From: Michael Rolnik <mrol...@gmail.com> This includes: - RJMP, IJMP, EIJMP, JMP - RCALL, ICALL, EICALL, CALL - RET, RETI - CPSE, CP, CPC, CPI - SBRC, SBRS, SBIC, SBIS - BRBC, BRBS
Signed-off-by: Michael Rolnik <mrol...@gmail.com> Signed-off-by: Richard Henderson <richard.hender...@linaro.org> Signed-off-by: Aleksandar Markovic <aleksandar.m.m...@gmail.com> Tested-by: Philippe Mathieu-Daudé <phi...@redhat.com> Reviewed-by: Aleksandar Markovic <aleksandar.m.m...@gmail.com> Signed-off-by: Thomas Huth <h...@tuxfamily.org> --- target/avr/insn.decode | 33 +++ target/avr/translate.c | 544 +++++++++++++++++++++++++++++++++++++++++ 2 files changed, 577 insertions(+) diff --git a/target/avr/insn.decode b/target/avr/insn.decode index 43baf6d92f..8b00d8807c 100644 --- a/target/avr/insn.decode +++ b/target/avr/insn.decode @@ -74,3 +74,36 @@ FMUL 0000 0011 0 ... 1 ... @fmul FMULS 0000 0011 1 ... 0 ... @fmul FMULSU 0000 0011 1 ... 1 ... @fmul DES 1001 0100 imm:4 1011 + +# +# Branch Instructions +# + +# The 22-bit immediate is partially in the opcode word, +# and partially in the next. Use append_16 to build the +# complete 22-bit value. +%imm_call 4:5 0:1 !function=append_16 + +@op_bit .... .... . bit:3 .... +@op_bit_imm .... .. imm:s7 bit:3 + +RJMP 1100 imm:s12 +IJMP 1001 0100 0000 1001 +EIJMP 1001 0100 0001 1001 +JMP 1001 010 ..... 110 . imm=%imm_call +RCALL 1101 imm:s12 +ICALL 1001 0101 0000 1001 +EICALL 1001 0101 0001 1001 +CALL 1001 010 ..... 111 . imm=%imm_call +RET 1001 0101 0000 1000 +RETI 1001 0101 0001 1000 +CPSE 0001 00 . ..... .... @op_rd_rr +CP 0001 01 . ..... .... @op_rd_rr +CPC 0000 01 . ..... .... @op_rd_rr +CPI 0011 .... .... .... @op_rd_imm8 +SBRC 1111 110 rr:5 0 bit:3 +SBRS 1111 111 rr:5 0 bit:3 +SBIC 1001 1001 reg:5 bit:3 +SBIS 1001 1011 reg:5 bit:3 +BRBS 1111 00 ....... ... @op_bit_imm +BRBC 1111 01 ....... ... @op_bit_imm diff --git a/target/avr/translate.c b/target/avr/translate.c index d81bfaa04b..6db9f74a02 100644 --- a/target/avr/translate.c +++ b/target/avr/translate.c @@ -145,6 +145,17 @@ static int to_regs_24_30_by_two(DisasContext *ctx, int indx) } +static uint16_t next_word(DisasContext *ctx) +{ + return cpu_lduw_code(ctx->env, ctx->npc++ * 2); +} + +static int append_16(DisasContext *ctx, int x) +{ + return x << 16 | next_word(ctx); +} + + static bool avr_have_feature(DisasContext *ctx, int feature) { if (!avr_feature(ctx->env, feature)) { @@ -964,3 +975,536 @@ static bool trans_DES(DisasContext *ctx, arg_DES *a) return true; } + +/* + * Branch Instructions + */ +static void gen_jmp_ez(DisasContext *ctx) +{ + tcg_gen_deposit_tl(cpu_pc, cpu_r[30], cpu_r[31], 8, 8); + tcg_gen_or_tl(cpu_pc, cpu_pc, cpu_eind); + ctx->bstate = DISAS_LOOKUP; +} + +static void gen_jmp_z(DisasContext *ctx) +{ + tcg_gen_deposit_tl(cpu_pc, cpu_r[30], cpu_r[31], 8, 8); + ctx->bstate = DISAS_LOOKUP; +} + +static void gen_push_ret(DisasContext *ctx, int ret) +{ + if (avr_feature(ctx->env, AVR_FEATURE_1_BYTE_PC)) { + + TCGv t0 = tcg_const_i32((ret & 0x0000ff)); + + tcg_gen_qemu_st_tl(t0, cpu_sp, MMU_DATA_IDX, MO_UB); + tcg_gen_subi_tl(cpu_sp, cpu_sp, 1); + + tcg_temp_free_i32(t0); + } else if (avr_feature(ctx->env, AVR_FEATURE_2_BYTE_PC)) { + + TCGv t0 = tcg_const_i32((ret & 0x00ffff)); + + tcg_gen_subi_tl(cpu_sp, cpu_sp, 1); + tcg_gen_qemu_st_tl(t0, cpu_sp, MMU_DATA_IDX, MO_BEUW); + tcg_gen_subi_tl(cpu_sp, cpu_sp, 1); + + tcg_temp_free_i32(t0); + + } else if (avr_feature(ctx->env, AVR_FEATURE_3_BYTE_PC)) { + + TCGv lo = tcg_const_i32((ret & 0x0000ff)); + TCGv hi = tcg_const_i32((ret & 0xffff00) >> 8); + + tcg_gen_qemu_st_tl(lo, cpu_sp, MMU_DATA_IDX, MO_UB); + tcg_gen_subi_tl(cpu_sp, cpu_sp, 