/*
 *  i386 translation
 *
 *  Copyright (c) 2003 Fabrice Bellard
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, see <http://www.gnu.org/licenses/>.
 */
#include "qemu/osdep.h"

#include "qemu/host-utils.h"
#include "cpu.h"
#include "disas/disas.h"
#include "exec/exec-all.h"
#include "tcg-op.h"
#include "exec/cpu_ldst.h"

#include "exec/helper-proto.h"
#include "exec/helper-gen.h"

#include "trace-tcg.h"
#include "exec/log.h"


#define PREFIX_REPZ   0x01
#define PREFIX_REPNZ  0x02
#define PREFIX_LOCK   0x04
#define PREFIX_DATA   0x08
#define PREFIX_ADR    0x10
#define PREFIX_VEX    0x20

#ifdef TARGET_X86_64
#define CODE64(s) ((s)->code64)
#define REX_X(s) ((s)->rex_x)
#define REX_B(s) ((s)->rex_b)
#else
#define CODE64(s) 0
#define REX_X(s) 0
#define REX_B(s) 0
#endif

#ifdef TARGET_X86_64
# define ctztl  ctz64
# define clztl  clz64
#else
# define ctztl  ctz32
# define clztl  clz32
#endif

/* For a switch indexed by MODRM, match all memory operands for a given OP.  */
#define CASE_MODRM_MEM_OP(OP) \
    case (0 << 6) | (OP << 3) | 0 ... (0 << 6) | (OP << 3) | 7: \
    case (1 << 6) | (OP << 3) | 0 ... (1 << 6) | (OP << 3) | 7: \
    case (2 << 6) | (OP << 3) | 0 ... (2 << 6) | (OP << 3) | 7

#define CASE_MODRM_OP(OP) \
    case (0 << 6) | (OP << 3) | 0 ... (0 << 6) | (OP << 3) | 7: \
    case (1 << 6) | (OP << 3) | 0 ... (1 << 6) | (OP << 3) | 7: \
    case (2 << 6) | (OP << 3) | 0 ... (2 << 6) | (OP << 3) | 7: \
    case (3 << 6) | (OP << 3) | 0 ... (3 << 6) | (OP << 3) | 7

//#define MACRO_TEST   1

/* global register indexes */
static TCGv_env cpu_env;
static TCGv cpu_A0;
static TCGv cpu_cc_dst, cpu_cc_src, cpu_cc_src2, cpu_cc_srcT;
static TCGv_i32 cpu_cc_op;
static TCGv cpu_regs[CPU_NB_REGS];
static TCGv cpu_seg_base[6];
static TCGv_i64 cpu_bndl[4];
static TCGv_i64 cpu_bndu[4];
/* local temps */
static TCGv cpu_T0, cpu_T1;
/* local register indexes (only used inside old micro ops) */
static TCGv cpu_tmp0, cpu_tmp4;
static TCGv_ptr cpu_ptr0, cpu_ptr1;
static TCGv_i32 cpu_tmp2_i32, cpu_tmp3_i32;
static TCGv_i64 cpu_tmp1_i64;

#include "exec/gen-icount.h"

#ifdef TARGET_X86_64
static int x86_64_hregs;
#endif

typedef struct DisasContext {
    /* current insn context */
    int override; /* -1 if no override */
    int prefix;
    TCGMemOp aflag;
    TCGMemOp dflag;
    target_ulong pc_start;
    target_ulong pc; /* pc = eip + cs_base */
    int is_jmp; /* 1 = means jump (stop translation), 2 means CPU
                   static state change (stop translation) */
    /* current block context */
    target_ulong cs_base; /* base of CS segment */
    int pe;     /* protected mode */
    int code32; /* 32 bit code segment */
#ifdef TARGET_X86_64
    int lma;    /* long mode active */
    int code64; /* 64 bit code segment */
    int rex_x, rex_b;
#endif
    int vex_l;  /* vex vector length */
    int vex_v;  /* vex vvvv register, without 1's compliment.  */
    int ss32;   /* 32 bit stack segment */
    CCOp cc_op;  /* current CC operation */
    bool cc_op_dirty;
    int addseg; /* non zero if either DS/ES/SS have a non zero base */
    int f_st;   /* currently unused */
    int vm86;   /* vm86 mode */
    int cpl;
    int iopl;
    int tf;     /* TF cpu flag */
    int singlestep_enabled; /* "hardware" single step enabled */
    int jmp_opt; /* use direct block chaining for direct jumps */
    int repz_opt; /* optimize jumps within repz instructions */
    int mem_index; /* select memory access functions */
    uint64_t flags; /* all execution flags */
    struct TranslationBlock *tb;
    int popl_esp_hack; /* for correct popl with esp base handling */
    int rip_offset; /* only used in x86_64, but left for simplicity */
    int cpuid_features;
    int cpuid_ext_features;
    int cpuid_ext2_features;
    int cpuid_ext3_features;
    int cpuid_7_0_ebx_features;
    int cpuid_xsave_features;
} DisasContext;

static void gen_eob(DisasContext *s);
static void gen_jmp(DisasContext *s, target_ulong eip);
static void gen_jmp_tb(DisasContext *s, target_ulong eip, int tb_num);
static void gen_op(DisasContext *s1, int op, TCGMemOp ot, int d);

/* i386 arith/logic operations */
enum {
    OP_ADDL,
    OP_ORL,
    OP_ADCL,
    OP_SBBL,
    OP_ANDL,
    OP_SUBL,
    OP_XORL,
    OP_CMPL,
};

/* i386 shift ops */
enum {
    OP_ROL,
    OP_ROR,
    OP_RCL,
    OP_RCR,
    OP_SHL,
    OP_SHR,
    OP_SHL1, /* undocumented */
    OP_SAR = 7,
};

enum {
    JCC_O,
    JCC_B,
    JCC_Z,
    JCC_BE,
    JCC_S,
    JCC_P,
    JCC_L,
    JCC_LE,
};

enum {
    /* I386 int registers */
    OR_EAX,   /* MUST be even numbered */
    OR_ECX,
    OR_EDX,
    OR_EBX,
    OR_ESP,
    OR_EBP,
    OR_ESI,
    OR_EDI,

    OR_TMP0 = 16,    /* temporary operand register */
    OR_TMP1,
    OR_A0, /* temporary register used when doing address evaluation */
};

enum {
    USES_CC_DST  = 1,
    USES_CC_SRC  = 2,
    USES_CC_SRC2 = 4,
    USES_CC_SRCT = 8,
};

/* Bit set if the global variable is live after setting CC_OP to X.  */
static const uint8_t cc_op_live[CC_OP_NB] = {
    [CC_OP_DYNAMIC] = USES_CC_DST | USES_CC_SRC | USES_CC_SRC2,
    [CC_OP_EFLAGS] = USES_CC_SRC,
    [CC_OP_MULB ... CC_OP_MULQ] = USES_CC_DST | USES_CC_SRC,
    [CC_OP_ADDB ... CC_OP_ADDQ] = USES_CC_DST | USES_CC_SRC,
    [CC_OP_ADCB ... CC_OP_ADCQ] = USES_CC_DST | USES_CC_SRC | USES_CC_SRC2,
    [CC_OP_SUBB ... CC_OP_SUBQ] = USES_CC_DST | USES_CC_SRC | USES_CC_SRCT,
    [CC_OP_SBBB ... CC_OP_SBBQ] = USES_CC_DST | USES_CC_SRC | USES_CC_SRC2,
    [CC_OP_LOGICB ... CC_OP_LOGICQ] = USES_CC_DST,
    [CC_OP_INCB ... CC_OP_INCQ] = USES_CC_DST | USES_CC_SRC,
    [CC_OP_DECB ... CC_OP_DECQ] = USES_CC_DST | USES_CC_SRC,
    [CC_OP_SHLB ... CC_OP_SHLQ] = USES_CC_DST | USES_CC_SRC,
    [CC_OP_SARB ... CC_OP_SARQ] = USES_CC_DST | USES_CC_SRC,
    [CC_OP_BMILGB ... CC_OP_BMILGQ] = USES_CC_DST | USES_CC_SRC,
    [CC_OP_ADCX] = USES_CC_DST | USES_CC_SRC,
    [CC_OP_ADOX] = USES_CC_SRC | USES_CC_SRC2,
    [CC_OP_ADCOX] = USES_CC_DST | USES_CC_SRC | USES_CC_SRC2,
    [CC_OP_CLR] = 0,
};

static void set_cc_op(DisasContext *s, CCOp op)
{
    int dead;

    if (s->cc_op == op) {
        return;
    }

    /* Discard CC computation that will no longer be used.  */
    dead = cc_op_live[s->cc_op] & ~cc_op_live[op];
    if (dead & USES_CC_DST) {
        tcg_gen_discard_tl(cpu_cc_dst);
    }
    if (dead & USES_CC_SRC) {
        tcg_gen_discard_tl(cpu_cc_src);
    }
    if (dead & USES_CC_SRC2) {
        tcg_gen_discard_tl(cpu_cc_src2);
    }
    if (dead & USES_CC_SRCT) {
        tcg_gen_discard_tl(cpu_cc_srcT);
    }

    if (op == CC_OP_DYNAMIC) {
        /* The DYNAMIC setting is translator only, and should never be
           stored.  Thus we always consider it clean.  */
        s->cc_op_dirty = false;
    } else {
        /* Discard any computed CC_OP value (see shifts).  */
        if (s->cc_op == CC_OP_DYNAMIC) {
            tcg_gen_discard_i32(cpu_cc_op);
        }
        s->cc_op_dirty = true;
    }
    s->cc_op = op;
}

static void gen_update_cc_op(DisasContext *s)
{
    if (s->cc_op_dirty) {
        tcg_gen_movi_i32(cpu_cc_op, s->cc_op);
        s->cc_op_dirty = false;
    }
}

#ifdef TARGET_X86_64

#define NB_OP_SIZES 4

#else /* !TARGET_X86_64 */

#define NB_OP_SIZES 3

#endif /* !TARGET_X86_64 */

#if defined(HOST_WORDS_BIGENDIAN)
#define REG_B_OFFSET (sizeof(target_ulong) - 1)
#define REG_H_OFFSET (sizeof(target_ulong) - 2)
#define REG_W_OFFSET (sizeof(target_ulong) - 2)
#define REG_L_OFFSET (sizeof(target_ulong) - 4)
#define REG_LH_OFFSET (sizeof(target_ulong) - 8)
#else
#define REG_B_OFFSET 0
#define REG_H_OFFSET 1
#define REG_W_OFFSET 0
#define REG_L_OFFSET 0
#define REG_LH_OFFSET 4
#endif

/* In instruction encodings for byte register accesses the
 * register number usually indicates "low 8 bits of register N";
 * however there are some special cases where N 4..7 indicates
 * [AH, CH, DH, BH], ie "bits 15..8 of register N-4". Return
 * true for this special case, false otherwise.
 */
static inline bool byte_reg_is_xH(int reg)
{
    if (reg < 4) {
        return false;
    }
#ifdef TARGET_X86_64
    if (reg >= 8 || x86_64_hregs) {
        return false;
    }
#endif
    return true;
}

/* Select the size of a push/pop operation.  */
static inline TCGMemOp mo_pushpop(DisasContext *s, TCGMemOp ot)
{
    if (CODE64(s)) {
        return ot == MO_16 ? MO_16 : MO_64;
    } else {
        return ot;
    }
}

/* Select the size of the stack pointer.  */
static inline TCGMemOp mo_stacksize(DisasContext *s)
{
    return CODE64(s) ? MO_64 : s->ss32 ? MO_32 : MO_16;
}

/* Select only size 64 else 32.  Used for SSE operand sizes.  */
static inline TCGMemOp mo_64_32(TCGMemOp ot)
{
#ifdef TARGET_X86_64
    return ot == MO_64 ? MO_64 : MO_32;
#else
    return MO_32;
#endif
}

/* Select size 8 if lsb of B is clear, else OT.  Used for decoding
   byte vs word opcodes.  */
static inline TCGMemOp mo_b_d(int b, TCGMemOp ot)
{
    return b & 1 ? ot : MO_8;
}

/* Select size 8 if lsb of B is clear, else OT capped at 32.
   Used for decoding operand size of port opcodes.  */
static inline TCGMemOp mo_b_d32(int b, TCGMemOp ot)
{
    return b & 1 ? (ot == MO_16 ? MO_16 : MO_32) : MO_8;
}

static void gen_op_mov_reg_v(TCGMemOp ot, int reg, TCGv t0)
{
    switch(ot) {
    case MO_8:
        if (!byte_reg_is_xH(reg)) {
            tcg_gen_deposit_tl(cpu_regs[reg], cpu_regs[reg], t0, 0, 8);
        } else {
            tcg_gen_deposit_tl(cpu_regs[reg - 4], cpu_regs[reg - 4], t0, 8, 8);
        }
        break;
    case MO_16:
        tcg_gen_deposit_tl(cpu_regs[reg], cpu_regs[reg], t0, 0, 16);
        break;
    case MO_32:
        /* For x86_64, this sets the higher half of register to zero.
           For i386, this is equivalent to a mov. */
        tcg_gen_ext32u_tl(cpu_regs[reg], t0);
        break;
#ifdef TARGET_X86_64
    case MO_64:
        tcg_gen_mov_tl(cpu_regs[reg], t0);
        break;
#endif
    default:
        tcg_abort();
    }
}

static inline void gen_op_mov_v_reg(TCGMemOp ot, TCGv t0, int reg)
{
    if (ot == MO_8 && byte_reg_is_xH(reg)) {
        tcg_gen_shri_tl(t0, cpu_regs[reg - 4], 8);
        tcg_gen_ext8u_tl(t0, t0);
    } else {
        tcg_gen_mov_tl(t0, cpu_regs[reg]);
    }
}

static void gen_add_A0_im(DisasContext *s, int val)
{
    tcg_gen_addi_tl(cpu_A0, cpu_A0, val);
    if (!CODE64(s)) {
        tcg_gen_ext32u_tl(cpu_A0, cpu_A0);
    }
}

static inline void gen_op_jmp_v(TCGv dest)
{
    tcg_gen_st_tl(dest, cpu_env, offsetof(CPUX86State, eip));
}

static inline void gen_op_add_reg_im(TCGMemOp size, int reg, int32_t val)
{
    tcg_gen_addi_tl(cpu_tmp0, cpu_regs[reg], val);
    gen_op_mov_reg_v(size, reg, cpu_tmp0);
}

static inline void gen_op_add_reg_T0(TCGMemOp size, int reg)
{
    tcg_gen_add_tl(cpu_tmp0, cpu_regs[reg], cpu_T0);
    gen_op_mov_reg_v(size, reg, cpu_tmp0);
}

static inline void gen_op_ld_v(DisasContext *s, int idx, TCGv t0, TCGv a0)
{
    tcg_gen_qemu_ld_tl(t0, a0, s->mem_index, idx | MO_LE);
}

static inline void gen_op_st_v(DisasContext *s, int idx, TCGv t0, TCGv a0)
{
    tcg_gen_qemu_st_tl(t0, a0, s->mem_index, idx | MO_LE);
}

static inline void gen_op_st_rm_T0_A0(DisasContext *s, int idx, int d)
{
    if (d == OR_TMP0) {
        gen_op_st_v(s, idx, cpu_T0, cpu_A0);
    } else {
        gen_op_mov_reg_v(idx, d, cpu_T0);
    }
}

static inline void gen_jmp_im(target_ulong pc)
{
    tcg_gen_movi_tl(cpu_tmp0, pc);
    gen_op_jmp_v(cpu_tmp0);
}

