diff --git a/include/exec/exec-all.h b/include/exec/exec-all.h index f9ac332f9d..0766e24f8e 100644 --- a/include/exec/exec-all.h +++ b/include/exec/exec-all.h @@ -145,7 +145,7 @@ struct TranslationBlock { #define CF_COUNT_MASK 0x7fff #define CF_LAST_IO 0x8000 /* Last insn may be an IO access. */ - uint8_t *tc_ptr; /* pointer to the translated code */ + void *tc_ptr; /* pointer to the translated code */ /* next matching tb for physical address. */ struct TranslationBlock *phys_hash_next; /* first and second physical page containing code. The lower bit @@ -229,7 +229,7 @@ void ppc_tb_set_jmp_target(unsigned long jmp_addr, unsigned long addr); static inline void tb_set_jmp_target1(uintptr_t jmp_addr, uintptr_t addr) { /* patch the branch destination */ - *(uint32_t *)jmp_addr = addr - (jmp_addr + 4); + stl_p((void*)jmp_addr, addr - (jmp_addr + 4)); /* no need to flush icache explicitly */ } #elif defined(__aarch64__) diff --git a/tcg/aarch64/tcg-target.c b/tcg/aarch64/tcg-target.c index 0a580b65ea..77bb6d97de 100644 --- a/tcg/aarch64/tcg-target.c +++ b/tcg/aarch64/tcg-target.c @@ -63,40 +63,34 @@ static const int tcg_target_call_oarg_regs[1] = { # endif #endif -static inline void reloc_pc26(void *code_ptr, intptr_t target) +static inline void reloc_pc26(tcg_insn_unit *code_ptr, tcg_insn_unit *target) { - intptr_t offset = (target - (intptr_t)code_ptr) / 4; + ptrdiff_t offset = target - code_ptr; + assert(offset == sextract64(offset, 0, 26)); /* read instruction, mask away previous PC_REL26 parameter contents, set the proper offset, then write back the instruction. */ - uint32_t insn = *(uint32_t *)code_ptr; - insn = deposit32(insn, 0, 26, offset); - *(uint32_t *)code_ptr = insn; + *code_ptr = deposit32(*code_ptr, 0, 26, offset); } -static inline void reloc_pc19(void *code_ptr, intptr_t target) +static inline void reloc_pc19(tcg_insn_unit *code_ptr, tcg_insn_unit *target) { - intptr_t offset = (target - (intptr_t)code_ptr) / 4; - /* read instruction, mask away previous PC_REL19 parameter contents, - set the proper offset, then write back the instruction. */ - uint32_t insn = *(uint32_t *)code_ptr; - insn = deposit32(insn, 5, 19, offset); - *(uint32_t *)code_ptr = insn; + ptrdiff_t offset = target - code_ptr; + assert(offset == sextract64(offset, 0, 19)); + *code_ptr = deposit32(*code_ptr, 5, 19, offset); } -static inline void patch_reloc(uint8_t *code_ptr, int type, +static inline void patch_reloc(tcg_insn_unit *code_ptr, int type, intptr_t value, intptr_t addend) { - value += addend; - + assert(addend == 0); switch (type) { case R_AARCH64_JUMP26: case R_AARCH64_CALL26: - reloc_pc26(code_ptr, value); + reloc_pc26(code_ptr, (tcg_insn_unit *)value); break; case R_AARCH64_CONDBR19: - reloc_pc19(code_ptr, value); + reloc_pc19(code_ptr, (tcg_insn_unit *)value); break; - default: tcg_abort(); } @@ -794,15 +788,10 @@ static void tcg_out_cmp(TCGContext *s, TCGType ext, TCGReg a, } } -static inline void tcg_out_goto(TCGContext *s, intptr_t target) +static inline void tcg_out_goto(TCGContext *s, tcg_insn_unit *target) { - intptr_t offset = (target - (intptr_t)s->code_ptr) / 4; - - if (offset < -0x02000000 || offset >= 0x02000000) { - /* out of 26bit range */ - tcg_abort(); - } - + ptrdiff_t offset = target - s->code_ptr; + assert(offset == sextract64(offset, 0, 26)); tcg_out_insn(s, 3206, B, offset); } @@ -828,29 +817,23 @@ static inline void tcg_out_callr(TCGContext *s, TCGReg reg) tcg_out_insn(s, 3207, BLR, reg); } -static inline void tcg_out_call(TCGContext *s, intptr_t target) +static inline void tcg_out_call(TCGContext *s, tcg_insn_unit *target) { - intptr_t offset = (target - (intptr_t)s->code_ptr) / 4; - - if (offset < -0x02000000 || offset >= 0x02000000) { /* out of 26bit rng */ - tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_TMP, target); - tcg_out_callr(s, TCG_REG_TMP); - } else { + ptrdiff_t offset = target - s->code_ptr; + if (offset == sextract64(offset, 0, 26)) { tcg_out_insn(s, 3206, BL, offset); + } else { + tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_TMP, (intptr_t)target); + tcg_out_callr(s, TCG_REG_TMP); } } void aarch64_tb_set_jmp_target(uintptr_t jmp_addr, uintptr_t addr) { - intptr_t target = addr; - intptr_t offset = (target - (intptr_t)jmp_addr) / 4; + tcg_insn_unit *code_ptr = (tcg_insn_unit *)jmp_addr; + tcg_insn_unit *target = (tcg_insn_unit *)addr; - if (offset < -0x02000000 || offset >= 0x02000000) { - /* out of 26bit range */ - tcg_abort(); - } - - patch_reloc((uint8_t *)jmp_addr, R_AARCH64_JUMP26, target, 0); + reloc_pc26(code_ptr, target); flush_icache_range(jmp_addr, jmp_addr + 4); } @@ -862,7 +845,7 @@ static inline void tcg_out_goto_label(TCGContext *s, int label_index) tcg_out_reloc(s, s->code_ptr, R_AARCH64_JUMP26, label_index, 0); tcg_out_goto_noaddr(s); } else { - tcg_out_goto(s, l->u.value); + tcg_out_goto(s, l->u.value_ptr); } } @@ -884,9 +867,8 @@ static void tcg_out_brcond(TCGContext *s, TCGMemOp ext, TCGCond c, TCGArg a, tcg_out_reloc(s, s->code_ptr, R_AARCH64_CONDBR19, label, 0); offset = tcg_in32(s) >> 5; } else { - offset = l->u.value - (uintptr_t)s->code_ptr; - offset >>= 2; - assert(offset >= -0x40000 && offset < 0x40000); + offset = l->u.value_ptr - s->code_ptr; + assert(offset == sextract64(offset, 0, 19)); } if (need_cmp) { @@ -982,7 +964,7 @@ static inline void tcg_out_addsub2(TCGContext *s, int ext, TCGReg rl, /* helper signature: helper_ret_ld_mmu(CPUState *env, target_ulong addr, * int mmu_idx, uintptr_t ra) */ -static const void * const qemu_ld_helpers[16] = { +static void * const qemu_ld_helpers[16] = { [MO_UB] = helper_ret_ldub_mmu, [MO_LEUW] = helper_le_lduw_mmu, [MO_LEUL] = helper_le_ldul_mmu, @@ -995,7 +977,7 @@ static const void * const qemu_ld_helpers[16] = { /* helper signature: helper_ret_st_mmu(CPUState *env, target_ulong addr, * uintxx_t val, int mmu_idx, uintptr_t ra) */ -static const void * const qemu_st_helpers[16] = { +static void * const qemu_st_helpers[16] = { [MO_UB] = helper_ret_stb_mmu, [MO_LEUW] = helper_le_stw_mmu, [MO_LEUL] = helper_le_stl_mmu, @@ -1005,11 +987,11 @@ static const void * const qemu_st_helpers[16] = { [MO_BEQ] = helper_be_stq_mmu, }; -static inline void tcg_out_adr(TCGContext *s, TCGReg rd, uintptr_t addr) +static inline void tcg_out_adr(TCGContext *s, TCGReg rd, void *target) { - addr -= (uintptr_t)s->code_ptr; - assert(addr == sextract64(addr, 0, 21)); - tcg_out_insn(s, 3406, ADR, rd, addr); + ptrdiff_t offset = tcg_pcrel_diff(s, target); + assert(offset == sextract64(offset, 0, 21)); + tcg_out_insn(s, 3406, ADR, rd, offset); } static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) @@ -1017,20 +999,20 @@ static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) TCGMemOp opc = lb->opc; TCGMemOp size = opc & MO_SIZE; - reloc_pc19(lb->label_ptr[0], (intptr_t)s->code_ptr); + reloc_pc19(lb->label_ptr[0], s->code_ptr); tcg_out_mov(s, TCG_TYPE_I64, TCG_REG_X0, TCG_AREG0); tcg_out_mov(s, TARGET_LONG_BITS == 64, TCG_REG_X1, lb->addrlo_reg); tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_X2, lb->mem_index); - tcg_out_adr(s, TCG_REG_X3, (intptr_t)lb->raddr); - tcg_out_call(s, (intptr_t)qemu_ld_helpers[opc & ~MO_SIGN]); + tcg_out_adr(s, TCG_REG_X3, lb->raddr); + tcg_out_call(s, qemu_ld_helpers[opc & ~MO_SIGN]); if (opc & MO_SIGN) { tcg_out_sxt(s, TCG_TYPE_I64, size, lb->datalo_reg, TCG_REG_X0); } else { tcg_out_mov(s, size == MO_64, lb->datalo_reg, TCG_REG_X0); } - tcg_out_goto(s, (intptr_t)lb->raddr); + tcg_out_goto(s, lb->raddr); } static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) @@ -1038,21 +1020,21 @@ static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) TCGMemOp opc = lb->opc; TCGMemOp size = opc & MO_SIZE; - reloc_pc19(lb->label_ptr[0], (intptr_t)s->code_ptr); + reloc_pc19(lb->label_ptr[0], s->code_ptr); tcg_out_mov(s, TCG_TYPE_I64, TCG_REG_X0, TCG_AREG0); tcg_out_mov(s, TARGET_LONG_BITS == 64, TCG_REG_X1, lb->addrlo_reg); tcg_out_mov(s, size == MO_64, TCG_REG_X2, lb->datalo_reg); tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_X3, lb->mem_index); - tcg_out_adr(s, TCG_REG_X4, (intptr_t)lb->raddr); - tcg_out_call(s, (intptr_t)qemu_st_helpers[opc]); - tcg_out_goto(s, (intptr_t)lb->raddr); + tcg_out_adr(s, TCG_REG_X4, lb->raddr); + tcg_out_call(s, qemu_st_helpers[opc]); + tcg_out_goto(s, lb->raddr); } static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOp opc, TCGReg data_reg, TCGReg addr_reg, - int mem_index, - uint8_t *raddr, uint8_t *label_ptr) + int mem_index, tcg_insn_unit *raddr, + tcg_insn_unit *label_ptr) { TCGLabelQemuLdst *label = new_ldst_label(s); @@ -1070,7 +1052,8 @@ static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOp opc, the slow path. Generated code returns the host addend in X1, clobbers X0,X2,X3,TMP. */ static void tcg_out_tlb_read(TCGContext *s, TCGReg addr_reg, TCGMemOp s_bits, - uint8_t **label_ptr, int mem_index, bool is_read) + tcg_insn_unit **label_ptr, int mem_index, + bool is_read) { TCGReg base = TCG_AREG0; int tlb_offset = is_read ? @@ -1218,7 +1201,7 @@ static void tcg_out_qemu_ld(TCGContext *s, TCGReg data_reg, TCGReg addr_reg, { #ifdef CONFIG_SOFTMMU TCGMemOp s_bits = memop & MO_SIZE; - uint8_t *label_ptr; + tcg_insn_unit *label_ptr; tcg_out_tlb_read(s, addr_reg, s_bits, &label_ptr, mem_index, 1); tcg_out_qemu_ld_direct(s, memop, data_reg, addr_reg, TCG_REG_X1); @@ -1235,7 +1218,7 @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg data_reg, TCGReg addr_reg, { #ifdef CONFIG_SOFTMMU TCGMemOp s_bits = memop & MO_SIZE; - uint8_t *label_ptr; + tcg_insn_unit *label_ptr; tcg_out_tlb_read(s, addr_reg, s_bits, &label_ptr, mem_index, 0); tcg_out_qemu_st_direct(s, memop, data_reg, addr_reg, TCG_REG_X1); @@ -1247,7 +1230,7 @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg data_reg, TCGReg addr_reg, #endif /* CONFIG_SOFTMMU */ } -static uint8_t *tb_ret_addr; +static tcg_insn_unit *tb_ret_addr; static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg args[TCG_MAX_OP_ARGS], @@ -1270,7 +1253,7 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, switch (opc) { case INDEX_op_exit_tb: tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_X0, a0); - tcg_out_goto(s, (intptr_t)tb_ret_addr); + tcg_out_goto(s, tb_ret_addr); break; case INDEX_op_goto_tb: @@ -1278,19 +1261,11 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, #error "USE_DIRECT_JUMP required for aarch64" #endif assert(s->tb_jmp_offset != NULL); /* consistency for USE_DIRECT_JUMP */ - s->tb_jmp_offset[a0] = s->code_ptr - s->code_buf; + s->tb_jmp_offset[a0] = tcg_current_code_size(s); /* actual branch destination will be patched by aarch64_tb_set_jmp_target later, beware retranslation. */ tcg_out_goto_noaddr(s); - s->tb_next_offset[a0] = s->code_ptr - s->code_buf; - break; - - case INDEX_op_call: - if (const_args[0]) { - tcg_out_call(s, a0); - } else { - tcg_out_callr(s, a0); - } + s->tb_next_offset[a0] = tcg_current_code_size(s); break; case INDEX_op_br: @@ -1613,13 +1588,12 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_insn(s, 3508, SMULH, TCG_TYPE_I64, a0, a1, a2); break; + case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_mov_i32: + case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ case INDEX_op_movi_i64: - case INDEX_op_movi_i32: - /* Always implemented with tcg_out_mov/i, never with tcg_out_op. */ + case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: - /* Opcode not implemented. */ tcg_abort(); } @@ -1629,15 +1603,8 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, static const TCGTargetOpDef aarch64_op_defs[] = { { INDEX_op_exit_tb, { } }, { INDEX_op_goto_tb, { } }, - { INDEX_op_call, { "ri" } }, { INDEX_op_br, { } }, - { INDEX_op_mov_i32, { "r", "r" } }, - { INDEX_op_mov_i64, { "r", "r" } }, - - { INDEX_op_movi_i32, { "r" } }, - { INDEX_op_movi_i64, { "r" } }, - { INDEX_op_ld8u_i32, { "r", "r" } }, { INDEX_op_ld8s_i32, { "r", "r" } }, { INDEX_op_ld16u_i32, { "r", "r" } }, diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h index a1d4322165..a32aea66fa 100644 --- a/tcg/aarch64/tcg-target.h +++ b/tcg/aarch64/tcg-target.h @@ -13,6 +13,7 @@ #ifndef TCG_TARGET_AARCH64 #define TCG_TARGET_AARCH64 1 +#define TCG_TARGET_INSN_UNIT_SIZE 4 #undef TCG_TARGET_STACK_GROWSUP typedef enum { diff --git a/tcg/arm/tcg-target.c b/tcg/arm/tcg-target.c index 7535175f9c..538ca2aed0 100644 --- a/tcg/arm/tcg-target.c +++ b/tcg/arm/tcg-target.c @@ -115,36 +115,18 @@ static const int tcg_target_call_oarg_regs[2] = { #define TCG_REG_TMP TCG_REG_R12 -static inline void reloc_abs32(void *code_ptr, intptr_t target) +static inline void reloc_pc24(tcg_insn_unit *code_ptr, tcg_insn_unit *target) { - *(uint32_t *) code_ptr = target; + ptrdiff_t offset = (tcg_ptr_byte_diff(target, code_ptr) - 8) >> 2; + *code_ptr = (*code_ptr & ~0xffffff) | (offset & 0xffffff); } -static inline void reloc_pc24(void *code_ptr, intptr_t target) -{ - uint32_t offset = ((target - ((intptr_t)code_ptr + 8)) >> 2); - - *(uint32_t *) code_ptr = ((*(uint32_t *) code_ptr) & ~0xffffff) - | (offset & 0xffffff); -} - -static void patch_reloc(uint8_t *code_ptr, int type, +static void patch_reloc(tcg_insn_unit *code_ptr, int type, intptr_t value, intptr_t addend) { - switch (type) { - case R_ARM_ABS32: - reloc_abs32(code_ptr, value); - break; - - case R_ARM_CALL: - case R_ARM_JUMP24: - default: - tcg_abort(); - - case R_ARM_PC24: - reloc_pc24(code_ptr, value); - break; - } + assert(type == R_ARM_PC24); + assert(addend == 0); + reloc_pc24(code_ptr, (tcg_insn_unit *)value); } #define TCG_CT_CONST_ARM 0x100 @@ -379,20 +361,18 @@ static inline void tcg_out_b(TCGContext *s, int cond, int32_t offset) static inline void tcg_out_b_noaddr(TCGContext *s, int cond) { - /* We pay attention here to not modify the branch target by skipping - the corresponding bytes. This ensure that caches and memory are + /* We pay attention here to not modify the branch target by masking + the corresponding bytes. This ensure that caches and memory are kept coherent during retranslation. */ - s->code_ptr += 3; - tcg_out8(s, (cond << 4) | 0x0a); + tcg_out32(s, deposit32(*s->code_ptr, 24, 8, (cond << 4) | 0x0a)); } static inline void tcg_out_bl_noaddr(TCGContext *s, int cond) { - /* We pay attention here to not modify the branch target by skipping - the corresponding bytes. This ensure that caches and memory are + /* We pay attention here to not modify the branch target by masking + the corresponding bytes. This ensure that caches and memory are kept coherent during retranslation. */ - s->code_ptr += 3; - tcg_out8(s, (cond << 4) | 0x0b); + tcg_out32(s, deposit32(*s->code_ptr, 24, 8, (cond << 4) | 0x0b)); } static inline void tcg_out_bl(TCGContext *s, int cond, int32_t offset) @@ -1010,20 +990,21 @@ static inline void tcg_out_st8(TCGContext *s, int cond, * with the code buffer limited to 16MB we wouldn't need the long case. * But we also use it for the tail-call to the qemu_ld/st helpers, which does. */ -static inline void tcg_out_goto(TCGContext *s, int cond, uint32_t addr) +static inline void tcg_out_goto(TCGContext *s, int cond, tcg_insn_unit *addr) { - int32_t disp = addr - (tcg_target_long) s->code_ptr; + intptr_t addri = (intptr_t)addr; + ptrdiff_t disp = tcg_pcrel_diff(s, addr); - if ((addr & 1) == 0 && disp - 8 < 0x01fffffd && disp - 8 > -0x01fffffd) { + if ((addri & 1) == 0 && disp - 8 < 0x01fffffd && disp - 8 > -0x01fffffd) { tcg_out_b(s, cond, disp); return; } - tcg_out_movi32(s, cond, TCG_REG_TMP, addr); + tcg_out_movi32(s, cond, TCG_REG_TMP, addri); if (use_armv5t_instructions) { tcg_out_bx(s, cond, TCG_REG_TMP); } else { - if (addr & 1) { + if (addri & 1) { tcg_abort(); } tcg_out_mov_reg(s, cond, TCG_REG_PC, TCG_REG_TMP); @@ -1032,39 +1013,28 @@ static inline void tcg_out_goto(TCGContext *s, int cond, uint32_t addr) /* The call case is mostly used for helpers - so it's not unreasonable * for them to be beyond branch range */ -static inline void tcg_out_call(TCGContext *s, uint32_t addr) +static void tcg_out_call(TCGContext *s, tcg_insn_unit *addr) { - int32_t val; + intptr_t addri = (intptr_t)addr; + ptrdiff_t disp = tcg_pcrel_diff(s, addr); - val = addr - (tcg_target_long) s->code_ptr; - if (val - 8 < 0x02000000 && val - 8 >= -0x02000000) { - if (addr & 1) { + if (disp - 8 < 0x02000000 && disp - 8 >= -0x02000000) { + if (addri & 1) { /* Use BLX if the target is in Thumb mode */ if (!use_armv5t_instructions) { tcg_abort(); } - tcg_out_blx_imm(s, val); + tcg_out_blx_imm(s, disp); } else { - tcg_out_bl(s, COND_AL, val); + tcg_out_bl(s, COND_AL, disp); } } else if (use_armv7_instructions) { - tcg_out_movi32(s, COND_AL, TCG_REG_TMP, addr); + tcg_out_movi32(s, COND_AL, TCG_REG_TMP, addri); tcg_out_blx(s, COND_AL, TCG_REG_TMP); } else { tcg_out_dat_imm(s, COND_AL, ARITH_ADD, TCG_REG_R14, TCG_REG_PC, 4); tcg_out_ld32_12(s, COND_AL, TCG_REG_PC, TCG_REG_PC, -4); - tcg_out32(s, addr); - } -} - -static inline void tcg_out_callr(TCGContext *s, int cond, int arg) -{ - if (use_armv5t_instructions) { - tcg_out_blx(s, cond, arg); - } else { - tcg_out_dat_reg(s, cond, ARITH_MOV, TCG_REG_R14, 0, - TCG_REG_PC, SHIFT_IMM_LSL(0)); - tcg_out_bx(s, cond, arg); + tcg_out32(s, addri); } } @@ -1073,9 +1043,9 @@ static inline void tcg_out_goto_label(TCGContext *s, int cond, int label_index) TCGLabel *l = &s->labels[label_index]; if (l->has_value) { - tcg_out_goto(s, cond, l->u.value); + tcg_out_goto(s, cond, l->u.value_ptr); } else { - tcg_out_reloc(s, s->code_ptr, R_ARM_PC24, label_index, 31337); + tcg_out_reloc(s, s->code_ptr, R_ARM_PC24, label_index, 0); tcg_out_b_noaddr(s, cond); } } @@ -1084,7 +1054,7 @@ static inline void tcg_out_goto_label(TCGContext *s, int cond, int label_index) /* helper signature: helper_ret_ld_mmu(CPUState *env, target_ulong addr, * int mmu_idx, uintptr_t ra) */ -static const void * const qemu_ld_helpers[16] = { +static void * const qemu_ld_helpers[16] = { [MO_UB] = helper_ret_ldub_mmu, [MO_SB] = helper_ret_ldsb_mmu, @@ -1104,7 +1074,7 @@ static const void * const qemu_ld_helpers[16] = { /* helper signature: helper_ret_st_mmu(CPUState *env, target_ulong addr, * uintxx_t val, int mmu_idx, uintptr_t ra) */ -static const void * const qemu_st_helpers[16] = { +static void * const qemu_st_helpers[16] = { [MO_UB] = helper_ret_stb_mmu, [MO_LEUW] = helper_le_stw_mmu, [MO_LEUL] = helper_le_stl_mmu, @@ -1256,7 +1226,7 @@ static TCGReg tcg_out_tlb_read(TCGContext *s, TCGReg addrlo, TCGReg addrhi, static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOp opc, TCGReg datalo, TCGReg datahi, TCGReg addrlo, TCGReg addrhi, int mem_index, - uint8_t *raddr, uint8_t *label_ptr) + tcg_insn_unit *raddr, tcg_insn_unit *label_ptr) { TCGLabelQemuLdst *label = new_ldst_label(s); @@ -1275,9 +1245,9 @@ static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) { TCGReg argreg, datalo, datahi; TCGMemOp opc = lb->opc; - uintptr_t func; + void *func; - reloc_pc24(lb->label_ptr[0], (tcg_target_long)s->code_ptr); + reloc_pc24(lb->label_ptr[0], s->code_ptr); argreg = tcg_out_arg_reg32(s, TCG_REG_R0, TCG_AREG0); if (TARGET_LONG_BITS == 64) { @@ -1292,9 +1262,9 @@ static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) icache usage. For pre-armv6, use the signed helpers since we do not have a single insn sign-extend. */ if (use_armv6_instructions) { - func = (uintptr_t)qemu_ld_helpers[opc & ~MO_SIGN]; + func = qemu_ld_helpers[opc & ~MO_SIGN]; } else { - func = (uintptr_t)qemu_ld_helpers[opc]; + func = qemu_ld_helpers[opc]; if (opc & MO_SIGN) { opc = MO_UL; } @@ -1328,7 +1298,7 @@ static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) break; } - tcg_out_goto(s, COND_AL, (tcg_target_long)lb->raddr); + tcg_out_goto(s, COND_AL, lb->raddr); } static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) @@ -1336,7 +1306,7 @@ static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) TCGReg argreg, datalo, datahi; TCGMemOp opc = lb->opc; - reloc_pc24(lb->label_ptr[0], (tcg_target_long)s->code_ptr); + reloc_pc24(lb->label_ptr[0], s->code_ptr); argreg = TCG_REG_R0; argreg = tcg_out_arg_reg32(s, argreg, TCG_AREG0); @@ -1368,7 +1338,7 @@ static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) argreg = tcg_out_arg_reg32(s, argreg, TCG_REG_R14); /* Tail-call to the helper, which will return to the fast path. */ - tcg_out_goto(s, COND_AL, (uintptr_t)qemu_st_helpers[opc]); + tcg_out_goto(s, COND_AL, qemu_st_helpers[opc]); } #endif /* SOFTMMU */ @@ -1499,7 +1469,7 @@ static void tcg_out_qemu_ld(TCGContext *s, const TCGArg *args, bool is64) #ifdef CONFIG_SOFTMMU int mem_index; TCGReg addend; - uint8_t *label_ptr; + tcg_insn_unit *label_ptr; #endif datalo = *args++; @@ -1628,7 +1598,7 @@ static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args, bool is64) #ifdef CONFIG_SOFTMMU int mem_index; TCGReg addend; - uint8_t *label_ptr; + tcg_insn_unit *label_ptr; #endif datalo = *args++; @@ -1660,7 +1630,7 @@ static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args, bool is64) #endif } -static uint8_t *tb_ret_addr; +static tcg_insn_unit *tb_ret_addr; static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, const int *const_args) @@ -1670,51 +1640,21 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, switch (opc) { case INDEX_op_exit_tb: - if (use_armv7_instructions || check_fit_imm(args[0])) { - tcg_out_movi32(s, COND_AL, TCG_REG_R0, args[0]); - tcg_out_goto(s, COND_AL, (tcg_target_ulong) tb_ret_addr); - } else { - uint8_t *ld_ptr = s->code_ptr; - tcg_out_ld32_12(s, COND_AL, TCG_REG_R0, TCG_REG_PC, 0); - tcg_out_goto(s, COND_AL, (tcg_target_ulong) tb_ret_addr); - *ld_ptr = (uint8_t) (s->code_ptr - ld_ptr) - 8; - tcg_out32(s, args[0]); - } + tcg_out_movi32(s, COND_AL, TCG_REG_R0, args[0]); + tcg_out_goto(s, COND_AL, tb_ret_addr); break; case INDEX_op_goto_tb: if (s->tb_jmp_offset) { /* Direct jump method */ -#if defined(USE_DIRECT_JUMP) - s->tb_jmp_offset[args[0]] = s->code_ptr - s->code_buf; + s->tb_jmp_offset[args[0]] = tcg_current_code_size(s); tcg_out_b_noaddr(s, COND_AL); -#else - tcg_out_ld32_12(s, COND_AL, TCG_REG_PC, TCG_REG_PC, -4); - s->tb_jmp_offset[args[0]] = s->code_ptr - s->code_buf; - tcg_out32(s, 0); -#endif } else { /* Indirect jump method */ -#if 1 - c = (int) (s->tb_next + args[0]) - ((int) s->code_ptr + 8); - if (c > 0xfff || c < -0xfff) { - tcg_out_movi32(s, COND_AL, TCG_REG_R0, - (tcg_target_long) (s->tb_next + args[0])); - tcg_out_ld32_12(s, COND_AL, TCG_REG_PC, TCG_REG_R0, 0); - } else - tcg_out_ld32_12(s, COND_AL, TCG_REG_PC, TCG_REG_PC, c); -#else - tcg_out_ld32_12(s, COND_AL, TCG_REG_R0, TCG_REG_PC, 0); - tcg_out_ld32_12(s, COND_AL, TCG_REG_PC, TCG_REG_R0, 0); - tcg_out32(s, (tcg_target_long) (s->tb_next + args[0])); -#endif + intptr_t ptr = (intptr_t)(s->tb_next + args[0]); + tcg_out_movi32(s, COND_AL, TCG_REG_R0, ptr & ~0xfff); + tcg_out_ld32_12(s, COND_AL, TCG_REG_PC, TCG_REG_R0, ptr & 0xfff); } - s->tb_next_offset[args[0]] = s->code_ptr - s->code_buf; - break; - case INDEX_op_call: - if (const_args[0]) - tcg_out_call(s, args[0]); - else - tcg_out_callr(s, COND_AL, args[0]); + s->tb_next_offset[args[0]] = tcg_current_code_size(s); break; case INDEX_op_br: tcg_out_goto_label(s, COND_AL, args[0]); @@ -1745,13 +1685,6 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_st32(s, COND_AL, args[0], args[1], args[2]); break; - case INDEX_op_mov_i32: - tcg_out_dat_reg(s, COND_AL, ARITH_MOV, - args[0], 0, args[1], SHIFT_IMM_LSL(0)); - break; - case INDEX_op_movi_i32: - tcg_out_movi32(s, COND_AL, args[0], args[1]); - break; case INDEX_op_movcond_i32: /* Constraints mean that v2 is always in the same register as dest, * so we only need to do "if condition passed, move v1 to dest". @@ -1967,6 +1900,9 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_udiv(s, COND_AL, args[0], args[1], args[2]); break; + case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ + case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ + case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); } @@ -1975,12 +1911,8 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, static const TCGTargetOpDef arm_op_defs[] = { { INDEX_op_exit_tb, { } }, { INDEX_op_goto_tb, { } }, - { INDEX_op_call, { "ri" } }, { INDEX_op_br, { } }, - { INDEX_op_mov_i32, { "r", "r" } }, - { INDEX_op_movi_i32, { "r" } }, - { INDEX_op_ld8u_i32, { "r", "r" } }, { INDEX_op_ld8s_i32, { "r", "r" } }, { INDEX_op_ld16u_i32, { "r", "r" } }, diff --git a/tcg/arm/tcg-target.h b/tcg/arm/tcg-target.h index 1bc5dacec0..73f10c4424 100644 --- a/tcg/arm/tcg-target.h +++ b/tcg/arm/tcg-target.h @@ -26,6 +26,7 @@ #define TCG_TARGET_ARM 1 #undef TCG_TARGET_STACK_GROWSUP +#define TCG_TARGET_INSN_UNIT_SIZE 4 typedef enum { TCG_REG_R0 = 0, diff --git a/tcg/i386/tcg-target.c b/tcg/i386/tcg-target.c index 34ece1f80b..a373073ff8 100644 --- a/tcg/i386/tcg-target.c +++ b/tcg/i386/tcg-target.c @@ -139,9 +139,9 @@ static bool have_bmi2; # define have_bmi2 0 #endif -static uint8_t *tb_ret_addr; +static tcg_insn_unit *tb_ret_addr; -static void patch_reloc(uint8_t *code_ptr, int type, +static void patch_reloc(tcg_insn_unit *code_ptr, int type, intptr_t value, intptr_t addend) { value += addend; @@ -151,14 +151,14 @@ static void patch_reloc(uint8_t *code_ptr, int type, if (value != (int32_t)value) { tcg_abort(); } - *(uint32_t *)code_ptr = value; + tcg_patch32(code_ptr, value); break; case R_386_PC8: value -= (uintptr_t)code_ptr; if (value != (int8_t)value) { tcg_abort(); } - *(uint8_t *)code_ptr = value; + tcg_patch8(code_ptr, value); break; default: tcg_abort(); @@ -859,7 +859,7 @@ static void tcg_out_jxx(TCGContext *s, int opc, int label_index, int small) TCGLabel *l = &s->labels[label_index]; if (l->has_value) { - val = l->u.value - (intptr_t)s->code_ptr; + val = tcg_pcrel_diff(s, l->u.value_ptr); val1 = val - 2; if ((int8_t)val1 == val1) { if (opc == -1) { @@ -1099,26 +1099,26 @@ static void tcg_out_movcond64(TCGContext *s, TCGCond cond, TCGArg dest, } #endif -static void tcg_out_branch(TCGContext *s, int call, uintptr_t dest) +static void tcg_out_branch(TCGContext *s, int call, tcg_insn_unit *dest) { - intptr_t disp = dest - (intptr_t)s->code_ptr - 5; + intptr_t disp = tcg_pcrel_diff(s, dest) - 5; if (disp == (int32_t)disp) { tcg_out_opc(s, call ? OPC_CALL_Jz : OPC_JMP_long, 0, 0, 0); tcg_out32(s, disp); } else { - tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R10, dest); + tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R10, (uintptr_t)dest); tcg_out_modrm(s, OPC_GRP5, call ? EXT5_CALLN_Ev : EXT5_JMPN_Ev, TCG_REG_R10); } } -static inline void tcg_out_calli(TCGContext *s, uintptr_t dest) +static inline void tcg_out_call(TCGContext *s, tcg_insn_unit *dest) { tcg_out_branch(s, 1, dest); } -static void tcg_out_jmp(TCGContext *s, uintptr_t dest) +static void tcg_out_jmp(TCGContext *s, tcg_insn_unit *dest) { tcg_out_branch(s, 0, dest); } @@ -1127,7 +1127,7 @@ static void tcg_out_jmp(TCGContext *s, uintptr_t dest) /* helper signature: helper_ret_ld_mmu(CPUState *env, target_ulong addr, * int mmu_idx, uintptr_t ra) */ -static const void * const qemu_ld_helpers[16] = { +static void * const qemu_ld_helpers[16] = { [MO_UB] = helper_ret_ldub_mmu, [MO_LEUW] = helper_le_lduw_mmu, [MO_LEUL] = helper_le_ldul_mmu, @@ -1140,7 +1140,7 @@ static const void * const qemu_ld_helpers[16] = { /* helper signature: helper_ret_st_mmu(CPUState *env, target_ulong addr, * uintxx_t val, int mmu_idx, uintptr_t ra) */ -static const void * const qemu_st_helpers[16] = { +static void * const qemu_st_helpers[16] = { [MO_UB] = helper_ret_stb_mmu, [MO_LEUW] = helper_le_stw_mmu, [MO_LEUL] = helper_le_stl_mmu, @@ -1173,7 +1173,7 @@ static const void * const qemu_st_helpers[16] = { static inline void tcg_out_tlb_load(TCGContext *s, TCGReg addrlo, TCGReg addrhi, int mem_index, TCGMemOp s_bits, - uint8_t **label_ptr, int which) + tcg_insn_unit **label_ptr, int which) { const TCGReg r0 = TCG_REG_L0; const TCGReg r1 = TCG_REG_L1; @@ -1247,8 +1247,8 @@ static inline void tcg_out_tlb_load(TCGContext *s, TCGReg addrlo, TCGReg addrhi, static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOp opc, TCGReg datalo, TCGReg datahi, TCGReg addrlo, TCGReg addrhi, - int mem_index, uint8_t *raddr, - uint8_t **label_ptr) + int mem_index, tcg_insn_unit *raddr, + tcg_insn_unit **label_ptr) { TCGLabelQemuLdst *label = new_ldst_label(s); @@ -1273,12 +1273,12 @@ static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) { TCGMemOp opc = l->opc; TCGReg data_reg; - uint8_t **label_ptr = &l->label_ptr[0]; + tcg_insn_unit **label_ptr = &l->label_ptr[0]; /* resolve label address */ - *(uint32_t *)label_ptr[0] = (uint32_t)(s->code_ptr - label_ptr[0] - 4); + tcg_patch32(label_ptr[0], s->code_ptr - label_ptr[0] - 4); if (TARGET_LONG_BITS > TCG_TARGET_REG_BITS) { - *(uint32_t *)label_ptr[1] = (uint32_t)(s->code_ptr - label_ptr[1] - 4); + tcg_patch32(label_ptr[1], s->code_ptr - label_ptr[1] - 4); } if (TCG_TARGET_REG_BITS == 32) { @@ -1308,7 +1308,7 @@ static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) (uintptr_t)l->raddr); } - tcg_out_calli(s, (uintptr_t)qemu_ld_helpers[opc & ~MO_SIGN]); + tcg_out_call(s, qemu_ld_helpers[opc & ~MO_SIGN]); data_reg = l->datalo_reg; switch (opc & MO_SSIZE) { @@ -1346,7 +1346,7 @@ static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) } /* Jump to the code corresponding to next IR of qemu_st */ - tcg_out_jmp(s, (uintptr_t)l->raddr); + tcg_out_jmp(s, l->raddr); } /* @@ -1356,13 +1356,13 @@ static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) { TCGMemOp opc = l->opc; TCGMemOp s_bits = opc & MO_SIZE; - uint8_t **label_ptr = &l->label_ptr[0]; + tcg_insn_unit **label_ptr = &l->label_ptr[0]; TCGReg retaddr; /* resolve label address */ - *(uint32_t *)label_ptr[0] = (uint32_t)(s->code_ptr - label_ptr[0] - 4); + tcg_patch32(label_ptr[0], s->code_ptr - label_ptr[0] - 4); if (TARGET_LONG_BITS > TCG_TARGET_REG_BITS) { - *(uint32_t *)label_ptr[1] = (uint32_t)(s->code_ptr - label_ptr[1] - 4); + tcg_patch32(label_ptr[1], s->code_ptr - label_ptr[1] - 4); } if (TCG_TARGET_REG_BITS == 32) { @@ -1413,7 +1413,7 @@ static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) /* "Tail call" to the helper, with the return address back inline. */ tcg_out_push(s, retaddr); - tcg_out_jmp(s, (uintptr_t)qemu_st_helpers[opc]); + tcg_out_jmp(s, qemu_st_helpers[opc]); } #elif defined(__x86_64__) && defined(__linux__) # include @@ -1534,7 +1534,7 @@ static void tcg_out_qemu_ld(TCGContext *s, const TCGArg *args, bool is64) #if defined(CONFIG_SOFTMMU) int mem_index; TCGMemOp s_bits; - uint8_t *label_ptr[2]; + tcg_insn_unit *label_ptr[2]; #endif datalo = *args++; @@ -1665,7 +1665,7 @@ static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args, bool is64) #if defined(CONFIG_SOFTMMU) int mem_index; TCGMemOp s_bits; - uint8_t *label_ptr[2]; + tcg_insn_unit *label_ptr[2]; #endif datalo = *args++; @@ -1731,35 +1731,24 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, switch(opc) { case INDEX_op_exit_tb: tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_EAX, args[0]); - tcg_out_jmp(s, (uintptr_t)tb_ret_addr); + tcg_out_jmp(s, tb_ret_addr); break; case INDEX_op_goto_tb: if (s->tb_jmp_offset) { /* direct jump method */ tcg_out8(s, OPC_JMP_long); /* jmp im */ - s->tb_jmp_offset[args[0]] = s->code_ptr - s->code_buf; + s->tb_jmp_offset[args[0]] = tcg_current_code_size(s); tcg_out32(s, 0); } else { /* indirect jump method */ tcg_out_modrm_offset(s, OPC_GRP5, EXT5_JMPN_Ev, -1, (intptr_t)(s->tb_next + args[0])); } - s->tb_next_offset[args[0]] = s->code_ptr - s->code_buf; - break; - case INDEX_op_call: - if (const_args[0]) { - tcg_out_calli(s, args[0]); - } else { - /* call *reg */ - tcg_out_modrm(s, OPC_GRP5, EXT5_CALLN_Ev, args[0]); - } + s->tb_next_offset[args[0]] = tcg_current_code_size(s); break; case INDEX_op_br: tcg_out_jxx(s, JCC_JMP, args[0], 0); break; - case INDEX_op_movi_i32: - tcg_out_movi(s, TCG_TYPE_I32, args[0], args[1]); - break; OP_32_64(ld8u): /* Note that we can ignore REXW for the zero-extend to 64-bit. */ tcg_out_modrm_offset(s, OPC_MOVZBL, args[0], args[1], args[2]); @@ -2009,9 +1998,6 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_setcond2(s, args, const_args); break; #else /* TCG_TARGET_REG_BITS == 64 */ - case INDEX_op_movi_i64: - tcg_out_movi(s, TCG_TYPE_I64, args[0], args[1]); - break; case INDEX_op_ld32s_i64: tcg_out_modrm_offset(s, OPC_MOVSLQ, args[0], args[1], args[2]); break; @@ -2068,6 +2054,11 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, } break; + case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ + case INDEX_op_mov_i64: + case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ + case INDEX_op_movi_i64: + case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); } @@ -2078,10 +2069,7 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, static const TCGTargetOpDef x86_op_defs[] = { { INDEX_op_exit_tb, { } }, { INDEX_op_goto_tb, { } }, - { INDEX_op_call, { "ri" } }, { INDEX_op_br, { } }, - { INDEX_op_mov_i32, { "r", "r" } }, - { INDEX_op_movi_i32, { "r" } }, { INDEX_op_ld8u_i32, { "r", "r" } }, { INDEX_op_ld8s_i32, { "r", "r" } }, { INDEX_op_ld16u_i32, { "r", "r" } }, @@ -2135,8 +2123,6 @@ static const TCGTargetOpDef x86_op_defs[] = { { INDEX_op_brcond2_i32, { "r", "r", "ri", "ri" } }, { INDEX_op_setcond2_i32, { "r", "r", "r", "ri", "ri" } }, #else - { INDEX_op_mov_i64, { "r", "r" } }, - { INDEX_op_movi_i64, { "r" } }, { INDEX_op_ld8u_i64, { "r", "r" } }, { INDEX_op_ld8s_i64, { "r", "r" } }, { INDEX_op_ld16u_i64, { "r", "r" } }, diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h index dbeb16d3ac..6c94e5cde0 100644 --- a/tcg/i386/tcg-target.h +++ b/tcg/i386/tcg-target.h @@ -24,6 +24,8 @@ #ifndef TCG_TARGET_I386 #define TCG_TARGET_I386 1 +#define TCG_TARGET_INSN_UNIT_SIZE 1 + #ifdef __x86_64__ # define TCG_TARGET_REG_BITS 64 # define TCG_TARGET_NB_REGS 16 diff --git a/tcg/ia64/tcg-target.c b/tcg/ia64/tcg-target.c index 1f523d6466..6bc9924641 100644 --- a/tcg/ia64/tcg-target.c +++ b/tcg/ia64/tcg-target.c @@ -692,112 +692,32 @@ static inline uint64_t tcg_opc_x4(int qp, uint64_t opc, int b1, uint64_t imm) /* - * Relocations + * Relocations - Note that we never encode branches elsewhere than slot 2. */ -static inline void reloc_pcrel21b(void *pc, intptr_t target) +static void reloc_pcrel21b_slot2(tcg_insn_unit *pc, tcg_insn_unit *target) { - uint64_t imm; - int64_t disp; - int slot; + uint64_t imm = target - pc; - slot = (intptr_t)pc & 3; - pc = (void *)((intptr_t)pc & ~3); - - disp = target - (intptr_t)pc; - imm = (uint64_t) disp >> 4; - - switch(slot) { - case 0: - *(uint64_t *)(pc + 0) = (*(uint64_t *)(pc + 8) & 0xfffffdc00003ffffull) - | ((imm & 0x100000) << 21) /* s */ - | ((imm & 0x0fffff) << 18); /* imm20b */ - break; - case 1: - *(uint64_t *)(pc + 8) = (*(uint64_t *)(pc + 8) & 0xfffffffffffb8000ull) - | ((imm & 0x100000) >> 