2); + tcg_gen_qemu_st_tl(hi, cpu_sp, MMU_DATA_IDX, MO_BEUW); + tcg_gen_subi_tl(cpu_sp, cpu_sp, 1); + + tcg_temp_free_i32(lo); + tcg_temp_free_i32(hi); + } +} + +static void gen_pop_ret(DisasContext *ctx, TCGv ret) +{ + if (avr_feature(ctx->env, AVR_FEATURE_1_BYTE_PC)) { + tcg_gen_addi_tl(cpu_sp, cpu_sp, 1); + tcg_gen_qemu_ld_tl(ret, cpu_sp, MMU_DATA_IDX, MO_UB); + } else if (avr_feature(ctx->env, AVR_FEATURE_2_BYTE_PC)) { + tcg_gen_addi_tl(cpu_sp, cpu_sp, 1); + tcg_gen_qemu_ld_tl(ret, cpu_sp, MMU_DATA_IDX, MO_BEUW); + tcg_gen_addi_tl(cpu_sp, cpu_sp, 1); + } else if (avr_feature(ctx->env, AVR_FEATURE_3_BYTE_PC)) { + TCGv lo = tcg_temp_new_i32(); + TCGv hi = tcg_temp_new_i32(); + + tcg_gen_addi_tl(cpu_sp, cpu_sp, 1); + tcg_gen_qemu_ld_tl(hi, cpu_sp, MMU_DATA_IDX, MO_BEUW); + + tcg_gen_addi_tl(cpu_sp, cpu_sp, 2); + tcg_gen_qemu_ld_tl(lo, cpu_sp, MMU_DATA_IDX, MO_UB); + + tcg_gen_deposit_tl(ret, lo, hi, 8, 16); + + tcg_temp_free_i32(lo); + tcg_temp_free_i32(hi); + } +} + +static void gen_goto_tb(DisasContext *ctx, int n, target_ulong dest) +{ + TranslationBlock *tb = ctx->tb; + + if (ctx->singlestep == 0) { + tcg_gen_goto_tb(n); + tcg_gen_movi_i32(cpu_pc, dest); + tcg_gen_exit_tb(tb, n); + } else { + tcg_gen_movi_i32(cpu_pc, dest); + gen_helper_debug(cpu_env); + tcg_gen_exit_tb(NULL, 0); + } + ctx->bstate = DISAS_NORETURN; +} + +/* + * Relative jump to an address within PC - 2K +1 and PC + 2K (words). For + * AVR microcontrollers with Program memory not exceeding 4K words (8KB) this + * instruction can address the entire memory from every address location. See + * also JMP. + */ +static bool trans_RJMP(DisasContext *ctx, arg_RJMP *a) +{ + int dst = ctx->npc + a->imm; + + gen_goto_tb(ctx, 0, dst); + + return true; +} + +/* + * Indirect jump to the address pointed to by the Z (16 bits) Pointer + * Register in the Register File. The Z-pointer Register is 16 bits wide and + * allows jump within the lowest 64K words (128KB) section of Program memory. + * This instruction is not available in all devices. Refer to the device + * specific instruction set summary. + */ +static bool trans_IJMP(DisasContext *ctx, arg_IJMP *a) +{ + if (!avr_have_feature(ctx, AVR_FEATURE_IJMP_ICALL)) { + return true; + } + + gen_jmp_z(ctx); + + return true; +} + +/* + * Indirect jump to the address pointed to by the Z (16 bits) Pointer + * Register in the Register File and the EIND Register in the I/O space. This + * instruction allows for indirect jumps to the entire 4M (words) Program + * memory space. See also IJMP. This instruction is not available in all + * devices. Refer to the device specific instruction set summary. + */ +static bool trans_EIJMP(DisasContext *ctx, arg_EIJMP *a) +{ + if (!avr_have_feature(ctx, AVR_FEATURE_EIJMP_EICALL)) { + return true; + } + + gen_jmp_ez(ctx); + return true; +} + +/* + * Jump to an address within the entire 4M (words) Program memory. See also + * RJMP. This instruction is not available in all devices. Refer to the device + * specific instruction set summary.0 + */ +static bool trans_JMP(DisasContext *ctx, arg_JMP *a) +{ + if (!avr_have_feature(ctx, AVR_FEATURE_JMP_CALL)) { + return true; + } + + gen_goto_tb(ctx, 0, a->imm); + + return true; +} + +/* + * Relative call to an address within PC - 2K + 1 and PC + 2K (words). The + * return address (the instruction after the RCALL) is stored onto the Stack. + * See also CALL. For AVR microcontrollers with Program memory not exceeding 4K + * words (8KB) this instruction can address the entire memory from every + * address location. The Stack Pointer uses a post-decrement scheme during + * RCALL. + */ +static bool