/* Compute SEG:REG into A0.  SEG is selected from the override segment
   (OVR_SEG) and the default segment (DEF_SEG).  OVR_SEG may be -1 to
   indicate no override.  */
static void gen_lea_v_seg(DisasContext *s, TCGMemOp aflag, TCGv a0,
                          int def_seg, int ovr_seg)
{
    switch (aflag) {
#ifdef TARGET_X86_64
    case MO_64:
        if (ovr_seg < 0) {
            tcg_gen_mov_tl(cpu_A0, a0);
            return;
        }
        break;
#endif
    case MO_32:
        /* 32 bit address */
        if (ovr_seg < 0) {
            if (s->addseg) {
                ovr_seg = def_seg;
            } else {
                tcg_gen_ext32u_tl(cpu_A0, a0);
                return;
            }
        }
        break;
    case MO_16:
        /* 16 bit address */
        tcg_gen_ext16u_tl(cpu_A0, a0);
        a0 = cpu_A0;
        if (ovr_seg < 0) {
            if (s->addseg) {
                ovr_seg = def_seg;
            } else {
                return;
            }
        }
        break;
    default:
        tcg_abort();
    }

    if (ovr_seg >= 0) {
        TCGv seg = cpu_seg_base[ovr_seg];

        if (aflag == MO_64) {
            tcg_gen_add_tl(cpu_A0, a0, seg);
        } else if (CODE64(s)) {
            tcg_gen_ext32u_tl(cpu_A0, a0);
            tcg_gen_add_tl(cpu_A0, cpu_A0, seg);
        } else {
            tcg_gen_add_tl(cpu_A0, a0, seg);
            tcg_gen_ext32u_tl(cpu_A0, cpu_A0);
        }
    }
}

static inline void gen_string_movl_A0_ESI(DisasContext *s)
{
    gen_lea_v_seg(s, s->aflag, cpu_regs[R_ESI], R_DS, s->override);
}

static inline void gen_string_movl_A0_EDI(DisasContext *s)
{
    gen_lea_v_seg(s, s->aflag, cpu_regs[R_EDI], R_ES, -1);
}

static inline void gen_op_movl_T0_Dshift(TCGMemOp ot)
{
    tcg_gen_ld32s_tl(cpu_T0, cpu_env, offsetof(CPUX86State, df));
    tcg_gen_shli_tl(cpu_T0, cpu_T0, ot);
};

static TCGv gen_ext_tl(TCGv dst, TCGv src, TCGMemOp size, bool sign)
{
    switch (size) {
    case MO_8:
        if (sign) {
            tcg_gen_ext8s_tl(dst, src);
        } else {
            tcg_gen_ext8u_tl(dst, src);
        }
        return dst;
    case MO_16:
        if (sign) {
            tcg_gen_ext16s_tl(dst, src);
        } else {
            tcg_gen_ext16u_tl(dst, src);
        }
        return dst;
#ifdef TARGET_X86_64
    case MO_32:
        if (sign) {
            tcg_gen_ext32s_tl(dst, src);
        } else {
            tcg_gen_ext32u_tl(dst, src);
        }
        return dst;
#endif
    default:
        return src;
    }
}

static void gen_extu(TCGMemOp ot, TCGv reg)
{
    gen_ext_tl(reg, reg, ot, false);
}

static void gen_exts(TCGMemOp ot, TCGv reg)
{
    gen_ext_tl(reg, reg, ot, true);
}

static inline void gen_op_jnz_ecx(TCGMemOp size, TCGLabel *label1)
{
    tcg_gen_mov_tl(cpu_tmp0, cpu_regs[R_ECX]);
    gen_extu(size, cpu_tmp0);
    tcg_gen_brcondi_tl(TCG_COND_NE, cpu_tmp0, 0, label1);
}

static inline void gen_op_jz_ecx(TCGMemOp size, TCGLabel *label1)
{
    tcg_gen_mov_tl(cpu_tmp0, cpu_regs[R_ECX]);
    gen_extu(size, cpu_tmp0);
    tcg_gen_brcondi_tl(TCG_COND_EQ, cpu_tmp0, 0, label1);
}

static void gen_helper_in_func(TCGMemOp ot, TCGv v, TCGv_i32 n)
{
    switch (ot) {
    case MO_8:
        gen_helper_inb(v, cpu_env, n);
        break;
    case MO_16:
        gen_helper_inw(v, cpu_env, n);
        break;
    case MO_32:
        gen_helper_inl(v, cpu_env, n);
        break;
    default:
        tcg_abort();
    }
}

static void gen_helper_out_func(TCGMemOp ot, TCGv_i32 v, TCGv_i32 n)
{
    switch (ot) {
    case MO_8:
        gen_helper_outb(cpu_env, v, n);
        break;
    case MO_16:
        gen_helper_outw(cpu_env, v, n);
        break;
    case MO_32:
        gen_helper_outl(cpu_env, v, n);
        break;
    default:
        tcg_abort();
    }
}

static void gen_check_io(DisasContext *s, TCGMemOp ot, target_ulong cur_eip,
                         uint32_t svm_flags)
{
    target_ulong next_eip;

    if (s->pe && (s->cpl > s->iopl || s->vm86)) {
        tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T0);
        switch (ot) {
        case MO_8:
            gen_helper_check_iob(cpu_env, cpu_tmp2_i32);
            break;
        case MO_16:
            gen_helper_check_iow(cpu_env, cpu_tmp2_i32);
            break;
        case MO_32:
            gen_helper_check_iol(cpu_env, cpu_tmp2_i32);
            break;
        default:
            tcg_abort();
        }
    }
    if(s->flags & HF_SVMI_MASK) {
        gen_update_cc_op(s);
        gen_jmp_im(cur_eip);
        svm_flags |= (1 << (4 + ot));
        next_eip = s->pc - s->cs_base;
        tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T0);
        gen_helper_svm_check_io(cpu_env, cpu_tmp2_i32,
                                tcg_const_i32(svm_flags),
                                tcg_const_i32(next_eip - cur_eip));
    }
}

static inline void gen_movs(DisasContext *s, TCGMemOp ot)
{
    gen_string_movl_A0_ESI(s);
    gen_op_ld_v(s, ot, cpu_T0, cpu_A0);
    gen_string_movl_A0_EDI(s);
    gen_op_st_v(s, ot, cpu_T0, cpu_A0);
    gen_op_movl_T0_Dshift(ot);
    gen_op_add_reg_T0(s->aflag, R_ESI);
    gen_op_add_reg_T0(s->aflag, R_EDI);
}

static void gen_op_update1_cc(void)
{
    tcg_gen_mov_tl(cpu_cc_dst, cpu_T0);
}

static void gen_op_update2_cc(void)
{
    tcg_gen_mov_tl(cpu_cc_src, cpu_T1);
    tcg_gen_mov_tl(cpu_cc_dst, cpu_T0);
}

static void gen_op_update3_cc(TCGv reg)
{
    tcg_gen_mov_tl(cpu_cc_src2, reg);
    tcg_gen_mov_tl(cpu_cc_src, cpu_T1);
    tcg_gen_mov_tl(cpu_cc_dst, cpu_T0);
}

static inline void gen_op_testl_T0_T1_cc(void)
{
    tcg_gen_and_tl(cpu_cc_dst, cpu_T0, cpu_T1);
}

static void gen_op_update_neg_cc(void)
{
    tcg_gen_mov_tl(cpu_cc_dst, cpu_T0);
    tcg_gen_neg_tl(cpu_cc_src, cpu_T0);
    tcg_gen_movi_tl(cpu_cc_srcT, 0);
}

/* compute all eflags to cc_src */
static void gen_compute_eflags(DisasContext *s)
{
    TCGv zero, dst, src1, src2;
    int live, dead;

    if (s->cc_op == CC_OP_EFLAGS) {
        return;
    }
    if (s->cc_op == CC_OP_CLR) {
        tcg_gen_movi_tl(cpu_cc_src, CC_Z | CC_P);
        set_cc_op(s, CC_OP_EFLAGS);
        return;
    }

    TCGV_UNUSED(zero);
    dst = cpu_cc_dst;
    src1 = cpu_cc_src;
    src2 = cpu_cc_src2;

    /* Take care to not read values that are not live.  */
    live = cc_op_live[s->cc_op] & ~USES_CC_SRCT;
    dead = live ^ (USES_CC_DST | USES_CC_SRC | USES_CC_SRC2);
    if (dead) {
        zero = tcg_const_tl(0);
        if (dead & USES_CC_DST) {
            dst = zero;
        }
        if (dead & USES_CC_SRC) {
            src1 = zero;
        }
        if (dead & USES_CC_SRC2) {
            src2 = zero;
        }
    }

    gen_update_cc_op(s);
    gen_helper_cc_compute_all(cpu_cc_src, dst, src1, src2, cpu_cc_op);
    set_cc_op(s, CC_OP_EFLAGS);

    if (dead) {
        tcg_temp_free(zero);
    }
}

typedef struct CCPrepare {
    TCGCond cond;
    TCGv reg;
    TCGv reg2;
    target_ulong imm;
    target_ulong mask;
    bool use_reg2;
    bool no_setcond;
} CCPrepare;

/* compute eflags.C to reg */
static CCPrepare gen_prepare_eflags_c(DisasContext *s, TCGv reg)
{
    TCGv t0, t1;
    int size, shift;

    switch (s->cc_op) {
    case CC_OP_SUBB ... CC_OP_SUBQ:
        /* (DATA_TYPE)CC_SRCT < (DATA_TYPE)CC_SRC */
        size = s->cc_op - CC_OP_SUBB;
        t1 = gen_ext_tl(cpu_tmp0, cpu_cc_src, size, false);
        /* If no temporary was used, be careful not to alias t1 and t0.  */
        t0 = TCGV_EQUAL(t1, cpu_cc_src) ? cpu_tmp0 : reg;
        tcg_gen_mov_tl(t0, cpu_cc_srcT);
        gen_extu(size, t0);
        goto add_sub;

    case CC_OP_ADDB ... CC_OP_ADDQ:
        /* (DATA_TYPE)CC_DST < (DATA_TYPE)CC_SRC */
        size = s->cc_op - CC_OP_ADDB;
        t1 = gen_ext_tl(cpu_tmp0, cpu_cc_src, size, false);
        t0 = gen_ext_tl(reg, cpu_cc_dst, size, false);
    add_sub:
        return (CCPrepare) { .cond = TCG_COND_LTU, .reg = t0,
                             .reg2 = t1, .mask = -1, .use_reg2 = true };

    case CC_OP_LOGICB ... CC_OP_LOGICQ:
    case CC_OP_CLR:
        return (CCPrepare) { .cond = TCG_COND_NEVER, .mask = -1 };

    case CC_OP_INCB ... CC_OP_INCQ:
    case CC_OP_DECB ... CC_OP_DECQ:
        return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src,
                             .mask = -1, .no_setcond = true };

    case CC_OP_SHLB ... CC_OP_SHLQ:
        /* (CC_SRC >> (DATA_BITS - 1)) & 1 */
        size = s->cc_op - CC_OP_SHLB;
        shift = (8 << size) - 1;
        return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src,
                             .mask = (target_ulong)1 << shift };

    case CC_OP_MULB ... CC_OP_MULQ:
        return (CCPrepare) { .cond = TCG_COND_NE,
                             .reg = cpu_cc_src, .mask = -1 };

    case CC_OP_BMILGB ... CC_OP_BMILGQ:
        size = s->cc_op - CC_OP_BMILGB;
        t0 = gen_ext_tl(reg, cpu_cc_src, size, false);
        return (CCPrepare) { .cond = TCG_COND_EQ, .reg = t0, .mask = -1 };

    case CC_OP_ADCX:
    case CC_OP_ADCOX:
        return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_dst,
                             .mask = -1, .no_setcond = true };

    case CC_OP_EFLAGS:
    case CC_OP_SARB ... CC_OP_SARQ:
        /* CC_SRC & 1 */
        return (CCPrepare) { .cond = TCG_COND_NE,
                             .reg = cpu_cc_src, .mask = CC_C };

    default:
       /* The need to compute only C from CC_OP_DYNAMIC is important
          in efficiently implementing e.g. INC at the start of a TB.  */
       gen_update_cc_op(s);
       gen_helper_cc_compute_c(reg, cpu_cc_dst, cpu_cc_src,
                               cpu_cc_src2, cpu_cc_op);
       return (CCPrepare) { .cond = TCG_COND_NE, .reg = reg,
                            .mask = -1, .no_setcond = true };
    }
}

/* compute eflags.P to reg */
static CCPrepare gen_prepare_eflags_p(DisasContext *s, TCGv reg)
{
    gen_compute_eflags(s);
    return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src,
                         .mask = CC_P };
}

/* compute eflags.S to reg */
static CCPrepare gen_prepare_eflags_s(DisasContext *s, TCGv reg)
{
    switch (s->cc_op) {
    case CC_OP_DYNAMIC:
        gen_compute_eflags(s);
        /* FALLTHRU */
    case CC_OP_EFLAGS:
    case CC_OP_ADCX:
    case CC_OP_ADOX:
    case CC_OP_ADCOX:
        return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src,
                             .mask = CC_S };
    case CC_OP_CLR:
        return (CCPrepare) { .cond = TCG_COND_NEVER, .mask = -1 };
    default:
        {
            TCGMemOp size = (s->cc_op - CC_OP_ADDB) & 3;
            TCGv t0 = gen_ext_tl(reg, cpu_cc_dst, size, true);
            return (CCPrepare) { .cond = TCG_COND_LT, .reg = t0, .mask = -1 };
        }
    }
}

/* compute eflags.O to reg */
static CCPrepare gen_prepare_eflags_o(DisasContext *s, TCGv reg)
{
    switch (s->cc_op) {
    case CC_OP_ADOX:
    case CC_OP_ADCOX:
        return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src2,
                             .mask = -1, .no_setcond = true };
    case CC_OP_CLR:
        return (CCPrepare) { .cond = TCG_COND_NEVER, .mask = -1 };
    default:
        gen_compute_eflags(s);
        return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src,
                             .mask = CC_O };
    }
}