2) /* s */ - | ((imm & 0x0fffe0) >> 5); /* imm20b */ - *(uint64_t *)(pc + 0) = (*(uint64_t *)(pc + 0) & 0x07ffffffffffffffull) - | ((imm & 0x00001f) << 59); /* imm20b */ - break; - case 2: - *(uint64_t *)(pc + 8) = (*(uint64_t *)(pc + 8) & 0xf700000fffffffffull) - | ((imm & 0x100000) << 39) /* s */ - | ((imm & 0x0fffff) << 36); /* imm20b */ - break; - } + pc->hi = (pc->hi & 0xf700000fffffffffull) + | ((imm & 0x100000) << 39) /* s */ + | ((imm & 0x0fffff) << 36); /* imm20b */ } -static inline uint64_t get_reloc_pcrel21b (void *pc) +static uint64_t get_reloc_pcrel21b_slot2(tcg_insn_unit *pc) { - int64_t low, high; - int slot; + int64_t high = pc->hi; - slot = (tcg_target_long) pc & 3; - pc = (void *)((tcg_target_long) pc & ~3); - - low = (*(uint64_t *)(pc + 0)); - high = (*(uint64_t *)(pc + 8)); - - switch(slot) { - case 0: - return ((low >> 21) & 0x100000) + /* s */ - ((low >> 18) & 0x0fffff); /* imm20b */ - case 1: - return ((high << 2) & 0x100000) + /* s */ - ((high << 5) & 0x0fffe0) + /* imm20b */ - ((low >> 59) & 0x00001f); /* imm20b */ - case 2: - return ((high >> 39) & 0x100000) + /* s */ - ((high >> 36) & 0x0fffff); /* imm20b */ - default: - tcg_abort(); - } + return ((high >> 39) & 0x100000) + /* s */ + ((high >> 36) & 0x0fffff); /* imm20b */ } -static inline void reloc_pcrel60b(void *pc, intptr_t target) -{ - int64_t disp; - uint64_t imm; - - disp = target - (intptr_t)pc; - imm = (uint64_t) disp >> 4; - - *(uint64_t *)(pc + 8) = (*(uint64_t *)(pc + 8) & 0xf700000fff800000ull) - | (imm & 0x0800000000000000ull) /* s */ - | ((imm & 0x07fffff000000000ull) >> 36) /* imm39 */ - | ((imm & 0x00000000000fffffull) << 36); /* imm20b */ - *(uint64_t *)(pc + 0) = (*(uint64_t *)(pc + 0) & 0x00003fffffffffffull) - | ((imm & 0x0000000ffff00000ull) << 28); /* imm39 */ -} - -static inline uint64_t get_reloc_pcrel60b (void *pc) -{ - int64_t low, high; - - low = (*(uint64_t *)(pc + 0)); - high = (*(uint64_t *)(pc + 8)); - - return ((high) & 0x0800000000000000ull) + /* s */ - ((high >> 36) & 0x00000000000fffffull) + /* imm20b */ - ((high << 36) & 0x07fffff000000000ull) + /* imm39 */ - ((low >> 28) & 0x0000000ffff00000ull); /* imm39 */ -} - - -static void patch_reloc(uint8_t *code_ptr, int type, +static void patch_reloc(tcg_insn_unit *code_ptr, int type, intptr_t value, intptr_t addend) { - value += addend; - switch (type) { - case R_IA64_PCREL21B: - reloc_pcrel21b(code_ptr, value); - break; - case R_IA64_PCREL60B: - reloc_pcrel60b(code_ptr, value); - default: - tcg_abort(); - } + assert(addend == 0); + assert(type == R_IA64_PCREL21B); + reloc_pcrel21b_slot2(code_ptr, (tcg_insn_unit *)value); } /* @@ -861,7 +781,7 @@ static inline int tcg_target_const_match(tcg_target_long val, TCGType type, * Code generation */ -static uint8_t *tb_ret_addr; +static tcg_insn_unit *tb_ret_addr; static inline void tcg_out_bundle(TCGContext *s, int template, uint64_t slot0, uint64_t slot1, @@ -872,9 +792,10 @@ static inline void tcg_out_bundle(TCGContext *s, int template, slot1 &= 0x1ffffffffffull; /* 41 bits */ slot2 &= 0x1ffffffffffull; /* 41 bits */ - *(uint64_t *)(s->code_ptr + 0) = (slot1 << 46) | (slot0 << 5) | template; - *(uint64_t *)(s->code_ptr + 8) = (slot2 << 23) | (slot1 >> 18); - s->code_ptr += 16; + *s->code_ptr++ = (tcg_insn_unit){ + (slot1 << 46) | (slot0 << 5) | template, + (slot2 << 23) | (slot1 >> 18) + }; } static inline uint64_t tcg_opc_mov_a(int qp, TCGReg dst, TCGReg src) @@ -909,33 +830,34 @@ static inline void tcg_out_movi(TCGContext *s, TCGType type, static void tcg_out_br(TCGContext *s, int label_index) { TCGLabel *l = &s->labels[label_index]; + uint64_t imm; /* We pay attention here to not modify the branch target by reading the existing value and using it again. This ensure that caches and memory are kept coherent during retranslation. */ + if (l->has_value) { + imm = l->u.value_ptr - s->code_ptr; + } else { + imm = get_reloc_pcrel21b_slot2(s->code_ptr); + tcg_out_reloc(s, s->code_ptr, R_IA64_PCREL21B, label_index, 0); + } + tcg_out_bundle(s, mmB, INSN_NOP_M, INSN_NOP_M, - tcg_opc_b1 (TCG_REG_P0, OPC_BR_SPTK_MANY_B1, - get_reloc_pcrel21b(s->code_ptr + 2))); - - if (l->has_value) { - reloc_pcrel21b((s->code_ptr - 16) + 2, l->u.value); - } else { - tcg_out_reloc(s, (s->code_ptr - 16) + 2, - R_IA64_PCREL21B, label_index, 0); - } + tcg_opc_b1(TCG_REG_P0, OPC_BR_SPTK_MANY_B1, imm)); } -static inline void tcg_out_calli(TCGContext *s, uintptr_t addr) +static inline void tcg_out_call(TCGContext *s, tcg_insn_unit *desc) { + uintptr_t func = desc->lo, gp = desc->hi, disp; + /* Look through the function descriptor. */ - uintptr_t disp, *desc = (uintptr_t *)addr; tcg_out_bundle(s, mlx, INSN_NOP_M, - tcg_opc_l2 (desc[1]), - tcg_opc_x2 (TCG_REG_P0, OPC_MOVL_X2, TCG_REG_R1, desc[1])); - disp = (desc[0] - (uintptr_t)s->code_ptr) >> 4; + tcg_opc_l2 (gp), + tcg_opc_x2 (TCG_REG_P0, OPC_MOVL_X2, TCG_REG_R1, gp)); + disp = (tcg_insn_unit *)func - s->code_ptr; tcg_out_bundle(s, mLX, INSN_NOP_M, tcg_opc_l4 (disp), @@ -943,23 +865,8 @@ static inline void tcg_out_calli(TCGContext *s, uintptr_t addr) TCG_REG_B0, disp)); } -static inline void tcg_out_callr(TCGContext *s, TCGReg addr) -{ - tcg_out_bundle(s, MmI, - tcg_opc_m1 (TCG_REG_P0, OPC_LD8_M1, TCG_REG_R2, addr), - tcg_opc_a4 (TCG_REG_P0, OPC_ADDS_A4, TCG_REG_R3, 8, addr), - tcg_opc_i21(TCG_REG_P0, OPC_MOV_I21, - TCG_REG_B6, TCG_REG_R2, 0)); - tcg_out_bundle(s, mmB, - tcg_opc_m1 (TCG_REG_P0, OPC_LD8_M1, TCG_REG_R1, TCG_REG_R3), - INSN_NOP_M, - tcg_opc_b5 (TCG_REG_P0, OPC_BR_CALL_SPTK_MANY_B5, - TCG_REG_B0, TCG_REG_B6)); -} - static void tcg_out_exit_tb(TCGContext *s, tcg_target_long arg) { - int64_t disp; uint64_t imm, opc1; /* At least arg == 0 is a common operation. */ @@ -970,8 +877,7 @@ static void tcg_out_exit_tb(TCGContext *s, tcg_target_long arg) opc1 = INSN_NOP_M; } - disp = tb_ret_addr - s->code_ptr; - imm = (uint64_t)disp >> 4; + imm = tb_ret_addr - s->code_ptr; tcg_out_bundle(s, mLX, opc1, @@ -1000,7 +906,7 @@ static inline void tcg_out_goto_tb(TCGContext *s, TCGArg arg) tcg_opc_b4 (TCG_REG_P0, OPC_BR_SPTK_MANY_B4, TCG_REG_B6)); } - s->tb_next_offset[arg] = s->code_ptr - s->code_buf; + s->tb_next_offset[arg] = tcg_current_code_size(s); } static inline void tcg_out_jmp(TCGContext *s, TCGArg addr) @@ -1521,19 +1427,22 @@ static inline void tcg_out_brcond(TCGContext *s, TCGCond cond, TCGReg arg1, TCGReg arg2, int label_index, int cmp4) { TCGLabel *l = &s->labels[label_index]; + uint64_t imm; + + /* We pay attention here to not modify the branch target by reading + the existing value and using it again. This ensure that caches and + memory are kept coherent during retranslation. */ + if (l->has_value) { + imm = l->u.value_ptr - s->code_ptr; + } else { + imm = get_reloc_pcrel21b_slot2(s->code_ptr); + tcg_out_reloc(s, s->code_ptr, R_IA64_PCREL21B, label_index, 0); + } tcg_out_bundle(s, miB, INSN_NOP_M, tcg_opc_cmp_a(TCG_REG_P0, cond, arg1, arg2, cmp4), - tcg_opc_b1(TCG_REG_P6, OPC_BR_DPTK_FEW_B1, - get_reloc_pcrel21b(s->code_ptr + 2))); - - if (l->has_value) { - reloc_pcrel21b((s->code_ptr - 16) + 2, l->u.value); - } else { - tcg_out_reloc(s, (s->code_ptr - 16) + 2, - R_IA64_PCREL21B, label_index, 0); - } + tcg_opc_b1(TCG_REG_P6, OPC_BR_DPTK_FEW_B1, imm)); } static inline void tcg_out_setcond(TCGContext *s, TCGCond cond, TCGArg ret, @@ -1646,7 +1555,7 @@ static inline void tcg_out_qemu_tlb(TCGContext *s, TCGReg addr_reg, typedef struct TCGLabelQemuLdst { bool is_ld; TCGMemOp size; - uint8_t *label_ptr; /* label pointers to be updated */ + tcg_insn_unit *label_ptr; /* label pointers to be updated */ } TCGLabelQemuLdst; typedef struct TCGBackendData { @@ -1660,7 +1569,7 @@ static inline void tcg_out_tb_init(TCGContext *s) } static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOp opc, - uint8_t *label_ptr) + tcg_insn_unit *label_ptr) { TCGBackendData *be = s->be; TCGLabelQemuLdst *l = &be->ldst_labels[be->nb_ldst_labels++]; @@ -1683,43 +1592,44 @@ static void tcg_out_tb_finalize(TCGContext *s) helper_le_ldul_mmu, helper_le_ldq_mmu, }; - uintptr_t thunks[8] = { }; + tcg_insn_unit *thunks[8] = { }; TCGBackendData *be = s->be; size_t i, n = be->nb_ldst_labels; for (i = 0; i < n; i++) { TCGLabelQemuLdst *l = &be->ldst_labels[i]; long x = l->is_ld * 4 + l->size; - uintptr_t dest = thunks[x]; + tcg_insn_unit *dest = thunks[x]; /* The out-of-line thunks are all the same; load the return address from B0, load the GP, and branch to the code. Note that we are always post-call, so the register window has rolled, so we're using incomming parameter register numbers, not outgoing. */ - if (dest == 0) { - uintptr_t disp, *desc = (uintptr_t *)helpers[x]; + if (dest == NULL) { + uintptr_t *desc = (uintptr_t *)helpers[x]; + uintptr_t func = desc[0], gp = desc[1], disp; - thunks[x] = dest = (uintptr_t)s->code_ptr; + thunks[x] = dest = s->code_ptr; tcg_out_bundle(s, mlx, INSN_NOP_M, - tcg_opc_l2 (desc[1]), + tcg_opc_l2 (gp), tcg_opc_x2 (TCG_REG_P0, OPC_MOVL_X2, - TCG_REG_R1, desc[1])); + TCG_REG_R1, gp)); tcg_out_bundle(s, mii, INSN_NOP_M, INSN_NOP_I, tcg_opc_i22(TCG_REG_P0, OPC_MOV_I22, l->is_ld ? TCG_REG_R35 : TCG_REG_R36, TCG_REG_B0)); - disp = (desc[0] - (uintptr_t)s->code_ptr) >> 4; + disp = (tcg_insn_unit *)func - s->code_ptr; tcg_out_bundle(s, mLX, INSN_NOP_M, tcg_opc_l3 (disp), tcg_opc_x3 (TCG_REG_P0, OPC_BRL_SPTK_MANY_X3, disp)); } - reloc_pcrel21b(l->label_ptr, dest); + reloc_pcrel21b_slot2(l->label_ptr, dest); } } @@ -1731,7 +1641,7 @@ static inline void tcg_out_qemu_ld(TCGContext *s, const TCGArg *args) int addr_reg, data_reg, mem_index; TCGMemOp opc, s_bits; uint64_t fin1, fin2; - uint8_t *label_ptr; + tcg_insn_unit *label_ptr; data_reg = args[0]; addr_reg = args[1]; @@ -1765,13 +1675,13 @@ static inline void tcg_out_qemu_ld(TCGContext *s, const TCGArg *args) tcg_opc_a1 (TCG_REG_P6, OPC_ADD_A1, TCG_REG_R2, TCG_REG_R2, TCG_REG_R57), tcg_opc_movi_a(TCG_REG_P7, TCG_REG_R58, mem_index)); - label_ptr = s->code_ptr + 2; + label_ptr = s->code_ptr; tcg_out_bundle(s, miB, tcg_opc_m1 (TCG_REG_P6, opc_ld_m1[s_bits], TCG_REG_R8, TCG_REG_R2), INSN_NOP_I, tcg_opc_b3 (TCG_REG_P7, OPC_BR_CALL_SPNT_FEW_B3, TCG_REG_B0, - get_reloc_pcrel21b(label_ptr))); + get_reloc_pcrel21b_slot2(label_ptr))); add_qemu_ldst_label(s, 1, opc, label_ptr); @@ -1792,7 +1702,7 @@ static inline void tcg_out_qemu_st(TCGContext *s, const TCGArg *args) int mem_index; uint64_t pre1, pre2; TCGMemOp opc, s_bits; - uint8_t *label_ptr; + tcg_insn_unit *label_ptr; data_reg = args[0]; addr_reg = args[1]; @@ -1827,13 +1737,13 @@ static inline void tcg_out_qemu_st(TCGContext *s, const TCGArg *args) tcg_opc_a1 (TCG_REG_P6, OPC_ADD_A1, TCG_REG_R2, TCG_REG_R2, TCG_REG_R57), tcg_opc_movi_a(TCG_REG_P7, TCG_REG_R59, mem_index)); - label_ptr = s->code_ptr + 2; + label_ptr = s->code_ptr; tcg_out_bundle(s, miB, tcg_opc_m4 (TCG_REG_P6, opc_st_m4[s_bits], TCG_REG_R58, TCG_REG_R2), INSN_NOP_I, tcg_opc_b3 (TCG_REG_P7, OPC_BR_CALL_SPNT_FEW_B3, TCG_REG_B0, - get_reloc_pcrel21b(label_ptr))); + get_reloc_pcrel21b_slot2(label_ptr))); add_qemu_ldst_label(s, 0, opc, label_ptr); } @@ -2085,24 +1995,10 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, case INDEX_op_br: tcg_out_br(s, args[0]); break; - case INDEX_op_call: - if (likely(const_args[0])) { - tcg_out_calli(s, args[0]); - } else { - tcg_out_callr(s, args[0]); - } - break; case INDEX_op_goto_tb: tcg_out_goto_tb(s, args[0]); break; - case INDEX_op_movi_i32: - tcg_out_movi(s, TCG_TYPE_I32, args[0], args[1]); - break; - case INDEX_op_movi_i64: - tcg_out_movi(s, TCG_TYPE_I64, args[0], args[1]); - break; - case INDEX_op_ld8u_i32: case INDEX_op_ld8u_i64: tcg_out_ld_rel(s, OPC_LD1_M1, args[0], args[1], args[2]); @@ -2312,6 +2208,11 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_qemu_st(s, args); break; + case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ + case INDEX_op_mov_i64: + case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ + case INDEX_op_movi_i64: + case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); } @@ -2319,13 +2220,9 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, static const TCGTargetOpDef ia64_op_defs[] = { { INDEX_op_br, { } }, - { INDEX_op_call, { "ri" } }, { INDEX_op_exit_tb, { } }, { INDEX_op_goto_tb, { } }, - { INDEX_op_mov_i32, { "r", "r" } }, - { INDEX_op_movi_i32, { "r" } }, - { INDEX_op_ld8u_i32, { "r", "r" } }, { INDEX_op_ld8s_i32, { "r", "r" } }, { INDEX_op_ld16u_i32, { "r", "r" } }, @@ -2367,9 +2264,6 @@ static const TCGTargetOpDef ia64_op_defs[] = { { INDEX_op_setcond_i32, { "r", "rZ", "rZ" } }, { INDEX_op_movcond_i32, { "r", "rZ", "rZ", "rI", "rI" } }, - { INDEX_op_mov_i64, { "r", "r" } }, - { INDEX_op_movi_i64, { "r" } }, - { INDEX_op_ld8u_i64, { "r", "r" } }, { INDEX_op_ld8s_i64, { "r", "r" } }, { INDEX_op_ld16u_i64, { "r", "r" } }, @@ -2442,8 +2336,11 @@ static void tcg_target_qemu_prologue(TCGContext *s) CPU_TEMP_BUF_NLONGS * sizeof(long)); /* First emit adhoc function descriptor */ - *(uint64_t *)(s->code_ptr) = (uint64_t)s->code_ptr + 16; /* entry point */ - s->code_ptr += 16; /* skip GP */ + *s->code_ptr = (tcg_insn_unit){ + (uint64_t)(s->code_ptr + 1), /* entry point */ + 0 /* skip gp */ + }; + s->code_ptr++; /* prologue */ tcg_out_bundle(s, miI, diff --git a/tcg/ia64/tcg-target.h b/tcg/ia64/tcg-target.h index d834beb323..3a59b50349 100644 --- a/tcg/ia64/tcg-target.h +++ b/tcg/ia64/tcg-target.h @@ -25,6 +25,12 @@ #ifndef TCG_TARGET_IA64 #define TCG_TARGET_IA64 1 +#define TCG_TARGET_INSN_UNIT_SIZE 16 +typedef struct { + uint64_t lo __attribute__((aligned(16))); + uint64_t hi; +} tcg_insn_unit; + /* We only map the first 64 registers */ #define TCG_TARGET_NB_REGS 64 typedef enum { diff --git a/tcg/mips/tcg-target.c b/tcg/mips/tcg-target.c index 37241b224b..0ae495c586 100644 --- a/tcg/mips/tcg-target.c +++ b/tcg/mips/tcg-target.c @@ -108,83 +108,38 @@ static const TCGReg tcg_target_call_oarg_regs[2] = { TCG_REG_V1 }; -static uint8_t *tb_ret_addr; +static tcg_insn_unit *tb_ret_addr; -static inline uint32_t reloc_lo16_val(void *pc, intptr_t target) +static inline uint32_t reloc_pc16_val(tcg_insn_unit *pc, tcg_insn_unit *target) { - return target & 0xffff; + /* Let the compiler perform the right-shift as part of the arithmetic. */ + ptrdiff_t disp = target - (pc + 1); + assert(disp == (int16_t)disp); + return disp & 0xffff; } -static inline void reloc_lo16(void *pc, intptr_t target) +static inline void reloc_pc16(tcg_insn_unit *pc, tcg_insn_unit *target) { - *(uint32_t *) pc = (*(uint32_t *) pc & ~0xffff) - | reloc_lo16_val(pc, target); + *pc = deposit32(*pc, 0, 16, reloc_pc16_val(pc, target)); } -static inline uint32_t reloc_hi16_val(void *pc, intptr_t target) +static inline uint32_t reloc_26_val(tcg_insn_unit *pc, tcg_insn_unit *target) { - return (target >> 16) & 0xffff; + assert((((uintptr_t)pc ^ (uintptr_t)target) & 0xf0000000) == 0); + return ((uintptr_t)target >> 2) & 0x3ffffff; } -static inline void reloc_hi16(void *pc, intptr_t target) +static inline void reloc_26(tcg_insn_unit *pc, tcg_insn_unit *target) { - *(uint32_t *) pc = (*(uint32_t *) pc & ~0xffff) - | reloc_hi16_val(pc, target); + *pc = deposit32(*pc, 0, 26, reloc_26_val(pc, target)); } -static inline uint32_t reloc_pc16_val(void *pc, intptr_t target) -{ - int32_t disp; - - disp = target - (intptr_t)pc - 4; - if (disp != (disp << 14) >> 14) { - tcg_abort (); - } - - return (disp >> 2) & 0xffff; -} - -static inline void reloc_pc16 (void *pc, tcg_target_long target) -{ - *(uint32_t *) pc = (*(uint32_t *) pc & ~0xffff) - | reloc_pc16_val(pc, target); -} - -static inline uint32_t reloc_26_val (void *pc, tcg_target_long target) -{ - if ((((tcg_target_long)pc + 4) & 0xf0000000) != (target & 0xf0000000)) { - tcg_abort (); - } - - return (target >> 2) & 0x3ffffff; -} - -static inline void reloc_pc26(void *pc, intptr_t target) -{ - *(uint32_t *) pc = (*(uint32_t *) pc & ~0x3ffffff) - | reloc_26_val(pc, target); -} - -static void patch_reloc(uint8_t *code_ptr, int type, +static void patch_reloc(tcg_insn_unit *code_ptr, int type, intptr_t value, intptr_t addend) { - value += addend; - switch(type) { - case R_MIPS_LO16: - reloc_lo16(code_ptr, value); - break; - case R_MIPS_HI16: - reloc_hi16(code_ptr, value); - break; - case R_MIPS_PC16: - reloc_pc16(code_ptr, value); - break; - case R_MIPS_26: - reloc_pc26(code_ptr, value); - break; - default: - tcg_abort(); - } + assert(type == R_MIPS_PC16); + assert(addend == 0); + reloc_pc16(code_ptr, (tcg_insn_unit *)value); } /* parse target specific constraints */ @@ -198,11 +153,6 @@ static int target_parse_constraint(TCGArgConstraint *ct, const char **pct_str) ct->ct |= TCG_CT_REG; tcg_regset_set(ct->u.regs, 0xffffffff); break; - case 'C': - ct->ct |= TCG_CT_REG; - tcg_regset_clear(ct->u.regs); - tcg_regset_set_reg(ct->u.regs, TCG_REG_T9); - break; case 'L': /* qemu_ld output arg constraint */ ct->ct |= TCG_CT_REG; tcg_regset_set(ct->u.regs, 0xffffffff); @@ -374,7 +324,7 @@ static inline void tcg_out_opc_br(TCGContext *s, int opc, /* We pay attention here to not modify the branch target by reading the existing value and using it again. This ensure that caches and memory are kept coherent during retranslation. */ - uint16_t offset = (uint16_t)(*(uint32_t *) s->code_ptr); + uint16_t offset = (uint16_t)*s->code_ptr; tcg_out_opc_imm(s, opc, rt, rs, offset); } @@ -663,9 +613,9 @@ static void tcg_out_brcond(TCGContext *s, TCGCond cond, TCGArg