diff options
Diffstat (limited to 'tcg/aarch64/tcg-target.c.inc')
| -rw-r--r-- | tcg/aarch64/tcg-target.c.inc | 1558 |
1 files changed, 899 insertions, 659 deletions
diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index 4645242..3b088b7 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -508,7 +508,9 @@ typedef enum { /* Add/subtract with carry instructions. */ I3503_ADC = 0x1a000000, + I3503_ADCS = 0x3a000000, I3503_SBC = 0x5a000000, + I3503_SBCS = 0x7a000000, /* Conditional select instructions. */ I3506_CSEL = 0x1a800000, @@ -1347,70 +1349,37 @@ static inline void tcg_out_extr(TCGContext *s, TCGType ext, TCGReg rd, tcg_out_insn(s, 3403, EXTR, ext, rd, rn, rm, a); } -static inline void tcg_out_shl(TCGContext *s, TCGType ext, - TCGReg rd, TCGReg rn, unsigned int m) +static void tgen_cmp(TCGContext *s, TCGType ext, TCGCond cond, + TCGReg a, TCGReg b) { - int bits = ext ? 64 : 32; - int max = bits - 1; - tcg_out_ubfm(s, ext, rd, rn, (bits - m) & max, (max - m) & max); -} - -static inline void tcg_out_shr(TCGContext *s, TCGType ext, - TCGReg rd, TCGReg rn, unsigned int m) -{ - int max = ext ? 63 : 31; - tcg_out_ubfm(s, ext, rd, rn, m & max, max); -} - -static inline void tcg_out_sar(TCGContext *s, TCGType ext, - TCGReg rd, TCGReg rn, unsigned int m) -{ - int max = ext ? 63 : 31; - tcg_out_sbfm(s, ext, rd, rn, m & max, max); -} - -static inline void tcg_out_rotr(TCGContext *s, TCGType ext, - TCGReg rd, TCGReg rn, unsigned int m) -{ - int max = ext ? 63 : 31; - tcg_out_extr(s, ext, rd, rn, rn, m & max); -} - -static inline void tcg_out_rotl(TCGContext *s, TCGType ext, - TCGReg rd, TCGReg rn, unsigned int m) -{ - int max = ext ? 63 : 31; - tcg_out_extr(s, ext, rd, rn, rn, -m & max); + if (is_tst_cond(cond)) { + tcg_out_insn(s, 3510, ANDS, ext, TCG_REG_XZR, a, b); + } else { + tcg_out_insn(s, 3502, SUBS, ext, TCG_REG_XZR, a, b); + } } -static inline void tcg_out_dep(TCGContext *s, TCGType ext, TCGReg rd, - TCGReg rn, unsigned lsb, unsigned width) +static void tgen_cmpi(TCGContext *s, TCGType ext, TCGCond cond, + TCGReg a, tcg_target_long b) { - unsigned size = ext ? 64 : 32; - unsigned a = (size - lsb) & (size - 1); - unsigned b = width - 1; - tcg_out_bfm(s, ext, rd, rn, a, b); + if (is_tst_cond(cond)) { + tcg_out_logicali(s, I3404_ANDSI, ext, TCG_REG_XZR, a, b); + } else if (b >= 0) { + tcg_debug_assert(is_aimm(b)); + tcg_out_insn(s, 3401, SUBSI, ext, TCG_REG_XZR, a, b); + } else { + tcg_debug_assert(is_aimm(-b)); + tcg_out_insn(s, 3401, ADDSI, ext, TCG_REG_XZR, a, -b); + } } static void tcg_out_cmp(TCGContext *s, TCGType ext, TCGCond cond, TCGReg a, tcg_target_long b, bool const_b) { - if (is_tst_cond(cond)) { - if (!const_b) { - tcg_out_insn(s, 3510, ANDS, ext, TCG_REG_XZR, a, b); - } else { - tcg_out_logicali(s, I3404_ANDSI, ext, TCG_REG_XZR, a, b); - } + if (const_b) { + tgen_cmpi(s, ext, cond, a, b); } else { - if (!const_b) { - tcg_out_insn(s, 3502, SUBS, ext, TCG_REG_XZR, a, b); - } else if (b >= 0) { - tcg_debug_assert(is_aimm(b)); - tcg_out_insn(s, 3401, SUBSI, ext, TCG_REG_XZR, a, b); - } else { - tcg_debug_assert(is_aimm(-b)); - tcg_out_insn(s, 3401, ADDSI, ext, TCG_REG_XZR, a, -b); - } + tgen_cmp(s, ext, cond, a, b); } } @@ -1438,7 +1407,7 @@ static void tcg_out_call(TCGContext *s, const tcg_insn_unit *target, tcg_out_call_int(s, target); } -static inline void tcg_out_goto_label(TCGContext *s, TCGLabel *l) +static void tcg_out_br(TCGContext *s, TCGLabel *l) { if (!l->has_value) { tcg_out_reloc(s, s->code_ptr, R_AARCH64_JUMP26, l, 0); @@ -1448,8 +1417,16 @@ static inline void tcg_out_goto_label(TCGContext *s, TCGLabel *l) } } -static void tcg_out_brcond(TCGContext *s, TCGType ext, TCGCond c, TCGArg a, - TCGArg b, bool b_const, TCGLabel *l) +static void tgen_brcond(TCGContext *s, TCGType type, TCGCond c, + TCGReg a, TCGReg b, TCGLabel *l) +{ + tgen_cmp(s, type, c, a, b); + tcg_out_reloc(s, s->code_ptr, R_AARCH64_CONDBR19, l, 0); + tcg_out_insn(s, 3202, B_C, c, 0); +} + +static void tgen_brcondi(TCGContext *s, TCGType ext, TCGCond c, + TCGReg a, tcg_target_long b, TCGLabel *l) { int tbit = -1; bool need_cmp = true; @@ -1458,14 +1435,14 @@ static void tcg_out_brcond(TCGContext *s, TCGType ext, TCGCond c, TCGArg a, case TCG_COND_EQ: case TCG_COND_NE: /* cmp xN,0; b.ne L -> cbnz xN,L */ - if (b_const && b == 0) { + if (b == 0) { need_cmp = false; } break; case TCG_COND_LT: case TCG_COND_GE: /* cmp xN,0; b.mi L -> tbnz xN,63,L */ - if (b_const && b == 0) { + if (b == 0) { c = (c == TCG_COND_LT ? TCG_COND_TSTNE : TCG_COND_TSTEQ); tbit = ext ? 63 : 31; need_cmp = false; @@ -1474,14 +1451,14 @@ static void tcg_out_brcond(TCGContext *s, TCGType ext, TCGCond c, TCGArg a, case TCG_COND_TSTEQ: case TCG_COND_TSTNE: /* tst xN,0xffffffff; b.ne L -> cbnz wN,L */ - if (b_const && b == UINT32_MAX) { + if (b == UINT32_MAX) { c = tcg_tst_eqne_cond(c); ext = TCG_TYPE_I32; need_cmp = false; break; } /* tst xN,1<<B; b.ne L -> tbnz xN,B,L */ - if (b_const && is_power_of_2(b)) { + if (is_power_of_2(b)) { tbit = ctz64(b); need_cmp = false; } @@ -1491,7 +1468,7 @@ static void tcg_out_brcond(TCGContext *s, TCGType ext, TCGCond c, TCGArg a, } if (need_cmp) { - tcg_out_cmp(s, ext, c, a, b, b_const); + tgen_cmpi(s, ext, c, a, b); tcg_out_reloc(s, s->code_ptr, R_AARCH64_CONDBR19, l, 0); tcg_out_insn(s, 3202, B_C, c, 0); return; @@ -1524,6 +1501,12 @@ static void tcg_out_brcond(TCGContext *s, TCGType ext, TCGCond c, TCGArg a, } } +static const TCGOutOpBrcond outop_brcond = { + .base.static_constraint = C_O0_I2(r, rC), + .out_rr = tgen_brcond, + .out_ri = tgen_brcondi, +}; + static inline void tcg_out_rev(TCGContext *s, int ext, MemOp s_bits, TCGReg rd, TCGReg rn) { @@ -1592,67 +1575,7 @@ static void tcg_out_extrl_i64_i32(TCGContext *s, TCGReg rd, TCGReg rn) tcg_out_mov(s, TCG_TYPE_I32, rd, rn); } -static void tcg_out_addsubi(TCGContext *s, int ext, TCGReg rd, - TCGReg