/* compute eflags.Z to reg */
static CCPrepare gen_prepare_eflags_z(DisasContext *s, TCGv reg)
{
    switch (s->cc_op) {
    case CC_OP_DYNAMIC:
        gen_compute_eflags(s);
        /* FALLTHRU */
    case CC_OP_EFLAGS:
    case CC_OP_ADCX:
    case CC_OP_ADOX:
    case CC_OP_ADCOX:
        return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src,
                             .mask = CC_Z };
    case CC_OP_CLR:
        return (CCPrepare) { .cond = TCG_COND_ALWAYS, .mask = -1 };
    default:
        {
            TCGMemOp size = (s->cc_op - CC_OP_ADDB) & 3;
            TCGv t0 = gen_ext_tl(reg, cpu_cc_dst, size, false);
            return (CCPrepare) { .cond = TCG_COND_EQ, .reg = t0, .mask = -1 };
        }
    }
}

/* perform a conditional store into register 'reg' according to jump opcode
   value 'b'. In the fast case, T0 is guaranted not to be used. */
static CCPrepare gen_prepare_cc(DisasContext *s, int b, TCGv reg)
{
    int inv, jcc_op, cond;
    TCGMemOp size;
    CCPrepare cc;
    TCGv t0;

    inv = b & 1;
    jcc_op = (b >> 1) & 7;

    switch (s->cc_op) {
    case CC_OP_SUBB ... CC_OP_SUBQ:
        /* We optimize relational operators for the cmp/jcc case.  */
        size = s->cc_op - CC_OP_SUBB;
        switch (jcc_op) {
        case JCC_BE:
            tcg_gen_mov_tl(cpu_tmp4, cpu_cc_srcT);
            gen_extu(size, cpu_tmp4);
            t0 = gen_ext_tl(cpu_tmp0, cpu_cc_src, size, false);
            cc = (CCPrepare) { .cond = TCG_COND_LEU, .reg = cpu_tmp4,
                               .reg2 = t0, .mask = -1, .use_reg2 = true };
            break;

        case JCC_L:
            cond = TCG_COND_LT;
            goto fast_jcc_l;
        case JCC_LE:
            cond = TCG_COND_LE;
        fast_jcc_l:
            tcg_gen_mov_tl(cpu_tmp4, cpu_cc_srcT);
            gen_exts(size, cpu_tmp4);
            t0 = gen_ext_tl(cpu_tmp0, cpu_cc_src, size, true);
            cc = (CCPrepare) { .cond = cond, .reg = cpu_tmp4,
                               .reg2 = t0, .mask = -1, .use_reg2 = true };
            break;

        default:
            goto slow_jcc;
        }
        break;

    default:
    slow_jcc:
        /* This actually generates good code for JC, JZ and JS.  */
        switch (jcc_op) {
        case JCC_O:
            cc = gen_prepare_eflags_o(s, reg);
            break;
        case JCC_B:
            cc = gen_prepare_eflags_c(s, reg);
            break;
        case JCC_Z:
            cc = gen_prepare_eflags_z(s, reg);
            break;
        case JCC_BE:
            gen_compute_eflags(s);
            cc = (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src,
                               .mask = CC_Z | CC_C };
            break;
        case JCC_S:
            cc = gen_prepare_eflags_s(s, reg);
            break;
        case JCC_P:
            cc = gen_prepare_eflags_p(s, reg);
            break;
        case JCC_L:
            gen_compute_eflags(s);
            if (TCGV_EQUAL(reg, cpu_cc_src)) {
                reg = cpu_tmp0;
            }
            tcg_gen_shri_tl(reg, cpu_cc_src, 4); /* CC_O -> CC_S */
            tcg_gen_xor_tl(reg, reg, cpu_cc_src);
            cc = (CCPrepare) { .cond = TCG_COND_NE, .reg = reg,
                               .mask = CC_S };
            break;
        default:
        case JCC_LE:
            gen_compute_eflags(s);
            if (TCGV_EQUAL(reg, cpu_cc_src)) {
                reg = cpu_tmp0;
            }
            tcg_gen_shri_tl(reg, cpu_cc_src, 4); /* CC_O -> CC_S */
            tcg_gen_xor_tl(reg, reg, cpu_cc_src);
            cc = (CCPrepare) { .cond = TCG_COND_NE, .reg = reg,
                               .mask = CC_S | CC_Z };
            break;
        }
        break;
    }

    if (inv) {
        cc.cond = tcg_invert_cond(cc.cond);
    }
    return cc;
}

static void gen_setcc1(DisasContext *s, int b, TCGv reg)
{
    CCPrepare cc = gen_prepare_cc(s, b, reg);

    if (cc.no_setcond) {
        if (cc.cond == TCG_COND_EQ) {
            tcg_gen_xori_tl(reg, cc.reg, 1);
        } else {
            tcg_gen_mov_tl(reg, cc.reg);
        }
        return;
    }

    if (cc.cond == TCG_COND_NE && !cc.use_reg2 && cc.imm == 0 &&
        cc.mask != 0 && (cc.mask & (cc.mask - 1)) == 0) {
        tcg_gen_shri_tl(reg, cc.reg, ctztl(cc.mask));
        tcg_gen_andi_tl(reg, reg, 1);
        return;
    }
    if (cc.mask != -1) {
        tcg_gen_andi_tl(reg, cc.reg, cc.mask);
        cc.reg = reg;
    }

    if (cc.use_reg2) {
        tcg_gen_setcond_tl(cc.cond, reg, cc.reg, cc.reg2);
    } else {
        tcg_gen_setcondi_tl(cc.cond, reg, cc.reg, cc.imm);
    }
}

static inline void gen_compute_eflags_c(DisasContext *s, TCGv reg)
{
    gen_setcc1(s, JCC_B << 1, reg);
}

/* generate a conditional jump to label 'l1' according to jump opcode
   value 'b'. In the fast case, T0 is guaranted not to be used. */
static inline void gen_jcc1_noeob(DisasContext *s, int b, TCGLabel *l1)
{
    CCPrepare cc = gen_prepare_cc(s, b, cpu_T0);

    if (cc.mask != -1) {
        tcg_gen_andi_tl(cpu_T0, cc.reg, cc.mask);
        cc.reg = cpu_T0;
    }
    if (cc.use_reg2) {
        tcg_gen_brcond_tl(cc.cond, cc.reg, cc.reg2, l1);
    } else {
        tcg_gen_brcondi_tl(cc.cond, cc.reg, cc.imm, l1);
    }
}

/* Generate a conditional jump to label 'l1' according to jump opcode
   value 'b'. In the fast case, T0 is guaranted not to be used.
   A translation block must end soon.  */
static inline void gen_jcc1(DisasContext *s, int b, TCGLabel *l1)
{
    CCPrepare cc = gen_prepare_cc(s, b, cpu_T0);

    gen_update_cc_op(s);
    if (cc.mask != -1) {
        tcg_gen_andi_tl(cpu_T0, cc.reg, cc.mask);
        cc.reg = cpu_T0;
    }
    set_cc_op(s, CC_OP_DYNAMIC);
    if (cc.use_reg2) {
        tcg_gen_brcond_tl(cc.cond, cc.reg, cc.reg2, l1);
    } else {
        tcg_gen_brcondi_tl(cc.cond, cc.reg, cc.imm, l1);
    }
}

/* XXX: does not work with gdbstub "ice" single step - not a
   serious problem */
static TCGLabel *gen_jz_ecx_string(DisasContext *s, target_ulong next_eip)
{
    TCGLabel *l1 = gen_new_label();
    TCGLabel *l2 = gen_new_label();
    gen_op_jnz_ecx(s->aflag, l1);
    gen_set_label(l2);
    gen_jmp_tb(s, next_eip, 1);
    gen_set_label(l1);
    return l2;
}

static inline void gen_stos(DisasContext *s, TCGMemOp ot)
{
    gen_op_mov_v_reg(MO_32, cpu_T0, R_EAX);
    gen_string_movl_A0_EDI(s);
    gen_op_st_v(s, ot, cpu_T0, cpu_A0);
    gen_op_movl_T0_Dshift(ot);
    gen_op_add_reg_T0(s->aflag, R_EDI);
}

static inline void gen_lods(DisasContext *s, TCGMemOp ot)
{
    gen_string_movl_A0_ESI(s);
    gen_op_ld_v(s, ot, cpu_T0, cpu_A0);
    gen_op_mov_reg_v(ot, R_EAX, cpu_T0);
    gen_op_movl_T0_Dshift(ot);
    gen_op_add_reg_T0(s->aflag, R_ESI);
}

static inline void gen_scas(DisasContext *s, TCGMemOp ot)
{
    gen_string_movl_A0_EDI(s);
    gen_op_ld_v(s, ot, cpu_T1, cpu_A0);
    gen_op(s, OP_CMPL, ot, R_EAX);
    gen_op_movl_T0_Dshift(ot);
    gen_op_add_reg_T0(s->aflag, R_EDI);
}

static inline void gen_cmps(DisasContext *s, TCGMemOp ot)
{
    gen_string_movl_A0_EDI(s);
    gen_op_ld_v(s, ot, cpu_T1, cpu_A0);
    gen_string_movl_A0_ESI(s);
    gen_op(s, OP_CMPL, ot, OR_TMP0);
    gen_op_movl_T0_Dshift(ot);
    gen_op_add_reg_T0(s->aflag, R_ESI);
    gen_op_add_reg_T0(s->aflag, R_EDI);
}

static void gen_bpt_io(DisasContext *s, TCGv_i32 t_port, int ot)
{
    if (s->flags & HF_IOBPT_MASK) {
        TCGv_i32 t_size = tcg_const_i32(1 << ot);
        TCGv t_next = tcg_const_tl(s->pc - s->cs_base);

        gen_helper_bpt_io(cpu_env, t_port, t_size, t_next);
        tcg_temp_free_i32(t_size);
        tcg_temp_free(t_next);
    }
}


static inline void gen_ins(DisasContext *s, TCGMemOp ot)
{
    if (s->tb->cflags & CF_USE_ICOUNT) {
        gen_io_start();
    }
    gen_string_movl_A0_EDI(s);
    /* Note: we must do this dummy write first to be restartable in
       case of page fault. */
    tcg_gen_movi_tl(cpu_T0, 0);
    gen_op_st_v(s, ot, cpu_T0, cpu_A0);
    tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_regs[R_EDX]);
    tcg_gen_andi_i32(cpu_tmp2_i32, cpu_tmp2_i32, 0xffff);
    gen_helper_in_func(ot, cpu_T0, cpu_tmp2_i32);
    gen_op_st_v(s, ot, cpu_T0, cpu_A0);
    gen_op_movl_T0_Dshift(ot);
    gen_op_add_reg_T0(s->aflag, R_EDI);
    gen_bpt_io(s, cpu_tmp2_i32, ot);
    if (s->tb->cflags & CF_USE_ICOUNT) {
        gen_io_end();
    }
}

static inline void gen_outs(DisasContext *s, TCGMemOp ot)
{
    if (s->tb->cflags & CF_USE_ICOUNT) {
        gen_io_start();
    }
    gen_string_movl_A0_ESI(s);
    gen_op_ld_v(s, ot, cpu_T0, cpu_A0);

    tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_regs[R_EDX]);
    tcg_gen_andi_i32(cpu_tmp2_i32, cpu_tmp2_i32, 0xffff);
    tcg_gen_trunc_tl_i32(cpu_tmp3_i32, cpu_T0);
    gen_helper_out_func(ot, cpu_tmp2_i32, cpu_tmp3_i32);
    gen_op_movl_T0_Dshift(ot);
    gen_op_add_reg_T0(s->aflag, R_ESI);
    gen_bpt_io(s, cpu_tmp2_i32, ot);
    if (s->tb->cflags & CF_USE_ICOUNT) {
        gen_io_end();
    }
}

/* same method as Valgrind : we generate jumps to current or next
   instruction */
#define GEN_REPZ(op)                                                          \
static inline void gen_repz_ ## op(DisasContext *s, TCGMemOp ot,              \
                                 target_ulong cur_eip, target_ulong next_eip) \
{                                                                             \
    TCGLabel *l2;                                                             \
    gen_update_cc_op(s);                                                      \
    l2 = gen_jz_ecx_string(s, next_eip);                                      \
    gen_ ## op(s, ot);                                                        \
    gen_op_add_reg_im(s->aflag, R_ECX, -1);                                   \
    /* a loop would cause two single step exceptions if ECX = 1               \
       before rep string_insn */                                              \
    if (s->repz_opt)                                                          \
        gen_op_jz_ecx(s->aflag, l2);                                          \
    gen_jmp(s, cur_eip);                                                      \
}

#define GEN_REPZ2(op)                                                         \
static inline void gen_repz_ ## op(DisasContext *s, TCGMemOp ot,              \
                                   target_ulong cur_eip,                      \
                                   target_ulong next_eip,                     \
                                   int nz)                                    \
{                                                                             \
    TCGLabel *l2;                                                             \
    gen_update_cc_op(s);                                                      \
    l2 = gen_jz_ecx_string(s, next_eip);                                      \
    gen_ ## op(s, ot);                                                        \
    gen_op_add_reg_im(s->aflag, R_ECX, -1);                                   \
    gen_update_cc_op(s);                                                      \
    gen_jcc1(s, (JCC_Z << 1) | (nz ^ 1), l2);                                 \
    if (s->repz_opt)                                                          \
        gen_op_jz_ecx(s->aflag, l2);                                          \
    gen_jmp(s, cur_eip);                                                      \
}

GEN_REPZ(movs)
GEN_REPZ(stos)
GEN_REPZ(lods)
GEN_REPZ(ins)
GEN_REPZ(outs)
GEN_REPZ2(scas)
GEN_REPZ2(cmps)

static void gen_helper_fp_arith_ST0_FT0(int op)
{
    switch (op) {
    case 0:
        gen_helper_fadd_ST0_FT0(cpu_env);
        break;
    case 1:
        gen_helper_fmul_ST0_FT0(cpu_env);
        break;
    case 2:
        gen_helper_fcom_ST0_FT0(cpu_env);
        break;
    case 3:
        gen_helper_fcom_ST0_FT0(cpu_env);
        break;
    case 4:
        gen_helper_fsub_ST0_FT0(cpu_env);
        break;
    case 5:
        gen_helper_fsubr_ST0_FT0(cpu_env);
        break;
    case 6:
        gen_helper_fdiv_ST0_FT0(cpu_env);
        break;
    case 7:
        gen_helper_fdivr_ST0_FT0(cpu_env);
        break;
    }
}