arg1, break; } if (l->has_value) { - reloc_pc16(s->code_ptr - 4, l->u.value); + reloc_pc16(s->code_ptr - 1, l->u.value_ptr); } else { - tcg_out_reloc(s, s->code_ptr - 4, R_MIPS_PC16, label_index, 0); + tcg_out_reloc(s, s->code_ptr - 1, R_MIPS_PC16, label_index, 0); } tcg_out_nop(s); } @@ -676,7 +626,7 @@ static void tcg_out_brcond2(TCGContext *s, TCGCond cond, TCGArg arg1, TCGArg arg2, TCGArg arg3, TCGArg arg4, int label_index) { - void *label_ptr; + tcg_insn_unit *label_ptr; switch(cond) { case TCG_COND_NE: @@ -733,7 +683,7 @@ static void tcg_out_brcond2(TCGContext *s, TCGCond cond, TCGArg arg1, tcg_abort(); } - reloc_pc16(label_ptr, (tcg_target_long) s->code_ptr); + reloc_pc16(label_ptr, s->code_ptr); } static void tcg_out_movcond(TCGContext *s, TCGCond cond, TCGReg ret, @@ -945,12 +895,12 @@ static void tcg_out_qemu_ld(TCGContext *s, const TCGArg *args, { TCGReg addr_regl, data_regl, data_regh, data_reg1, data_reg2; #if defined(CONFIG_SOFTMMU) - void *label1_ptr, *label2_ptr; + tcg_insn_unit *label1_ptr, *label2_ptr; int arg_num; int mem_index, s_bits; int addr_meml; # if TARGET_LONG_BITS == 64 - uint8_t *label3_ptr; + tcg_insn_unit *label3_ptr; TCGReg addr_regh; int addr_memh; # endif @@ -1011,7 +961,7 @@ static void tcg_out_qemu_ld(TCGContext *s, const TCGArg *args, tcg_out_opc_br(s, OPC_BEQ, addr_regh, TCG_REG_AT); tcg_out_nop(s); - reloc_pc16(label3_ptr, (tcg_target_long) s->code_ptr); + reloc_pc16(label3_ptr, s->code_ptr); # else label1_ptr = s->code_ptr; tcg_out_opc_br(s, OPC_BEQ, TCG_REG_T0, TCG_REG_AT); @@ -1060,7 +1010,7 @@ static void tcg_out_qemu_ld(TCGContext *s, const TCGArg *args, tcg_out_nop(s); /* label1: fast path */ - reloc_pc16(label1_ptr, (tcg_target_long) s->code_ptr); + reloc_pc16(label1_ptr, s->code_ptr); tcg_out_opc_imm(s, OPC_LW, TCG_REG_A0, TCG_REG_A0, offsetof(CPUArchState, tlb_table[mem_index][0].addend)); @@ -1121,7 +1071,7 @@ static void tcg_out_qemu_ld(TCGContext *s, const TCGArg *args, } #if defined(CONFIG_SOFTMMU) - reloc_pc16(label2_ptr, (tcg_target_long) s->code_ptr); + reloc_pc16(label2_ptr, s->code_ptr); #endif } @@ -1130,14 +1080,14 @@ static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args, { TCGReg addr_regl, data_regl, data_regh, data_reg1, data_reg2; #if defined(CONFIG_SOFTMMU) - uint8_t *label1_ptr, *label2_ptr; + tcg_insn_unit *label1_ptr, *label2_ptr; int arg_num; int mem_index, s_bits; int addr_meml; #endif #if TARGET_LONG_BITS == 64 # if defined(CONFIG_SOFTMMU) - uint8_t *label3_ptr; + tcg_insn_unit *label3_ptr; TCGReg addr_regh; int addr_memh; # endif @@ -1200,7 +1150,7 @@ static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args, tcg_out_opc_br(s, OPC_BEQ, addr_regh, TCG_REG_AT); tcg_out_nop(s); - reloc_pc16(label3_ptr, (tcg_target_long) s->code_ptr); + reloc_pc16(label3_ptr, s->code_ptr); # else label1_ptr = s->code_ptr; tcg_out_opc_br(s, OPC_BEQ, TCG_REG_T0, TCG_REG_AT); @@ -1241,7 +1191,7 @@ static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args, tcg_out_nop(s); /* label1: fast path */ - reloc_pc16(label1_ptr, (tcg_target_long) s->code_ptr); + reloc_pc16(label1_ptr, s->code_ptr); tcg_out_opc_imm(s, OPC_LW, TCG_REG_A0, TCG_REG_A0, offsetof(CPUArchState, tlb_table[mem_index][0].addend)); @@ -1293,17 +1243,24 @@ static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args, } #if defined(CONFIG_SOFTMMU) - reloc_pc16(label2_ptr, (tcg_target_long) s->code_ptr); + reloc_pc16(label2_ptr, s->code_ptr); #endif } +static void tcg_out_call(TCGContext *s, tcg_insn_unit *target) +{ + tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_T9, (intptr_t)target); + tcg_out_opc_reg(s, OPC_JALR, TCG_REG_RA, TCG_REG_T9, 0); + tcg_out_nop(s); +} + static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, const int *const_args) { switch(opc) { case INDEX_op_exit_tb: tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_V0, args[0]); - tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_AT, (tcg_target_long)tb_ret_addr); + tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_AT, (uintptr_t)tb_ret_addr); tcg_out_opc_reg(s, OPC_JR, 0, TCG_REG_AT, 0); tcg_out_nop(s); break; @@ -1313,28 +1270,18 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_abort(); } else { /* indirect jump method */ - tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_AT, (tcg_target_long)(s->tb_next + args[0])); + tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_AT, + (uintptr_t)(s->tb_next + args[0])); tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_AT, TCG_REG_AT, 0); tcg_out_opc_reg(s, OPC_JR, 0, TCG_REG_AT, 0); } tcg_out_nop(s); - s->tb_next_offset[args[0]] = s->code_ptr - s->code_buf; - break; - case INDEX_op_call: - tcg_out_opc_reg(s, OPC_JALR, TCG_REG_RA, args[0], 0); - tcg_out_nop(s); + s->tb_next_offset[args[0]] = tcg_current_code_size(s); break; case INDEX_op_br: tcg_out_brcond(s, TCG_COND_EQ, TCG_REG_ZERO, TCG_REG_ZERO, args[0]); break; - case INDEX_op_mov_i32: - tcg_out_mov(s, TCG_TYPE_I32, args[0], args[1]); - break; - case INDEX_op_movi_i32: - tcg_out_movi(s, TCG_TYPE_I32, args[0], args[1]); - break; - case INDEX_op_ld8u_i32: tcg_out_ldst(s, OPC_LBU, args[0], args[1], args[2]); break; @@ -1582,6 +1529,9 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_qemu_st(s, args, 3); break; + case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ + case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ + case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); } @@ -1590,11 +1540,8 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, static const TCGTargetOpDef mips_op_defs[] = { { INDEX_op_exit_tb, { } }, { INDEX_op_goto_tb, { } }, - { INDEX_op_call, { "C" } }, { INDEX_op_br, { } }, - { INDEX_op_mov_i32, { "r", "r" } }, - { INDEX_op_movi_i32, { "r" } }, { INDEX_op_ld8u_i32, { "r", "r" } }, { INDEX_op_ld8s_i32, { "r", "r" } }, { INDEX_op_ld16u_i32, { "r", "r" } }, diff --git a/tcg/mips/tcg-target.h b/tcg/mips/tcg-target.h index 9576db514d..c6d2267d77 100644 --- a/tcg/mips/tcg-target.h +++ b/tcg/mips/tcg-target.h @@ -26,6 +26,7 @@ #ifndef TCG_TARGET_MIPS #define TCG_TARGET_MIPS 1 +#define TCG_TARGET_INSN_UNIT_SIZE 4 #define TCG_TARGET_NB_REGS 32 typedef enum { diff --git a/tcg/optimize.c b/tcg/optimize.c index 0302f4f99a..3a504a1961 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -513,12 +513,8 @@ static bool swap_commutative2(TCGArg *p1, TCGArg *p2) static TCGArg *tcg_constant_folding(TCGContext *s, uint16_t *tcg_opc_ptr, TCGArg *args, TCGOpDef *tcg_op_defs) { - int i, nb_ops, op_index, nb_temps, nb_globals, nb_call_args; - tcg_target_ulong mask, affected; - TCGOpcode op; - const TCGOpDef *def; + int nb_ops, op_index, nb_temps, nb_globals; TCGArg *gen_args; - TCGArg tmp; /* Array VALS has an element for each temp. If this temp holds a constant then its value is kept in VALS' element. @@ -532,22 +528,27 @@ static TCGArg *tcg_constant_folding(TCGContext *s, uint16_t *tcg_opc_ptr, nb_ops = tcg_opc_ptr - s->gen_opc_buf; gen_args = args; for (op_index = 0; op_index < nb_ops; op_index++) { - op = s->gen_opc_buf[op_index]; - def = &tcg_op_defs[op]; - /* Do copy propagation */ + TCGOpcode op = s->gen_opc_buf[op_index]; + const TCGOpDef *def = &tcg_op_defs[op]; + tcg_target_ulong mask, affected; + int nb_oargs, nb_iargs, nb_args, i; + TCGArg tmp; + if (op == INDEX_op_call) { - int nb_oargs = args[0] >> 16; - int nb_iargs = args[0] & 0xffff; - for (i = nb_oargs + 1; i < nb_oargs + nb_iargs + 1; i++) { - if (temps[args[i]].state == TCG_TEMP_COPY) { - args[i] = find_better_copy(s, args[i]); - } - } + *gen_args++ = tmp = *args++; + nb_oargs = tmp >> 16; + nb_iargs = tmp & 0xffff; + nb_args = nb_oargs + nb_iargs + def->nb_cargs; } else { - for (i = def->nb_oargs; i < def->nb_oargs + def->nb_iargs; i++) { - if (temps[args[i]].state == TCG_TEMP_COPY) { - args[i] = find_better_copy(s, args[i]); - } + nb_oargs = def->nb_oargs; + nb_iargs = def->nb_iargs; + nb_args = def->nb_args; + } + + /* Do copy propagation */ + for (i = nb_oargs; i < nb_oargs + nb_iargs; i++) { + if (temps[args[i]].state == TCG_TEMP_COPY) { + args[i] = find_better_copy(s, args[i]); } } @@ -882,7 +883,7 @@ static TCGArg *tcg_constant_folding(TCGContext *s, uint16_t *tcg_opc_ptr, CASE_OP_32_64(qemu_ld): { - TCGMemOp mop = args[def->nb_oargs + def->nb_iargs]; + TCGMemOp mop = args[nb_oargs + nb_iargs]; if (!(mop & MO_SIGN)) { mask = (2ULL << ((8 << (mop & MO_SIZE)) - 1)) - 1; } @@ -900,15 +901,15 @@ static TCGArg *tcg_constant_folding(TCGContext *s, uint16_t *tcg_opc_ptr, } if (mask == 0) { - assert(def->nb_oargs == 1); + assert(nb_oargs == 1); s->gen_opc_buf[op_index] = op_to_movi(op); tcg_opt_gen_movi(gen_args, args[0], 0); - args += def->nb_oargs + def->nb_iargs + def->nb_cargs; + args += nb_args; gen_args += 2; continue; } if (affected == 0) { - assert(def->nb_oargs == 1); + assert(nb_oargs == 1); if (temps_are_copies(args[0], args[1])) { s->gen_opc_buf[op_index] = INDEX_op_nop; } else if (temps[args[1]].state != TCG_TEMP_CONST) { @@ -920,7 +921,7 @@ static TCGArg *tcg_constant_folding(TCGContext *s, uint16_t *tcg_opc_ptr, tcg_opt_gen_movi(gen_args, args[0], temps[args[1]].val); gen_args += 2; } - args += def->nb_iargs + 1; + args += nb_args; continue; } @@ -1246,24 +1247,13 @@ static TCGArg *tcg_constant_folding(TCGContext *s, uint16_t *tcg_opc_ptr, break; case INDEX_op_call: - nb_call_args = (args[0] >> 16) + (args[0] & 0xffff); - if (!(args[nb_call_args + 1] & (TCG_CALL_NO_READ_GLOBALS | - TCG_CALL_NO_WRITE_GLOBALS))) { + if (!(args[nb_oargs + nb_iargs + 1] + & (TCG_CALL_NO_READ_GLOBALS | TCG_CALL_NO_WRITE_GLOBALS))) { for (i = 0; i < nb_globals; i++) { reset_temp(i); } } - for (i = 0; i < (args[0] >> 16); i++) { - reset_temp(args[i + 1]); - } - i = nb_call_args + 3; - while (i) { - *gen_args = *args; - args++; - gen_args++; - i--; - } - break; + goto do_reset_output; default: do_default: @@ -1275,7 +1265,8 @@ static TCGArg *tcg_constant_folding(TCGContext *s, uint16_t *tcg_opc_ptr, if (def->flags & TCG_OPF_BB_END) { reset_all_temps(nb_temps); } else { - for (i = 0; i < def->nb_oargs; i++) { + do_reset_output: + for (i = 0; i < nb_oargs; i++) { reset_temp(args[i]); /* Save the corresponding known-zero bits mask for the first output argument (only one supported so far). */ @@ -1284,11 +1275,11 @@ static TCGArg *tcg_constant_folding(TCGContext *s, uint16_t *tcg_opc_ptr, } } } - for (i = 0; i < def->nb_args; i++) { + for (i = 0; i < nb_args; i++) { gen_args[i] = args[i]; } - args += def->nb_args; - gen_args += def->nb_args; + args += nb_args; + gen_args += nb_args; break; } } diff --git a/tcg/ppc/tcg-target.c b/tcg/ppc/tcg-target.c index 83d9340fca..436b65bffb 100644 --- a/tcg/ppc/tcg-target.c +++ b/tcg/ppc/tcg-target.c @@ -24,7 +24,7 @@ #include "tcg-be-ldst.h" -static uint8_t *tb_ret_addr; +static tcg_insn_unit *tb_ret_addr; #if defined _CALL_DARWIN || defined __APPLE__ #define TCG_TARGET_CALL_DARWIN @@ -171,50 +171,47 @@ static const int tcg_target_callee_save_regs[] = { TCG_REG_R31 }; -static uint32_t reloc_pc24_val (void *pc, tcg_target_long target) +static inline bool in_range_b(tcg_target_long target) { - tcg_target_long disp; - - disp = target - (tcg_target_long) pc; - if ((disp << 6) >> 6 != disp) - tcg_abort (); + return target == sextract32(target, 0, 26); +} +static uint32_t reloc_pc24_val(tcg_insn_unit *pc, tcg_insn_unit *target) +{ + ptrdiff_t disp = tcg_ptr_byte_diff(target, pc); + assert(in_range_b(disp)); return disp & 0x3fffffc; } -static void reloc_pc24 (void *pc, tcg_target_long target) +static void reloc_pc24(tcg_insn_unit *pc, tcg_insn_unit *target) { - *(uint32_t *) pc = (*(uint32_t *) pc & ~0x3fffffc) - | reloc_pc24_val (pc, target); + *pc = (*pc & ~0x3fffffc) | reloc_pc24_val(pc, target); } -static uint16_t reloc_pc14_val (void *pc, tcg_target_long target) +static uint16_t reloc_pc14_val(tcg_insn_unit *pc, tcg_insn_unit *target) { - tcg_target_long disp; - - disp = target - (tcg_target_long) pc; - if (disp != (int16_t) disp) - tcg_abort (); - + ptrdiff_t disp = tcg_ptr_byte_diff(target, pc); + assert(disp == (int16_t) disp); return disp & 0xfffc; } -static void reloc_pc14 (void *pc, tcg_target_long target) +static void reloc_pc14(tcg_insn_unit *pc, tcg_insn_unit *target) { - *(uint32_t *) pc = (*(uint32_t *) pc & ~0xfffc) - | reloc_pc14_val (pc, target); + *pc = (*pc & ~0xfffc) | reloc_pc14_val(pc, target); } -static void patch_reloc(uint8_t *code_ptr, int type, +static void patch_reloc(tcg_insn_unit *code_ptr, int type, intptr_t value, intptr_t addend) { - value += addend; + tcg_insn_unit *target = (tcg_insn_unit *)value; + + assert(addend == 0); switch (type) { case R_PPC_REL14: - reloc_pc14 (code_ptr, value); + reloc_pc14(code_ptr, target); break; case R_PPC_REL24: - reloc_pc24 (code_ptr, value); + reloc_pc24(code_ptr, target); break; default: tcg_abort(); @@ -480,47 +477,36 @@ static void tcg_out_ldst (TCGContext *s, int ret, int addr, } } -static void tcg_out_b (TCGContext *s, int mask, tcg_target_long target) +static void tcg_out_b(TCGContext *s, int mask, tcg_insn_unit *target) { - tcg_target_long disp; - - disp = target - (tcg_target_long) s->code_ptr; - if ((disp << 6) >> 6 == disp) - tcg_out32 (s, B | (disp & 0x3fffffc) | mask); - else { - tcg_out_movi (s, TCG_TYPE_I32, 0, (tcg_target_long) target); - tcg_out32 (s, MTSPR | RS (0) | CTR); - tcg_out32 (s, BCCTR | BO_ALWAYS | mask); + ptrdiff_t disp = tcg_pcrel_diff(s, target); + if (in_range_b(disp)) { + tcg_out32(s, B | (disp & 0x3fffffc) | mask); + } else { + tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R0, (uintptr_t)target); + tcg_out32(s, MTSPR | RS(TCG_REG_R0) | CTR); + tcg_out32(s, BCCTR | BO_ALWAYS | mask); } } -static void tcg_out_call (TCGContext *s, tcg_target_long arg, int const_arg, - int lk) +static void tcg_out_call1(TCGContext *s, tcg_insn_unit *target, int lk) { #ifdef _CALL_AIX - int reg; - - if (const_arg) { - reg = 2; - tcg_out_movi (s, TCG_TYPE_I32, reg, arg); - } - else reg = arg; - - tcg_out32 (s, LWZ | RT (0) | RA (reg)); - tcg_out32 (s, MTSPR | RA (0) | CTR); - tcg_out32 (s, LWZ | RT (2) | RA (reg) | 4); - tcg_out32 (s, BCCTR | BO_ALWAYS | lk); + tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R2, (uintptr_t)target); + tcg_out32(s, LWZ | RT(TCG_REG_R0) | RA(reg)); + tcg_out32(s, MTSPR | RA(TCG_REG_R0) | CTR); + tcg_out32(s, LWZ | RT(TCG_REG_R2) | RA(reg) | 4); + tcg_out32(s, BCCTR | BO_ALWAYS | lk); #else - if (const_arg) { - tcg_out_b (s, lk, arg); - } - else { - tcg_out32 (s, MTSPR | RS (arg) | LR); - tcg_out32 (s, BCLR | BO_ALWAYS | lk); - } + tcg_out_b(s, lk, target); #endif } +static void tcg_out_call(TCGContext *s, tcg_insn_unit *target) +{ + tcg_out_call1(s, target, LK); +} + #if defined(CONFIG_SOFTMMU) static void add_qemu_ldst_label (TCGContext *s, @@ -531,8 +517,8 @@ static void add_qemu_ldst_label (TCGContext *s, int addrlo_reg, int addrhi_reg, int mem_index, - uint8_t *raddr, - uint8_t *label_ptr) + tcg_insn_unit *raddr, + tcg_insn_unit *label_ptr) { TCGLabelQemuLdst *label = new_ldst_label(s); @@ -550,7 +536,7 @@ static void add_qemu_ldst_label (TCGContext *s, /* helper signature: helper_ret_ld_mmu(CPUState *env, target_ulong addr, * int mmu_idx, uintptr_t ra) */ -static const void * const qemu_ld_helpers[16] = { +static void * const qemu_ld_helpers[16] = { [MO_UB] = helper_ret_ldub_mmu, [MO_LEUW] = helper_le_lduw_mmu, [MO_LEUL] = helper_le_ldul_mmu, @@ -563,7 +549,7 @@ static const void * const qemu_ld_helpers[16] = { /* helper signature: helper_ret_st_mmu(CPUState *env, target_ulong addr, * uintxx_t val, int mmu_idx, uintptr_t ra) */ -static const void * const qemu_st_helpers[16] = { +static void * const qemu_st_helpers[16] = { [MO_UB] = helper_ret_stb_mmu, [MO_LEUW] = helper_le_stw_mmu, [MO_LEUL] = helper_le_stl_mmu, @@ -573,8 +559,8 @@ static const void * const qemu_st_helpers[16] = { [MO_BEQ] = helper_be_stq_mmu, }; -static void *ld_trampolines[16]; -static void *st_trampolines[16]; +static tcg_insn_unit *ld_trampolines[16]; +static tcg_insn_unit *st_trampolines[16]; /* Perform the TLB load and compare. Branches to the slow path, placing the address of the branch in *LABEL_PTR. Loads the addend of the TLB into R0. @@ -582,14 +568,15 @@ static void *st_trampolines[16]; static void tcg_out_tlb_check(TCGContext *s, TCGReg r0, TCGReg r1, TCGReg r2, TCGReg addrlo, TCGReg addrhi, TCGMemOp s_bits, - int mem_index, int is_load, uint8_t **label_ptr) + int mem_index, int is_load, + tcg_insn_unit **label_ptr) { int cmp_off = (is_load ? offsetof(CPUArchState, tlb_table[mem_index][0].addr_read) : offsetof(CPUArchState, tlb_table[mem_index][0].addr_write)); int add_off = offsetof(CPUArchState, tlb_table[mem_index][0].addend); - uint16_t retranst; + tcg_insn_unit retranst; TCGReg base = TCG_AREG0; /* Extract the page index, shifted into place for tlb index. */ @@ -648,7 +635,7 @@ static void tcg_out_tlb_check(TCGContext *s, TCGReg r0, TCGReg r1, TCGReg r2, This address cannot be used for a tail call, but it's shorter than forming an address from scratch. */ *label_ptr = s->code_ptr; - retranst = ((uint16_t *) s->code_ptr)[1] & ~3; + retranst = *s->code_ptr & 0xfffc; tcg_out32(s, BC | BI(7, CR_EQ) | retranst | BO_COND_FALSE | LK); } #endif @@ -659,7 +646,7 @@ static void tcg_out_qemu_ld(TCGContext *s, const TCGArg *args, bool is64) TCGMemOp opc, bswap; #ifdef CONFIG_SOFTMMU int mem_index; - uint8_t *label_ptr; + tcg_insn_unit *label_ptr; #endif