rn, int64_t aimm) -{ - if (aimm >= 0) { - tcg_out_insn(s, 3401, ADDI, ext, rd, rn, aimm); - } else { - tcg_out_insn(s, 3401, SUBI, ext, rd, rn, -aimm); - } -} - -static void tcg_out_addsub2(TCGContext *s, TCGType ext, TCGReg rl, - TCGReg rh, TCGReg al, TCGReg ah, - tcg_target_long bl, tcg_target_long bh, - bool const_bl, bool const_bh, bool sub) -{ - TCGReg orig_rl = rl; - AArch64Insn insn; - - if (rl == ah || (!const_bh && rl == bh)) { - rl = TCG_REG_TMP0; - } - - if (const_bl) { - if (bl < 0) { - bl = -bl; - insn = sub ? I3401_ADDSI : I3401_SUBSI; - } else { - insn = sub ? I3401_SUBSI : I3401_ADDSI; - } - - if (unlikely(al == TCG_REG_XZR)) { - /* ??? We want to allow al to be zero for the benefit of - negation via subtraction. However, that leaves open the - possibility of adding 0+const in the low part, and the - immediate add instructions encode XSP not XZR. Don't try - anything more elaborate here than loading another zero. */ - al = TCG_REG_TMP0; - tcg_out_movi(s, ext, al, 0); - } - tcg_out_insn_3401(s, insn, ext, rl, al, bl); - } else { - tcg_out_insn_3502(s, sub ? I3502_SUBS : I3502_ADDS, ext, rl, al, bl); - } - - insn = I3503_ADC; - if (const_bh) { - /* Note that the only two constants we support are 0 and -1, and - that SBC = rn + ~rm + c, so adc -1 is sbc 0, and vice-versa. */ - if ((bh != 0) ^ sub) { - insn = I3503_SBC; - } - bh = TCG_REG_XZR; - } else if (sub) { - insn = I3503_SBC; - } - tcg_out_insn_3503(s, insn, ext, rh, ah, bh); - - tcg_out_mov(s, ext, orig_rl, rl); -} - -static inline void tcg_out_mb(TCGContext *s, TCGArg a0) +static void tcg_out_mb(TCGContext *s, unsigned a0) { static const uint32_t sync[] = { [0 ... TCG_MO_ALL] = DMB_ISH | DMB_LD | DMB_ST, @@ -1664,37 +1587,6 @@ static inline void tcg_out_mb(TCGContext *s, TCGArg a0) tcg_out32(s, sync[a0 & TCG_MO_ALL]); } -static void tcg_out_cltz(TCGContext *s, TCGType ext, TCGReg d, - TCGReg a0, TCGArg b, bool const_b, bool is_ctz) -{ - TCGReg a1 = a0; - if (is_ctz) { - a1 = TCG_REG_TMP0; - tcg_out_insn(s, 3507, RBIT, ext, a1, a0); - } - if (const_b && b == (ext ? 64 : 32)) { - tcg_out_insn(s, 3507, CLZ, ext, d, a1); - } else { - AArch64Insn sel = I3506_CSEL; - - tcg_out_cmp(s, ext, TCG_COND_NE, a0, 0, 1); - tcg_out_insn(s, 3507, CLZ, ext, TCG_REG_TMP0, a1); - - if (const_b) { - if (b == -1) { - b = TCG_REG_XZR; - sel = I3506_CSINV; - } else if (b == 0) { - b = TCG_REG_XZR; - } else { - tcg_out_movi(s, ext, d, b); - b = d; - } - } - tcg_out_insn_3506(s, sel, ext, d, TCG_REG_TMP0, b, TCG_COND_NE); - } -} - typedef struct { TCGReg base; TCGReg index; @@ -1769,7 +1661,6 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, unsigned s_mask = (1u << s_bits) - 1; unsigned mem_index = get_mmuidx(oi); TCGReg addr_adj; - TCGType mask_type; uint64_t compare_mask; ldst = new_ldst_label(s); @@ -1777,9 +1668,6 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst->oi = oi; ldst->addr_reg = addr_reg; - mask_type = (s->page_bits + s->tlb_dyn_max_bits > 32 - ? TCG_TYPE_I64 : TCG_TYPE_I32); - /* Load cpu->neg.tlb.f[mmu_idx].{mask,table} into {tmp0,tmp1}. */ QEMU_BUILD_BUG_ON(offsetof(CPUTLBDescFast, mask) != 0); QEMU_BUILD_BUG_ON(offsetof(CPUTLBDescFast, table) != 8); @@ -1787,9 +1675,9 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tlb_mask_table_ofs(s, mem_index), 1, 0); /* Extract the TLB index from the address into X0. */ - tcg_out_insn(s, 3502S, AND_LSR, mask_type == TCG_TYPE_I64, + tcg_out_insn(s, 3502S, AND_LSR, TCG_TYPE_I64, TCG_REG_TMP0, TCG_REG_TMP0, addr_reg, - s->page_bits - CPU_TLB_ENTRY_BITS); + TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); /* Add the tlb_table pointer, forming the CPUTLBEntry address. */ tcg_out_insn(s, 3502, ADD, 1, TCG_REG_TMP1, TCG_REG_TMP1, TCG_REG_TMP0); @@ -1815,7 +1703,7 @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_insn(s, 3401, ADDI, addr_type, addr_adj, addr_reg, s_mask - a_mask); } - compare_mask = (uint64_t)s->page_mask | a_mask; + compare_mask = (uint64_t)TARGET_PAGE_MASK | a_mask; /* Store the page mask part of the address into TMP2. */ tcg_out_logicali(s, I3404_ANDI, addr_type, TCG_REG_TMP2, @@ -1914,8 +1802,8 @@ static void tcg_out_qemu_st_direct(TCGContext *s, MemOp memop, } } -static void tcg_out_qemu_ld(TCGContext *s, TCGReg data_reg, TCGReg addr_reg, - MemOpIdx oi, TCGType data_type) +static void tgen_qemu_ld(TCGContext *s, TCGType data_type, TCGReg data_reg, + TCGReg addr_reg, MemOpIdx oi) { TCGLabelQemuLdst *ldst; HostAddress h; @@ -1930,8 +1818,13 @@ static void tcg_out_qemu_ld(TCGContext *s, TCGReg data_reg, TCGReg addr_reg, } } -static void tcg_out_qemu_st(TCGContext *s, TCGReg data_reg, TCGReg addr_reg, - MemOpIdx oi, TCGType data_type) +static const TCGOutOpQemuLdSt outop_qemu_ld = { + .base.static_constraint = C_O1_I1(r, r), + .out = tgen_qemu_ld, +}; + +static void tgen_qemu_st(TCGContext *s, TCGType data_type, TCGReg data_reg, + TCGReg addr_reg, MemOpIdx oi) { TCGLabelQemuLdst *ldst; HostAddress h; @@ -1946,6 +1839,11 @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg data_reg, TCGReg addr_reg, } } +static const TCGOutOpQemuLdSt outop_qemu_st = { + .base.static_constraint = C_O0_I2(rz, r), + .out = tgen_qemu_st, +}; + static void tcg_out_qemu_ldst_i128(TCGContext *s, TCGReg datalo, TCGReg datahi, TCGReg addr_reg, MemOpIdx oi, bool is_ld) { @@ -2048,6 +1946,28 @@ static void tcg_out_qemu_ldst_i128(TCGContext *s, TCGReg datalo, TCGReg datahi, } } +static void tgen_qemu_ld2(TCGContext *s, TCGType type, TCGReg datalo, + TCGReg datahi, TCGReg addr_reg, MemOpIdx oi) +{ + tcg_out_qemu_ldst_i128(s, datalo, datahi, addr_reg, oi, true); +} + +static const TCGOutOpQemuLdSt2 outop_qemu_ld2 = { + .base.static_constraint = C_O2_I1(r, r, r), + .out = tgen_qemu_ld2, +}; + +static void tgen_qemu_st2(TCGContext *s, TCGType type, TCGReg datalo, + TCGReg datahi, TCGReg addr_reg, MemOpIdx oi) +{ + tcg_out_qemu_ldst_i128(s, datalo, datahi, addr_reg, oi, false); +} + +static const TCGOutOpQemuLdSt2 outop_qemu_st2 = { + .base.static_constraint = C_O0_I3(rz, rz, r), + .out = tgen_qemu_st2, +}; + static const tcg_insn_unit *tb_ret_addr; static void tcg_out_exit_tb(TCGContext *s, uintptr_t a0) @@ -2094,6 +2014,11 @@ static void tcg_out_goto_tb(TCGContext *s, int which) tcg_out_bti(s, BTI_J); } +static void tcg_out_goto_ptr(TCGContext *s, TCGReg a0) +{ + tcg_out_insn(s, 3207, BR, a0); +} + void tb_target_set_jmp_target(const TranslationBlock *tb, int n, uintptr_t jmp_rx, uintptr_t jmp_rw) { @@ -2115,402 +2040,859 @@ void tb_target_set_jmp_target(const TranslationBlock *tb, int n, flush_idcache_range(jmp_rx, jmp_rw, 4); } -static void tcg_out_op(TCGContext *s, TCGOpcode opc, TCGType ext, - const TCGArg args[TCG_MAX_OP_ARGS], - const int const_args[TCG_MAX_OP_ARGS]) + +static void tgen_add(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) { - /* Hoist the loads of the most common arguments. */ - TCGArg a0 = args[0]; - TCGArg a1 = args[1]; - TCGArg a2 = args[2]; - int c2 = const_args[2]; + tcg_out_insn(s, 3502, ADD, type, a0, a1, a2); +} - switch (opc) { - case INDEX_op_goto_ptr: - tcg_out_insn(s, 3207, BR, a0); - break; +static void tgen_addi(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + if (a2 >= 0) { + tcg_out_insn(s, 3401, ADDI, type, a0, a1, a2); + } else { + tcg_out_insn(s, 3401, SUBI, type, a0, a1, -a2); + } +} - case INDEX_op_br: - tcg_out_goto_label(s, arg_label(a0)); - break; +static const TCGOutOpBinary outop_add = { + .base.static_constraint = C_O1_I2(r, r, rA), + .out_rrr = tgen_add, + .out_rri = tgen_addi, +}; - case INDEX_op_ld8u_i32: - case INDEX_op_ld8u_i64: - tcg_out_ldst(s, I3312_LDRB, a0, a1, a2, 0); - break; - case INDEX_op_ld8s_i32: - tcg_out_ldst(s, I3312_LDRSBW, a0, a1, a2, 0); - break; - case INDEX_op_ld8s_i64: - tcg_out_ldst(s, I3312_LDRSBX, a0, a1, a2, 0); - break; - case INDEX_op_ld16u_i32: - case INDEX_op_ld16u_i64: - tcg_out_ldst(s, I3312_LDRH, a0, a1, a2, 1); - break; - case INDEX_op_ld16s_i32: - tcg_out_ldst(s, I3312_LDRSHW, a0, a1, a2, 1); - break; - case INDEX_op_ld16s_i64: - tcg_out_ldst(s, I3312_LDRSHX, a0, a1, a2, 1); - break; - case INDEX_op_ld_i32: - case INDEX_op_ld32u_i64: - tcg_out_ldst(s, I3312_LDRW, a0, a1, a2, 2); - break; - case INDEX_op_ld32s_i64: - tcg_out_ldst(s, I3312_LDRSWX, a0, a1, a2, 2); - break; - case INDEX_op_ld_i64: - tcg_out_ldst(s, I3312_LDRX, a0, a1, a2, 3); - break; +static void tgen_addco(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3502, ADDS, type, a0, a1, a2); +} - case INDEX_op_st8_i32: - case INDEX_op_st8_i64: - tcg_out_ldst(s, I3312_STRB, a0, a1, a2, 0); - break; - case INDEX_op_st16_i32: - case INDEX_op_st16_i64: - tcg_out_ldst(s, I3312_STRH, a0, a1, a2, 1); +static void tgen_addco_imm(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + if (a2 >= 0) { + tcg_out_insn(s, 3401, ADDSI, type, a0, a1, a2); + } else { + tcg_out_insn(s, 3401, SUBSI, type, a0, a1, -a2); + } +} + +static const TCGOutOpBinary outop_addco = { + .base.static_constraint = C_O1_I2(r, r, rA), + .out_rrr = tgen_addco, + .out_rri = tgen_addco_imm, +}; + +static void tgen_addci_rrr(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3503, ADC, type, a0, a1, a2); +} + +static void tgen_addci_rri(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + /* + * Note that the only two constants we support are 0 and -1, and + * that SBC = rn + ~rm + c, so adc -1 is sbc 0, and vice-versa. + */ + if (a2) { + tcg_out_insn(s, 3503, SBC, type, a0, a1, TCG_REG_XZR); + } else { + tcg_out_insn(s, 3503, ADC, type, a0, a1, TCG_REG_XZR); + } +} + +static const TCGOutOpAddSubCarry outop_addci = { + .base.static_constraint = C_O1_I2(r, rz, rMZ), + .out_rrr = tgen_addci_rrr, + .out_rri = tgen_addci_rri, +}; + +static void tgen_addcio(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3503, ADCS, type, a0, a1, a2); +} + +static void tgen_addcio_imm(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + /* Use SBCS w/0 for ADCS w/-1 -- see above. */ + if (a2) { + tcg_out_insn(s, 3503, SBCS, type, a0, a1, TCG_REG_XZR); + } else { + tcg_out_insn(s, 3503, ADCS, type, a0, a1, TCG_REG_XZR); + } +} + +static const TCGOutOpBinary outop_addcio = { + .base.static_constraint = C_O1_I2(r, rz, rMZ), + .out_rrr = tgen_addcio, + .out_rri = tgen_addcio_imm, +}; + +static void tcg_out_set_carry(TCGContext *s) +{ + tcg_out_insn(s, 3502, SUBS, TCG_TYPE_I32, + TCG_REG_XZR, TCG_REG_XZR, TCG_REG_XZR); +} + +static void tgen_and(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3510, AND, type, a0, a1, a2); +} + +static void tgen_andi(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + tcg_out_logicali(s, I3404_ANDI, type, a0, a1, a2); +} + +static const TCGOutOpBinary outop_and = { + .base.static_constraint = C_O1_I2(r, r, rL), + .out_rrr = tgen_and, + .out_rri = tgen_andi, +}; + +static void tgen_andc(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3510, BIC, type, a0, a1, a2); +} + +static const TCGOutOpBinary outop_andc = { + .base.static_constraint = C_O1_I2(r, r, r), + .out_rrr = tgen_andc, +}; + +static void tgen_clz(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_cmp(s, type, TCG_COND_NE, a1, 0, true); + tcg_out_insn(s, 3507, CLZ, type, TCG_REG_TMP0, a1); + tcg_out_insn(s, 3506, CSEL, type, a0, TCG_REG_TMP0, a2, TCG_COND_NE); +} + +static void tgen_clzi(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + if (a2 == (type == TCG_TYPE_I32 ? 