/* NOTE the exception in "r" op ordering */
static void gen_helper_fp_arith_STN_ST0(int op, int opreg)
{
    TCGv_i32 tmp = tcg_const_i32(opreg);
    switch (op) {
    case 0:
        gen_helper_fadd_STN_ST0(cpu_env, tmp);
        break;
    case 1:
        gen_helper_fmul_STN_ST0(cpu_env, tmp);
        break;
    case 4:
        gen_helper_fsubr_STN_ST0(cpu_env, tmp);
        break;
    case 5:
        gen_helper_fsub_STN_ST0(cpu_env, tmp);
        break;
    case 6:
        gen_helper_fdivr_STN_ST0(cpu_env, tmp);
        break;
    case 7:
        gen_helper_fdiv_STN_ST0(cpu_env, tmp);
        break;
    }
}

/* if d == OR_TMP0, it means memory operand (address in A0) */
static void gen_op(DisasContext *s1, int op, TCGMemOp ot, int d)
{
    if (d != OR_TMP0) {
        gen_op_mov_v_reg(ot, cpu_T0, d);
    } else {
        gen_op_ld_v(s1, ot, cpu_T0, cpu_A0);
    }
    switch(op) {
    case OP_ADCL:
        gen_compute_eflags_c(s1, cpu_tmp4);
        tcg_gen_add_tl(cpu_T0, cpu_T0, cpu_T1);
        tcg_gen_add_tl(cpu_T0, cpu_T0, cpu_tmp4);
        gen_op_st_rm_T0_A0(s1, ot, d);
        gen_op_update3_cc(cpu_tmp4);
        set_cc_op(s1, CC_OP_ADCB + ot);
        break;
    case OP_SBBL:
        gen_compute_eflags_c(s1, cpu_tmp4);
        tcg_gen_sub_tl(cpu_T0, cpu_T0, cpu_T1);
        tcg_gen_sub_tl(cpu_T0, cpu_T0, cpu_tmp4);
        gen_op_st_rm_T0_A0(s1, ot, d);
        gen_op_update3_cc(cpu_tmp4);
        set_cc_op(s1, CC_OP_SBBB + ot);
        break;
    case OP_ADDL:
        tcg_gen_add_tl(cpu_T0, cpu_T0, cpu_T1);
        gen_op_st_rm_T0_A0(s1, ot, d);
        gen_op_update2_cc();
        set_cc_op(s1, CC_OP_ADDB + ot);
        break;
    case OP_SUBL:
        tcg_gen_mov_tl(cpu_cc_srcT, cpu_T0);
        tcg_gen_sub_tl(cpu_T0, cpu_T0, cpu_T1);
        gen_op_st_rm_T0_A0(s1, ot, d);
        gen_op_update2_cc();
        set_cc_op(s1, CC_OP_SUBB + ot);
        break;
    default:
    case OP_ANDL:
        tcg_gen_and_tl(cpu_T0, cpu_T0, cpu_T1);
        gen_op_st_rm_T0_A0(s1, ot, d);
        gen_op_update1_cc();
        set_cc_op(s1, CC_OP_LOGICB + ot);
        break;
    case OP_ORL:
        tcg_gen_or_tl(cpu_T0, cpu_T0, cpu_T1);
        gen_op_st_rm_T0_A0(s1, ot, d);
        gen_op_update1_cc();
        set_cc_op(s1, CC_OP_LOGICB + ot);
        break;
    case OP_XORL:
        tcg_gen_xor_tl(cpu_T0, cpu_T0, cpu_T1);
        gen_op_st_rm_T0_A0(s1, ot, d);
        gen_op_update1_cc();
        set_cc_op(s1, CC_OP_LOGICB + ot);
        break;
    case OP_CMPL:
        tcg_gen_mov_tl(cpu_cc_src, cpu_T1);
        tcg_gen_mov_tl(cpu_cc_srcT, cpu_T0);
        tcg_gen_sub_tl(cpu_cc_dst, cpu_T0, cpu_T1);
        set_cc_op(s1, CC_OP_SUBB + ot);
        break;
    }
}

/* if d == OR_TMP0, it means memory operand (address in A0) */
static void gen_inc(DisasContext *s1, TCGMemOp ot, int d, int c)
{
    if (d != OR_TMP0) {
        gen_op_mov_v_reg(ot, cpu_T0, d);
    } else {
        gen_op_ld_v(s1, ot, cpu_T0, cpu_A0);
    }
    gen_compute_eflags_c(s1, cpu_cc_src);
    if (c > 0) {
        tcg_gen_addi_tl(cpu_T0, cpu_T0, 1);
        set_cc_op(s1, CC_OP_INCB + ot);
    } else {
        tcg_gen_addi_tl(cpu_T0, cpu_T0, -1);
        set_cc_op(s1, CC_OP_DECB + ot);
    }
    gen_op_st_rm_T0_A0(s1, ot, d);
    tcg_gen_mov_tl(cpu_cc_dst, cpu_T0);
}

static void gen_shift_flags(DisasContext *s, TCGMemOp ot, TCGv result,
                            TCGv shm1, TCGv count, bool is_right)
{
    TCGv_i32 z32, s32, oldop;
    TCGv z_tl;

    /* Store the results into the CC variables.  If we know that the
       variable must be dead, store unconditionally.  Otherwise we'll
       need to not disrupt the current contents.  */
    z_tl = tcg_const_tl(0);
    if (cc_op_live[s->cc_op] & USES_CC_DST) {
        tcg_gen_movcond_tl(TCG_COND_NE, cpu_cc_dst, count, z_tl,
                           result, cpu_cc_dst);
    } else {
        tcg_gen_mov_tl(cpu_cc_dst, result);
    }
    if (cc_op_live[s->cc_op] & USES_CC_SRC) {
        tcg_gen_movcond_tl(TCG_COND_NE, cpu_cc_src, count, z_tl,
                           shm1, cpu_cc_src);
    } else {
        tcg_gen_mov_tl(cpu_cc_src, shm1);
    }
    tcg_temp_free(z_tl);

    /* Get the two potential CC_OP values into temporaries.  */
    tcg_gen_movi_i32(cpu_tmp2_i32, (is_right ? CC_OP_SARB : CC_OP_SHLB) + ot);
    if (s->cc_op == CC_OP_DYNAMIC) {
        oldop = cpu_cc_op;
    } else {
        tcg_gen_movi_i32(cpu_tmp3_i32, s->cc_op);
        oldop = cpu_tmp3_i32;
    }

    /* Conditionally store the CC_OP value.  */
    z32 = tcg_const_i32(0);
    s32 = tcg_temp_new_i32();
    tcg_gen_trunc_tl_i32(s32, count);
    tcg_gen_movcond_i32(TCG_COND_NE, cpu_cc_op, s32, z32, cpu_tmp2_i32, oldop);
    tcg_temp_free_i32(z32);
    tcg_temp_free_i32(s32);

    /* The CC_OP value is no longer predictable.  */
    set_cc_op(s, CC_OP_DYNAMIC);
}

static void gen_shift_rm_T1(DisasContext *s, TCGMemOp ot, int op1,
                            int is_right, int is_arith)
{
    target_ulong mask = (ot == MO_64 ? 0x3f : 0x1f);

    /* load */
    if (op1 == OR_TMP0) {
        gen_op_ld_v(s, ot, cpu_T0, cpu_A0);
    } else {
        gen_op_mov_v_reg(ot, cpu_T0, op1);
    }

    tcg_gen_andi_tl(cpu_T1, cpu_T1, mask);
    tcg_gen_subi_tl(cpu_tmp0, cpu_T1, 1);

    if (is_right) {
        if (is_arith) {
            gen_exts(ot, cpu_T0);
            tcg_gen_sar_tl(cpu_tmp0, cpu_T0, cpu_tmp0);
            tcg_gen_sar_tl(cpu_T0, cpu_T0, cpu_T1);
        } else {
            gen_extu(ot, cpu_T0);
            tcg_gen_shr_tl(cpu_tmp0, cpu_T0, cpu_tmp0);
            tcg_gen_shr_tl(cpu_T0, cpu_T0, cpu_T1);
        }
    } else {
        tcg_gen_shl_tl(cpu_tmp0, cpu_T0, cpu_tmp0);
        tcg_gen_shl_tl(cpu_T0, cpu_T0, cpu_T1);
    }

    /* store */
    gen_op_st_rm_T0_A0(s, ot, op1);

    gen_shift_flags(s, ot, cpu_T0, cpu_tmp0, cpu_T1, is_right);
}

static void gen_shift_rm_im(DisasContext *s, TCGMemOp ot, int op1, int op2,
                            int is_right, int is_arith)
{
    int mask = (ot == MO_64 ? 0x3f : 0x1f);

    /* load */
    if (op1 == OR_TMP0)
        gen_op_ld_v(s, ot, cpu_T0, cpu_A0);
    else
        gen_op_mov_v_reg(ot, cpu_T0, op1);

    op2 &= mask;
    if (op2 != 0) {
        if (is_right) {
            if (is_arith) {
                gen_exts(ot, cpu_T0);
                tcg_gen_sari_tl(cpu_tmp4, cpu_T0, op2 - 1);
                tcg_gen_sari_tl(cpu_T0, cpu_T0, op2);
            } else {
                gen_extu(ot, cpu_T0);
                tcg_gen_shri_tl(cpu_tmp4, cpu_T0, op2 - 1);
                tcg_gen_shri_tl(cpu_T0, cpu_T0, op2);
            }
        } else {
            tcg_gen_shli_tl(cpu_tmp4, cpu_T0, op2 - 1);
            tcg_gen_shli_tl(cpu_T0, cpu_T0, op2);
        }
    }

    /* store */
    gen_op_st_rm_T0_A0(s, ot, op1);

    /* update eflags if non zero shift */
    if (op2 != 0) {
        tcg_gen_mov_tl(cpu_cc_src, cpu_tmp4);
        tcg_gen_mov_tl(cpu_cc_dst, cpu_T0);
        set_cc_op(s, (is_right ? CC_OP_SARB : CC_OP_SHLB) + ot);
    }
}

static void gen_rot_rm_T1(DisasContext *s, TCGMemOp ot, int op1, int is_right)
{
    target_ulong mask = (ot == MO_64 ? 0x3f : 0x1f);
    TCGv_i32 t0, t1;

    /* load */
    if (op1 == OR_TMP0) {
        gen_op_ld_v(s, ot, cpu_T0, cpu_A0);
    } else {
        gen_op_mov_v_reg(ot, cpu_T0, op1);
    }

    tcg_gen_andi_tl(cpu_T1, cpu_T1, mask);

    switch (ot) {
    case MO_8:
        /* Replicate the 8-bit input so that a 32-bit rotate works.  */
        tcg_gen_ext8u_tl(cpu_T0, cpu_T0);
        tcg_gen_muli_tl(cpu_T0, cpu_T0, 0x01010101);
        goto do_long;
    case MO_16:
        /* Replicate the 16-bit input so that a 32-bit rotate works.  */
        tcg_gen_deposit_tl(cpu_T0, cpu_T0, cpu_T0, 16, 16);
        goto do_long;
    do_long:
#ifdef TARGET_X86_64
    case MO_32:
        tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T0);
        tcg_gen_trunc_tl_i32(cpu_tmp3_i32, cpu_T1);
        if (is_right) {
            tcg_gen_rotr_i32(cpu_tmp2_i32, cpu_tmp2_i32, cpu_tmp3_i32);
        } else {
            tcg_gen_rotl_i32(cpu_tmp2_i32, cpu_tmp2_i32, cpu_tmp3_i32);
        }
        tcg_gen_extu_i32_tl(cpu_T0, cpu_tmp2_i32);
        break;
#endif
    default:
        if (is_right) {
            tcg_gen_rotr_tl(cpu_T0, cpu_T0, cpu_T1);
        } else {
            tcg_gen_rotl_tl(cpu_T0, cpu_T0, cpu_T1);
        }
        break;
    }

    /* store */
    gen_op_st_rm_T0_A0(s, ot, op1);

    /* We'll need the flags computed into CC_SRC.  */
    gen_compute_eflags(s);

    /* The value that was "rotated out" is now present at the other end
       of the word.  Compute C into CC_DST and O into CC_SRC2.  Note that
       since we've computed the flags into CC_SRC, these variables are
       currently dead.  */
    if (is_right) {
        tcg_gen_shri_tl(cpu_cc_src2, cpu_T0, mask - 1);
        tcg_gen_shri_tl(cpu_cc_dst, cpu_T0, mask);
        tcg_gen_andi_tl(cpu_cc_dst, cpu_cc_dst, 1);
    } else {
        tcg_gen_shri_tl(cpu_cc_src2, cpu_T0, mask);
        tcg_gen_andi_tl(cpu_cc_dst, cpu_T0, 1);
    }
    tcg_gen_andi_tl(cpu_cc_src2, cpu_cc_src2, 1);
    tcg_gen_xor_tl(cpu_cc_src2, cpu_cc_src2, cpu_cc_dst);

    /* Now conditionally store the new CC_OP value.  If the shift count
       is 0 we keep the CC_OP_EFLAGS setting so that only CC_SRC is live.
       Otherwise reuse CC_OP_ADCOX which have the C and O flags split out
       exactly as we computed above.  */
    t0 = tcg_const_i32(0);
    t1 = tcg_temp_new_i32();
    tcg_gen_trunc_tl_i32(t1, cpu_T1);
    tcg_gen_movi_i32(cpu_tmp2_i32, CC_OP_ADCOX); 
    tcg_gen_movi_i32(cpu_tmp3_i32, CC_OP_EFLAGS);
    tcg_gen_movcond_i32(TCG_COND_NE, cpu_cc_op, t1, t0,
                        cpu_tmp2_i32, cpu_tmp3_i32);
    tcg_temp_free_i32(t0);
    tcg_temp_free_i32(t1);

    /* The CC_OP value is no longer predictable.  */ 
    set_cc_op(s, CC_OP_DYNAMIC);
}

static void gen_rot_rm_im(DisasContext *s, TCGMemOp ot, int op1, int op2,
                          int is_right)
{
    int mask = (ot == MO_64 ? 0x3f : 0x1f);
    int shift;