datalo = *args++; @@ -731,7 +718,7 @@ static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args, bool is64) TCGMemOp opc, bswap, s_bits; #ifdef CONFIG_SOFTMMU int mem_index; - uint8_t *label_ptr; + tcg_insn_unit *label_ptr; #endif datalo = *args++; @@ -790,7 +777,7 @@ static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) TCGReg ir, datalo, datahi; TCGMemOp opc = l->opc; - reloc_pc14 (l->label_ptr[0], (uintptr_t)s->code_ptr); + reloc_pc14(l->label_ptr[0], s->code_ptr); ir = TCG_REG_R4; if (TARGET_LONG_BITS == 32) { @@ -804,7 +791,7 @@ static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) } tcg_out_movi(s, TCG_TYPE_I32, ir++, l->mem_index); tcg_out32(s, MFSPR | RT(ir++) | LR); - tcg_out_b(s, LK, (uintptr_t)ld_trampolines[opc & ~MO_SIGN]); + tcg_out_b(s, LK, ld_trampolines[opc & ~MO_SIGN]); datalo = l->datalo_reg; switch (opc & MO_SSIZE) { @@ -832,7 +819,7 @@ static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) } break; } - tcg_out_b (s, 0, (uintptr_t)l->raddr); + tcg_out_b(s, 0, l->raddr); } static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) @@ -840,7 +827,7 @@ static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) TCGReg ir, datalo; TCGMemOp opc = l->opc; - reloc_pc14 (l->label_ptr[0], (tcg_target_long) s->code_ptr); + reloc_pc14(l->label_ptr[0], s->code_ptr); ir = TCG_REG_R4; if (TARGET_LONG_BITS == 32) { @@ -878,16 +865,16 @@ static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) tcg_out_movi(s, TCG_TYPE_I32, ir++, l->mem_index); tcg_out32(s, MFSPR | RT(ir++) | LR); - tcg_out_b(s, LK, (uintptr_t)st_trampolines[opc]); - tcg_out_b(s, 0, (uintptr_t)l->raddr); + tcg_out_b(s, LK, st_trampolines[opc]); + tcg_out_b(s, 0, l->raddr); } #endif #ifdef CONFIG_SOFTMMU -static void emit_ldst_trampoline (TCGContext *s, const void *ptr) +static void emit_ldst_trampoline(TCGContext *s, tcg_insn_unit *ptr) { - tcg_out_mov (s, TCG_TYPE_I32, 3, TCG_AREG0); - tcg_out_call (s, (tcg_target_long) ptr, 1, 0); + tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_R3, TCG_AREG0); + tcg_out_call1(s, ptr, 0); } #endif @@ -909,12 +896,13 @@ static void tcg_target_qemu_prologue (TCGContext *s) #ifdef _CALL_AIX { - uint32_t addr; + uintptr_t addr; /* First emit adhoc function descriptor */ - addr = (uint32_t) s->code_ptr + 12; - tcg_out32 (s, addr); /* entry point */ - s->code_ptr += 8; /* skip TOC and environment pointer */ + addr = (uintptr_t)s->code_ptr + 12; + tcg_out32(s, addr); /* entry point */ + tcg_out32(s, 0); /* toc */ + tcg_out32(s, 0); /* environment pointer */ } #endif tcg_out32 (s, MFSPR | RT (0) | LR); @@ -1065,18 +1053,17 @@ static void tcg_out_cmp (TCGContext *s, int cond, TCGArg arg1, TCGArg arg2, } -static void tcg_out_bc (TCGContext *s, int bc, int label_index) +static void tcg_out_bc(TCGContext *s, int bc, int label_index) { TCGLabel *l = &s->labels[label_index]; - if (l->has_value) - tcg_out32 (s, bc | reloc_pc14_val (s->code_ptr, l->u.value)); - else { - uint16_t val = *(uint16_t *) &s->code_ptr[2]; - + if (l->has_value) { + tcg_out32(s, bc | reloc_pc14_val(s->code_ptr, l->u.value_ptr)); + } else { /* Thanks to Andrzej Zaborowski */ - tcg_out32 (s, bc | (val & 0xfffc)); - tcg_out_reloc (s, s->code_ptr - 4, R_PPC_REL14, label_index, 0); + tcg_insn_unit retrans = *s->code_ptr & 0xfffc; + tcg_out_reloc(s, s->code_ptr, R_PPC_REL14, label_index, 0); + tcg_out32(s, bc | retrans); } } @@ -1367,43 +1354,33 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, { switch (opc) { case INDEX_op_exit_tb: - tcg_out_movi (s, TCG_TYPE_I32, TCG_REG_R3, args[0]); - tcg_out_b (s, 0, (tcg_target_long) tb_ret_addr); + tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_R3, args[0]); + tcg_out_b(s, 0, tb_ret_addr); break; case INDEX_op_goto_tb: if (s->tb_jmp_offset) { /* direct jump method */ - - s->tb_jmp_offset[args[0]] = s->code_ptr - s->code_buf; - s->code_ptr += 16; - } - else { + s->tb_jmp_offset[args[0]] = tcg_current_code_size(s); + s->code_ptr += 4; + } else { tcg_abort (); } - s->tb_next_offset[args[0]] = s->code_ptr - s->code_buf; + s->tb_next_offset[args[0]] = tcg_current_code_size(s); break; case INDEX_op_br: { TCGLabel *l = &s->labels[args[0]]; if (l->has_value) { - tcg_out_b (s, 0, l->u.value); - } - else { - uint32_t val = *(uint32_t *) s->code_ptr; - + tcg_out_b(s, 0, l->u.value_ptr); + } else { /* Thanks to Andrzej Zaborowski */ - tcg_out32 (s, B | (val & 0x3fffffc)); - tcg_out_reloc (s, s->code_ptr - 4, R_PPC_REL24, args[0], 0); + tcg_insn_unit retrans = *s->code_ptr & 0x3fffffc; + tcg_out_reloc(s, s->code_ptr, R_PPC_REL24, args[0], 0); + tcg_out32(s, B | retrans); } } break; - case INDEX_op_call: - tcg_out_call (s, args[0], const_args[0], LK); - break; - case INDEX_op_movi_i32: - tcg_out_movi(s, TCG_TYPE_I32, args[0], args[1]); - break; case INDEX_op_ld8u_i32: tcg_out_ldst (s, args[0], args[1], args[2], LBZ, LBZX); break; @@ -1839,20 +1816,19 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, const_args[2]); break; + case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ + case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ + case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: - tcg_dump_ops (s); - tcg_abort (); + tcg_abort(); } } static const TCGTargetOpDef ppc_op_defs[] = { { INDEX_op_exit_tb, { } }, { INDEX_op_goto_tb, { } }, - { INDEX_op_call, { "ri" } }, { INDEX_op_br, { } }, - { INDEX_op_mov_i32, { "r", "r" } }, - { INDEX_op_movi_i32, { "r" } }, { INDEX_op_ld8u_i32, { "r", "r" } }, { INDEX_op_ld8s_i32, { "r", "r" } }, { INDEX_op_ld16u_i32, { "r", "r" } }, diff --git a/tcg/ppc/tcg-target.h b/tcg/ppc/tcg-target.h index 0d4f5959a7..dd7e557948 100644 --- a/tcg/ppc/tcg-target.h +++ b/tcg/ppc/tcg-target.h @@ -25,6 +25,7 @@ #define TCG_TARGET_PPC 1 #define TCG_TARGET_NB_REGS 32 +#define TCG_TARGET_INSN_UNIT_SIZE 4 typedef enum { TCG_REG_R0 = 0, diff --git a/tcg/ppc64/tcg-target.c b/tcg/ppc64/tcg-target.c index 45b1c06910..c90ddcd03a 100644 --- a/tcg/ppc64/tcg-target.c +++ b/tcg/ppc64/tcg-target.c @@ -31,7 +31,7 @@ #define TCG_CT_CONST_ZERO 0x1000 #define TCG_CT_CONST_MONE 0x2000 -static uint8_t *tb_ret_addr; +static tcg_insn_unit *tb_ret_addr; #if TARGET_LONG_BITS == 32 #define LD_ADDR LWZ @@ -168,61 +168,54 @@ static inline bool in_range_b(tcg_target_long target) return target == sextract64(target, 0, 26); } -static uint32_t reloc_pc24_val(void *pc, tcg_target_long target) +static uint32_t reloc_pc24_val(tcg_insn_unit *pc, tcg_insn_unit *target) { - tcg_target_long disp; - - disp = target - (tcg_target_long)pc; + ptrdiff_t disp = tcg_ptr_byte_diff(target, pc); assert(in_range_b(disp)); - return disp & 0x3fffffc; } -static void reloc_pc24(void *pc, tcg_target_long target) +static void reloc_pc24(tcg_insn_unit *pc, tcg_insn_unit *target) { - *(uint32_t *)pc = (*(uint32_t *)pc & ~0x3fffffc) - | reloc_pc24_val(pc, target); + *pc = (*pc & ~0x3fffffc) | reloc_pc24_val(pc, target); } -static uint16_t reloc_pc14_val(void *pc, tcg_target_long target) +static uint16_t reloc_pc14_val(tcg_insn_unit *pc, tcg_insn_unit *target) { - tcg_target_long disp; - - disp = target - (tcg_target_long)pc; - if (disp != (int16_t) disp) { - tcg_abort(); - } - + ptrdiff_t disp = tcg_ptr_byte_diff(target, pc); + assert(disp == (int16_t) disp); return disp & 0xfffc; } -static void reloc_pc14(void *pc, tcg_target_long target) +static void reloc_pc14(tcg_insn_unit *pc, tcg_insn_unit *target) { - *(uint32_t *)pc = (*(uint32_t *)pc & ~0xfffc) | reloc_pc14_val(pc, target); + *pc = (*pc & ~0xfffc) | reloc_pc14_val(pc, target); } static inline void tcg_out_b_noaddr(TCGContext *s, int insn) { - unsigned retrans = *(uint32_t *)s->code_ptr & 0x3fffffc; + unsigned retrans = *s->code_ptr & 0x3fffffc; tcg_out32(s, insn | retrans); } static inline void tcg_out_bc_noaddr(TCGContext *s, int insn) { - unsigned retrans = *(uint32_t *)s->code_ptr & 0xfffc; + unsigned retrans = *s->code_ptr & 0xfffc; tcg_out32(s, insn | retrans); } -static void patch_reloc(uint8_t *code_ptr, int type, +static void patch_reloc(tcg_insn_unit *code_ptr, int type, intptr_t value, intptr_t addend) { - value += addend; + tcg_insn_unit *target = (tcg_insn_unit *)value; + + assert(addend == 0); switch (type) { case R_PPC_REL14: - reloc_pc14(code_ptr, value); + reloc_pc14(code_ptr, target); break; case R_PPC_REL24: - reloc_pc24(code_ptr, value); + reloc_pc24(code_ptr, target); break; default: tcg_abort(); @@ -702,61 +695,48 @@ static void tcg_out_xori32(TCGContext *s, TCGReg dst, TCGReg src, uint32_t c) tcg_out_zori32(s, dst, src, c, XORI, XORIS); } -static void tcg_out_b(TCGContext *s, int mask, tcg_target_long target) +static void tcg_out_b(TCGContext *s, int mask, tcg_insn_unit *target) { - tcg_target_long disp; - - disp = target - (tcg_target_long)s->code_ptr; + ptrdiff_t disp = tcg_pcrel_diff(s, target); if (in_range_b(disp)) { tcg_out32(s, B | (disp & 0x3fffffc) | mask); } else { - tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_R0, (tcg_target_long)target); + tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_R0, (uintptr_t)target); tcg_out32(s, MTSPR | RS(TCG_REG_R0) | CTR); tcg_out32(s, BCCTR | BO_ALWAYS | mask); } } -static void tcg_out_call(TCGContext *s, tcg_target_long arg, int const_arg) +static void tcg_out_call(TCGContext *s, tcg_insn_unit *target) { #ifdef __APPLE__ - if (const_arg) { - tcg_out_b(s, LK, arg); - } else { - tcg_out32(s, MTSPR | RS(arg) | LR); - tcg_out32(s, BCLR | BO_ALWAYS | LK); - } + tcg_out_b(s, LK, target); #else - TCGReg reg = arg; - int ofs = 0; - - if (const_arg) { - /* Look through the descriptor. If the branch is in range, and we - don't have to spend too much effort on building the toc. */ - intptr_t tgt = ((intptr_t *)arg)[0]; - intptr_t toc = ((intptr_t *)arg)[1]; - intptr_t diff = tgt - (intptr_t)s->code_ptr; - - if (in_range_b(diff) && toc == (uint32_t)toc) { - tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_R2, toc); - tcg_out_b(s, LK, tgt); - return; - } + /* Look through the descriptor. If the branch is in range, and we + don't have to spend too much effort on building the toc. */ + void *tgt = ((void **)target)[0]; + uintptr_t toc = ((uintptr_t *)target)[1]; + intptr_t diff = tcg_pcrel_diff(s, tgt); + if (in_range_b(diff) && toc == (uint32_t)toc) { + tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_R2, toc); + tcg_out_b(s, LK, tgt); + } else { /* Fold the low bits of the constant into the addresses below. */ - ofs = (int16_t)arg; + intptr_t arg = (intptr_t)target; + int ofs = (int16_t)arg; + if (ofs + 8 < 0x8000) { arg -= ofs; } else { ofs = 0; } - reg = TCG_REG_R2; - tcg_out_movi(s, TCG_TYPE_I64, reg, arg); + tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_R2, arg); + tcg_out32(s, LD | TAI(TCG_REG_R0, TCG_REG_R2, ofs)); + tcg_out32(s, MTSPR | RA(TCG_REG_R0) | CTR); + tcg_out32(s, LD | TAI(TCG_REG_R2, TCG_REG_R2, ofs + 8)); + tcg_out32(s, BCCTR | BO_ALWAYS | LK); } - - tcg_out32(s, LD | TAI(TCG_REG_R0, reg, ofs)); - tcg_out32(s, MTSPR | RA(TCG_REG_R0) | CTR); - tcg_out32(s, LD | TAI(TCG_REG_R2, reg, ofs + 8)); - tcg_out32(s, BCCTR | BO_ALWAYS | LK); #endif } @@ -844,7 +824,7 @@ static const uint32_t qemu_exts_opc[4] = { /* helper signature: helper_ld_mmu(CPUState *env, target_ulong addr, * int mmu_idx, uintptr_t ra) */ -static const void * const qemu_ld_helpers[16] = { +static void * const qemu_ld_helpers[16] = { [MO_UB] = helper_ret_ldub_mmu, [MO_LEUW] = helper_le_lduw_mmu, [MO_LEUL] = helper_le_ldul_mmu, @@ -857,7 +837,7 @@ static const void * const qemu_ld_helpers[16] = { /* helper signature: helper_st_mmu(CPUState *env, target_ulong addr, * uintxx_t val, int mmu_idx, uintptr_t ra) */ -static const void * const qemu_st_helpers[16] = { +static void * const qemu_st_helpers[16] = { [MO_UB] = helper_ret_stb_mmu, [MO_LEUW] = helper_le_stw_mmu, [MO_LEUL] = helper_le_stl_mmu, @@ -946,7 +926,7 @@ static TCGReg tcg_out_tlb_read(TCGContext *s, TCGMemOp s_bits, TCGReg addr_reg, helper code. */ static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOp opc, int data_reg, int addr_reg, int mem_index, - uint8_t *raddr, uint8_t *label_ptr) + tcg_insn_unit *raddr, tcg_insn_unit *label_ptr) { TCGLabelQemuLdst *label = new_ldst_label(s); @@ -963,7 +943,7 @@ static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) { TCGMemOp opc = lb->opc; - reloc_pc14(lb->label_ptr[0], (uintptr_t)s->code_ptr); + reloc_pc14(lb->label_ptr[0], s->code_ptr); tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_R3, TCG_AREG0); @@ -974,7 +954,7 @@ static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_R5, lb->mem_index); tcg_out32(s, MFSPR | RT(TCG_REG_R6) | LR); - tcg_out_call(s, (tcg_target_long)qemu_ld_helpers[opc & ~MO_SIGN], 1); + tcg_out_call(s, qemu_ld_helpers[opc & ~MO_SIGN]); if (opc & MO_SIGN) { uint32_t insn = qemu_exts_opc[opc & MO_SIZE]; @@ -983,7 +963,7 @@ static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) tcg_out_mov(s, TCG_TYPE_I64, lb->datalo_reg, TCG_REG_R3); } - tcg_out_b(s, 0, (uintptr_t)lb->raddr); + tcg_out_b(s, 0, lb->raddr); } static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) @@ -991,7 +971,7 @@ static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) TCGMemOp opc = lb->opc; TCGMemOp s_bits = opc & MO_SIZE; - reloc_pc14(lb->label_ptr[0], (uintptr_t)s->code_ptr); + reloc_pc14(lb->label_ptr[0], s->code_ptr); tcg_out_mov(s, TCG_TYPE_I64, TCG_REG_R3, TCG_AREG0); @@ -1004,9 +984,9 @@ static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_R6, lb->mem_index); tcg_out32(s, MFSPR | RT(TCG_REG_R7) | LR); - tcg_out_call(s, (tcg_target_long)qemu_st_helpers[opc], 1); + tcg_out_call(s, qemu_st_helpers[opc]); - tcg_out_b(s, 0, (uintptr_t)lb->raddr); + tcg_out_b(s, 0, lb->raddr); } #endif /* SOFTMMU */ @@ -1017,7 +997,7 @@ static void tcg_out_qemu_ld(TCGContext *s, TCGReg data_reg, TCGReg addr_reg, uint32_t insn; TCGMemOp s_bits = opc & MO_SIZE; #ifdef CONFIG_SOFTMMU - void *label_ptr; + tcg_insn_unit *label_ptr; #endif #ifdef CONFIG_SOFTMMU @@ -1063,7 +1043,7 @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg data_reg, TCGReg addr_reg, TCGReg rbase; uint32_t insn; #ifdef CONFIG_SOFTMMU - void *label_ptr; + tcg_insn_unit *label_ptr; #endif #ifdef CONFIG_SOFTMMU @@ -1123,7 +1103,8 @@ static void tcg_target_qemu_prologue(TCGContext *s) #ifndef __APPLE__ /* First emit adhoc function descriptor */ tcg_out64(s, (uint64_t)s->code_ptr + 24); /* entry point */ - s->code_ptr += 16; /* skip TOC and environment pointer */ + tcg_out64(s, 0); /* toc */ + tcg_out64(s, 0); /* environment pointer */ #endif /* Prologue */ @@ -1415,7 +1396,7 @@ static void tcg_out_bc(TCGContext *s, int bc, int label_index) TCGLabel *l = &s->labels[label_index]; if (l->has_value) { - tcg_out32(s, bc | reloc_pc14_val(s->code_ptr, l->u.value)); + tcg_out32(s, bc | reloc_pc14_val(s->code_ptr, l->u.value_ptr)); } else { tcg_out_reloc(s, s->code_ptr, R_PPC_REL14, label_index, 0); tcg_out_bc_noaddr(s, bc); @@ -1478,15 +1459,13 @@ static void tcg_out_movcond(TCGContext *s, TCGType type, TCGCond cond, } } -void ppc_tb_set_jmp_target(unsigned long jmp_addr, unsigned long addr) +void ppc_tb_set_jmp_target(uintptr_t jmp_addr, uintptr_t addr) { TCGContext s; - unsigned long patch_size; - s.code_ptr = (uint8_t *) jmp_addr; - tcg_out_b(&s, 0, addr); - patch_size = s.code_ptr - (uint8_t *) jmp_addr; - flush_icache_range(jmp_addr, jmp_addr + patch_size); + s.code_buf = s.code_ptr = (tcg_insn_unit *)jmp_addr; + tcg_out_b(&s, 0, (tcg_insn_unit *)addr); + flush_icache_range(jmp_addr, jmp_addr + tcg_current_code_size(&s)); } static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, @@ -1498,40 +1477,31 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, switch (opc) { case INDEX_op_exit_tb: tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_R3, args[0]); - tcg_out_b(s, 0, (tcg_target_long)tb_ret_addr); + tcg_out_b(s, 0, tb_ret_addr); break; case INDEX_op_goto_tb: if (s->tb_jmp_offset) { /* Direct jump method. */ - s->tb_jmp_offset[args[0]] = s->code_ptr - s->code_buf; - s->code_ptr += 28; + s->tb_jmp_offset[args[0]] = tcg_current_code_size(s); + s->code_ptr += 7; } else { /* Indirect jump method. */ tcg_abort(); } - s->tb_next_offset[args[0]] = s->code_ptr - s->code_buf; + s->tb_next_offset[args[0]] = tcg_current_code_size(s); break; case INDEX_op_br: { TCGLabel *l = &s->labels[args[0]]; if (l->has_value) { - tcg_out_b(s, 0, l->u.value); + tcg_out_b(s, 0, l->u.value_ptr); } else { tcg_out_reloc(s, s->code_ptr, R_PPC_REL24, args[0], 0); tcg_out_b_noaddr(s, B); } } break; - case INDEX_op_call: - tcg_out_call(s, args[0], const_args[0]); - break; - case INDEX_op_movi_i32: - tcg_out_movi(s, TCG_TYPE_I32, args[0], args[1]); - break; - case INDEX_op_movi_i64: - tcg_out_movi(s, TCG_TYPE_I64, args[0], args[1]); - break; case INDEX_op_ld8u_i32: case INDEX_op_ld8u_i64: tcg_out_mem_long(s, LBZ, LBZX, args[0], args[1], args[2]); @@ -2012,8 +1982,12 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, tcg_out32(s, MULHD | TAB(args[0], args[1], args[2])); break; + case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ + case INDEX_op_mov_i64: + case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ + case INDEX_op_movi_i64: + case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: - tcg_dump_ops(s); tcg_abort(); } } @@ -2021,14 +1995,8 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, static const TCGTargetOpDef