32 : 64)) { + tcg_out_insn(s, 3507, CLZ, type, a0, a1); + return; + } + + tcg_out_cmp(s, type, TCG_COND_NE, a1, 0, true); + tcg_out_insn(s, 3507, CLZ, type, a0, a1); + + switch (a2) { + case -1: + tcg_out_insn(s, 3506, CSINV, type, a0, a0, TCG_REG_XZR, TCG_COND_NE); break; - case INDEX_op_st_i32: - case INDEX_op_st32_i64: - tcg_out_ldst(s, I3312_STRW, a0, a1, a2, 2); + case 0: + tcg_out_insn(s, 3506, CSEL, type, a0, a0, TCG_REG_XZR, TCG_COND_NE); break; - case INDEX_op_st_i64: - tcg_out_ldst(s, I3312_STRX, a0, a1, a2, 3); + default: + tcg_out_movi(s, type, TCG_REG_TMP0, a2); + tcg_out_insn(s, 3506, CSEL, type, a0, a0, TCG_REG_TMP0, TCG_COND_NE); break; + } +} - case INDEX_op_add_i32: - a2 = (int32_t)a2; - /* FALLTHRU */ - case INDEX_op_add_i64: - if (c2) { - tcg_out_addsubi(s, ext, a0, a1, a2); - } else { - tcg_out_insn(s, 3502, ADD, ext, a0, a1, a2); - } - break; +static const TCGOutOpBinary outop_clz = { + .base.static_constraint = C_O1_I2(r, r, rAL), + .out_rrr = tgen_clz, + .out_rri = tgen_clzi, +}; - case INDEX_op_sub_i32: - a2 = (int32_t)a2; - /* FALLTHRU */ - case INDEX_op_sub_i64: - if (c2) { - tcg_out_addsubi(s, ext, a0, a1, -a2); - } else { - tcg_out_insn(s, 3502, SUB, ext, a0, a1, a2); - } - break; +static const TCGOutOpUnary outop_ctpop = { + .base.static_constraint = C_NotImplemented, +}; - case INDEX_op_neg_i64: - case INDEX_op_neg_i32: - tcg_out_insn(s, 3502, SUB, ext, a0, TCG_REG_XZR, a1); - break; +static void tgen_ctz(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3507, RBIT, type, TCG_REG_TMP0, a1); + tgen_clz(s, type, a0, TCG_REG_TMP0, a2); +} - case INDEX_op_and_i32: - a2 = (int32_t)a2; - /* FALLTHRU */ - case INDEX_op_and_i64: - if (c2) { - tcg_out_logicali(s, I3404_ANDI, ext, a0, a1, a2); - } else { - tcg_out_insn(s, 3510, AND, ext, a0, a1, a2); - } - break; +static void tgen_ctzi(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + tcg_out_insn(s, 3507, RBIT, type, TCG_REG_TMP0, a1); + tgen_clzi(s, type, a0, TCG_REG_TMP0, a2); +} - case INDEX_op_andc_i32: - a2 = (int32_t)a2; - /* FALLTHRU */ - case INDEX_op_andc_i64: - if (c2) { - tcg_out_logicali(s, I3404_ANDI, ext, a0, a1, ~a2); - } else { - tcg_out_insn(s, 3510, BIC, ext, a0, a1, a2); - } - break; +static const TCGOutOpBinary outop_ctz = { + .base.static_constraint = C_O1_I2(r, r, rAL), + .out_rrr = tgen_ctz, + .out_rri = tgen_ctzi, +}; - case INDEX_op_or_i32: - a2 = (int32_t)a2; - /* FALLTHRU */ - case INDEX_op_or_i64: - if (c2) { - tcg_out_logicali(s, I3404_ORRI, ext, a0, a1, a2); - } else { - tcg_out_insn(s, 3510, ORR, ext, a0, a1, a2); - } - break; +static void tgen_divs(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3508, SDIV, type, a0, a1, a2); +} - case INDEX_op_orc_i32: - a2 = (int32_t)a2; - /* FALLTHRU */ - case INDEX_op_orc_i64: - if (c2) { - tcg_out_logicali(s, I3404_ORRI, ext, a0, a1, ~a2); - } else { - tcg_out_insn(s, 3510, ORN, ext, a0, a1, a2); - } - break; +static const TCGOutOpBinary outop_divs = { + .base.static_constraint = C_O1_I2(r, r, r), + .out_rrr = tgen_divs, +}; - case INDEX_op_xor_i32: - a2 = (int32_t)a2; - /* FALLTHRU */ - case INDEX_op_xor_i64: - if (c2) { - tcg_out_logicali(s, I3404_EORI, ext, a0, a1, a2); - } else { - tcg_out_insn(s, 3510, EOR, ext, a0, a1, a2); - } - break; +static const TCGOutOpDivRem outop_divs2 = { + .base.static_constraint = C_NotImplemented, +}; - case INDEX_op_eqv_i32: - a2 = (int32_t)a2; - /* FALLTHRU */ - case INDEX_op_eqv_i64: - if (c2) { - tcg_out_logicali(s, I3404_EORI, ext, a0, a1, ~a2); - } else { - tcg_out_insn(s, 3510, EON, ext, a0, a1, a2); - } - break; +static void tgen_divu(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3508, UDIV, type, a0, a1, a2); +} - case INDEX_op_not_i64: - case INDEX_op_not_i32: - tcg_out_insn(s, 3510, ORN, ext, a0, TCG_REG_XZR, a1); - break; +static const TCGOutOpBinary outop_divu = { + .base.static_constraint = C_O1_I2(r, r, r), + .out_rrr = tgen_divu, +}; - case INDEX_op_mul_i64: - case INDEX_op_mul_i32: - tcg_out_insn(s, 3509, MADD, ext, a0, a1, a2, TCG_REG_XZR); - break; +static const TCGOutOpDivRem outop_divu2 = { + .base.static_constraint = C_NotImplemented, +}; - case INDEX_op_div_i64: - case INDEX_op_div_i32: - tcg_out_insn(s, 3508, SDIV, ext, a0, a1, a2); - break; - case INDEX_op_divu_i64: - case INDEX_op_divu_i32: - tcg_out_insn(s, 3508, UDIV, ext, a0, a1, a2); - break; +static void tgen_eqv(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3510, EON, type, a0, a1, a2); +} - case INDEX_op_rem_i64: - case INDEX_op_rem_i32: - tcg_out_insn(s, 3508, SDIV, ext, TCG_REG_TMP0, a1, a2); - tcg_out_insn(s, 3509, MSUB, ext, a0, TCG_REG_TMP0, a2, a1); - break; - case INDEX_op_remu_i64: - case INDEX_op_remu_i32: - tcg_out_insn(s, 3508, UDIV, ext, TCG_REG_TMP0, a1, a2); - tcg_out_insn(s, 3509, MSUB, ext, a0, TCG_REG_TMP0, a2, a1); - break; +static const TCGOutOpBinary outop_eqv = { + .base.static_constraint = C_O1_I2(r, r, r), + .out_rrr = tgen_eqv, +}; - case INDEX_op_shl_i64: - case INDEX_op_shl_i32: - if (c2) { - tcg_out_shl(s, ext, a0, a1, a2); - } else { - tcg_out_insn(s, 3508, LSLV, ext, a0, a1, a2); - } - break; +static void tgen_extrh_i64_i32(TCGContext *s, TCGType t, TCGReg a0, TCGReg a1) +{ + tcg_out_ubfm(s, TCG_TYPE_I64, a0, a1, 32, 63); +} - case INDEX_op_shr_i64: - case INDEX_op_shr_i32: - if (c2) { - tcg_out_shr(s, ext, a0, a1, a2); - } else { - tcg_out_insn(s, 3508, LSRV, ext, a0, a1, a2); - } - break; +static const TCGOutOpUnary outop_extrh_i64_i32 = { + .base.static_constraint = C_O1_I1(r, r), + .out_rr = tgen_extrh_i64_i32, +}; - case INDEX_op_sar_i64: - case INDEX_op_sar_i32: - if (c2) { - tcg_out_sar(s, ext, a0, a1, a2); - } else { - tcg_out_insn(s, 3508, ASRV, ext, a0, a1, a2); - } - break; +static void tgen_mul(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3509, MADD, type, a0, a1, a2, TCG_REG_XZR); +} - case INDEX_op_rotr_i64: - case INDEX_op_rotr_i32: - if (c2) { - tcg_out_rotr(s, ext, a0, a1, a2); - } else { - tcg_out_insn(s, 3508, RORV, ext, a0, a1, a2); - } - break; +static const TCGOutOpBinary outop_mul = { + .base.static_constraint = C_O1_I2(r, r, r), + .out_rrr = tgen_mul, +}; - case INDEX_op_rotl_i64: - case INDEX_op_rotl_i32: - if (c2) { - tcg_out_rotl(s, ext, a0, a1, a2); - } else { - tcg_out_insn(s, 