    /* load */
    if (op1 == OR_TMP0) {
        gen_op_ld_v(s, ot, cpu_T0, cpu_A0);
    } else {
        gen_op_mov_v_reg(ot, cpu_T0, op1);
    }

    op2 &= mask;
    if (op2 != 0) {
        switch (ot) {
#ifdef TARGET_X86_64
        case MO_32:
            tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T0);
            if (is_right) {
                tcg_gen_rotri_i32(cpu_tmp2_i32, cpu_tmp2_i32, op2);
            } else {
                tcg_gen_rotli_i32(cpu_tmp2_i32, cpu_tmp2_i32, op2);
            }
            tcg_gen_extu_i32_tl(cpu_T0, cpu_tmp2_i32);
            break;
#endif
        default:
            if (is_right) {
                tcg_gen_rotri_tl(cpu_T0, cpu_T0, op2);
            } else {
                tcg_gen_rotli_tl(cpu_T0, cpu_T0, op2);
            }
            break;
        case MO_8:
            mask = 7;
            goto do_shifts;
        case MO_16:
            mask = 15;
        do_shifts:
            shift = op2 & mask;
            if (is_right) {
                shift = mask + 1 - shift;
            }
            gen_extu(ot, cpu_T0);
            tcg_gen_shli_tl(cpu_tmp0, cpu_T0, shift);
            tcg_gen_shri_tl(cpu_T0, cpu_T0, mask + 1 - shift);
            tcg_gen_or_tl(cpu_T0, cpu_T0, cpu_tmp0);
            break;
        }
    }

    /* store */
    gen_op_st_rm_T0_A0(s, ot, op1);

    if (op2 != 0) {
        /* Compute the flags into CC_SRC.  */
        gen_compute_eflags(s);

        /* The value that was "rotated out" is now present at the other end
           of the word.  Compute C into CC_DST and O into CC_SRC2.  Note that
           since we've computed the flags into CC_SRC, these variables are
           currently dead.  */
        if (is_right) {
            tcg_gen_shri_tl(cpu_cc_src2, cpu_T0, mask - 1);
            tcg_gen_shri_tl(cpu_cc_dst, cpu_T0, mask);
            tcg_gen_andi_tl(cpu_cc_dst, cpu_cc_dst, 1);
        } else {
            tcg_gen_shri_tl(cpu_cc_src2, cpu_T0, mask);
            tcg_gen_andi_tl(cpu_cc_dst, cpu_T0, 1);
        }
        tcg_gen_andi_tl(cpu_cc_src2, cpu_cc_src2, 1);
        tcg_gen_xor_tl(cpu_cc_src2, cpu_cc_src2, cpu_cc_dst);
        set_cc_op(s, CC_OP_ADCOX);
    }
}

/* XXX: add faster immediate = 1 case */
static void gen_rotc_rm_T1(DisasContext *s, TCGMemOp ot, int op1,
                           int is_right)
{
    gen_compute_eflags(s);
    assert(s->cc_op == CC_OP_EFLAGS);

    /* load */
    if (op1 == OR_TMP0)
        gen_op_ld_v(s, ot, cpu_T0, cpu_A0);
    else
        gen_op_mov_v_reg(ot, cpu_T0, op1);
    
    if (is_right) {
        switch (ot) {
        case MO_8:
            gen_helper_rcrb(cpu_T0, cpu_env, cpu_T0, cpu_T1);
            break;
        case MO_16:
            gen_helper_rcrw(cpu_T0, cpu_env, cpu_T0, cpu_T1);
            break;
        case MO_32:
            gen_helper_rcrl(cpu_T0, cpu_env, cpu_T0, cpu_T1);
            break;
#ifdef TARGET_X86_64
        case MO_64:
            gen_helper_rcrq(cpu_T0, cpu_env, cpu_T0, cpu_T1);
            break;
#endif
        default:
            tcg_abort();
        }
    } else {
        switch (ot) {
        case MO_8:
            gen_helper_rclb(cpu_T0, cpu_env, cpu_T0, cpu_T1);
            break;
        case MO_16:
            gen_helper_rclw(cpu_T0, cpu_env, cpu_T0, cpu_T1);
            break;
        case MO_32:
            gen_helper_rcll(cpu_T0, cpu_env, cpu_T0, cpu_T1);
            break;
#ifdef TARGET_X86_64
        case MO_64:
            gen_helper_rclq(cpu_T0, cpu_env, cpu_T0, cpu_T1);
            break;
#endif
        default:
            tcg_abort();
        }
    }
    /* store */
    gen_op_st_rm_T0_A0(s, ot, op1);
}

/* XXX: add faster immediate case */
static void gen_shiftd_rm_T1(DisasContext *s, TCGMemOp ot, int op1,
                             bool is_right, TCGv count_in)
{
    target_ulong mask = (ot == MO_64 ? 63 : 31);
    TCGv count;

    /* load */
    if (op1 == OR_TMP0) {
        gen_op_ld_v(s, ot, cpu_T0, cpu_A0);
    } else {
        gen_op_mov_v_reg(ot, cpu_T0, op1);
    }

    count = tcg_temp_new();
    tcg_gen_andi_tl(count, count_in, mask);

    switch (ot) {
    case MO_16:
        /* Note: we implement the Intel behaviour for shift count > 16.
           This means "shrdw C, B, A" shifts A:B:A >> C.  Build the B:A
           portion by constructing it as a 32-bit value.  */
        if (is_right) {
            tcg_gen_deposit_tl(cpu_tmp0, cpu_T0, cpu_T1, 16, 16);
            tcg_gen_mov_tl(cpu_T1, cpu_T0);
            tcg_gen_mov_tl(cpu_T0, cpu_tmp0);
        } else {
            tcg_gen_deposit_tl(cpu_T1, cpu_T0, cpu_T1, 16, 16);
        }
        /* FALLTHRU */
#ifdef TARGET_X86_64
    case MO_32:
        /* Concatenate the two 32-bit values and use a 64-bit shift.  */
        tcg_gen_subi_tl(cpu_tmp0, count, 1);
        if (is_right) {
            tcg_gen_concat_tl_i64(cpu_T0, cpu_T0, cpu_T1);
            tcg_gen_shr_i64(cpu_tmp0, cpu_T0, cpu_tmp0);
            tcg_gen_shr_i64(cpu_T0, cpu_T0, count);
        } else {
            tcg_gen_concat_tl_i64(cpu_T0, cpu_T1, cpu_T0);
            tcg_gen_shl_i64(cpu_tmp0, cpu_T0, cpu_tmp0);
            tcg_gen_shl_i64(cpu_T0, cpu_T0, count);
            tcg_gen_shri_i64(cpu_tmp0, cpu_tmp0, 32);
            tcg_gen_shri_i64(cpu_T0, cpu_T0, 32);
        }
        break;
#endif
    default:
        tcg_gen_subi_tl(cpu_tmp0, count, 1);
        if (is_right) {
            tcg_gen_shr_tl(cpu_tmp0, cpu_T0, cpu_tmp0);

            tcg_gen_subfi_tl(cpu_tmp4, mask + 1, count);
            tcg_gen_shr_tl(cpu_T0, cpu_T0, count);
            tcg_gen_shl_tl(cpu_T1, cpu_T1, cpu_tmp4);
        } else {
            tcg_gen_shl_tl(cpu_tmp0, cpu_T0, cpu_tmp0);
            if (ot == MO_16) {
                /* Only needed if count > 16, for Intel behaviour.  */
                tcg_gen_subfi_tl(cpu_tmp4, 33, count);
                tcg_gen_shr_tl(cpu_tmp4, cpu_T1, cpu_tmp4);
                tcg_gen_or_tl(cpu_tmp0, cpu_tmp0, cpu_tmp4);
            }

            tcg_gen_subfi_tl(cpu_tmp4, mask + 1, count);
            tcg_gen_shl_tl(cpu_T0, cpu_T0, count);
            tcg_gen_shr_tl(cpu_T1, cpu_T1, cpu_tmp4);
        }
        tcg_gen_movi_tl(cpu_tmp4, 0);
        tcg_gen_movcond_tl(TCG_COND_EQ, cpu_T1, count, cpu_tmp4,
                           cpu_tmp4, cpu_T1);
        tcg_gen_or_tl(cpu_T0, cpu_T0, cpu_T1);
        break;
    }

    /* store */
    gen_op_st_rm_T0_A0(s, ot, op1);

    gen_shift_flags(s, ot, cpu_T0, cpu_tmp0, count, is_right);
    tcg_temp_free(count);
}

static void gen_shift(DisasContext *s1, int op, TCGMemOp ot, int d, int s)
{
    if (s != OR_TMP1)
        gen_op_mov_v_reg(ot, cpu_T1, s);
    switch(op) {
    case OP_ROL:
        gen_rot_rm_T1(s1, ot, d, 0);
        break;
    case OP_ROR:
        gen_rot_rm_T1(s1, ot, d, 1);
        break;
    case OP_SHL:
    case OP_SHL1:
        gen_shift_rm_T1(s1, ot, d, 0, 0);
        break;
    case OP_SHR:
        gen_shift_rm_T1(s1, ot, d, 1, 0);
        break;
    case OP_SAR:
        gen_shift_rm_T1(s1, ot, d, 1, 1);
        break;
    case OP_RCL:
        gen_rotc_rm_T1(s1, ot, d, 0);
        break;
    case OP_RCR:
        gen_rotc_rm_T1(s1, ot, d, 1);
        break;
    }
}

static void gen_shifti(DisasContext *s1, int op, TCGMemOp ot, int d, int c)
{
    switch(op) {
    case OP_ROL:
        gen_rot_rm_im(s1, ot, d, c, 0);
        break;
    case OP_ROR:
        gen_rot_rm_im(s1, ot, d, c, 1);
        break;
    case OP_SHL:
    case OP_SHL1:
        gen_shift_rm_im(s1, ot, d, c, 0, 0);
        break;
    case OP_SHR:
        gen_shift_rm_im(s1, ot, d, c, 1, 0);
        break;
    case OP_SAR:
        gen_shift_rm_im(s1, ot, d, c, 1, 1);
        break;
    default:
        /* currently not optimized */
        tcg_gen_movi_tl(cpu_T1, c);
        gen_shift(s1, op, ot, d, OR_TMP1);
        break;
    }
}

/* Decompose an address.  */

typedef struct AddressParts {
    int def_seg;
    int base;
    int index;
    int scale;
    target_long disp;
} AddressParts;

static AddressParts gen_lea_modrm_0(CPUX86State *env, DisasContext *s,
                                    int modrm)
{
    int def_seg, base, index, scale, mod, rm;
    target_long disp;
    bool havesib;

    def_seg = R_DS;
    index = -1;
    scale = 0;
    disp = 0;

    mod = (modrm >> 6) & 3;
    rm = modrm & 7;
    base = rm | REX_B(s);

    if (mod == 3) {
        /* Normally filtered out earlier, but including this path
           simplifies multi-byte nop, as well as bndcl, bndcu, bndcn.  */
        goto done;
    }

    switch (s->aflag) {
    case MO_64:
    case MO_32:
        havesib = 0;
        if (rm == 4) {
            int code = cpu_ldub_code(env, s->pc++);
            scale = (code >> 6) & 3;
            index = ((code >> 3) & 7) | REX_X(s);
            if (index == 4) {
                index = -1;  /* no index */
            }
            base = (code & 7) | REX_B(s);
            havesib = 1;
        }

        switch (mod) {
        case 0:
            if ((base & 7) == 5) {
                base = -1;
                disp = (int32_t)cpu_ldl_code(env, s->pc);
                s->pc += 4;
                if (CODE64(s) && !havesib) {
                    base = -2;
                    disp += s->pc + s->rip_offset;
                }
            }
            break;
        case 1:
            disp = (int8_t)cpu_ldub_code(env, s->pc++);
            break;
        default:
        case 2:
            disp = (int32_t)cpu_ldl_code(env, s->pc);
            s->pc += 4;
            break;
        }

        /* For correct popl handling with esp.  */
        if (base == R_ESP && s->popl_esp_hack) {
            disp += s->popl_esp_hack;
        }
        if (base == R_EBP || base == R_ESP) {
            def_seg = R_SS;
        }
        break;

    case MO_16:
        if (mod == 0) {
            if (rm == 6) {
                base = -1;
                disp = cpu_lduw_code(env, s->pc);
                s->pc += 2;
                break;
            }
        } else if (mod == 1) {
            disp = (int8_t)cpu_ldub_code(env, s->pc++);
        } else {
            disp = (int16_t)cpu_lduw_code(env, s->pc);
            s->pc += 2;
        }

        switch (rm) {
        case 0:
            base = R_EBX;
            index = R_ESI;
            break;
        case 1:
            base = R_EBX;
            index = R_EDI;
            break;
        case 2:
            base = R_EBP;
            index = R_ESI;
            def_seg = R_SS;
            break;
        case 3:
            base = R_EBP;
            index = R_EDI;
            def_seg = R_SS;
            break;
        case 4:
            base = R_ESI;
            break;
        case 5:
            base = R_EDI;
            break;
        case 6:
            base = R_EBP;
            def_seg = R_SS;
            break;
        default:
        case 7:
            base = R_EBX;
            break;
        }
        break;

    default:
        tcg_abort();
    }

 done:
    return (AddressParts){ def_seg, base, index, scale, disp };
}

/* Compute the address, with a minimum number of TCG ops.  */
static TCGv gen_lea_modrm_1(AddressParts a)
{
    TCGv ea;

    TCGV_UNUSED(ea);
    if (a.index >= 0) {
        if (a.scale == 0) {
            ea = cpu_regs[a.index];
        } else {
            tcg_gen_shli_tl(cpu_A0, cpu_regs[a.index], a.scale);
            ea = cpu_A0;
        }
        if (a.base >= 0) {
            tcg_gen_add_tl(cpu_A0, ea, cpu_regs[a.base]);
            ea = cpu_A0;
        }
    } else if (a.base >= 0) {
        ea = cpu_regs[a.base];
    }
    if (TCGV_IS_UNUSED(ea)) {
        tcg_gen_movi_tl(cpu_A0, a.disp);
        ea = cpu_A0;
    } else if (a.disp != 0) {
        tcg_gen_addi_tl(cpu_A0, ea, a.disp);
        ea = cpu_A0;
    }

    return ea;
}

static void gen_lea_modrm(CPUX86State *env, DisasContext *s, int modrm)
{
    AddressParts a = gen_lea_modrm_0(env, s, modrm);
    TCGv ea = gen_lea_modrm_1(a);
    gen_lea_v_seg(s, s->aflag, ea, a.def_seg, s->override);
}

static void gen_nop_modrm(CPUX86State *env, DisasContext *s, int modrm)
{
    (void)gen_lea_modrm_0(env, s, modrm);
}

/* Used for BNDCL, BNDCU, BNDCN.  */

static void gen_bndck(CPUX86State *env, DisasContext *s, int modrm,
                      TCGCond cond, TCGv_i64 bndv)
{
    TCGv ea = gen_lea_modrm_1(gen_lea_modrm_0(env, s, modrm));

    tcg_gen_extu_tl_i64(cpu_tmp1_i64, ea);
    if (!CODE64(s)) {
        tcg_gen_ext32u_i64(cpu_tmp1_i64, cpu_tmp1_i64);
    }
    tcg_gen_setcond_i64(cond, cpu_tmp1_i64, cpu_tmp1_i64, bndv);
    tcg_gen_extrl_i64_i32(cpu_tmp2_i32, cpu_tmp1_i64);
    gen_helper_bndck(cpu_env, cpu_tmp2_i32);
}

