld16u:
+        tcg_out_qemu_ld(s, args, 1);
+        break;
+    case INDEX_op_qemu_ld16s:
+        tcg_out_qemu_ld(s, args, 1 | 4);
+        break;
+    case INDEX_op_qemu_ld32u:
+        tcg_out_qemu_ld(s, args, 2);
+        break;
+    case INDEX_op_qemu_ld32s:
+        tcg_out_qemu_ld(s, args, 2 | 4);
+        break;
+    case INDEX_op_qemu_ld64:
+        tcg_out_qemu_ld(s, args, 3);
+        break;
+        
+    case INDEX_op_qemu_st8:
+        tcg_out_qemu_st(s, args, 0);
+        break;
+    case INDEX_op_qemu_st16:
+        tcg_out_qemu_st(s, args, 1);
+        break;
+    case INDEX_op_qemu_st32:
+        tcg_out_qemu_st(s, args, 2);
+        break;
+    case INDEX_op_qemu_st64:
+        tcg_out_qemu_st(s, args, 3);
+        break;
+
+    default:
+        tcg_abort();
+    }
+}
+
+static const TCGTargetOpDef x86_64_op_defs[] = {
+    { INDEX_op_exit_tb, { } },
+    { INDEX_op_goto_tb, { } },
+    { INDEX_op_call, { "ri" } }, /* XXX: might need a specific constant constraint */
+    { INDEX_op_jmp, { "ri" } }, /* XXX: might need a specific constant constraint */
+    { INDEX_op_br, { } },
+
+    { INDEX_op_mov_i32, { "r", "r" } },
+    { INDEX_op_movi_i32, { "r" } },
+    { INDEX_op_ld8u_i32, { "r", "r" } },
+    { INDEX_op_ld8s_i32, { "r", "r" } },
+    { INDEX_op_ld16u_i32, { "r", "r" } },
+    { INDEX_op_ld16s_i32, { "r", "r" } },
+    { INDEX_op_ld_i32, { "r", "r" } },
+    { INDEX_op_st8_i32, { "r", "r" } },
+    { INDEX_op_st16_i32, { "r", "r" } },
+    { INDEX_op_st_i32, { "r", "r" } },
+
+    { INDEX_op_add_i32, { "r", "0", "ri" } },
+    { INDEX_op_mul_i32, { "r", "0", "ri" } },
+    { INDEX_op_div2_i32, { "a", "d", "0", "1", "r" } },
+    { INDEX_op_divu2_i32, { "a", "d", "0", "1", "r" } },
+    { INDEX_op_sub_i32, { "r", "0", "ri" } },
+    { INDEX_op_and_i32, { "r", "0", "ri" } },
+    { INDEX_op_or_i32, { "r", "0", "ri" } },
+    { INDEX_op_xor_i32, { "r", "0", "ri" } },
+
+    { INDEX_op_shl_i32, { "r", "0", "ci" } },
+    { INDEX_op_shr_i32, { "r", "0", "ci" } },
+    { INDEX_op_sar_i32, { "r", "0", "ci" } },
+
+    { INDEX_op_brcond_i32, { "r", "ri" } },
+
+    { INDEX_op_mov_i64, { "r", "r" } },
+    { INDEX_op_movi_i64, { "r" } },
+    { INDEX_op_ld8u_i64, { "r", "r" } },
+    { INDEX_op_ld8s_i64, { "r", "r" } },
+    { INDEX_op_ld16u_i64, { "r", "r" } },
+    { INDEX_op_ld16s_i64, { "r", "r" } },
+    { INDEX_op_ld32u_i64, { "r", "r" } },
+    { INDEX_op_ld32s_i64, { "r", "r" } },
+    { INDEX_op_ld_i64, { "r", "r" } },
+    { INDEX_op_st8_i64, { "r", "r" } },
+    { INDEX_op_st16_i64, { "r", "r" } },
+    { INDEX_op_st32_i64, { "r", "r" } },
+    { INDEX_op_st_i64, { "r", "r" } },
+
+    { INDEX_op_add_i64, { "r", "0", "re" } },
+    { INDEX_op_mul_i64, { "r", "0", "re" } },
+    { INDEX_op_div2_i64, { "a", "d", "0", "1", "r" } },
+    { INDEX_op_divu2_i64, { "a", "d", "0", "1", "r" } },
+    { INDEX_op_sub_i64, { "r", "0", "re" } },
+    { INDEX_op_and_i64, { "r", "0", "reZ" } },
+    { INDEX_op_or_i64, { "r", "0", "re" } },