3502, SUB, 0, TCG_REG_TMP0, TCG_REG_XZR, a2); - tcg_out_insn(s, 3508, RORV, ext, a0, a1, TCG_REG_TMP0); - } - break; +static const TCGOutOpMul2 outop_muls2 = { + .base.static_constraint = C_NotImplemented, +}; - case INDEX_op_clz_i64: - case INDEX_op_clz_i32: - tcg_out_cltz(s, ext, a0, a1, a2, c2, false); - break; - case INDEX_op_ctz_i64: - case INDEX_op_ctz_i32: - tcg_out_cltz(s, ext, a0, a1, a2, c2, true); - break; +static TCGConstraintSetIndex cset_mulh(TCGType type, unsigned flags) +{ + return type == TCG_TYPE_I64 ? C_O1_I2(r, r, r) : C_NotImplemented; +} - case INDEX_op_brcond_i32: - a1 = (int32_t)a1; - /* FALLTHRU */ - case INDEX_op_brcond_i64: - tcg_out_brcond(s, ext, a2, a0, a1, const_args[1], arg_label(args[3])); - break; +static void tgen_mulsh(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3508, SMULH, TCG_TYPE_I64, a0, a1, a2); +} - case INDEX_op_setcond_i32: - a2 = (int32_t)a2; - /* FALLTHRU */ - case INDEX_op_setcond_i64: - tcg_out_cmp(s, ext, args[3], a1, a2, c2); - /* Use CSET alias of CSINC Wd, WZR, WZR, invert(cond). */ - tcg_out_insn(s, 3506, CSINC, TCG_TYPE_I32, a0, TCG_REG_XZR, - TCG_REG_XZR, tcg_invert_cond(args[3])); - break; +static const TCGOutOpBinary outop_mulsh = { + .base.static_constraint = C_Dynamic, + .base.dynamic_constraint = cset_mulh, + .out_rrr = tgen_mulsh, +}; - case INDEX_op_negsetcond_i32: - a2 = (int32_t)a2; - /* FALLTHRU */ - case INDEX_op_negsetcond_i64: - tcg_out_cmp(s, ext, args[3], a1, a2, c2); - /* Use CSETM alias of CSINV Wd, WZR, WZR, invert(cond). */ - tcg_out_insn(s, 3506, CSINV, ext, a0, TCG_REG_XZR, - TCG_REG_XZR, tcg_invert_cond(args[3])); - break; +static const TCGOutOpMul2 outop_mulu2 = { + .base.static_constraint = C_NotImplemented, +}; - case INDEX_op_movcond_i32: - a2 = (int32_t)a2; - /* FALLTHRU */ - case INDEX_op_movcond_i64: - tcg_out_cmp(s, ext, args[5], a1, a2, c2); - tcg_out_insn(s, 3506, CSEL, ext, a0, args[3], args[4], args[5]); - break; +static void tgen_muluh(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3508, UMULH, TCG_TYPE_I64, a0, a1, a2); +} - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_ld_i64: - tcg_out_qemu_ld(s, a0, a1, a2, ext); - break; - case INDEX_op_qemu_st_i32: - case INDEX_op_qemu_st_i64: - tcg_out_qemu_st(s, a0, a1, a2, ext); - break; - case INDEX_op_qemu_ld_i128: - tcg_out_qemu_ldst_i128(s, a0, a1, a2, args[3], true); - break; - case INDEX_op_qemu_st_i128: - tcg_out_qemu_ldst_i128(s, a0, a1, a2, args[3], false); - break; +static const TCGOutOpBinary outop_muluh = { + .base.static_constraint = C_Dynamic, + .base.dynamic_constraint = cset_mulh, + .out_rrr = tgen_muluh, +}; - case INDEX_op_bswap64_i64: - tcg_out_rev(s, TCG_TYPE_I64, MO_64, a0, a1); - break; - case INDEX_op_bswap32_i64: - tcg_out_rev(s, TCG_TYPE_I32, MO_32, a0, a1); - if (a2 & TCG_BSWAP_OS) { - tcg_out_ext32s(s, a0, a0); - } - break; - case INDEX_op_bswap32_i32: - tcg_out_rev(s, TCG_TYPE_I32, MO_32, a0, a1); - break; - case INDEX_op_bswap16_i64: - case INDEX_op_bswap16_i32: - tcg_out_rev(s, TCG_TYPE_I32, MO_16, a0, a1); - if (a2 & TCG_BSWAP_OS) { - /* Output must be sign-extended. */ - tcg_out_ext16s(s, ext, a0, a0); - } else if ((a2 & (TCG_BSWAP_IZ | TCG_BSWAP_OZ)) == TCG_BSWAP_OZ) { - /* Output must be zero-extended, but input isn't. */ - tcg_out_ext16u(s, a0, a0); - } - break; +static const TCGOutOpBinary outop_nand = { + .base.static_constraint = C_NotImplemented, +}; - case INDEX_op_deposit_i64: - case INDEX_op_deposit_i32: - tcg_out_dep(s, ext, a0, a2, args[3], args[4]); - break; +static const TCGOutOpBinary outop_nor = { + .base.static_constraint = C_NotImplemented, +}; - case INDEX_op_extract_i64: - case INDEX_op_extract_i32: - if (a2 == 0) { - uint64_t mask = MAKE_64BIT_MASK(0, args[3]); - tcg_out_logicali(s, I3404_ANDI, ext, a0, a1, mask); - } else { - tcg_out_ubfm(s, ext, a0, a1, a2, a2 + args[3] - 1); - } - break; +static void tgen_or(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3510, ORR, type, a0, a1, a2); +} - case INDEX_op_sextract_i64: - case INDEX_op_sextract_i32: - tcg_out_sbfm(s, ext, a0, a1, a2, a2 + args[3] - 1); - break; +static void tgen_ori(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + tcg_out_logicali(s, I3404_ORRI, type, a0, a1, a2); +} - case INDEX_op_extract2_i64: - case INDEX_op_extract2_i32: - tcg_out_extr(s, ext, a0, a2, a1, args[3]); - break; +static const TCGOutOpBinary outop_or = { + .base.static_constraint = C_O1_I2(r, r, rL), + .out_rrr = tgen_or, + .out_rri = tgen_ori, +}; - case INDEX_op_add2_i32: - tcg_out_addsub2(s, TCG_TYPE_I32, a0, a1, a2, args[3], - (int32_t)args[4], args[5], const_args[4], - const_args[5], false); - break; - case INDEX_op_add2_i64: - tcg_out_addsub2(s, TCG_TYPE_I64, a0, a1, a2, args[3], args[4], - args[5], const_args[4], const_args[5], false); - break; - case INDEX_op_sub2_i32: - tcg_out_addsub2(s, TCG_TYPE_I32, a0, a1, a2, args[3], - (int32_t)args[4], args[5], const_args[4], - const_args[5], true); - break; - case INDEX_op_sub2_i64: - tcg_out_addsub2(s, TCG_TYPE_I64, a0, a1, a2, args[3], args[4], - args[5], const_args[4], const_args[5], true); - break; +static void tgen_orc(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3510, ORN, type, a0, a1, a2); +} - case INDEX_op_muluh_i64: - tcg_out_insn(s, 3508, UMULH, TCG_TYPE_I64, a0, a1, a2); - break; - case INDEX_op_mulsh_i64: - tcg_out_insn(s, 3508, SMULH, TCG_TYPE_I64, a0, a1, a2); - break; +static const TCGOutOpBinary outop_orc = { + .base.static_constraint = C_O1_I2(r, r, r), + .out_rrr = tgen_orc, +}; - case INDEX_op_mb: - tcg_out_mb(s, a0); - break; +static void tgen_rems(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3508, SDIV, type, TCG_REG_TMP0, a1, a2); + tcg_out_insn(s, 3509, MSUB, type, a0, TCG_REG_TMP0, a2, a1); +} - case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ - case INDEX_op_mov_i64: - case INDEX_op_call: /* Always