/* used for LEA and MOV AX, mem */
static void gen_add_A0_ds_seg(DisasContext *s)
{
    gen_lea_v_seg(s, s->aflag, cpu_A0, R_DS, s->override);
}

/* generate modrm memory load or store of 'reg'. TMP0 is used if reg ==
   OR_TMP0 */
static void gen_ldst_modrm(CPUX86State *env, DisasContext *s, int modrm,
                           TCGMemOp ot, int reg, int is_store)
{
    int mod, rm;

    mod = (modrm >> 6) & 3;
    rm = (modrm & 7) | REX_B(s);
    if (mod == 3) {
        if (is_store) {
            if (reg != OR_TMP0)
                gen_op_mov_v_reg(ot, cpu_T0, reg);
            gen_op_mov_reg_v(ot, rm, cpu_T0);
        } else {
            gen_op_mov_v_reg(ot, cpu_T0, rm);
            if (reg != OR_TMP0)
                gen_op_mov_reg_v(ot, reg, cpu_T0);
        }
    } else {
        gen_lea_modrm(env, s, modrm);
        if (is_store) {
            if (reg != OR_TMP0)
                gen_op_mov_v_reg(ot, cpu_T0, reg);
            gen_op_st_v(s, ot, cpu_T0, cpu_A0);
        } else {
            gen_op_ld_v(s, ot, cpu_T0, cpu_A0);
            if (reg != OR_TMP0)
                gen_op_mov_reg_v(ot, reg, cpu_T0);
        }
    }
}

static inline uint32_t insn_get(CPUX86State *env, DisasContext *s, TCGMemOp ot)
{
    uint32_t ret;

    switch (ot) {
    case MO_8:
        ret = cpu_ldub_code(env, s->pc);
        s->pc++;
        break;
    case MO_16:
        ret = cpu_lduw_code(env, s->pc);
        s->pc += 2;
        break;
    case MO_32:
#ifdef TARGET_X86_64
    case MO_64:
#endif
        ret = cpu_ldl_code(env, s->pc);
        s->pc += 4;
        break;
    default:
        tcg_abort();
    }
    return ret;
}

static inline int insn_const_size(TCGMemOp ot)
{
    if (ot <= MO_32) {
        return 1 << ot;
    } else {
        return 4;
    }
}

static inline bool use_goto_tb(DisasContext *s, target_ulong pc)
{
#ifndef CONFIG_USER_ONLY
    return (pc & TARGET_PAGE_MASK) == (s->tb->pc & TARGET_PAGE_MASK) ||
           (pc & TARGET_PAGE_MASK) == (s->pc_start & TARGET_PAGE_MASK);
#else
    return true;
#endif
}

static inline void gen_goto_tb(DisasContext *s, int tb_num, target_ulong eip)
{
    target_ulong pc = s->cs_base + eip;

    if (use_goto_tb(s, pc))  {
        /* jump to same page: we can use a direct jump */
        tcg_gen_goto_tb(tb_num);
        gen_jmp_im(eip);
        tcg_gen_exit_tb((uintptr_t)s->tb + tb_num);
    } else {
        /* jump to another page: currently not optimized */
        gen_jmp_im(eip);
        gen_eob(s);
    }
}

static inline void gen_jcc(DisasContext *s, int b,
                           target_ulong val, target_ulong next_eip)
{
    TCGLabel *l1, *l2;

    if (s->jmp_opt) {
        l1 = gen_new_label();
        gen_jcc1(s, b, l1);

        gen_goto_tb(s, 0, next_eip);

        gen_set_label(l1);
        gen_goto_tb(s, 1, val);
        s->is_jmp = DISAS_TB_JUMP;
    } else {
        l1 = gen_new_label();
        l2 = gen_new_label();
        gen_jcc1(s, b, l1);

        gen_jmp_im(next_eip);
        tcg_gen_br(l2);

        gen_set_label(l1);
        gen_jmp_im(val);
        gen_set_label(l2);
        gen_eob(s);
    }
}

static void gen_cmovcc1(CPUX86State *env, DisasContext *s, TCGMemOp ot, int b,
                        int modrm, int reg)
{
    CCPrepare cc;

    gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);

    cc = gen_prepare_cc(s, b, cpu_T1);
    if (cc.mask != -1) {
        TCGv t0 = tcg_temp_new();
        tcg_gen_andi_tl(t0, cc.reg, cc.mask);
        cc.reg = t0;
    }
    if (!cc.use_reg2) {
        cc.reg2 = tcg_const_tl(cc.imm);
    }

    tcg_gen_movcond_tl(cc.cond, cpu_T0, cc.reg, cc.reg2,
                       cpu_T0, cpu_regs[reg]);
    gen_op_mov_reg_v(ot, reg, cpu_T0);

    if (cc.mask != -1) {
        tcg_temp_free(cc.reg);
    }
    if (!cc.use_reg2) {
        tcg_temp_free(cc.reg2);
    }
}

static inline void gen_op_movl_T0_seg(int seg_reg)
{
    tcg_gen_ld32u_tl(cpu_T0, cpu_env,
                     offsetof(CPUX86State,segs[seg_reg].selector));
}

static inline void gen_op_movl_seg_T0_vm(int seg_reg)
{
    tcg_gen_ext16u_tl(cpu_T0, cpu_T0);
    tcg_gen_st32_tl(cpu_T0, cpu_env,
                    offsetof(CPUX86State,segs[seg_reg].selector));
    tcg_gen_shli_tl(cpu_seg_base[seg_reg], cpu_T0, 4);
}

/* move T0 to seg_reg and compute if the CPU state may change. Never
   call this function with seg_reg == R_CS */
static void gen_movl_seg_T0(DisasContext *s, int seg_reg)
{
    if (s->pe && !s->vm86) {
        tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T0);
        gen_helper_load_seg(cpu_env, tcg_const_i32(seg_reg), cpu_tmp2_i32);
        /* abort translation because the addseg value may change or
           because ss32 may change. For R_SS, translation must always
           stop as a special handling must be done to disable hardware
           interrupts for the next instruction */
        if (seg_reg == R_SS || (s->code32 && seg_reg < R_FS))
            s->is_jmp = DISAS_TB_JUMP;
    } else {
        gen_op_movl_seg_T0_vm(seg_reg);
        if (seg_reg == R_SS)
            s->is_jmp = DISAS_TB_JUMP;
    }
}

static inline int svm_is_rep(int prefixes)
{
    return ((prefixes & (PREFIX_REPZ | PREFIX_REPNZ)) ? 8 : 0);
}

static inline void
gen_svm_check_intercept_param(DisasContext *s, target_ulong pc_start,
                              uint32_t type, uint64_t param)
{
    /* no SVM activated; fast case */
    if (likely(!(s->flags & HF_SVMI_MASK)))
        return;
    gen_update_cc_op(s);
    gen_jmp_im(pc_start - s->cs_base);
    gen_helper_svm_check_intercept_param(cpu_env, tcg_const_i32(type),
                                         tcg_const_i64(param));
}

static inline void
gen_svm_check_intercept(DisasContext *s, target_ulong pc_start, uint64_t type)
{
    gen_svm_check_intercept_param(s, pc_start, type, 0);
}

static inline void gen_stack_update(DisasContext *s, int addend)
{
    gen_op_add_reg_im(mo_stacksize(s), R_ESP, addend);
}

/* Generate a push. It depends on ss32, addseg and dflag.  */
static void gen_push_v(DisasContext *s, TCGv val)
{
    TCGMemOp d_ot = mo_pushpop(s, s->dflag);
    TCGMemOp a_ot = mo_stacksize(s);
    int size = 1 << d_ot;
    TCGv new_esp = cpu_A0;

    tcg_gen_subi_tl(cpu_A0, cpu_regs[R_ESP], size);

    if (!CODE64(s)) {
        if (s->addseg) {
            new_esp = cpu_tmp4;
            tcg_gen_mov_tl(new_esp, cpu_A0);
        }
        gen_lea_v_seg(s, a_ot, cpu_A0, R_SS, -1);
    }

    gen_op_st_v(s, d_ot, val, cpu_A0);
    gen_op_mov_reg_v(a_ot, R_ESP, new_esp);
}

/* two step pop is necessary for precise exceptions */
static TCGMemOp gen_pop_T0(DisasContext *s)
{
    TCGMemOp d_ot = mo_pushpop(s, s->dflag);

    gen_lea_v_seg(s, mo_stacksize(s), cpu_regs[R_ESP], R_SS, -1);
    gen_op_ld_v(s, d_ot, cpu_T0, cpu_A0);

    return d_ot;
}

static inline void gen_pop_update(DisasContext *s, TCGMemOp ot)
{
    gen_stack_update(s, 1 << ot);
}

static inline void gen_stack_A0(DisasContext *s)
{
    gen_lea_v_seg(s, s->ss32 ? MO_32 : MO_16, cpu_regs[R_ESP], R_SS, -1);
}

static void gen_pusha(DisasContext *s)
{
    TCGMemOp s_ot = s->ss32 ? MO_32 : MO_16;
    TCGMemOp d_ot = s->dflag;
    int size = 1 << d_ot;
    int i;

    for (i = 0; i < 8; i++) {
        tcg_gen_addi_tl(cpu_A0, cpu_regs[R_ESP], (i - 8) * size);
        gen_lea_v_seg(s, s_ot, cpu_A0, R_SS, -1);
        gen_op_st_v(s, d_ot, cpu_regs[7 - i], cpu_A0);
    }

    gen_stack_update(s, -8 * size);
}

static void gen_popa(DisasContext *s)
{
    TCGMemOp s_ot = s->ss32 ? MO_32 : MO_16;
    TCGMemOp d_ot = s->dflag;
    int size = 1 << d_ot;
    int i;

    for (i = 0; i < 8; i++) {
        /* ESP is not reloaded */
        if (7 - i == R_ESP) {
            continue;
        }
        tcg_gen_addi_tl(cpu_A0, cpu_regs[R_ESP], i * size);
        gen_lea_v_seg(s, s_ot, cpu_A0, R_SS, -1);
        gen_op_ld_v(s, d_ot, cpu_T0, cpu_A0);
        gen_op_mov_reg_v(d_ot, 7 - i, cpu_T0);
    }

    gen_stack_update(s, 8 * size);
}

static void gen_enter(DisasContext *s, int esp_addend, int level)
{
    TCGMemOp d_ot = mo_pushpop(s, s->dflag);
    TCGMemOp a_ot = CODE64(s) ? MO_64 : s->ss32 ? MO_32 : MO_16;
    int size = 1 << d_ot;

    /* Push BP; compute FrameTemp into T1.  */
    tcg_gen_subi_tl(cpu_T1, cpu_regs[R_ESP], size);
    gen_lea_v_seg(s, a_ot, cpu_T1, R_SS, -1);
    gen_op_st_v(s, d_ot, cpu_regs[R_EBP], cpu_A0);

    level &= 31;
    if (level != 0) {
        int i;

        /* Copy level-1 pointers from the previous frame.  */
        for (i = 1; i < level; ++i) {
            tcg_gen_subi_tl(cpu_A0, cpu_regs[R_EBP], size * i);
            gen_lea_v_seg(s, a_ot, cpu_A0, R_SS, -1);
            gen_op_ld_v(s, d_ot, cpu_tmp0, cpu_A0);

            tcg_gen_subi_tl(cpu_A0, cpu_T1, size * i);
            gen_lea_v_seg(s, a_ot, cpu_A0, R_SS, -1);
            gen_op_st_v(s, d_ot, cpu_tmp0, cpu_A0);
        }

        /* Push the current FrameTemp as the last level.  */
        tcg_gen_subi_tl(cpu_A0, cpu_T1, size * level);
        gen_lea_v_seg(s, a_ot, cpu_A0, R_SS, -1);
        gen_op_st_v(s, d_ot, cpu_T1, cpu_A0);
    }

    /* Copy the FrameTemp value to EBP.  */
    gen_op_mov_reg_v(a_ot, R_EBP, cpu_T1);

    /* Compute the final value of ESP.  */
    tcg_gen_subi_tl(cpu_T1, cpu_T1, esp_addend + size * level);
    gen_op_mov_reg_v(a_ot, R_ESP, cpu_T1);
}

static void gen_leave(DisasContext *s)
{
    TCGMemOp d_ot = mo_pushpop(s, s->dflag);
    TCGMemOp a_ot = mo_stacksize(s);

    gen_lea_v_seg(s, a_ot, cpu_regs[R_EBP], R_SS, -1);
    gen_op_ld_v(s, d_ot, cpu_T0, cpu_A0);

    tcg_gen_addi_tl(cpu_T1, cpu_regs[R_EBP], 1 << d_ot);

    gen_op_mov_reg_v(d_ot, R_EBP, cpu_T0);
    gen_op_mov_reg_v(a_ot, R_ESP, cpu_T1);
}

static void gen_exception(DisasContext *s, int trapno, target_ulong cur_eip)
{
    gen_update_cc_op(s);
    gen_jmp_im(cur_eip);
    gen_helper_raise_exception(cpu_env, tcg_const_i32(trapno));
    s->is_jmp = DISAS_TB_JUMP;
}

/* Generate #UD for the current instruction.  The assumption here is that
   the instruction is known, but it isn't allowed in the current cpu mode.  */
static void gen_illegal_opcode(DisasContext *s)
{
    gen_exception(s, EXCP06_ILLOP, s->pc_start - s->cs_base);
}

/* Similarly, except that the assumption here is that we don't decode
   the instruction at all -- either a missing opcode, an unimplemented
   feature, or just a bogus instruction stream.  */
static void gen_unknown_opcode(CPUX86State *env, DisasContext *s)
{
    gen_illegal_opcode(s);

    if (qemu_loglevel_mask(LOG_UNIMP)) {
        target_ulong pc = s->pc_start, end = s->pc;
        qemu_log("ILLOPC: " TARGET_FMT_lx ":", pc);
        for (; pc < end; ++pc) {
            qemu_log(" %02x", cpu_ldub_code(env, pc));
        }
        qemu_log("\n");
    }
}