ppc_op_defs[] = { { INDEX_op_exit_tb, { } }, { INDEX_op_goto_tb, { } }, - { INDEX_op_call, { "ri" } }, { INDEX_op_br, { } }, - { INDEX_op_mov_i32, { "r", "r" } }, - { INDEX_op_mov_i64, { "r", "r" } }, - { INDEX_op_movi_i32, { "r" } }, - { INDEX_op_movi_i64, { "r" } }, - { INDEX_op_ld8u_i32, { "r", "r" } }, { INDEX_op_ld8s_i32, { "r", "r" } }, { INDEX_op_ld16u_i32, { "r", "r" } }, diff --git a/tcg/ppc64/tcg-target.h b/tcg/ppc64/tcg-target.h index 3815b840a6..29f479a3cf 100644 --- a/tcg/ppc64/tcg-target.h +++ b/tcg/ppc64/tcg-target.h @@ -25,6 +25,7 @@ #define TCG_TARGET_PPC64 1 #define TCG_TARGET_NB_REGS 32 +#define TCG_TARGET_INSN_UNIT_SIZE 4 typedef enum { TCG_REG_R0 = 0, diff --git a/tcg/s390/tcg-target.c b/tcg/s390/tcg-target.c index 1d912a7937..ebdd0743cf 100644 --- a/tcg/s390/tcg-target.c +++ b/tcg/s390/tcg-target.c @@ -320,7 +320,7 @@ static const uint8_t tcg_cond_to_ltr_cond[] = { #ifdef CONFIG_SOFTMMU /* helper signature: helper_ld_mmu(CPUState *env, target_ulong addr, int mmu_idx) */ -static const void * const qemu_ld_helpers[4] = { +static void * const qemu_ld_helpers[4] = { helper_ldb_mmu, helper_ldw_mmu, helper_ldl_mmu, @@ -329,7 +329,7 @@ static const void * const qemu_ld_helpers[4] = { /* helper signature: helper_st_mmu(CPUState *env, target_ulong addr, uintxx_t val, int mmu_idx) */ -static const void * const qemu_st_helpers[4] = { +static void * const qemu_st_helpers[4] = { helper_stb_mmu, helper_stw_mmu, helper_stl_mmu, @@ -337,7 +337,7 @@ static const void * const qemu_st_helpers[4] = { }; #endif -static uint8_t *tb_ret_addr; +static tcg_insn_unit *tb_ret_addr; /* A list of relevant facilities used by this translator. Some of these are required for proper operation, and these are checked at startup. */ @@ -350,23 +350,20 @@ static uint8_t *tb_ret_addr; static uint64_t facilities; -static void patch_reloc(uint8_t *code_ptr, int type, +static void patch_reloc(tcg_insn_unit *code_ptr, int type, intptr_t value, intptr_t addend) { - intptr_t code_ptr_tl = (intptr_t)code_ptr; - intptr_t pcrel2; - - /* ??? Not the usual definition of "addend". */ - pcrel2 = (value - (code_ptr_tl + addend)) >> 1; + intptr_t pcrel2 = (tcg_insn_unit *)value - (code_ptr - 1); + assert(addend == -2); switch (type) { case R_390_PC16DBL: assert(pcrel2 == (int16_t)pcrel2); - *(int16_t *)code_ptr = pcrel2; + tcg_patch16(code_ptr, pcrel2); break; case R_390_PC32DBL: assert(pcrel2 == (int32_t)pcrel2); - *(int32_t *)code_ptr = pcrel2; + tcg_patch32(code_ptr, pcrel2); break; default: tcg_abort(); @@ -672,7 +669,7 @@ static void tcg_out_movi(TCGContext *s, TCGType type, /* Try for PC-relative address load. */ if ((sval & 1) == 0) { - intptr_t off = (sval - (intptr_t)s->code_ptr) >> 1; + ptrdiff_t off = tcg_pcrel_diff(s, (void *)sval) >> 1; if (off == (int32_t)off) { tcg_out_insn(s, RIL, LARL, ret, off); return; @@ -789,10 +786,10 @@ static inline void tcg_out_st(TCGContext *s, TCGType type, TCGReg data, /* load data from an absolute host address */ static void tcg_out_ld_abs(TCGContext *s, TCGType type, TCGReg dest, void *abs) { - tcg_target_long addr = (tcg_target_long)abs; + intptr_t addr = (intptr_t)abs; - if (facilities & FACILITY_GEN_INST_EXT) { - tcg_target_long disp = (addr - (tcg_target_long)s->code_ptr) >> 1; + if ((facilities & FACILITY_GEN_INST_EXT) && !(addr & 1)) { + ptrdiff_t disp = tcg_pcrel_diff(s, abs) >> 1; if (disp == (int32_t)disp) { if (type == TCG_TYPE_I32) { tcg_out_insn(s, RIL, LRL, dest, disp); @@ -1154,15 +1151,15 @@ static void tgen_deposit(TCGContext *s, TCGReg dest, TCGReg src, tcg_out_risbg(s, dest, src, msb, lsb, ofs, 0); } -static void tgen_gotoi(TCGContext *s, int cc, tcg_target_long dest) +static void tgen_gotoi(TCGContext *s, int cc, tcg_insn_unit *dest) { - tcg_target_long off = (dest - (tcg_target_long)s->code_ptr) >> 1; - if (off > -0x8000 && off < 0x7fff) { + ptrdiff_t off = dest - s->code_ptr; + if (off == (int16_t)off) { tcg_out_insn(s, RI, BRC, cc, off); } else if (off == (int32_t)off) { tcg_out_insn(s, RIL, BRCL, cc, off); } else { - tcg_out_movi(s, TCG_TYPE_PTR, TCG_TMP0, dest); + tcg_out_movi(s, TCG_TYPE_PTR, TCG_TMP0, (uintptr_t)dest); tcg_out_insn(s, RR, BCR, cc, TCG_TMP0); } } @@ -1171,15 +1168,15 @@ static void tgen_branch(TCGContext *s, int cc, int labelno) { TCGLabel* l = &s->labels[labelno]; if (l->has_value) { - tgen_gotoi(s, cc, l->u.value); + tgen_gotoi(s, cc, l->u.value_ptr); } else if (USE_LONG_BRANCHES) { tcg_out16(s, RIL_BRCL | (cc << 4)); tcg_out_reloc(s, s->code_ptr, R_390_PC32DBL, labelno, -2); - s->code_ptr += 4; + s->code_ptr += 2; } else { tcg_out16(s, RI_BRC | (cc << 4)); tcg_out_reloc(s, s->code_ptr, R_390_PC16DBL, labelno, -2); - s->code_ptr += 2; + s->code_ptr += 1; } } @@ -1187,14 +1184,14 @@ static void tgen_compare_branch(TCGContext *s, S390Opcode opc, int cc, TCGReg r1, TCGReg r2, int labelno) { TCGLabel* l = &s->labels[labelno]; - tcg_target_long off; + intptr_t off; if (l->has_value) { - off = (l->u.value - (tcg_target_long)s->code_ptr) >> 1; + off = l->u.value_ptr - s->code_ptr; } else { /* We need to keep the offset unchanged for retranslation. */ - off = ((int16_t *)s->code_ptr)[1]; - tcg_out_reloc(s, s->code_ptr + 2, R_390_PC16DBL, labelno, -2); + off = s->code_ptr[1]; + tcg_out_reloc(s, s->code_ptr + 1, R_390_PC16DBL, labelno, -2); } tcg_out16(s, (opc & 0xff00) | (r1 << 4) | r2); @@ -1209,11 +1206,11 @@ static void tgen_compare_imm_branch(TCGContext *s, S390Opcode opc, int cc, tcg_target_long off; if (l->has_value) { - off = (l->u.value - (tcg_target_long)s->code_ptr) >> 1; + off = l->u.value_ptr - s->code_ptr; } else { /* We need to keep the offset unchanged for retranslation. */ - off = ((int16_t *)s->code_ptr)[1]; - tcg_out_reloc(s, s->code_ptr + 2, R_390_PC16DBL, labelno, -2); + off = s->code_ptr[1]; + tcg_out_reloc(s, s->code_ptr + 1, R_390_PC16DBL, labelno, -2); } tcg_out16(s, (opc & 0xff00) | (r1 << 4) | cc); @@ -1272,13 +1269,13 @@ static void tgen_brcond(TCGContext *s, TCGType type, TCGCond c, tgen_branch(s, cc, labelno); } -static void tgen_calli(TCGContext *s, tcg_target_long dest) +static void tcg_out_call(TCGContext *s, tcg_insn_unit *dest) { - tcg_target_long off = (dest - (tcg_target_long)s->code_ptr) >> 1; + ptrdiff_t off = dest - s->code_ptr; if (off == (int32_t)off) { tcg_out_insn(s, RIL, BRASL, TCG_REG_R14, off); } else { - tcg_out_movi(s, TCG_TYPE_PTR, TCG_TMP0, dest); + tcg_out_movi(s, TCG_TYPE_PTR, TCG_TMP0, (uintptr_t)dest); tcg_out_insn(s, RR, BASR, TCG_REG_R14, TCG_TMP0); } } @@ -1395,14 +1392,14 @@ static void tcg_out_qemu_st_direct(TCGContext *s, int opc, TCGReg data, #if defined(CONFIG_SOFTMMU) static TCGReg tcg_prepare_qemu_ldst(TCGContext* s, TCGReg data_reg, TCGReg addr_reg, int mem_index, int opc, - uint16_t **label2_ptr_p, int is_store) + tcg_insn_unit **label2_ptr_p, int is_store) { const TCGReg arg0 = tcg_target_call_iarg_regs[0]; const TCGReg arg1 = tcg_target_call_iarg_regs[1]; const TCGReg arg2 = tcg_target_call_iarg_regs[2]; const TCGReg arg3 = tcg_target_call_iarg_regs[3]; int s_bits = opc & 3; - uint16_t *label1_ptr; + tcg_insn_unit *label1_ptr; tcg_target_long ofs; if (TARGET_LONG_BITS == 32) { @@ -1436,7 +1433,7 @@ static TCGReg tcg_prepare_qemu_ldst(TCGContext* s, TCGReg data_reg, tcg_out_mov(s, TCG_TYPE_I64, arg1, addr_reg); } - label1_ptr = (uint16_t*)s->code_ptr; + label1_ptr = s->code_ptr; /* je label1 (offset will be patched in later) */ tcg_out_insn(s, RI, BRC, S390_CC_EQ, 0); @@ -1463,11 +1460,11 @@ static TCGReg tcg_prepare_qemu_ldst(TCGContext* s, TCGReg data_reg, } tcg_out_movi(s, TCG_TYPE_I32, arg3, mem_index); tcg_out_mov(s, TCG_TYPE_I64, arg0, TCG_AREG0); - tgen_calli(s, (tcg_target_ulong)qemu_st_helpers[s_bits]); + tcg_out_call(s, qemu_st_helpers[s_bits]); } else { tcg_out_movi(s, TCG_TYPE_I32, arg2, mem_index); tcg_out_mov(s, TCG_TYPE_I64, arg0, TCG_AREG0); - tgen_calli(s, (tcg_target_ulong)qemu_ld_helpers[s_bits]); + tcg_out_call(s, qemu_ld_helpers[s_bits]); /* sign extension */ switch (opc) { @@ -1488,13 +1485,12 @@ static TCGReg tcg_prepare_qemu_ldst(TCGContext* s, TCGReg data_reg, } /* jump to label2 (end) */ - *label2_ptr_p = (uint16_t*)s->code_ptr; + *label2_ptr_p = s->code_ptr; tcg_out_insn(s, RI, BRC, S390_CC_ALWAYS, 0); /* this is label1, patch branch */ - *(label1_ptr + 1) = ((unsigned long)s->code_ptr - - (unsigned long)label1_ptr) >> 1; + label1_ptr[1] = s->code_ptr - label1_ptr; ofs = offsetof(CPUArchState, tlb_table[mem_index][0].addend); assert(ofs < 0x80000); @@ -1504,11 +1500,10 @@ static TCGReg tcg_prepare_qemu_ldst(TCGContext* s, TCGReg data_reg, return arg1; } -static void tcg_finish_qemu_ldst(TCGContext* s, uint16_t *label2_ptr) +static void tcg_finish_qemu_ldst(TCGContext* s, tcg_insn_unit *label2_ptr) { /* patch branch */ - *(label2_ptr + 1) = ((unsigned long)s->code_ptr - - (unsigned long)label2_ptr) >> 1; + label2_ptr[1] = s->code_ptr - label2_ptr; } #else static void tcg_prepare_user_ldst(TCGContext *s, TCGReg *addr_reg, @@ -1535,7 +1530,7 @@ static void tcg_out_qemu_ld(TCGContext* s, const TCGArg* args, int opc) TCGReg addr_reg, data_reg; #if defined(CONFIG_SOFTMMU) int mem_index; - uint16_t *label2_ptr; + tcg_insn_unit *label2_ptr; #else TCGReg index_reg; tcg_target_long disp; @@ -1564,7 +1559,7 @@ static void tcg_out_qemu_st(TCGContext* s, const TCGArg* args, int opc) TCGReg addr_reg, data_reg; #if defined(CONFIG_SOFTMMU) int mem_index; - uint16_t *label2_ptr; + tcg_insn_unit *label2_ptr; #else TCGReg index_reg; tcg_target_long disp; @@ -1602,7 +1597,7 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, case INDEX_op_exit_tb: /* return value */ tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R2, args[0]); - tgen_gotoi(s, S390_CC_ALWAYS, (unsigned long)tb_ret_addr); + tgen_gotoi(s, S390_CC_ALWAYS, tb_ret_addr); break; case INDEX_op_goto_tb: @@ -1614,22 +1609,7 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, /* and go there */ tcg_out_insn(s, RR, BCR, S390_CC_ALWAYS, TCG_TMP0); } - s->tb_next_offset[args[0]] = s->code_ptr - s->code_buf; - break; - - case INDEX_op_call: - if (const_args[0]) { - tgen_calli(s, args[0]); - } else { - tcg_out_insn(s, RR, BASR, TCG_REG_R14, args[0]); - } - break; - - case INDEX_op_mov_i32: - tcg_out_mov(s, TCG_TYPE_I32, args[0], args[1]); - break; - case INDEX_op_movi_i32: - tcg_out_movi(s, TCG_TYPE_I32, args[0], args[1]); + s->tb_next_offset[args[0]] = tcg_current_code_size(s); break; OP_32_64(ld8u): @@ -1864,13 +1844,6 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_qemu_st(s, args, LD_UINT64); break; - case INDEX_op_mov_i64: - tcg_out_mov(s, TCG_TYPE_I64, args[0], args[1]); - break; - case INDEX_op_movi_i64: - tcg_out_movi(s, TCG_TYPE_I64, args[0], args[1]); - break; - case INDEX_op_ld16s_i64: tcg_out_mem(s, 0, RXY_LGH, args[0], args[1], TCG_REG_NONE, args[2]); break; @@ -2075,8 +2048,12 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, tgen_deposit(s, args[0], args[2], args[3], args[4]); break; + case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ + case INDEX_op_mov_i64: + case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ + case INDEX_op_movi_i64: + case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: - fprintf(stderr,"unimplemented opc 0x%x\n",opc); tcg_abort(); } } @@ -2084,12 +2061,8 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, static const TCGTargetOpDef s390_op_defs[] = { { INDEX_op_exit_tb, { } }, { INDEX_op_goto_tb, { } }, - { INDEX_op_call, { "ri" } }, { INDEX_op_br, { } }, - { INDEX_op_mov_i32, { "r", "r" } }, - { INDEX_op_movi_i32, { "r" } }, - { INDEX_op_ld8u_i32, { "r", "r" } }, { INDEX_op_ld8s_i32, { "r", "r" } }, { INDEX_op_ld16u_i32, { "r", "r" } }, @@ -2147,9 +2120,6 @@ static const TCGTargetOpDef s390_op_defs[] = { { INDEX_op_qemu_st32, { "L", "L" } }, { INDEX_op_qemu_st64, { "L", "L" } }, - { INDEX_op_mov_i64, { "r", "r" } }, - { INDEX_op_movi_i64, { "r" } }, - { INDEX_op_ld8u_i64, { "r", "r" } }, { INDEX_op_ld8s_i64, { "r", "r" } }, { INDEX_op_ld16u_i64, { "r", "r" } }, diff --git a/tcg/s390/tcg-target.h b/tcg/s390/tcg-target.h index 755c00237e..5bf733eb3d 100644 --- a/tcg/s390/tcg-target.h +++ b/tcg/s390/tcg-target.h @@ -24,6 +24,8 @@ #ifndef TCG_TARGET_S390 #define TCG_TARGET_S390 1 +#define TCG_TARGET_INSN_UNIT_SIZE 2 + typedef enum TCGReg { TCG_REG_R0 = 0, TCG_REG_R1, diff --git a/tcg/sparc/tcg-target.c b/tcg/sparc/tcg-target.c index 5468ff5c16..17ff5773ad 100644 --- a/tcg/sparc/tcg-target.c +++ b/tcg/sparc/tcg-target.c @@ -279,37 +279,32 @@ static inline int check_fit_i32(int32_t val, unsigned int bits) # define check_fit_ptr check_fit_i32 #endif -static void patch_reloc(uint8_t *code_ptr, int type, +static void patch_reloc(tcg_insn_unit *code_ptr, int type, intptr_t value, intptr_t addend) { uint32_t insn; - value += addend; + + assert(addend == 0); + value = tcg_ptr_byte_diff((tcg_insn_unit *)value, code_ptr); + switch (type) { - case R_SPARC_32: - if (value != (uint32_t)value) { - tcg_abort(); - } - *(uint32_t *)code_ptr = value; - break; case R_SPARC_WDISP16: - value -= (intptr_t)code_ptr; if (!check_fit_ptr(value >> 2, 16)) { tcg_abort(); } - insn = *(uint32_t *)code_ptr; + insn = *code_ptr; insn &= ~INSN_OFF16(-1); insn |= INSN_OFF16(value); - *(uint32_t *)code_ptr = insn; + *code_ptr = insn; break; case R_SPARC_WDISP19: - value -= (intptr_t)code_ptr; if (!check_fit_ptr(value >> 2, 19)) { tcg_abort(); } - insn = *(uint32_t *)code_ptr; + insn = *code_ptr; insn &= ~INSN_OFF19(-1); insn |= INSN_OFF19(value); - *(uint32_t *)code_ptr = insn; + *code_ptr = insn; break; default: tcg_abort(); @@ -573,10 +568,10 @@ static void tcg_out_bpcc(TCGContext *s, int scond, int flags, int label) int off19; if (l->has_value) { - off19 = INSN_OFF19(l->u.value - (unsigned long)s->code_ptr); + off19 = INSN_OFF19(tcg_pcrel_diff(s, l->u.value_ptr)); } else { /* Make sure to preserve destinations during retranslation. */ - off19 = *(uint32_t *)s->code_ptr & INSN_OFF19(-1); + off19 = *s->code_ptr & INSN_OFF19(-1); tcg_out_reloc(s, s->code_ptr, R_SPARC_WDISP19, label, 0); } tcg_out_bpcc0(s, scond, flags, off19); @@ -620,10 +615,10 @@ static void tcg_out_brcond_i64(TCGContext *s, TCGCond cond, TCGReg arg1, int off16; if (l->has_value) { - off16 = INSN_OFF16(l->u.value - (unsigned long)s->code_ptr); + off16 = INSN_OFF16(tcg_pcrel_diff(s, l->u.value_ptr)); } else { /* Make sure to preserve destinations during retranslation. */ - off16 = *(uint32_t *)s->code_ptr & INSN_OFF16(-1); + off16 = *s->code_ptr & INSN_OFF16(-1); tcg_out_reloc(s, s->code_ptr, R_SPARC_WDISP16, label, 0); } tcg_out32(s, INSN_OP(0) | INSN_OP2(3) | BPR_PT | INSN_RS1(arg1) @@ -740,62 +735,66 @@ static void tcg_out_addsub2(TCGContext *s, TCGReg rl, TCGReg rh, tcg_out_mov(s, TCG_TYPE_I32, rl, tmp); } -static void tcg_out_calli(TCGContext *s, uintptr_t dest) +static void tcg_out_call_nodelay(TCGContext *s, tcg_insn_unit *dest) { - intptr_t disp = dest - (uintptr_t)s->code_ptr; + ptrdiff_t disp = tcg_pcrel_diff(s, dest); if (disp == (int32_t)disp) { tcg_out32(s, CALL | (uint32_t)disp >> 2); } else { - tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_T1, dest & ~0xfff); - tcg_out_arithi(s, TCG_REG_O7, TCG_REG_T1, dest & 0xfff, JMPL); + uintptr_t desti = (uintptr_t)dest; + tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_T1, desti & ~0xfff); + tcg_out_arithi(s, TCG_REG_O7, TCG_REG_T1, desti & 0xfff, JMPL); } } +static void tcg_out_call(TCGContext *s, tcg_insn_unit *dest) +{ + tcg_out_call_nodelay(s, dest); + tcg_out_nop(s); +} + #ifdef CONFIG_SOFTMMU -static uintptr_t qemu_ld_trampoline[16]; -static uintptr_t qemu_st_trampoline[16]; +static tcg_insn_unit *qemu_ld_trampoline[16]; +static tcg_insn_unit *qemu_st_trampoline[16]; static void build_trampolines(TCGContext *s) { - static uintptr_t const qemu_ld_helpers[16] = { - [MO_UB] = (uintptr_t)helper_ret_ldub_mmu, - [MO_SB] = (uintptr_t)helper_ret_ldsb_mmu, - [MO_LEUW] = (uintptr_t)helper_le_lduw_mmu, - [MO_LESW] = (uintptr_t)helper_le_ldsw_mmu, - [MO_LEUL] = (uintptr_t)helper_le_ldul_mmu, - [MO_LEQ] = (uintptr_t)helper_le_ldq_mmu, - [MO_BEUW] = (uintptr_t)helper_be_lduw_mmu, - [MO_BESW] = (uintptr_t)helper_be_ldsw_mmu, - [MO_BEUL] = (uintptr_t)helper_be_ldul_mmu, - [MO_BEQ] = (uintptr_t)helper_be_ldq_mmu, + static void * const qemu_ld_helpers[16] = { + [MO_UB] = helper_ret_ldub_mmu, + [MO_SB] = helper_ret_ldsb_mmu, + [MO_LEUW] = helper_le_lduw_mmu, + [MO_LESW] = helper_le_ldsw_mmu, + [MO_LEUL] = helper_le_ldul_mmu, + [MO_LEQ] = helper_le_ldq_mmu, + [MO_BEUW] = helper_be_lduw_mmu, + [MO_BESW] = helper_be_ldsw_mmu, + [MO_BEUL] = helper_be_ldul_mmu, + [MO_BEQ] = helper_be_ldq_mmu, }; - static uintptr_t const qemu_st_helpers[16] = { - [MO_UB] = (uintptr_t)helper_ret_stb_mmu, - [MO_LEUW] = (uintptr_t)helper_le_stw_mmu, - [MO_LEUL] = (uintptr_t)helper_le_stl_mmu, - [MO_LEQ] = (uintptr_t)helper_le_stq_mmu, - [MO_BEUW] = (uintptr_t)helper_be_stw_mmu, - [MO_BEUL] = (uintptr_t)helper_be_stl_mmu, - [MO_BEQ] = (uintptr_t)helper_be_stq_mmu, + static void * const qemu_st_helpers[16] = { + [MO_UB] = helper_ret_stb_mmu, + [MO_LEUW] = helper_le_stw_mmu, + [MO_LEUL] = helper_le_stl_mmu, + [MO_LEQ] = helper_le_stq_mmu, + [MO_BEUW] = helper_be_stw_mmu, + [MO_BEUL] = helper_be_stl_mmu, + [MO_BEQ] = helper_be_stq_mmu, }; int i; TCGReg ra; - uintptr_t tramp; for (i = 0; i < 16; ++i) { - if (qemu_ld_helpers[i] == 0) { + if (qemu_ld_helpers[i] == NULL) { continue; } /* May as well align the trampoline. */ - tramp = (uintptr_t)s->code_ptr; - while (tramp & 15) { + while ((uintptr_t)s->code_ptr & 15) { tcg_out_nop(s); - tramp += 4; } - qemu_ld_trampoline[i] = tramp; + qemu_ld_trampoline[i] = s->code_ptr; if (SPARC64 || TARGET_LONG_BITS == 32) { ra = TCG_REG_O3; @@ -810,22 +809,20 @@ static void build_trampolines(TCGContext *s) /* Set the env operand. */ tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_O0, TCG_AREG0); /* Tail call. */ - tcg_out_calli(s, qemu_ld_helpers[i]); + tcg_out_call_nodelay(s, qemu_ld_helpers[i]); tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_O7, ra); } for (i = 0; i < 16; ++i) { - if (qemu_st_helpers[i] == 0) { + if (qemu_st_helpers[i] == NULL) { continue; } /* May as well align the trampoline. */ - tramp = (uintptr_t)s->code_ptr; - while (tramp & 15) { + while ((uintptr_t)s->code_ptr & 15) { tcg_out_nop(s); - tramp += 4; } - qemu_st_trampoline[i] = tramp; + qemu_st_trampoline[i] = s->code_ptr; if (SPARC64) { ra = TCG_REG_O4; @@ -859,7 +856,7 @@ static void build_trampolines(TCGContext *s) /* Set the env operand. */ tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_O0, TCG_AREG0); /* Tail call. */ - tcg_out_calli(s, qemu_st_helpers[i]); + tcg_out_call_nodelay(s, qemu_st_helpers[i]); tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_O7, ra); } } @@ -1005,8 +1002,8 @@ static void tcg_out_qemu_ld(TCGContext *s, TCGReg data, TCGReg addr, #ifdef CONFIG_SOFTMMU TCGMemOp s_bits = memop & MO_SIZE; TCGReg addrz, param; - uintptr_t func; - uint32_t *label_ptr; + tcg_insn_unit *func; + tcg_insn_unit *label_ptr; addrz = tcg_out_tlb_load(s, addr, memi, s_bits, offsetof(CPUTLBEntry, addr_read)); @@ -1016,7 +1013,7 @@ static void tcg_out_qemu_ld(TCGContext *s, TCGReg data, TCGReg addr, over the TLB Miss case. */ /* beq,a,pt %[xi]cc, label0 */ - label_ptr = (uint32_t *)s->code_ptr; + label_ptr = s->code_ptr; tcg_out_bpcc0(s, COND_E, BPCC_A | BPCC_PT | (TARGET_LONG_BITS == 64 ? BPCC_XCC : BPCC_ICC), 0); /* delay slot */ @@ -1038,8 +1035,8 @@ static void tcg_out_qemu_ld(TCGContext *s, TCGReg data, TCGReg addr, } else { func = qemu_ld_trampoline[memop]; } - assert(func != 0); - tcg_out_calli(s, func); + assert(func != NULL); + tcg_out_call_nodelay(s, func); /* delay slot */ tcg_out_movi(s, TCG_TYPE_I32, param, memi); @@ -1067,7 +1064,7 @@ static void tcg_out_qemu_ld(TCGContext *s, TCGReg data, TCGReg addr, } } - *label_ptr |= INSN_OFF19((uintptr_t)s->code_ptr - (uintptr_t)label_ptr); + *label_ptr |= INSN_OFF19(tcg_ptr_byte_diff(s->code_ptr, label_ptr)); #else if (SPARC64 && TARGET_LONG_BITS == 32) { tcg_out_arithi(s, TCG_REG_T1, addr, 0, SHIFT_SRL); @@ -1085,8 +1082,8 @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg data, TCGReg addr, #ifdef CONFIG_SOFTMMU TCGMemOp s_bits = memop & MO_SIZE; TCGReg addrz, param; - uintptr_t func; - uint32_t *label_ptr; + tcg_insn_unit *func; + tcg_insn_unit *label_ptr; addrz = tcg_out_tlb_load(s, addr, memi, s_bits, offsetof(CPUTLBEntry, addr_write)); @@ -1094,7 +1091,7 @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg data, TCGReg addr, /* The fast path is exactly one insn. Thus we can perform the entire TLB Hit in the (annulled) delay slot of the branch over TLB Miss. */ /* beq,a,pt %[xi]cc, label0 */ - label_ptr = (uint32_t *)s->code_ptr; + label_ptr = s->code_ptr; tcg_out_bpcc0(s, COND_E, BPCC_A | BPCC_PT | (TARGET_LONG_BITS == 64 ? BPCC_XCC : BPCC_ICC), 0); /* delay slot */ @@ -1115,12 +1112,12 @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg data, TCGReg addr, tcg_out_mov(s, TCG_TYPE_REG, param++, data); func = qemu_st_trampoline[memop]; - assert(func != 0); - tcg_out_calli(s, func); + assert(func != NULL); + tcg_out_call_nodelay(s, func); /* delay slot */ tcg_out_movi(s, TCG_TYPE_REG, param, memi); - *label_ptr |= INSN_OFF19((uintptr_t)s->code_ptr - (uintptr_t)label_ptr); + *label_ptr |= INSN_OFF19(tcg_ptr_byte_diff(s->code_ptr, label_ptr)); #else if (SPARC64 && TARGET_LONG_BITS == 32) { tcg_out_arithi(s, TCG_REG_T1, addr, 0, SHIFT_SRL); @@ -1159,26 +1156,16 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, case INDEX_op_goto_tb: if (s->tb_jmp_offset) { /* direct jump method */ - uint32_t old_insn = *(uint32_t *)s->code_ptr; - s->tb_jmp_offset[a0] = s->code_ptr - s->code_buf; + s->tb_jmp_offset[a0] = tcg_current_code_size(s); /* Make sure to preserve links during retranslation. */ - tcg_out32(s, CALL | (old_insn & ~INSN_OP(-1))); + tcg_out32(s, CALL | (*s->code_ptr & ~INSN_OP(-1))); } else { /* indirect jump method */ tcg_out_ld_ptr(s, TCG_REG_T1, (uintptr_t)(s->tb_next + a0)); tcg_out_arithi(s, TCG_REG_G0, TCG_REG_T1, 0, JMPL); } tcg_out_nop(s); - s->tb_next_offset[a0] = s->code_ptr - s->code_buf; - break; - case INDEX_op_call: - if (const_args[0]) { - tcg_out_calli(s, a0); - } else { - tcg_out_arithi(s, TCG_REG_O7, a0, 0, JMPL); - } - /* delay slot */ - tcg_out_nop(s); + s->tb_next_offset[a0] = tcg_current_code_size(s); break; case INDEX_op_br: tcg_out_bpcc(s, COND_A, BPCC_PT, a0); @@ -1373,13 +1360,12 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_arithc(s, a0, TCG_REG_G0, a1, const_args[1], c); break; + case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_mov_i32: + case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ case INDEX_op_movi_i64: - case INDEX_op_movi_i32: - /* Always implemented with tcg_out_mov/i, never with tcg_out_op. */ + case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: - /* Opcode not implemented. */ tcg_abort(); } } @@ -1387,11 +1373,8 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, static const TCGTargetOpDef sparc_op_defs[] = { { INDEX_op_exit_tb, { } }, { INDEX_op_goto_tb, { } }, - { INDEX_op_call, { "ri" } }, { INDEX_op_br, { } }, - { INDEX_op_mov_i32, { "r", "r" } }, - { INDEX_op_movi_i32, { "r" } }, { INDEX_op_ld8u_i32, { "r", "r" } }, { INDEX_op_ld8s_i32, { "r", "r" } }, { INDEX_op_ld16u_i32, { "r", "r" } }, @@ -1428,8 +1411,6 @@ static const TCGTargetOpDef sparc_op_defs[] = { { INDEX_op_mulu2_i32, { "r", "r", "rZ", "rJ" } }, { INDEX_op_muls2_i32, { "r", "r", "rZ", "rJ" } }, - { INDEX_op_mov_i64, { "R", "R" } }, - { INDEX_op_movi_i64, { "R" } }, { INDEX_op_ld8u_i64, { "R", "r" } }, { INDEX_op_ld8s_i64, { "R", "r" } }, { INDEX_op_ld16u_i64, { "R", "r" } }, diff --git a/tcg/sparc/tcg-target.h b/tcg/sparc/tcg-target.h index 3a903dbfd0..473bfc71fd 100644 --- a/tcg/sparc/tcg-target.h +++ b/tcg/sparc/tcg-target.h @@ -26,6 +26,7 @@ #define TCG_TARGET_REG_BITS 64 +#define TCG_TARGET_INSN_UNIT_SIZE 4 #define TCG_TARGET_NB_REGS 32 typedef enum { diff --git a/tcg/tcg-be-ldst.h b/tcg/tcg-be-ldst.h index ad94c0ca51..49b3de61ea 100644 --- a/tcg/tcg-be-ldst.h +++ b/tcg/tcg-be-ldst.h @@ -31,8 +31,8 @@ typedef struct TCGLabelQemuLdst { TCGReg datalo_reg; /* reg index for low word to be loaded or stored */ TCGReg datahi_reg; /* reg index for high word to be loaded or stored */ int mem_index; /* soft MMU memory index */ - uint8_t *raddr; /* gen code addr of the next IR of qemu_ld/st IR */ - uint8_t *label_ptr[2]; /* label pointers to be updated */ + tcg_insn_unit *raddr; /* gen code addr of the next IR of qemu_ld/st IR */ + tcg_insn_unit *label_ptr[2]; /* label pointers to be updated */ } TCGLabelQemuLdst; typedef struct TCGBackendData { diff --git a/tcg/tcg-op.h b/tcg/tcg-op.h index 6dbbb38838..bdd0139482 100644 --- a/tcg/tcg-op.h +++ b/tcg/tcg-op.h @@ -390,11 +390,7 @@ static inline int tcg_gen_sizemask(int n, int is_64bit, int is_signed) static inline void tcg_gen_helperN(void *func, int flags, int sizemask, TCGArg ret, int nargs, TCGArg *args) { - TCGv_ptr fn; - fn = tcg_const_ptr(func); - tcg_gen_callN(&tcg_ctx, fn, flags, sizemask, ret, - nargs, args); - tcg_temp_free_ptr(fn); + tcg_gen_callN(&tcg_ctx, func, flags, sizemask, ret, nargs, args); } /* Note: Both tcg_gen_helper32() and tcg_gen_helper64() are currently @@ -405,29 +401,23 @@ static inline void tcg_gen_helperN(void *func, int flags, int sizemask, static inline void tcg_gen_helper32(void *func, int sizemask, TCGv_i32 ret, TCGv_i32 a, TCGv_i32 b) { - TCGv_ptr fn; TCGArg args[2]; - fn = tcg_const_ptr(func); args[0] = GET_TCGV_I32(a); args[1] = GET_TCGV_I32(b); - tcg_gen_callN(&tcg_ctx, fn, + tcg_gen_callN(&tcg_ctx, func, TCG_CALL_NO_READ_GLOBALS | TCG_CALL_NO_SIDE_EFFECTS, sizemask, GET_TCGV_I32(ret), 2, args); - tcg_temp_free_ptr(fn); } static inline void tcg_gen_helper64(void *func, int sizemask, TCGv_i64 ret, TCGv_i64 a, TCGv_i64 b) { - TCGv_ptr fn; TCGArg args[2]; - fn = tcg_const_ptr(func); args[0] = GET_TCGV_I64(a); args[1] = GET_TCGV_I64(b); - tcg_gen_callN(&tcg_ctx, fn, + tcg_gen_callN(&tcg_ctx, func, TCG_CALL_NO_READ_GLOBALS | TCG_CALL_NO_SIDE_EFFECTS, sizemask, GET_TCGV_I64(ret), 2, args); - tcg_temp_free_ptr(fn); } /* 32 bit ops */ diff --git a/tcg/tcg-opc.h b/tcg/tcg-opc.h index adb6ce1bb9..71ba64ac16 100644 --- a/tcg/tcg-opc.h +++ b/tcg/tcg-opc.h @@ -40,7 +40,7 @@ DEF(discard, 1, 0, 0, TCG_OPF_NOT_PRESENT) DEF(set_label, 0, 0, 1, TCG_OPF_BB_END | TCG_OPF_NOT_PRESENT) /* variable number of parameters */ -DEF(call, 0, 1, 2, TCG_OPF_CALL_CLOBBER) +DEF(call, 0, 0, 3, TCG_OPF_CALL_CLOBBER | TCG_OPF_NOT_PRESENT) DEF(br, 0, 0, 1, TCG_OPF_BB_END) @@ -51,8 +51,8 @@ DEF(br, 0, 0, 1, TCG_OPF_BB_END) # define IMPL64 TCG_OPF_64BIT #endif -DEF(mov_i32, 1, 1, 0, 0) -DEF(movi_i32, 1, 0, 1, 0) +DEF(mov_i32, 1, 1, 0, TCG_OPF_NOT_PRESENT) +DEF(movi_i32, 1, 0, 1, TCG_OPF_NOT_PRESENT) DEF(setcond_i32, 1, 2, 1, 0) DEF(movcond_i32, 1, 4, 1, IMPL(TCG_TARGET_HAS_movcond_i32)) /* load/store */ @@ -110,8 +110,8 @@ DEF(eqv_i32, 1, 2, 0, IMPL(TCG_TARGET_HAS_eqv_i32)) DEF(nand_i32, 1, 2, 0, IMPL(TCG_TARGET_HAS_nand_i32)) DEF(nor_i32, 1, 2, 0, IMPL(TCG_TARGET_HAS_nor_i32)) -DEF(mov_i64, 1, 1, 0, IMPL64) -DEF(movi_i64, 1, 0, 1, IMPL64) +DEF(mov_i64, 1, 1, 0, TCG_OPF_64BIT | TCG_OPF_NOT_PRESENT) +DEF(movi_i64, 1, 0, 1, TCG_OPF_64BIT | TCG_OPF_NOT_PRESENT) DEF(setcond_i64, 1, 2, 1, IMPL64) DEF(movcond_i64, 1, 4, 1, IMPL64 | IMPL(TCG_TARGET_HAS_movcond_i64)) /* load/store */ diff --git a/tcg/tcg.c b/tcg/tcg.c index e71f7a0975..0670affe2f 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -65,7 +65,7 @@ /* Forward declarations for functions declared in tcg-target.c and used here. */ static void tcg_target_init(TCGContext *s); static void tcg_target_qemu_prologue(TCGContext *s); -static void patch_reloc(uint8_t *code_ptr, int type, +static void patch_reloc(tcg_insn_unit *code_ptr, int type, intptr_t value, intptr_t addend); /* The CIE and FDE header definitions will be common to all hosts. */ @@ -101,6 +101,7 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, const int *const_args); static void tcg_out_st(TCGContext *s, TCGType type, TCGReg arg, TCGReg arg1, intptr_t arg2); +static void tcg_out_call(TCGContext *s, tcg_insn_unit *target); static int tcg_target_const_match(tcg_target_long val, TCGType type, const TCGArgConstraint *arg_ct); static void tcg_out_tb_init(TCGContext *s); @@ -117,35 +118,87 @@ const size_t tcg_op_defs_max = ARRAY_SIZE(tcg_op_defs); static TCGRegSet tcg_target_available_regs[2]; static TCGRegSet tcg_target_call_clobber_regs; +#if TCG_TARGET_INSN_UNIT_SIZE == 1 static inline void tcg_out8(TCGContext *s, uint8_t v) { *s->code_ptr++ = v; } +static inline void tcg_patch8(tcg_insn_unit *p, uint8_t v) +{ + *p = v; +} +#endif + +#if TCG_TARGET_INSN_UNIT_SIZE <= 2 static inline void tcg_out16(TCGContext *s, uint16_t v) { - uint8_t *p = s->code_ptr; - *(uint16_t *)p = v; - s->code_ptr = p + 2; + if (TCG_TARGET_INSN_UNIT_SIZE == 2) { + *s->code_ptr++ = v; + } else { + tcg_insn_unit *p = s->code_ptr; + memcpy(p, &v, sizeof(v)); + s->code_ptr = p + (2 / TCG_TARGET_INSN_UNIT_SIZE); + } } +static inline void tcg_patch16(tcg_insn_unit *p, uint16_t v) +{ + if (TCG_TARGET_INSN_UNIT_SIZE == 2) { + *p = v; + } else { + memcpy(p, &v, sizeof(v)); + } +} +#endif + +#if TCG_TARGET_INSN_UNIT_SIZE <= 4 static inline void tcg_out32(TCGContext *s, uint32_t v) { - uint8_t *p = s->code_ptr; - *(uint32_t *)p = v; - s->code_ptr = p + 4; + if (TCG_TARGET_INSN_UNIT_SIZE == 4) { + *s->code_ptr++ = v; + } else { + tcg_insn_unit *p = s->code_ptr; + memcpy(p, &v, sizeof(v)); + s->code_ptr = p + (4 / TCG_TARGET_INSN_UNIT_SIZE); + } } +static inline void tcg_patch32(tcg_insn_unit *p, uint32_t v) +{ + if (TCG_TARGET_INSN_UNIT_SIZE == 4) { + *p = v; + } else { + memcpy(p, &v, sizeof(v)); + } +} +#endif + +#if TCG_TARGET_INSN_UNIT_SIZE <= 8 static inline void tcg_out64(TCGContext *s, uint64_t v) { - uint8_t *p = s->code_ptr; - *(uint64_t *)p = v; - s->code_ptr = p + 8; + if (TCG_TARGET_INSN_UNIT_SIZE == 8) { + *s->code_ptr++ = v; + } else { + tcg_insn_unit *p = s->code_ptr; + memcpy(p, &v, sizeof(v)); + s->code_ptr = p + (8 / TCG_TARGET_INSN_UNIT_SIZE); + } } +static inline void tcg_patch64(tcg_insn_unit *p, uint64_t v) +{ + if (TCG_TARGET_INSN_UNIT_SIZE == 8) { + *p = v; + } else { + memcpy(p, &v, sizeof(v)); + } +} +#endif + /* label relocation processing */ -static void tcg_out_reloc(TCGContext *s, uint8_t *code_ptr, int type, +static void tcg_out_reloc(TCGContext *s, tcg_insn_unit *code_ptr, int type, int label_index, intptr_t addend) { TCGLabel *l; @@ -168,23 +221,20 @@ static void tcg_out_reloc(TCGContext *s, uint8_t *code_ptr, int type, } } -static void tcg_out_label(TCGContext *s, int label_index, void *ptr) +static void tcg_out_label(TCGContext *s, int label_index, tcg_insn_unit *ptr) { - TCGLabel *l; - TCGRelocation *r; + TCGLabel *l = &s->labels[label_index]; intptr_t value = (intptr_t)ptr; + TCGRelocation *r; - l = &s->labels[label_index]; - if (l->has_value) { - tcg_abort(); - } - r = l->u.first_reloc; - while (r != NULL) { + assert(!l->has_value); + + for (r = l->u.first_reloc; r != NULL; r = r->next) { patch_reloc(r->ptr, r->type, value, r->addend); - r = r->next; } + l->has_value = 1; - l->u.value = value; + l->u.value_ptr = ptr; } int gen_new_label(void) @@ -339,7 +389,7 @@ void tcg_prologue_init(TCGContext *s) #ifdef DEBUG_DISAS if (qemu_loglevel_mask(CPU_LOG_TB_OUT_ASM)) { - size_t size = s->code_ptr - s->code_buf; + size_t size = tcg_current_code_size(s); qemu_log("PROLOGUE: [size=%zu]\n", size); log_disas(s->code_buf, size); qemu_log("\n"); @@ -656,7 +706,7 @@ int tcg_check_temp_count(void) /* Note: we convert the 64 bit args to 32 bit and do some alignment and endian swap. Maybe it would be better to do the alignment and endian swap in tcg_reg_alloc_call(). */ -void tcg_gen_callN(TCGContext *s, TCGv_ptr func, unsigned int flags, +void tcg_gen_callN(TCGContext *s, void *func, unsigned int flags, int sizemask, TCGArg ret, int nargs, TCGArg *args) { int i; @@ -783,11 +833,10 @@ void tcg_gen_callN(TCGContext *s, TCGv_ptr func, unsigned int flags, *s->gen_opparam_ptr++ = args[i]; real_args++; } - *s->gen_opparam_ptr++ = GET_TCGV_PTR(func); - + *s->gen_opparam_ptr++ = (uintptr_t)func; *s->gen_opparam_ptr++ = flags; - *nparam = (nb_rets << 16) | (real_args + 1); + *nparam = (nb_rets << 16) | real_args; /* total parameters, needed to go backward in the instruction stream */ *s->gen_opparam_ptr++ = 1 + nb_rets + real_args + 3; @@ -1194,49 +1243,21 @@ void tcg_dump_ops(TCGContext *s) nb_iargs = arg & 0xffff; nb_cargs = def->nb_cargs; - qemu_log(" %s ", def->name); - - /* function name */ - qemu_log("%s", - tcg_get_arg_str_idx(s, buf, sizeof(buf), - args[nb_oargs + nb_iargs - 1])); - /* flags */ - qemu_log(",$0x%" TCG_PRIlx, args[nb_oargs + nb_iargs]); - /* nb out args */ - qemu_log(",$%d", nb_oargs); - for(i = 0; i < nb_oargs; i++) { - qemu_log(","); - qemu_log("%s", tcg_get_arg_str_idx(s, buf, sizeof(buf), + /* function name, flags, out args */ + qemu_log(" %s %s,$0x%" TCG_PRIlx ",$%d", def->name, + tcg_find_helper(s, args[nb_oargs + nb_iargs]), + args[nb_oargs + nb_iargs + 1], nb_oargs); + for (i = 0; i < nb_oargs; i++) { + qemu_log(",%s", tcg_get_arg_str_idx(s, buf, sizeof(buf), args[i])); } - for(i = 0; i < (nb_iargs - 1); i++) { - qemu_log(","); - if (args[nb_oargs + i] == TCG_CALL_DUMMY_ARG) { - qemu_log(""); - } else { - qemu_log("%s", tcg_get_arg_str_idx(s, buf, sizeof(buf), - args[nb_oargs + i])); - } - } - } else if (c == INDEX_op_movi_i32 || c == INDEX_op_movi_i64) { - tcg_target_ulong val; - const char *name; - - nb_oargs = def->nb_oargs; - nb_iargs = def->nb_iargs; - nb_cargs = def->nb_cargs; - qemu_log(" %s %s,$", def->name, - tcg_get_arg_str_idx(s, buf, sizeof(buf), args[0])); - val = args[1]; - name = tcg_find_helper(s, val); - if (name) { - qemu_log("%s", name); - } else { - if (c == INDEX_op_movi_i32) { - qemu_log("0x%x", (uint32_t)val); - } else { - qemu_log("0x%" PRIx64 , (uint64_t)val); + for (i = 0; i < nb_iargs; i++) { + TCGArg arg = args[nb_oargs + i]; + const char *t = ""; + if (arg != TCG_CALL_DUMMY_ARG) { + t = tcg_get_arg_str_idx(s, buf, sizeof(buf), arg); } + qemu_log(",%s", t); } } else { qemu_log(" %s ", def->name); @@ -1499,9 +1520,9 @@ static inline void tcg_la_bb_end(TCGContext *s, uint8_t *dead_temps, temporaries are removed. */ static void tcg_liveness_analysis(TCGContext *s) { - int i, op_index, nb_args, nb_iargs, nb_oargs, arg, nb_ops; + int i, op_index, nb_args, nb_iargs, nb_oargs, nb_ops; TCGOpcode op, op_new, op_new2; - TCGArg *args; + TCGArg *args, arg; const TCGOpDef *def; uint8_t *dead_temps, *mem_temps; uint16_t dead_args; @@ -1531,15 +1552,15 @@ static void tcg_liveness_analysis(TCGContext *s) nb_args = args[-1]; args -= nb_args; - nb_iargs = args[0] & 0xffff; - nb_oargs = args[0] >> 16; - args++; - call_flags = args[nb_oargs + nb_iargs]; + arg = *args++; + nb_iargs = arg & 0xffff; + nb_oargs = arg >> 16; + call_flags = args[nb_oargs + nb_iargs + 1]; /* pure functions can be removed if their result is not used */ if (call_flags & TCG_CALL_NO_SIDE_EFFECTS) { - for(i = 0; i < nb_oargs; i++) { + for (i = 0; i < nb_oargs; i++) { arg = args[i]; if (!dead_temps[arg] || mem_temps[arg]) { goto do_not_remove_call; @@ -1553,7 +1574,7 @@ static void tcg_liveness_analysis(TCGContext *s) /* output args are dead */ dead_args = 0; sync_args = 0; - for(i = 0; i < nb_oargs; i++) { + for (i = 0; i < nb_oargs; i++) { arg = args[i]; if (dead_temps[arg]) { dead_args |= (1 << i); @@ -1576,7 +1597,7 @@ static void tcg_liveness_analysis(TCGContext *s) } /* input args are live */ - for(i = nb_oargs; i < nb_iargs + nb_oargs; i++) { + for (i = nb_oargs; i < nb_iargs + nb_oargs; i++) { arg = args[i]; if (arg != TCG_CALL_DUMMY_ARG) { if (dead_temps[arg]) { @@ -2075,13 +2096,12 @@ static void tcg_reg_alloc_mov(TCGContext *s, const TCGOpDef *def, { TCGRegSet allocated_regs; TCGTemp *ts, *ots; - const TCGArgConstraint *arg_ct, *oarg_ct; + TCGType type; tcg_regset_set(allocated_regs, s->reserved_regs); ots = &s->temps[args[0]]; ts = &s->temps[args[1]]; - oarg_ct = &def->args_ct[0]; - arg_ct = &def->args_ct[1]; + type = ots->type; /* If the source value is not in a register, and we're going to be forced to have it in a register in order to perform the copy, @@ -2089,12 +2109,13 @@ static void tcg_reg_alloc_mov(TCGContext *s, const TCGOpDef *def, we don't have to reload SOURCE the next time it is used. */ if (((NEED_SYNC_ARG(0) || ots->fixed_reg) && ts->val_type != TEMP_VAL_REG) || ts->val_type == TEMP_VAL_MEM) { - ts->reg = tcg_reg_alloc(s, arg_ct->u.regs, allocated_regs); + ts->reg = tcg_reg_alloc(s, tcg_target_available_regs[type], + allocated_regs); if (ts->val_type == TEMP_VAL_MEM) { - tcg_out_ld(s, ts->type, ts->reg, ts->mem_reg, ts->mem_offset); + tcg_out_ld(s, type, ts->reg, ts->mem_reg, ts->mem_offset); ts->mem_coherent = 1; } else if (ts->val_type == TEMP_VAL_CONST) { - tcg_out_movi(s, ts->type, ts->reg, ts->val); + tcg_out_movi(s, type, ts->reg, ts->val); } s->reg_to_temp[ts->reg] = args[1]; ts->val_type = TEMP_VAL_REG; @@ -2109,7 +2130,7 @@ static void tcg_reg_alloc_mov(TCGContext *s, const TCGOpDef *def, if (!ots->mem_allocated) { temp_allocate_frame(s, args[0]); } - tcg_out_st(s, ots->type, ts->reg, ots->mem_reg, ots->mem_offset); + tcg_out_st(s, type, ts->reg, ots->mem_reg, ots->mem_offset); if (IS_DEAD_ARG(1)) { temp_dead(s, args[1]); } @@ -2137,9 +2158,10 @@ static void tcg_reg_alloc_mov(TCGContext *s, const TCGOpDef *def, /* When allocating a new register, make sure to not spill the input one. */ tcg_regset_set_reg(allocated_regs, ts->reg); - ots->reg = tcg_reg_alloc(s, oarg_ct->u.regs, allocated_regs); + ots->reg = tcg_reg_alloc(s, tcg_target_available_regs[type], + allocated_regs); } - tcg_out_mov(s, ots->type, ots->reg, ts->reg); + tcg_out_mov(s, type, ots->reg, ts->reg); } ots->val_type = TEMP_VAL_REG; ots->mem_coherent = 0; @@ -2323,26 +2345,27 @@ static int tcg_reg_alloc_call(TCGContext *s, const TCGOpDef *def, uint16_t dead_args, uint8_t sync_args) { int nb_iargs, nb_oargs, flags, nb_regs, i, reg, nb_params; - TCGArg arg, func_arg; + TCGArg arg; TCGTemp *ts; intptr_t stack_offset; size_t call_stack_size; - uintptr_t func_addr; - int const_func_arg, allocate_args; + tcg_insn_unit *func_addr; + int allocate_args; TCGRegSet allocated_regs; - const TCGArgConstraint *arg_ct; arg = *args++; nb_oargs = arg >> 16; nb_iargs = arg & 0xffff; - nb_params = nb_iargs - 1; + nb_params = nb_iargs; - flags = args[nb_oargs + nb_iargs]; + func_addr = (tcg_insn_unit *)(intptr_t)args[nb_oargs + nb_iargs]; + flags = args[nb_oargs + nb_iargs + 1]; nb_regs = ARRAY_SIZE(tcg_target_call_iarg_regs); - if (nb_regs > nb_params) + if (nb_regs > nb_params) { nb_regs = nb_params; + } /* assign stack slots first */ call_stack_size = (nb_params - nb_regs) * sizeof(tcg_target_long); @@ -2410,40 +2433,6 @@ static int tcg_reg_alloc_call(TCGContext *s, const TCGOpDef *def, } } - /* assign function address */ - func_arg = args[nb_oargs + nb_iargs - 1]; - arg_ct = &def->args_ct[0]; - ts = &s->temps[func_arg]; - func_addr = ts->val; - const_func_arg = 0; - if (ts->val_type == TEMP_VAL_MEM) { - reg = tcg_reg_alloc(s, arg_ct->u.regs, allocated_regs); - tcg_out_ld(s, ts->type, reg, ts->mem_reg, ts->mem_offset); - func_arg = reg; - tcg_regset_set_reg(allocated_regs, reg); - } else if (ts->val_type == TEMP_VAL_REG) { - reg = ts->reg; - if (!tcg_regset_test_reg(arg_ct->u.regs, reg)) { - reg = tcg_reg_alloc(s, arg_ct->u.regs, allocated_regs); - tcg_out_mov(s, ts->type, reg, ts->reg); - } - func_arg = reg; - tcg_regset_set_reg(allocated_regs, reg); - } else if (ts->val_type == TEMP_VAL_CONST) { - if (tcg_target_const_match(func_addr, ts->type, arg_ct)) { - const_func_arg = 1; - func_arg = func_addr; - } else { - reg = tcg_reg_alloc(s, arg_ct->u.regs, allocated_regs); - tcg_out_movi(s, ts->type, reg, func_addr); - func_arg = reg; - tcg_regset_set_reg(allocated_regs, reg); - } - } else { - tcg_abort(); - } - - /* mark dead temporaries and free the associated registers */ for(i = nb_oargs; i < nb_iargs + nb_oargs; i++) { if (IS_DEAD_ARG(i)) { @@ -2468,7 +2457,7 @@ static int tcg_reg_alloc_call(TCGContext *s, const TCGOpDef *def, save_globals(s, allocated_regs); } - tcg_out_op(s, opc, &func_arg, &const_func_arg); + tcg_out_call(s, func_addr); /* assign output registers and emit moves if needed */ for(i = 0; i < nb_oargs; i++) { @@ -2518,7 +2507,8 @@ static void dump_op_count(void) #endif -static inline int tcg_gen_code_common(TCGContext *s, uint8_t *gen_code_buf, +static inline int tcg_gen_code_common(TCGContext *s, + tcg_insn_unit *gen_code_buf, long search_pc) { TCGOpcode opc; @@ -2633,7 +2623,7 @@ static inline int tcg_gen_code_common(TCGContext *s, uint8_t *gen_code_buf, } args += def->nb_args; next: - if (search_pc >= 0 && search_pc < s->code_ptr - gen_code_buf) { + if (search_pc >= 0 && search_pc < tcg_current_code_size(s)) { return op_index; } op_index++; @@ -2647,7 +2637,7 @@ static inline int tcg_gen_code_common(TCGContext *s, uint8_t *gen_code_buf, return -1; } -int tcg_gen_code(TCGContext *s, uint8_t *gen_code_buf) +int tcg_gen_code(TCGContext *s, tcg_insn_unit *gen_code_buf) { #ifdef CONFIG_PROFILER { @@ -2666,16 +2656,17 @@ int tcg_gen_code(TCGContext *s, uint8_t *gen_code_buf) tcg_gen_code_common(s, gen_code_buf, -1); /* flush instruction cache */ - flush_icache_range((uintptr_t)gen_code_buf, (uintptr_t)s->code_ptr); + flush_icache_range((uintptr_t)s->code_buf, (uintptr_t)s->code_ptr); - return s->code_ptr - gen_code_buf; + return tcg_current_code_size(s); } /* Return the index of the micro operation such as the pc after is < offset bytes from the start of the TB. The contents of gen_code_buf must not be changed, though writing the same values is ok. Return -1 if not found. */ -int tcg_gen_code_search_pc(TCGContext *s, uint8_t *gen_code_buf, long offset) +int tcg_gen_code_search_pc(TCGContext *s, tcg_insn_unit *gen_code_buf, + long offset) { return tcg_gen_code_common(s, gen_code_buf, offset); } diff --git a/tcg/tcg.h b/tcg/tcg.h index a6a2d06080..fbc93101cf 100644 --- a/tcg/tcg.h +++ b/tcg/tcg.h @@ -146,10 +146,25 @@ typedef enum TCGOpcode { #define tcg_regset_andnot(d, a, b) (d) = (a) & ~(b) #define tcg_regset_not(d, a) (d) = ~(a) +#ifndef TCG_TARGET_INSN_UNIT_SIZE +# error "Missing TCG_TARGET_INSN_UNIT_SIZE" +#elif TCG_TARGET_INSN_UNIT_SIZE == 1 +typedef uint8_t tcg_insn_unit; +#elif TCG_TARGET_INSN_UNIT_SIZE == 2 +typedef uint16_t tcg_insn_unit; +#elif TCG_TARGET_INSN_UNIT_SIZE == 4 +typedef uint32_t tcg_insn_unit; +#elif TCG_TARGET_INSN_UNIT_SIZE == 8 +typedef uint64_t tcg_insn_unit; +#else +/* The port better have done this. */ +#endif + + typedef struct TCGRelocation { struct TCGRelocation *next; int type; - uint8_t *ptr; + tcg_insn_unit *ptr; intptr_t addend; } TCGRelocation; @@ -157,6 +172,7 @@ typedef struct TCGLabel { int has_value; union { uintptr_t value; + tcg_insn_unit *value_ptr; TCGRelocation *first_reloc; } u; } TCGLabel; @@ -464,7 +480,7 @@ struct TCGContext { int nb_temps; /* goto_tb support */ - uint8_t *code_buf; + tcg_insn_unit *code_buf; uintptr_t *tb_next; uint16_t *tb_next_offset; uint16_t *tb_jmp_offset; /* != NULL if USE_DIRECT_JUMP */ @@ -485,7 +501,7 @@ struct TCGContext { intptr_t frame_end; int frame_reg; - uint8_t *code_ptr; + tcg_insn_unit *code_ptr; TCGTemp temps[TCG_MAX_TEMPS]; /* globals first, temps after */ TCGTempSet free_temps[TCG_TYPE_COUNT * 2]; @@ -524,14 +540,17 @@ struct TCGContext { uint16_t gen_opc_icount[OPC_BUF_SIZE]; uint8_t gen_opc_instr_start[OPC_BUF_SIZE]; - /* Code generation */ + /* Code generation. Note that we specifically do not use tcg_insn_unit + here, because there's too much arithmetic throughout that relies + on addition and subtraction working on bytes. Rely on the GCC + extension that allows arithmetic on void*. */ int code_gen_max_blocks; - uint8_t *code_gen_prologue; - uint8_t *code_gen_buffer; + void *code_gen_prologue; + void *code_gen_buffer; size_t code_gen_buffer_size; /* threshold to flush the translated code buffer */ size_t code_gen_buffer_max_size; - uint8_t *code_gen_ptr; + void *code_gen_ptr; TBContext tb_ctx; @@ -566,8 +585,9 @@ void tcg_context_init(TCGContext *s); void tcg_prologue_init(TCGContext *s); void tcg_func_start(TCGContext *s); -int tcg_gen_code(TCGContext *s, uint8_t *gen_code_buf); -int tcg_gen_code_search_pc(TCGContext *s, uint8_t *gen_code_buf, long offset); +int tcg_gen_code(TCGContext *s, tcg_insn_unit *gen_code_buf); +int tcg_gen_code_search_pc(TCGContext *s, tcg_insn_unit *gen_code_buf, + long offset); void tcg_set_frame(TCGContext *s, int reg, intptr_t start, intptr_t size); @@ -705,7 +725,7 @@ void tcg_add_target_add_op_defs(const TCGTargetOpDef *tdefs); #define tcg_temp_free_ptr(T) tcg_temp_free_i64(TCGV_PTR_TO_NAT(T)) #endif -void tcg_gen_callN(TCGContext *s, TCGv_ptr func, unsigned int flags, +void tcg_gen_callN(TCGContext *s, void *func, unsigned int flags, int sizemask, TCGArg ret, int nargs, TCGArg *args); void tcg_gen_shifti_i64(TCGv_i64 ret, TCGv_i64 arg1, @@ -723,6 +743,51 @@ TCGv_i64 tcg_const_i64(int64_t val); TCGv_i32 tcg_const_local_i32(int32_t val); TCGv_i64 tcg_const_local_i64(int64_t val); +/** + * tcg_ptr_byte_diff + * @a, @b: addresses to be differenced + * + * There are many places within the TCG backends where we need a byte + * difference between two pointers. While this can be accomplished + * with local casting, it's easy to get wrong -- especially if one is + * concerned with the signedness of the result. + * + * This version relies on GCC's void pointer arithmetic to get the + * correct result. + */ + +static inline ptrdiff_t tcg_ptr_byte_diff(void *a, void *b) +{ + return a - b; +} + +/** + * tcg_pcrel_diff + * @s: the tcg context + * @target: address of the target + * + * Produce a pc-relative difference, from the current code_ptr + * to the destination address. + */ + +static inline ptrdiff_t tcg_pcrel_diff(TCGContext *s, void *target) +{ + return tcg_ptr_byte_diff(target, s->code_ptr); +} + +/** + * tcg_current_code_size + * @s: the tcg context + * + * Compute the current code size within the translation block. + * This is used to fill in qemu's data structures for goto_tb. + */ + +static inline size_t tcg_current_code_size(TCGContext *s) +{ + return tcg_ptr_byte_diff(s->code_ptr, s->code_buf); +} + /** * tcg_qemu_tb_exec: * @env: CPUArchState * for the CPU diff --git a/tcg/tci/tcg-target.c b/tcg/tci/tcg-target.c index 47c0b85d95..9b39231c15 100644 --- a/tcg/tci/tcg-target.c +++ b/tcg/tci/tcg-target.c @@ -59,12 +59,8 @@ static const TCGTargetOpDef tcg_target_op_defs[] = { { INDEX_op_exit_tb, { NULL } }, { INDEX_op_goto_tb, { NULL } }, - { INDEX_op_call, { RI } }, { INDEX_op_br, { NULL } }, - { INDEX_op_mov_i32, { R, R } }, - { INDEX_op_movi_i32, { R } }, - { INDEX_op_ld8u_i32, { R, R } }, { INDEX_op_ld8s_i32, { R, R } }, { INDEX_op_ld16u_i32, { R, R } }, @@ -141,9 +137,6 @@ static const TCGTargetOpDef tcg_target_op_defs[] = { #endif #if TCG_TARGET_REG_BITS == 64 - { INDEX_op_mov_i64, { R, R } }, - { INDEX_op_movi_i64, { R } }, - { INDEX_op_ld8u_i64, { R, R } }, { INDEX_op_ld8s_i64, { R, R } }, { INDEX_op_ld16u_i64, { R, R } }, @@ -371,14 +364,18 @@ static const char *const tcg_target_reg_names[TCG_TARGET_NB_REGS] = { }; #endif -static void patch_reloc(uint8_t *code_ptr, int type, +static void patch_reloc(tcg_insn_unit *code_ptr, int type, intptr_t value, intptr_t addend) { /* tcg_out_reloc always uses the same type, addend. */ assert(type == sizeof(tcg_target_long)); assert(addend == 0); assert(value != 0); - *(tcg_target_long *)code_ptr = value; + if (TCG_TARGET_REG_BITS == 32) { + tcg_patch32(code_ptr, value); + } else { + tcg_patch64(code_ptr, value); + } } /* Parse target specific constraints. */ @@ -413,8 +410,11 @@ void tci_disas(uint8_t opc) /* Write value (native size). */ static void tcg_out_i(TCGContext *s, tcg_target_ulong v) { - *(tcg_target_ulong *)s->code_ptr = v; - s->code_ptr += sizeof(tcg_target_ulong); + if (TCG_TARGET_REG_BITS == 32) { + tcg_out32(s, v); + } else { + tcg_out64(s, v); + } } /* Write opcode. */ @@ -542,6 +542,11 @@ static void tcg_out_movi(TCGContext *s, TCGType type, old_code_ptr[1] = s->code_ptr - old_code_ptr; } +static inline void tcg_out_call(TCGContext *s, tcg_insn_unit *arg) +{ + tcg_out_ri(s, 1, (uintptr_t)arg); +} + static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, const int *const_args) { @@ -557,21 +562,18 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, if (s->tb_jmp_offset) { /* Direct jump method. */ assert(args[0] < ARRAY_SIZE(s->tb_jmp_offset)); - s->tb_jmp_offset[args[0]] = s->code_ptr - s->code_buf; + s->tb_jmp_offset[args[0]] = tcg_current_code_size(s); tcg_out32(s, 0); } else { /* Indirect jump method. */ TODO(); } assert(args[0] < ARRAY_SIZE(s->tb_next_offset)); - s->tb_next_offset[args[0]] = s->code_ptr - s->code_buf; + s->tb_next_offset[args[0]] = tcg_current_code_size(s); break; case INDEX_op_br: tci_out_label(s, args[0]); break; - case INDEX_op_call: - tcg_out_ri(s, const_args[0], args[0]); - break; case INDEX_op_setcond_i32: tcg_out_r(s, args[0]); tcg_out_r(s, args[1]); @@ -596,9 +598,6 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, tcg_out8(s, args[3]); /* condition */ break; #endif - case INDEX_op_movi_i32: - TODO(); /* Handled by tcg_out_movi? */ - break; case INDEX_op_ld8u_i32: case INDEX_op_ld8s_i32: case INDEX_op_ld16u_i32: @@ -654,10 +653,6 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, break; #if TCG_TARGET_REG_BITS == 64 - case INDEX_op_mov_i64: - case INDEX_op_movi_i64: - TODO(); - break; case INDEX_op_add_i64: case INDEX_op_sub_i64: case INDEX_op_mul_i64: @@ -825,11 +820,12 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, tcg_out_i(s, *args); #endif break; - case INDEX_op_end: - TODO(); - break; + case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ + case INDEX_op_mov_i64: + case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ + case INDEX_op_movi_i64: + case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: - fprintf(stderr, "Missing: %s\n", tcg_op_defs[opc].name); tcg_abort(); } old_code_ptr[1] = s->code_ptr - old_code_ptr; diff --git a/tcg/tci/tcg-target.h b/tcg/tci/tcg-target.h index 37719e875c..0be5acdb81 100644 --- a/tcg/tci/tcg-target.h +++ b/tcg/tci/tcg-target.h @@ -43,6 +43,7 @@ #include "config-host.h" #define TCG_TARGET_INTERPRETER 1 +#define TCG_TARGET_INSN_UNIT_SIZE 1 #if UINTPTR_MAX == UINT32_MAX # define TCG_TARGET_REG_BITS 32 diff --git a/translate-all.c b/translate-all.c index 5759974d42..5549a85ed5 100644 --- a/translate-all.c +++ b/translate-all.c @@ -143,7 +143,7 @@ void cpu_gen_init(void) int cpu_gen_code(CPUArchState *env, TranslationBlock *tb, int *gen_code_size_ptr) { TCGContext *s = &tcg_ctx; - uint8_t *gen_code_buf; + tcg_insn_unit *gen_code_buf; int gen_code_size; #ifdef CONFIG_PROFILER int64_t ti; @@ -186,8 +186,8 @@ int cpu_gen_code(CPUArchState *env, TranslationBlock *tb, int *gen_code_size_ptr #ifdef DEBUG_DISAS if (qemu_loglevel_mask(CPU_LOG_TB_OUT_ASM)) { - qemu_log("OUT: [size=%d]\n", *gen_code_size_ptr); - log_disas(tb->tc_ptr, *gen_code_size_ptr); + qemu_log("OUT: [size=%d]\n", gen_code_size); + log_disas(tb->tc_ptr, gen_code_size); qemu_log("\n"); qemu_log_flush(); } @@ -235,7 +235,8 @@ static int cpu_restore_state_from_tb(CPUState *cpu, TranslationBlock *tb, s->tb_jmp_offset = NULL; s->tb_next = tb->tb_next; #endif - j = tcg_gen_code_search_pc(s, (uint8_t *)tc_ptr, searched_pc - tc_ptr); + j = tcg_gen_code_search_pc(s, (tcg_insn_unit *)tc_ptr, + searched_pc - tc_ptr); if (j < 0) return -1; /* now find start of instruction before */ @@ -944,7 +945,6 @@ TranslationBlock *tb_gen_code(CPUState *cpu, { CPUArchState *env = cpu->env_ptr; TranslationBlock *tb; - uint8_t *tc_ptr; tb_page_addr_t phys_pc, phys_page2; target_ulong virt_page2; int code_gen_size; @@ -959,8 +959,7 @@ TranslationBlock *tb_gen_code(CPUState *cpu, /* Don't forget to invalidate previous TB info. */ tcg_ctx.tb_ctx.tb_invalidated_flag = 1; } - tc_ptr = tcg_ctx.code_gen_ptr; - tb->tc_ptr = tc_ptr; + tb->tc_ptr = tcg_ctx.code_gen_ptr; tb->cs_base = cs_base; tb->flags = flags; tb->cflags = cflags;