emitted via tcg_out_call. */ - case INDEX_op_exit_tb: /* Always emitted via tcg_out_exit_tb. */ - case INDEX_op_goto_tb: /* Always emitted via tcg_out_goto_tb. */ - case INDEX_op_ext8s_i32: /* Always emitted via tcg_reg_alloc_op. */ - case INDEX_op_ext8s_i64: - case INDEX_op_ext8u_i32: - case INDEX_op_ext8u_i64: - case INDEX_op_ext16s_i64: - case INDEX_op_ext16s_i32: - case INDEX_op_ext16u_i64: - case INDEX_op_ext16u_i32: - case INDEX_op_ext32s_i64: - case INDEX_op_ext32u_i64: - case INDEX_op_ext_i32_i64: - case INDEX_op_extu_i32_i64: - case INDEX_op_extrl_i64_i32: - default: - g_assert_not_reached(); +static const TCGOutOpBinary outop_rems = { + .base.static_constraint = C_O1_I2(r, r, r), + .out_rrr = tgen_rems, +}; + +static void tgen_remu(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3508, UDIV, type, TCG_REG_TMP0, a1, a2); + tcg_out_insn(s, 3509, MSUB, type, a0, TCG_REG_TMP0, a2, a1); +} + +static const TCGOutOpBinary outop_remu = { + .base.static_constraint = C_O1_I2(r, r, r), + .out_rrr = tgen_remu, +}; + +static const TCGOutOpBinary outop_rotl = { + .base.static_constraint = C_NotImplemented, +}; + +static void tgen_rotr(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3508, RORV, type, a0, a1, a2); +} + +static void tgen_rotri(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + int max = type == TCG_TYPE_I32 ? 31 : 63; + tcg_out_extr(s, type, a0, a1, a1, a2 & max); +} + +static const TCGOutOpBinary outop_rotr = { + .base.static_constraint = C_O1_I2(r, r, ri), + .out_rrr = tgen_rotr, + .out_rri = tgen_rotri, +}; + +static void tgen_sar(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3508, ASRV, type, a0, a1, a2); +} + +static void tgen_sari(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + int max = type == TCG_TYPE_I32 ? 31 : 63; + tcg_out_sbfm(s, type, a0, a1, a2 & max, max); +} + +static const TCGOutOpBinary outop_sar = { + .base.static_constraint = C_O1_I2(r, r, ri), + .out_rrr = tgen_sar, + .out_rri = tgen_sari, +}; + +static void tgen_shl(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3508, LSLV, type, a0, a1, a2); +} + +static void tgen_shli(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + int max = type == TCG_TYPE_I32 ? 31 : 63; + tcg_out_ubfm(s, type, a0, a1, -a2 & max, ~a2 & max); +} + +static const TCGOutOpBinary outop_shl = { + .base.static_constraint = C_O1_I2(r, r, ri), + .out_rrr = tgen_shl, + .out_rri = tgen_shli, +}; + +static void tgen_shr(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3508, LSRV, type, a0, a1, a2); +} + +static void tgen_shri(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + int max = type == TCG_TYPE_I32 ? 31 : 63; + tcg_out_ubfm(s, type, a0, a1, a2 & max, max); +} + +static const TCGOutOpBinary outop_shr = { + .base.static_constraint = C_O1_I2(r, r, ri), + .out_rrr = tgen_shr, + .out_rri = tgen_shri, +}; + +static void tgen_sub(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3502, SUB, type, a0, a1, a2); +} + +static const TCGOutOpSubtract outop_sub = { + .base.static_constraint = C_O1_I2(r, r, r), + .out_rrr = tgen_sub, +}; + +static void tgen_subbo_rrr(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3502, SUBS, type, a0, a1, a2); +} + +static void tgen_subbo_rri(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + if (a2 >= 0) { + tcg_out_insn(s, 3401, SUBSI, type, a0, a1, a2); + } else { + tcg_out_insn(s, 3401, ADDSI, type, a0, a1, -a2); + } +} + +static void tgen_subbo_rir(TCGContext *s, TCGType type, + TCGReg a0, tcg_target_long a1, TCGReg a2) +{ + tgen_subbo_rrr(s, type, a0, TCG_REG_XZR, a2); +} + +static void tgen_subbo_rii(TCGContext *s, TCGType type, + TCGReg a0, tcg_target_long a1, tcg_target_long a2) +{ + if (a2 == 0) { + tgen_subbo_rrr(s, type, a0, TCG_REG_XZR, TCG_REG_XZR); + return; } + + /* + * We want to allow a1 to be zero for the benefit of negation via + * subtraction. However, that leaves open the possibility of + * adding 0 +/- const, and the immediate add/sub instructions + * encode XSP not XZR. Since we have 0 - non-zero, borrow is + * always set. + */ + tcg_out_movi(s, type, a0, -a2); + tcg_out_set_borrow(s); } +static const TCGOutOpAddSubCarry outop_subbo = { + .base.static_constraint = C_O1_I2(r, rZ, rA), + .out_rrr = tgen_subbo_rrr, + .out_rri = tgen_subbo_rri, + .out_rir = tgen_subbo_rir, + .out_rii = tgen_subbo_rii, +}; + +static void tgen_subbi_rrr(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3503, SBC, type, a0, a1, a2); +} + +static void tgen_subbi_rri(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + tgen_addci_rri(s, type, a0, a1, ~a2); +} + +static const TCGOutOpAddSubCarry outop_subbi = { + .base.static_constraint = C_O1_I2(r, rz, rMZ), + .out_rrr = tgen_subbi_rrr, + .out_rri = tgen_subbi_rri, +}; + +static void tgen_subbio_rrr(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3503, SBCS, type, a0, a1, a2); +} + +static void tgen_subbio_rri(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + tgen_addcio_imm(s, type, a0, a1, ~a2); +} + +static const TCGOutOpAddSubCarry outop_subbio = { + .base.static_constraint = C_O1_I2(r, rz, rMZ), + .out_rrr = tgen_subbio_rrr, + .out_rri = tgen_subbio_rri, +}; + +static void tcg_out_set_borrow(TCGContext *s) +{ + tcg_out_insn(s, 3502, ADDS, TCG_TYPE_I32, + TCG_REG_XZR, TCG_REG_XZR, TCG_REG_XZR); +} + +static void tgen_xor(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tcg_out_insn(s, 3510, EOR, type, a0, a1, a2); +} + +static void tgen_xori(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + tcg_out_logicali(s, I3404_EORI, type, a0, a1, a2); +} + +static const TCGOutOpBinary outop_xor = { + .base.static_constraint = C_O1_I2(r, r, rL), + .out_rrr = tgen_xor, + .out_rri = tgen_xori, +}; + +static void tgen_bswap16(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, unsigned flags) +{ + tcg_out_rev(s, TCG_TYPE_I32, MO_16, a0, a1); + if (flags & TCG_BSWAP_OS) { + /* Output must be sign-extended. */ + tcg_out_ext16s(s, type, a0, a0); + } else if ((flags & (TCG_BSWAP_IZ | TCG_BSWAP_OZ)) == TCG_BSWAP_OZ) { + /* Output must be zero-extended, but input isn't. */ + tcg_out_ext16u(s, a0, a0); + } +} + +static const TCGOutOpBswap outop_bswap16 = { + .base.static_constraint = C_O1_I1(r, r), + .out_rr = tgen_bswap16, +}; + +static void tgen_bswap32(TCGContext *s, TCGType type, + TCGReg a0, TCGReg a1, unsigned flags) +{ + tcg_out_rev(s, TCG_TYPE_I32, MO_32, a0, a1); + if (flags & TCG_BSWAP_OS) { + tcg_out_ext32s(s, a0, a0); + } +} + +static const TCGOutOpBswap outop_bswap32 = { + .base.static_constraint = C_O1_I1(r, r), + .out_rr = tgen_bswap32, +}; + +static void tgen_bswap64(TCGContext *s, TCGType type, TCGReg a0, TCGReg a1) +{ + tcg_out_rev(s, TCG_TYPE_I64, MO_64, a0, a1); +} + +static const TCGOutOpUnary outop_bswap64 = { + .base.static_constraint = C_O1_I1(r, r), + .out_rr = tgen_bswap64, +}; + +static void tgen_neg(TCGContext *s, TCGType type, TCGReg a0, TCGReg a1) +{ + tgen_sub(s, type, a0, TCG_REG_XZR, a1); +} + +static const TCGOutOpUnary outop_neg = { + .base.static_constraint = C_O1_I1(r, r), + .out_rr = tgen_neg, +}; + +static void tgen_not(TCGContext *s, TCGType type, TCGReg a0, TCGReg a1) +{ + tgen_orc(s, type, a0, TCG_REG_XZR, a1); +} + +static const TCGOutOpUnary outop_not = { + .base.static_constraint = C_O1_I1(r, r), + .out_rr = tgen_not, +}; + +static void tgen_cset(TCGContext *s, TCGCond cond, TCGReg ret) +{ + /* Use CSET alias of CSINC Wd, WZR, WZR, invert(cond). */ + tcg_out_insn(s, 3506, CSINC, TCG_TYPE_I32, ret, TCG_REG_XZR, + TCG_REG_XZR, tcg_invert_cond(cond)); +} + +static void tgen_setcond(TCGContext *s, TCGType type, TCGCond cond, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tgen_cmp(s, type, cond, a1, a2); + tgen_cset(s, cond, a0); +} + +static void tgen_setcondi(TCGContext *s, TCGType type, TCGCond cond, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + tgen_cmpi(s, type, cond, a1, a2); + tgen_cset(s, cond, a0); +} + +static const TCGOutOpSetcond outop_setcond = { + .base.static_constraint = C_O1_I2(r, r, rC), + .out_rrr = tgen_setcond, + .out_rri = tgen_setcondi, +}; + +static void tgen_csetm(TCGContext *s, TCGType ext, TCGCond cond, TCGReg ret) +{ + /* Use CSETM alias of CSINV Wd, WZR, WZR, invert(cond). */ + tcg_out_insn(s, 3506, CSINV, ext, ret, TCG_REG_XZR, + TCG_REG_XZR, tcg_invert_cond(cond)); +} + +static void tgen_negsetcond(TCGContext *s, TCGType type, TCGCond cond, + TCGReg a0, TCGReg a1, TCGReg a2) +{ + tgen_cmp(s, type, cond, a1, a2); + tgen_csetm(s, type, cond, a0); +} + +static void tgen_negsetcondi(TCGContext *s, TCGType type, TCGCond cond, + TCGReg a0, TCGReg a1, tcg_target_long a2) +{ + tgen_cmpi(s, type, cond, a1, a2); + tgen_csetm(s, type, cond, a0); +} + +static const TCGOutOpSetcond outop_negsetcond = { + .base.static_constraint = C_O1_I2(r, r, rC), + .out_rrr = tgen_negsetcond, + .out_rri = tgen_negsetcondi, +}; + +static void tgen_movcond(TCGContext *s, TCGType type, TCGCond cond, + TCGReg ret, TCGReg c1, TCGArg c2, bool const_c2, + TCGArg vt, bool const_vt, TCGArg vf, bool const_vf) +{ + tcg_out_cmp(s, type, cond, c1, c2, const_c2); + tcg_out_insn(s, 3506, CSEL, type, ret, vt, vf, cond); +} + +static const TCGOutOpMovcond outop_movcond = { + .base.static_constraint = C_O1_I4(r, r, rC, rz, rz), + .out = tgen_movcond, +}; + +static void tgen_deposit(TCGContext *s, TCGType type, TCGReg a0, TCGReg a1, + TCGReg a2, unsigned ofs, unsigned len) +{ + unsigned mask = type == TCG_TYPE_I32 ? 31 : 63; + + /* + * Since we can't support "0Z" as a constraint, we allow a1 in + * any register. Fix things up as if a matching constraint. + */ + if (a0 != a1) { + if (a0 == a2) { + tcg_out_mov(s, type, TCG_REG_TMP0, a2); + a2 = TCG_REG_TMP0; + } + tcg_out_mov(s, type, a0, a1); + } + tcg_out_bfm(s, type, a0, a2, -ofs & mask, len - 1); +} + +static void tgen_depositi(TCGContext *s, TCGType type, TCGReg a0, TCGReg a1, + tcg_target_long a2, unsigned ofs, unsigned len) +{ + tgen_andi(s, type, a0, a1, ~MAKE_64BIT_MASK(ofs, len)); +} + +static void tgen_depositz(TCGContext *s, TCGType type, TCGReg a0, TCGReg a2, + unsigned ofs, unsigned len) +{ + int max = type == TCG_TYPE_I32 ? 31 : 63; + tcg_out_ubfm(s, type, a0, a2, -ofs & max, len - 1); +} + +static const TCGOutOpDeposit outop_deposit = { + .base.static_constraint = C_O1_I2(r, rZ, rZ), + .out_rrr = tgen_deposit, + .out_rri = tgen_depositi, + .out_rzr = tgen_depositz, +}; + +static void tgen_extract(TCGContext *s, TCGType type, TCGReg a0, TCGReg a1, + unsigned ofs, unsigned len) +{ + if (ofs == 0) { + uint64_t mask = MAKE_64BIT_MASK(0, len); + tcg_out_logicali(s, I3404_ANDI, type, a0, a1, mask); + } else { + tcg_out_ubfm(s, type, a0, a1, ofs, ofs + len - 1); + } +} + +static const TCGOutOpExtract outop_extract = { + .base.static_constraint = C_O1_I1(r, r), + .out_rr = tgen_extract, +}; + +static void tgen_sextract(TCGContext *s, TCGType type, TCGReg a0, TCGReg a1, + unsigned ofs, unsigned len) +{ + tcg_out_sbfm(s, type, a0, a1, ofs, ofs + len - 1); +} + +static const TCGOutOpExtract outop_sextract = { + .base.static_constraint = C_O1_I1(r, r), + .out_rr = tgen_sextract, +}; + +static void tgen_extract2(TCGContext *s, TCGType type, TCGReg a0, + TCGReg a1, TCGReg a2, unsigned shr) +{ + tcg_out_extr(s, type, a0, a2, a1, shr); +} + +static const TCGOutOpExtract2 outop_extract2 = { + .base.static_constraint = C_O1_I2(r, rz, rz), + .out_rrr = tgen_extract2, +}; + +static void tgen_ld8u(TCGContext *s, TCGType type, TCGReg dest, + TCGReg base, ptrdiff_t offset) +{ + tcg_out_ldst(s, I3312_LDRB, dest, base, offset, 0); +} + +static const TCGOutOpLoad outop_ld8u = { + .base.static_constraint = C_O1_I1(r, r), + .out = tgen_ld8u, +}; + +static void tgen_ld8s(TCGContext *s, TCGType type, TCGReg dest, + TCGReg base, ptrdiff_t