/* an interrupt is different from an exception because of the
   privilege checks */
static void gen_interrupt(DisasContext *s, int intno,
                          target_ulong cur_eip, target_ulong next_eip)
{
    gen_update_cc_op(s);
    gen_jmp_im(cur_eip);
    gen_helper_raise_interrupt(cpu_env, tcg_const_i32(intno),
                               tcg_const_i32(next_eip - cur_eip));
    s->is_jmp = DISAS_TB_JUMP;
}

static void gen_debug(DisasContext *s, target_ulong cur_eip)
{
    gen_update_cc_op(s);
    gen_jmp_im(cur_eip);
    gen_helper_debug(cpu_env);
    s->is_jmp = DISAS_TB_JUMP;
}

static void gen_set_hflag(DisasContext *s, uint32_t mask)
{
    if ((s->flags & mask) == 0) {
        TCGv_i32 t = tcg_temp_new_i32();
        tcg_gen_ld_i32(t, cpu_env, offsetof(CPUX86State, hflags));
        tcg_gen_ori_i32(t, t, mask);
        tcg_gen_st_i32(t, cpu_env, offsetof(CPUX86State, hflags));
        tcg_temp_free_i32(t);
        s->flags |= mask;
    }
}

static void gen_reset_hflag(DisasContext *s, uint32_t mask)
{
    if (s->flags & mask) {
        TCGv_i32 t = tcg_temp_new_i32();
        tcg_gen_ld_i32(t, cpu_env, offsetof(CPUX86State, hflags));
        tcg_gen_andi_i32(t, t, ~mask);
        tcg_gen_st_i32(t, cpu_env, offsetof(CPUX86State, hflags));
        tcg_temp_free_i32(t);
        s->flags &= ~mask;
    }
}

/* Clear BND registers during legacy branches.  */
static void gen_bnd_jmp(DisasContext *s)
{
    /* Clear the registers only if BND prefix is missing, MPX is enabled,
       and if the BNDREGs are known to be in use (non-zero) already.
       The helper itself will check BNDPRESERVE at runtime.  */
    if ((s->prefix & PREFIX_REPNZ) == 0
        && (s->flags & HF_MPX_EN_MASK) != 0
        && (s->flags & HF_MPX_IU_MASK) != 0) {
        gen_helper_bnd_jmp(cpu_env);
    }
}

/* Generate an end of block. Trace exception is also generated if needed.
   If IIM, set HF_INHIBIT_IRQ_MASK if it isn't already set.  */
static void gen_eob_inhibit_irq(DisasContext *s, bool inhibit)
{
    gen_update_cc_op(s);

    /* If several instructions disable interrupts, only the first does it.  */
    if (inhibit && !(s->flags & HF_INHIBIT_IRQ_MASK)) {
        gen_set_hflag(s, HF_INHIBIT_IRQ_MASK);
    } else {
        gen_reset_hflag(s, HF_INHIBIT_IRQ_MASK);
    }

    if (s->tb->flags & HF_RF_MASK) {
        gen_helper_reset_rf(cpu_env);
    }
    if (s->singlestep_enabled) {
        gen_helper_debug(cpu_env);
    } else if (s->tf) {
        gen_helper_single_step(cpu_env);
    } else {
        tcg_gen_exit_tb(0);
    }
    s->is_jmp = DISAS_TB_JUMP;
}

/* End of block, resetting the inhibit irq flag.  */
static void gen_eob(DisasContext *s)
{
    gen_eob_inhibit_irq(s, false);
}

/* generate a jump to eip. No segment change must happen before as a
   direct call to the next block may occur */
static void gen_jmp_tb(DisasContext *s, target_ulong eip, int tb_num)
{
    gen_update_cc_op(s);
    set_cc_op(s, CC_OP_DYNAMIC);
    if (s->jmp_opt) {
        gen_goto_tb(s, tb_num, eip);
        s->is_jmp = DISAS_TB_JUMP;
    } else {
        gen_jmp_im(eip);
        gen_eob(s);
    }
}

static void gen_jmp(DisasContext *s, target_ulong eip)
{
    gen_jmp_tb(s, eip, 0);
}

static inline void gen_ldq_env_A0(DisasContext *s, int offset)
{
    tcg_gen_qemu_ld_i64(cpu_tmp1_i64, cpu_A0, s->mem_index, MO_LEQ);
    tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, offset);
}

static inline void gen_stq_env_A0(DisasContext *s, int offset)
{
    tcg_gen_ld_i64(cpu_tmp1_i64, cpu_env, offset);
    tcg_gen_qemu_st_i64(cpu_tmp1_i64, cpu_A0, s->mem_index, MO_LEQ);
}

static inline void gen_ldo_env_A0(DisasContext *s, int offset)
{
    int mem_index = s->mem_index;
    tcg_gen_qemu_ld_i64(cpu_tmp1_i64, cpu_A0, mem_index, MO_LEQ);
    tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, offset + offsetof(ZMMReg, ZMM_Q(0)));
    tcg_gen_addi_tl(cpu_tmp0, cpu_A0, 8);
    tcg_gen_qemu_ld_i64(cpu_tmp1_i64, cpu_tmp0, mem_index, MO_LEQ);
    tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, offset + offsetof(ZMMReg, ZMM_Q(1)));
}

static inline void gen_sto_env_A0(DisasContext *s, int offset)
{
    int mem_index = s->mem_index;
    tcg_gen_ld_i64(cpu_tmp1_i64, cpu_env, offset + offsetof(ZMMReg, ZMM_Q(0)));
    tcg_gen_qemu_st_i64(cpu_tmp1_i64, cpu_A0, mem_index, MO_LEQ);
    tcg_gen_addi_tl(cpu_tmp0, cpu_A0, 8);
    tcg_gen_ld_i64(cpu_tmp1_i64, cpu_env, offset + offsetof(ZMMReg, ZMM_Q(1)));
    tcg_gen_qemu_st_i64(cpu_tmp1_i64, cpu_tmp0, mem_index, MO_LEQ);
}

static inline void gen_op_movo(int d_offset, int s_offset)
{
    tcg_gen_ld_i64(cpu_tmp1_i64, cpu_env, s_offset + offsetof(ZMMReg, ZMM_Q(0)));
    tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, d_offset + offsetof(ZMMReg, ZMM_Q(0)));
    tcg_gen_ld_i64(cpu_tmp1_i64, cpu_env, s_offset + offsetof(ZMMReg, ZMM_Q(1)));
    tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, d_offset + offsetof(ZMMReg, ZMM_Q(1)));
}

static inline void gen_op_movq(int d_offset, int s_offset)
{
    tcg_gen_ld_i64(cpu_tmp1_i64, cpu_env, s_offset);
    tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, d_offset);
}

static inline void gen_op_movl(int d_offset, int s_offset)
{
    tcg_gen_ld_i32(cpu_tmp2_i32, cpu_env, s_offset);
    tcg_gen_st_i32(cpu_tmp2_i32, cpu_env, d_offset);
}

static inline void gen_op_movq_env_0(int d_offset)
{
    tcg_gen_movi_i64(cpu_tmp1_i64, 0);
    tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, d_offset);
}

typedef void (*SSEFunc_i_ep)(TCGv_i32 val, TCGv_ptr env, TCGv_ptr reg);
typedef void (*SSEFunc_l_ep)(TCGv_i64 val, TCGv_ptr env, TCGv_ptr reg);
typedef void (*SSEFunc_0_epi)(TCGv_ptr env, TCGv_ptr reg, TCGv_i32 val);
typedef void (*SSEFunc_0_epl)(TCGv_ptr env, TCGv_ptr reg, TCGv_i64 val);
typedef void (*SSEFunc_0_epp)(TCGv_ptr env, TCGv_ptr reg_a, TCGv_ptr reg_b);
typedef void (*SSEFunc_0_eppi)(TCGv_ptr env, TCGv_ptr reg_a, TCGv_ptr reg_b,
                               TCGv_i32 val);
typedef void (*SSEFunc_0_ppi)(TCGv_ptr reg_a, TCGv_ptr reg_b, TCGv_i32 val);
typedef void (*SSEFunc_0_eppt)(TCGv_ptr env, TCGv_ptr reg_a, TCGv_ptr reg_b,
                               TCGv val);

#define SSE_SPECIAL ((void *)1)
#define SSE_DUMMY ((void *)2)

#define MMX_OP2(x) { gen_helper_ ## x ## _mmx, gen_helper_ ## x ## _xmm }
#define SSE_FOP(x) { gen_helper_ ## x ## ps, gen_helper_ ## x ## pd, \
                     gen_helper_ ## x ## ss, gen_helper_ ## x ## sd, }

static const SSEFunc_0_epp sse_op_table1[256][4] = {
    /* 3DNow! extensions */
    [0x0e] = { SSE_DUMMY }, /* femms */
    [0x0f] = { SSE_DUMMY }, /* pf... */
    /* pure SSE operations */
    [0x10] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movups, movupd, movss, movsd */
    [0x11] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movups, movupd, movss, movsd */
    [0x12] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movlps, movlpd, movsldup, movddup */
    [0x13] = { SSE_SPECIAL, SSE_SPECIAL },  /* movlps, movlpd */
    [0x14] = { gen_helper_punpckldq_xmm, gen_helper_punpcklqdq_xmm },
    [0x15] = { gen_helper_punpckhdq_xmm, gen_helper_punpckhqdq_xmm },
    [0x16] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL },  /* movhps, movhpd, movshdup */
    [0x17] = { SSE_SPECIAL, SSE_SPECIAL },  /* movhps, movhpd */

    [0x28] = { SSE_SPECIAL, SSE_SPECIAL },  /* movaps, movapd */
    [0x29] = { SSE_SPECIAL, SSE_SPECIAL },  /* movaps, movapd */
    [0x2a] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* cvtpi2ps, cvtpi2pd, cvtsi2ss, cvtsi2sd */
    [0x2b] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movntps, movntpd, movntss, movntsd */
    [0x2c] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* cvttps2pi, cvttpd2pi, cvttsd2si, cvttss2si */
    [0x2d] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* cvtps2pi, cvtpd2pi, cvtsd2si, cvtss2si */
    [0x2e] = { gen_helper_ucomiss, gen_helper_ucomisd },
    [0x2f] = { gen_helper_comiss, gen_helper_comisd },
    [0x50] = { SSE_SPECIAL, SSE_SPECIAL }, /* movmskps, movmskpd */
    [0x51] = SSE_FOP(sqrt),
    [0x52] = { gen_helper_rsqrtps, NULL, gen_helper_rsqrtss, NULL },
    [0x53] = { gen_helper_rcpps, NULL, gen_helper_rcpss, NULL },
    [0x54] = { gen_helper_pand_xmm, gen_helper_pand_xmm }, /* andps, andpd */
    [0x55] = { gen_helper_pandn_xmm, gen_helper_pandn_xmm }, /* andnps, andnpd */
    [0x56] = { gen_helper_por_xmm, gen_helper_por_xmm }, /* orps, orpd */
    [0x57] = { gen_helper_pxor_xmm, gen_helper_pxor_xmm }, /* xorps, xorpd */
    [0x58] = SSE_FOP(add),
    [0x59] = SSE_FOP(mul),
    [0x5a] = { gen_helper_cvtps2pd, gen_helper_cvtpd2ps,
               gen_helper_cvtss2sd, gen_helper_cvtsd2ss },
    [0x5b] = { gen_helper_cvtdq2ps, gen_helper_cvtps2dq, gen_helper_cvttps2dq },
    [0x5c] = SSE_FOP(sub),
    [0x5d] = SSE_FOP(min),
    [0x5e] = SSE_FOP(div),
    [0x5f] = SSE_FOP(max),

    [0xc2] = SSE_FOP(cmpeq),
    [0xc6] = { (SSEFunc_0_epp)gen_helper_shufps,
               (SSEFunc_0_epp)gen_helper_shufpd }, /* XXX: casts */

    /* SSSE3, SSE4, MOVBE, CRC32, BMI1, BMI2, ADX.  */
    [0x38] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL },
    [0x3a] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL },

    /* MMX ops and their SSE extensions */
    [0x60] = MMX_OP2(punpcklbw),
    [0x61] = MMX_OP2(punpcklwd),
    [0x62] = MMX_OP2(punpckldq),
    [0x63] = MMX_OP2(packsswb),
    [0x64] = MMX_OP2(pcmpgtb),
    [0x65] = MMX_OP2(pcmpgtw),
    [0x66] = MMX_OP2(pcmpgtl),
    [0x67] = MMX_OP2(packuswb),
    [0x68] = MMX_OP2(punpckhbw),
    [0x69] = MMX_OP2(punpckhwd),
    [0x6a] = MMX_OP2(punpckhdq),
    [0x6b] = MMX_OP2(packssdw),
    [0x6c] = { NULL, gen_helper_punpcklqdq_xmm },
    [0x6d] = { NULL, gen_helper_punpckhqdq_xmm },
    [0x6e] = { SSE_SPECIAL, SSE_SPECIAL }, /* movd mm, ea */
    [0x6f] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movq, movdqa, , movqdu */
    [0x70] = { (SSEFunc_0_epp)gen_helper_pshufw_mmx,
               (SSEFunc_0_epp)gen_helper_pshufd_xmm,
               (SSEFunc_0_epp)gen_helper_pshufhw_xmm,
               (SSEFunc_0_epp)gen_helper_pshuflw_xmm }, /* XXX: casts */
    [0x71] = { SSE_SPECIAL, SSE_SPECIAL }, /* shiftw */
    [0x72] = { SSE_SPECIAL, SSE_SPECIAL }, /* shiftd */
    [0x73] = { SSE_SPECIAL, SSE_SPECIAL }, /* shiftq */
    [0x74] = MMX_OP2(pcmpeqb),
    [0x75] = MMX_OP2(pcmpeqw),
    [0x76] = MMX_OP2(pcmpeql),
    [0x77] = { SSE_DUMMY }, /* emms */
    [0x78] = { NULL, SSE_SPECIAL, NULL, SSE_SPECIAL }, /* extrq_i, insertq_i */
    [0x79] = { NULL, gen_helper_extrq_r, NULL, gen_helper_insertq_r },
    [0x7c] = { NULL, gen_helper_haddpd, NULL, gen_helper_haddps },
    [0x7d] = { NULL, gen_helper_hsubpd, NULL, gen_helper_hsubps },
    [0x7e] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movd, movd, , movq */
    [0x7f] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movq, movdqa, movdqu */
    [0xc4] = { SSE_SPECIAL, SSE_SPECIAL }, /* pinsrw */
    [0xc5] = { SSE_SPECIAL, SSE_SPECIAL }, /* pextrw */
    [0xd0] = { NULL, gen_helper_addsubpd, NULL, gen_helper_addsubps },
    [0xd1] = MMX_OP2(psrlw),
    [0xd2] = MMX_OP2(psrld),
    [0xd3] = MMX_OP2(psrlq),
    [0xd4] = MMX_OP2(paddq),
    [0xd5] = MMX_OP2(pmullw),
    [0xd6] = { NULL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL },
    [0xd7] = { SSE_SPECIAL, SSE_SPECIAL }, /* pmovmskb */
    [0xd8] = MMX_OP2(psubusb),
    [0xd9] = MMX_OP2(psubusw),
    [0xda] = MMX_OP2(pminub),
    [0xdb] = MMX_OP2(pand),
    [0xdc] = MMX_OP2(paddusb),
    [0xdd] = MMX_OP2(paddusw),
    [0xde] = MMX_OP2(pmaxub),
    [0xdf] = MMX_OP2(pandn),
    [0xe0] = MMX_OP2(pavgb),
    [0xe1] = MMX_OP2(psraw),
    [0xe2] = MMX_OP2(psrad),
    [0xe3] = MMX_OP2(pavgw),
    [0xe4] = MMX_OP2(pmulhuw),
    [0xe5] = MMX_OP2(pmulhw),
    [0xe6] = { NULL, gen_helper_cvttpd2dq, gen_helper_cvtdq2pd, gen_helper_cvtpd2dq },
    [0xe7] = { SSE_SPECIAL , SSE_SPECIAL },  /* movntq, movntq */
    [0xe8] = MMX_OP2(psubsb),
    [0xe9] = MMX_OP2(psubsw),
    [0xea] = MMX_OP2(pminsw),
    [0xeb] = MMX_OP2(por),
    [0xec] = MMX_OP2(paddsb),
    [0xed] = MMX_OP2(paddsw),
    [0xee] = MMX_OP2(pmaxsw),
    [0xef] = MMX_OP2(pxor),
    [0xf0] = { NULL, NULL, NULL, SSE_SPECIAL }, /* lddqu */
    [0xf1] = MMX_OP2(psllw),
    [0xf2] = MMX_OP2(pslld),
    [0xf3] = MMX_OP2(psllq),
    [0xf4] = MMX_OP2(pmuludq),
    [0xf5] = MMX_OP2(pmaddwd),
    [0xf6] = MMX_OP2(psadbw),
    [0xf7] = { (SSEFunc_0_epp)gen_helper_maskmov_mmx,
               (SSEFunc_0_epp)gen_helper_maskmov_xmm }, /* XXX: casts */
    [0xf8] = MMX_OP2(psubb),
    [0xf9] = MMX_OP2(psubw),
    [0xfa] = MMX_OP2(psubl),
    [0xfb] = MMX_OP2(psubq),
    [0xfc] = MMX_OP2(paddb),
    [0xfd] = MMX_OP2(paddw),
    [0xfe] = MMX_OP2(paddl),
};