+    { INDEX_op_xor_i64, { "r", "0", "re" } },
+
+    { INDEX_op_shl_i64, { "r", "0", "ci" } },
+    { INDEX_op_shr_i64, { "r", "0", "ci" } },
+    { INDEX_op_sar_i64, { "r", "0", "ci" } },
+
+    { INDEX_op_brcond_i64, { "r", "re" } },
+
+    { INDEX_op_bswap_i32, { "r", "0" } },
+    { INDEX_op_bswap_i64, { "r", "0" } },
+
+    { INDEX_op_qemu_ld8u, { "r", "L" } },
+    { INDEX_op_qemu_ld8s, { "r", "L" } },
+    { INDEX_op_qemu_ld16u, { "r", "L" } },
+    { INDEX_op_qemu_ld16s, { "r", "L" } },
+    { INDEX_op_qemu_ld32u, { "r", "L" } },
+    { INDEX_op_qemu_ld32s, { "r", "L" } },
+    { INDEX_op_qemu_ld64, { "r", "L" } },
+
+    { 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" } },
+
+    { -1 },
+};
+
+void tcg_target_init(TCGContext *s)
+{
+    tcg_regset_set32(tcg_target_available_regs[TCG_TYPE_I32], 0, 0xffff);
+    tcg_regset_set32(tcg_target_available_regs[TCG_TYPE_I64], 0, 0xffff);
+    tcg_regset_set32(tcg_target_call_clobber_regs, 0,
+                     (1 << TCG_REG_RDI) | 
+                     (1 << TCG_REG_RSI) | 
+                     (1 << TCG_REG_RDX) |
+                     (1 << TCG_REG_RCX) |
+                     (1 << TCG_REG_R8) |
+                     (1 << TCG_REG_R9) |
+                     (1 << TCG_REG_RAX) |
+                     (1 << TCG_REG_R10) |
+                     (1 << TCG_REG_R11));
+    
+    tcg_regset_clear(s->reserved_regs);
+    tcg_regset_set_reg(s->reserved_regs, TCG_REG_RSP);
+    /* XXX: will be suppresed when proper global TB entry code will be
+       generated */
+    tcg_regset_set_reg(s->reserved_regs, TCG_REG_RBX);
+    tcg_regset_set_reg(s->reserved_regs, TCG_REG_RBP);
+    
+    tcg_add_target_add_op_defs(x86_64_op_defs);
+}
diff --git a/tcg/x86_64/tcg-target.h b/tcg/x86_64/tcg-target.h
new file mode 100644
index 0000000000..c8421f7f03
--- /dev/null
+++ b/tcg/x86_64/tcg-target.h
@@ -0,0 +1,69 @@
+/*
+ * Tiny Code Generator for QEMU
+ *
+ * Copyright (c) 2008 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#define TCG_TARGET_X86_64 1
+
+#define TCG_TARGET_REG_BITS 64
+//#define TCG_TARGET_WORDS_BIGENDIAN
+
+#define TCG_TARGET_NB_REGS 16
+
+enum {
+    TCG_REG_RAX = 0,
+    TCG_REG_RCX,
+    TCG_REG_RDX,
+    TCG_REG_RBX,
+    TCG_REG_RSP,
+    TCG_REG_RBP,
+    TCG_REG_RSI,
+    TCG_REG_RDI,
+    TCG_REG_R8,
+    TCG_REG_R9,
+    TCG_REG_R10,
+    TCG_REG_R11,
+    TCG_REG_R12,
+    TCG_REG_R13,
+    TCG_REG_R14,
+    TCG_REG_R15,
+};
+
+#define TCG_CT_CONST_S32 0x100
+#define TCG_CT_CONST_U32 0x200
+
+/* used for function call generation */
+#define TCG_REG_CALL_STACK TCG_REG_RSP 
+#define TCG_TARGET_STACK_ALIGN 16
+
+/* optional instructions */
+#define TCG_TARGET_HAS_bswap_i32
+#define TCG_TARGET_HAS_bswap_i64
+
+/* Note: must be synced with dyngen-exec.h */
+#define TCG_AREG0 TCG_REG_R14
+#define TCG_AREG1 TCG_REG_R15
+#define TCG_AREG2 TCG_REG_R12
+#define TCG_AREG3 TCG_REG_R13
+
+static inline void flush_icache_range(unsigned long start, unsigned long stop)
+{
+}