offset) +{ + AArch64Insn insn = type == TCG_TYPE_I32 ? I3312_LDRSBW : I3312_LDRSBX; + tcg_out_ldst(s, insn, dest, base, offset, 0); +} + +static const TCGOutOpLoad outop_ld8s = { + .base.static_constraint = C_O1_I1(r, r), + .out = tgen_ld8s, +}; + +static void tgen_ld16u(TCGContext *s, TCGType type, TCGReg dest, + TCGReg base, ptrdiff_t offset) +{ + tcg_out_ldst(s, I3312_LDRH, dest, base, offset, 1); +} + +static const TCGOutOpLoad outop_ld16u = { + .base.static_constraint = C_O1_I1(r, r), + .out = tgen_ld16u, +}; + +static void tgen_ld16s(TCGContext *s, TCGType type, TCGReg dest, + TCGReg base, ptrdiff_t offset) +{ + AArch64Insn insn = type == TCG_TYPE_I32 ? I3312_LDRSHW : I3312_LDRSHX; + tcg_out_ldst(s, insn, dest, base, offset, 1); +} + +static const TCGOutOpLoad outop_ld16s = { + .base.static_constraint = C_O1_I1(r, r), + .out = tgen_ld16s, +}; + +static void tgen_ld32u(TCGContext *s, TCGType type, TCGReg dest, + TCGReg base, ptrdiff_t offset) +{ + tcg_out_ldst(s, I3312_LDRW, dest, base, offset, 2); +} + +static const TCGOutOpLoad outop_ld32u = { + .base.static_constraint = C_O1_I1(r, r), + .out = tgen_ld32u, +}; + +static void tgen_ld32s(TCGContext *s, TCGType type, TCGReg dest, + TCGReg base, ptrdiff_t offset) +{ + tcg_out_ldst(s, I3312_LDRSWX, dest, base, offset, 2); +} + +static const TCGOutOpLoad outop_ld32s = { + .base.static_constraint = C_O1_I1(r, r), + .out = tgen_ld32s, +}; + +static void tgen_st8_r(TCGContext *s, TCGType type, TCGReg data, + TCGReg base, ptrdiff_t offset) +{ + tcg_out_ldst(s, I3312_STRB, data, base, offset, 0); +} + +static const TCGOutOpStore outop_st8 = { + .base.static_constraint = C_O0_I2(rz, r), + .out_r = tgen_st8_r, +}; + +static void tgen_st16_r(TCGContext *s, TCGType type, TCGReg data, + TCGReg base, ptrdiff_t offset) +{ + tcg_out_ldst(s, I3312_STRH, data, base, offset, 1); +} + +static const TCGOutOpStore outop_st16 = { + .base.static_constraint = C_O0_I2(rz, r), + .out_r = tgen_st16_r, +}; + +static const TCGOutOpStore outop_st = { + .base.static_constraint = C_O0_I2(rz, r), + .out_r = tcg_out_st, +}; + static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc, unsigned vecl, unsigned vece, const TCGArg args[TCG_MAX_OP_ARGS], @@ -2955,148 +3337,6 @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op, TCGType type, unsigned flags) { switch (op) { - case INDEX_op_goto_ptr: - return C_O0_I1(r); - - case INDEX_op_ld8u_i32: - case INDEX_op_ld8s_i32: - case INDEX_op_ld16u_i32: - case INDEX_op_ld16s_i32: - case INDEX_op_ld_i32: - case INDEX_op_ld8u_i64: - case INDEX_op_ld8s_i64: - case INDEX_op_ld16u_i64: - case INDEX_op_ld16s_i64: - case INDEX_op_ld32u_i64: - case INDEX_op_ld32s_i64: - case INDEX_op_ld_i64: - case INDEX_op_neg_i32: - case INDEX_op_neg_i64: - case INDEX_op_not_i32: - case INDEX_op_not_i64: - case INDEX_op_bswap16_i32: - case INDEX_op_bswap32_i32: - case INDEX_op_bswap16_i64: - case INDEX_op_bswap32_i64: - case INDEX_op_bswap64_i64: - case INDEX_op_ext8s_i32: - case INDEX_op_ext16s_i32: - case INDEX_op_ext8u_i32: - case INDEX_op_ext16u_i32: - case INDEX_op_ext8s_i64: - case INDEX_op_ext16s_i64: - case INDEX_op_ext32s_i64: - case INDEX_op_ext8u_i64: - case INDEX_op_ext16u_i64: - case INDEX_op_ext32u_i64: - case INDEX_op_ext_i32_i64: - case INDEX_op_extu_i32_i64: - case INDEX_op_extract_i32: - case INDEX_op_extract_i64: - case INDEX_op_sextract_i32: - case INDEX_op_sextract_i64: - return C_O1_I1(r, r); - - case INDEX_op_st8_i32: - case INDEX_op_st16_i32: - case INDEX_op_st_i32: - case INDEX_op_st8_i64: - case INDEX_op_st16_i64: - case INDEX_op_st32_i64: - case INDEX_op_st_i64: - return C_O0_I2(rz, r); - - case INDEX_op_add_i32: - case INDEX_op_add_i64: - case INDEX_op_sub_i32: - case INDEX_op_sub_i64: - return C_O1_I2(r, r, rA); - - case INDEX_op_setcond_i32: - case INDEX_op_setcond_i64: - case INDEX_op_negsetcond_i32: - case INDEX_op_negsetcond_i64: - return C_O1_I2(r, r, rC); - - case INDEX_op_mul_i32: - case INDEX_op_mul_i64: - case INDEX_op_div_i32: - case INDEX_op_div_i64: - case INDEX_op_divu_i32: - case INDEX_op_divu_i64: - case INDEX_op_rem_i32: - case INDEX_op_rem_i64: - case INDEX_op_remu_i32: - case INDEX_op_remu_i64: - case INDEX_op_muluh_i64: - case INDEX_op_mulsh_i64: - return C_O1_I2(r, r, r); - - case INDEX_op_and_i32: - case INDEX_op_and_i64: - case INDEX_op_or_i32: - case INDEX_op_or_i64: - case INDEX_op_xor_i32: - case INDEX_op_xor_i64: - case INDEX_op_andc_i32: - case INDEX_op_andc_i64: - case INDEX_op_orc_i32: - case INDEX_op_orc_i64: - case INDEX_op_eqv_i32: - case INDEX_op_eqv_i64: - return C_O1_I2(r, r, rL); - - case INDEX_op_shl_i32: - case INDEX_op_shr_i32: - case INDEX_op_sar_i32: - case INDEX_op_rotl_i32: - case INDEX_op_rotr_i32: - case INDEX_op_shl_i64: - case INDEX_op_shr_i64: - case INDEX_op_sar_i64: - case INDEX_op_rotl_i64: - case INDEX_op_rotr_i64: - return C_O1_I2(r, r, ri); - - case INDEX_op_clz_i32: - case INDEX_op_ctz_i32: - case INDEX_op_clz_i64: - case INDEX_op_ctz_i64: - return C_O1_I2(r, r, rAL); - - case INDEX_op_brcond_i32: - case INDEX_op_brcond_i64: - return C_O0_I2(r, rC); - - case INDEX_op_movcond_i32: - case INDEX_op_movcond_i64: - return C_O1_I4(r, r, rC, rz, rz); - - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_ld_i64: - return C_O1_I1(r, r); - case INDEX_op_qemu_ld_i128: - return C_O2_I1(r, r, r); - case INDEX_op_qemu_st_i32: - case INDEX_op_qemu_st_i64: - return C_O0_I2(rz, r); - case INDEX_op_qemu_st_i128: - return C_O0_I3(rz, rz, r); - - case INDEX_op_deposit_i32: - case INDEX_op_deposit_i64: - return C_O1_I2(r, 0, rz); - - case INDEX_op_extract2_i32: - case INDEX_op_extract2_i64: - return C_O1_I2(r, rz, rz); - - case INDEX_op_add2_i32: - case INDEX_op_add2_i64: - case INDEX_op_sub2_i32: - case INDEX_op_sub2_i64: - return C_O2_I4(r, r, rz, rz, rA, rMZ); - case INDEX_op_add_vec: case INDEX_op_sub_vec: case INDEX_op_mul_vec: |