static const SSEFunc_0_epp sse_op_table2[3 * 8][2] = {
    [0 + 2] = MMX_OP2(psrlw),
    [0 + 4] = MMX_OP2(psraw),
    [0 + 6] = MMX_OP2(psllw),
    [8 + 2] = MMX_OP2(psrld),
    [8 + 4] = MMX_OP2(psrad),
    [8 + 6] = MMX_OP2(pslld),
    [16 + 2] = MMX_OP2(psrlq),
    [16 + 3] = { NULL, gen_helper_psrldq_xmm },
    [16 + 6] = MMX_OP2(psllq),
    [16 + 7] = { NULL, gen_helper_pslldq_xmm },
};

static const SSEFunc_0_epi sse_op_table3ai[] = {
    gen_helper_cvtsi2ss,
    gen_helper_cvtsi2sd
};

#ifdef TARGET_X86_64
static const SSEFunc_0_epl sse_op_table3aq[] = {
    gen_helper_cvtsq2ss,
    gen_helper_cvtsq2sd
};
#endif

static const SSEFunc_i_ep sse_op_table3bi[] = {
    gen_helper_cvttss2si,
    gen_helper_cvtss2si,
    gen_helper_cvttsd2si,
    gen_helper_cvtsd2si
};

#ifdef TARGET_X86_64
static const SSEFunc_l_ep sse_op_table3bq[] = {
    gen_helper_cvttss2sq,
    gen_helper_cvtss2sq,
    gen_helper_cvttsd2sq,
    gen_helper_cvtsd2sq
};
#endif

static const SSEFunc_0_epp sse_op_table4[8][4] = {
    SSE_FOP(cmpeq),
    SSE_FOP(cmplt),
    SSE_FOP(cmple),
    SSE_FOP(cmpunord),
    SSE_FOP(cmpneq),
    SSE_FOP(cmpnlt),
    SSE_FOP(cmpnle),
    SSE_FOP(cmpord),
};

static const SSEFunc_0_epp sse_op_table5[256] = {
    [0x0c] = gen_helper_pi2fw,
    [0x0d] = gen_helper_pi2fd,
    [0x1c] = gen_helper_pf2iw,
    [0x1d] = gen_helper_pf2id,
    [0x8a] = gen_helper_pfnacc,
    [0x8e] = gen_helper_pfpnacc,
    [0x90] = gen_helper_pfcmpge,
    [0x94] = gen_helper_pfmin,
    [0x96] = gen_helper_pfrcp,
    [0x97] = gen_helper_pfrsqrt,
    [0x9a] = gen_helper_pfsub,
    [0x9e] = gen_helper_pfadd,
    [0xa0] = gen_helper_pfcmpgt,
    [0xa4] = gen_helper_pfmax,
    [0xa6] = gen_helper_movq, /* pfrcpit1; no need to actually increase precision */
    [0xa7] = gen_helper_movq, /* pfrsqit1 */
    [0xaa] = gen_helper_pfsubr,
    [0xae] = gen_helper_pfacc,
    [0xb0] = gen_helper_pfcmpeq,
    [0xb4] = gen_helper_pfmul,
    [0xb6] = gen_helper_movq, /* pfrcpit2 */
    [0xb7] = gen_helper_pmulhrw_mmx,
    [0xbb] = gen_helper_pswapd,
    [0xbf] = gen_helper_pavgb_mmx /* pavgusb */
};

struct SSEOpHelper_epp {
    SSEFunc_0_epp op[2];
    uint32_t ext_mask;
};

struct SSEOpHelper_eppi {
    SSEFunc_0_eppi op[2];
    uint32_t ext_mask;
};

#define SSSE3_OP(x) { MMX_OP2(x), CPUID_EXT_SSSE3 }
#define SSE41_OP(x) { { NULL, gen_helper_ ## x ## _xmm }, CPUID_EXT_SSE41 }
#define SSE42_OP(x) { { NULL, gen_helper_ ## x ## _xmm }, CPUID_EXT_SSE42 }
#define SSE41_SPECIAL { { NULL, SSE_SPECIAL }, CPUID_EXT_SSE41 }
#define PCLMULQDQ_OP(x) { { NULL, gen_helper_ ## x ## _xmm }, \
        CPUID_EXT_PCLMULQDQ }
#define AESNI_OP(x) { { NULL, gen_helper_ ## x ## _xmm }, CPUID_EXT_AES }

static const struct SSEOpHelper_epp sse_op_table6[256] = {
    [0x00] = SSSE3_OP(pshufb),
    [0x01] = SSSE3_OP(phaddw),
    [0x02] = SSSE3_OP(phaddd),
    [0x03] = SSSE3_OP(phaddsw),
    [0x04] = SSSE3_OP(pmaddubsw),
    [0x05] = SSSE3_OP(phsubw),
    [0x06] = SSSE3_OP(phsubd),
    [0x07] = SSSE3_OP(phsubsw),
    [0x08] = SSSE3_OP(psignb),
    [0x09] = SSSE3_OP(psignw),
    [0x0a] = SSSE3_OP(psignd),
    [0x0b] = SSSE3_OP(pmulhrsw),
    [0x10] = SSE41_OP(pblendvb),
    [0x14] = SSE41_OP(blendvps),
    [0x15] = SSE41_OP(blendvpd),
    [0x17] = SSE41_OP(ptest),
    [0x1c] = SSSE3_OP(pabsb),
    [0x1d] = SSSE3_OP(pabsw),
    [0x1e] = SSSE3_OP(pabsd),
    [0x20] = SSE41_OP(pmovsxbw),
    [0x21] = SSE41_OP(pmovsxbd),
    [0x22] = SSE41_OP(pmovsxbq),
    [0x23] = SSE41_OP(pmovsxwd),
    [0x24] = SSE41_OP(pmovsxwq),
    [0x25] = SSE41_OP(pmovsxdq),
    [0x28] = SSE41_OP(pmuldq),
    [0x29] = SSE41_OP(pcmpeqq),
    [0x2a] = SSE41_SPECIAL, /* movntqda */
    [0x2b] = SSE41_OP(packusdw),
    [0x30] = SSE41_OP(pmovzxbw),
    [0x31] = SSE41_OP(pmovzxbd),
    [0x32] = SSE41_OP(pmovzxbq),
    [0x33] = SSE41_OP(pmovzxwd),
    [0x34] = SSE41_OP(pmovzxwq),
    [0x35] = SSE41_OP(pmovzxdq),
    [0x37] = SSE42_OP(pcmpgtq),
    [0x38] = SSE41_OP(pminsb),
    [0x39] = SSE41_OP(pminsd),
    [0x3a] = SSE41_OP(pminuw),
    [0x3b] = SSE41_OP(pminud),
    [0x3c] = SSE41_OP(pmaxsb),
    [0x3d] = SSE41_OP(pmaxsd),
    [0x3e] = SSE41_OP(pmaxuw),
    [0x3f] = SSE41_OP(pmaxud),
    [0x40] = SSE41_OP(pmulld),
    [0x41] = SSE41_OP(phminposuw),
    [0xdb] = AESNI_OP(aesimc),
    [0xdc] = AESNI_OP(aesenc),
    [0xdd] = AESNI_OP(aesenclast),
    [0xde] = AESNI_OP(aesdec),
    [0xdf] = AESNI_OP(aesdeclast),
};

static const struct SSEOpHelper_eppi sse_op_table7[256] = {
    [0x08] = SSE41_OP(roundps),
    [0x09] = SSE41_OP(roundpd),
    [0x0a] = SSE41_OP(roundss),
    [0x0b] = SSE41_OP(roundsd),
    [0x0c] = SSE41_OP(blendps),
    [0x0d] = SSE41_OP(blendpd),
    [0x0e] = SSE41_OP(pblendw),
    [0x0f] = SSSE3_OP(palignr),
    [0x14] = SSE41_SPECIAL, /* pextrb */
    [0x15] = SSE41_SPECIAL, /* pextrw */
    [0x16] = SSE41_SPECIAL, /* pextrd/pextrq */
    [0x17] = SSE41_SPECIAL, /* extractps */
    [0x20] = SSE41_SPECIAL, /* pinsrb */
    [0x21] = SSE41_SPECIAL, /* insertps */
    [0x22] = SSE41_SPECIAL, /* pinsrd/pinsrq */
    [0x40] = SSE41_OP(dpps),
    [0x41] = SSE41_OP(dppd),
    [0x42] = SSE41_OP(mpsadbw),
    [0x44] = PCLMULQDQ_OP(pclmulqdq),
    [0x60] = SSE42_OP(pcmpestrm),
    [0x61] = SSE42_OP(pcmpestri),
    [0x62] = SSE42_OP(pcmpistrm),
    [0x63] = SSE42_OP(pcmpistri),
    [0xdf] = AESNI_OP(aeskeygenassist),
};

static void gen_sse(CPUX86State *env, DisasContext *s, int b,
                    target_ulong pc_start, int rex_r)
{
    int b1, op1_offset, op2_offset, is_xmm, val;
    int modrm, mod, rm, reg;
    SSEFunc_0_epp sse_fn_epp;
    SSEFunc_0_eppi sse_fn_eppi;
    SSEFunc_0_ppi sse_fn_ppi;
    SSEFunc_0_eppt sse_fn_eppt;
    TCGMemOp ot;

    b &= 0xff;
    if (s->prefix & PREFIX_DATA)
        b1 = 1;
    else if (s->prefix & PREFIX_REPZ)
        b1 = 2;
    else if (s->prefix & PREFIX_REPNZ)
        b1 = 3;
    else
        b1 = 0;
    sse_fn_epp = sse_op_table1[b][b1];
    if (!sse_fn_epp) {
        goto unknown_op;
    }
    if ((b <= 0x5f && b >= 0x10) || b == 0xc6 || b == 0xc2) {
        is_xmm = 1;
    } else {
        if (b1 == 0) {
            /* MMX case */
            is_xmm = 0;
        } else {
            is_xmm = 1;
        }
    }
    /* simple MMX/SSE operation */
    if (s->flags & HF_TS_MASK) {
        gen_exception(s, EXCP07_PREX, pc_start - s->cs_base);
        return;
    }
    if (s->flags & HF_EM_MASK) {
    illegal_op:
        gen_illegal_opcode(s);
        return;
    }
    if (is_xmm
        && !(s->flags & HF_OSFXSR_MASK)
        && ((b != 0x38 && b != 0x3a) || (s->prefix & PREFIX_DATA))) {
        goto unknown_op;
    }
    if (b == 0x0e) {
        if (!(s->cpuid_ext2_features & CPUID_EXT2_3DNOW)) {
            /* If we were fully decoding this we might use illegal_op.  */
            goto unknown_op;
        }
        /* femms */
        gen_helper_emms(cpu_env);
        return;
    }
    if (b == 0x77) {
        /* emms */
        gen_helper_emms(cpu_env);
        return;
    }
    /* prepare MMX state (XXX: optimize by storing fptt and fptags in
       the static cpu state) */
    if (!is_xmm) {
        gen_helper_enter_mmx(cpu_env);
    }

    modrm = cpu_ldub_code(env, s->pc++);
    reg = ((modrm >> 3) & 7);
    if (is_xmm)
        reg |= rex_r;
    mod = (modrm >> 6) & 3;
    if (sse_fn_epp == SSE_SPECIAL) {
        b |= (b1 << 8);
        switch(b) {
        case 0x0e7: /* movntq */
            if (mod == 3) {
                goto illegal_op;
            }
            gen_lea_modrm(env, s, modrm);
            gen_stq_env_A0(s, offsetof(CPUX86State, fpregs[reg].mmx));
            break;
        case 0x1e7: /* movntdq */
        case 0x02b: /* movntps */
        case 0x12b: /* movntps */
            if (mod == 3)
                goto illegal_op;
            gen_lea_modrm(env, s, modrm);
            gen_sto_env_A0(s, offsetof(CPUX86State, xmm_regs[reg]));
            break;
        case 0x3f0: /* lddqu */
            if (mod == 3)
                goto illegal_op;
            gen_lea_modrm(env, s, modrm);
            gen_ldo_env_A0(s, offsetof(CPUX86State, xmm_regs[reg]));
            break;
        case 0x22b: /* movntss */
        case 0x32b: /* movntsd */
            if (mod == 3)
                goto illegal_op;