trans_RCALL(DisasContext *ctx, arg_RCALL *a) +{ + int ret = ctx->npc; + int dst = ctx->npc + a->imm; + + gen_push_ret(ctx, ret); + gen_goto_tb(ctx, 0, dst); + + return true; +} + +/* + * Calls to a subroutine within the entire 4M (words) Program memory. The + * return address (to the instruction after the CALL) will be stored onto the + * Stack. See also RCALL. The Stack Pointer uses a post-decrement scheme during + * CALL. This instruction is not available in all devices. Refer to the device + * specific instruction set summary. + */ +static bool trans_ICALL(DisasContext *ctx, arg_ICALL *a) +{ + if (!avr_have_feature(ctx, AVR_FEATURE_IJMP_ICALL)) { + return true; + } + + int ret = ctx->npc; + + gen_push_ret(ctx, ret); + gen_jmp_z(ctx); + + return true; +} + +/* + * Indirect call of a subroutine pointed to by the Z (16 bits) Pointer + * Register in the Register File and the EIND Register in the I/O space. This + * instruction allows for indirect calls to the entire 4M (words) Program + * memory space. See also ICALL. The Stack Pointer uses a post-decrement scheme + * during EICALL. This instruction is not available in all devices. Refer to + * the device specific instruction set summary. + */ +static bool trans_EICALL(DisasContext *ctx, arg_EICALL *a) +{ + if (!avr_have_feature(ctx, AVR_FEATURE_EIJMP_EICALL)) { + return true; + } + + int ret = ctx->npc; + + gen_push_ret(ctx, ret); + gen_jmp_ez(ctx); + return true; +} + +/* + * Calls to a subroutine within the entire Program memory. The return + * address (to the instruction after the CALL) will be stored onto the Stack. + * (See also RCALL). The Stack Pointer uses a post-decrement scheme during + * CALL. This instruction is not available in all devices. Refer to the device + * specific instruction set summary. + */ +static bool trans_CALL(DisasContext *ctx, arg_CALL *a) +{ + if (!avr_have_feature(ctx, AVR_FEATURE_JMP_CALL)) { + return true; + } + + int Imm = a->imm; + int ret = ctx->npc; + + gen_push_ret(ctx, ret); + gen_goto_tb(ctx, 0, Imm); + + return true; +} + +/* + * Returns from subroutine. The return address is loaded from the STACK. + * The Stack Pointer uses a preincrement scheme during RET. + */ +static bool trans_RET(DisasContext *ctx, arg_RET *a) +{ + gen_pop_ret(ctx, cpu_pc); + + ctx->bstate = DISAS_LOOKUP; + return true; +} + +/* + * Returns from interrupt. The return address is loaded from the STACK and + * the Global Interrupt Flag is set. Note that the Status Register is not + * automatically stored when entering an interrupt routine, and it is not + * restored when returning from an interrupt routine. This must be handled by + * the application program. The Stack Pointer uses a pre-increment scheme + * during RETI. + */ +static bool trans_RETI(DisasContext *ctx, arg_RETI *a) +{ + gen_pop_ret(ctx, cpu_pc); + tcg_gen_movi_tl(cpu_If, 1); + + /* Need to return to main loop to re-evaluate interrupts. */ + ctx->bstate = DISAS_EXIT; + return true; +} + +/* + * This instruction performs a compare between two registers Rd and Rr, and + * skips the next instruction if Rd = Rr. + */ +static bool trans_CPSE(DisasContext *ctx, arg_CPSE *a) +{ + ctx->skip_cond = TCG_COND_EQ; + ctx->skip_var0 = cpu_r[a->rd]; + ctx->skip_var1 = cpu_r[a->rr]; + return true; +} + +/* + * This instruction performs a compare between two registers Rd and Rr. + * None of the registers are changed. All conditional branches can be used + * after this instruction. + */ +static bool trans_CP(DisasContext *ctx, arg_CP *a) +{ + TCGv Rd = cpu_r[a->rd]; + TCGv Rr = cpu_r[a->rr]; + TCGv R = tcg_temp_new_i32(); + + tcg_gen_sub_tl(R, Rd, Rr); /* R = Rd - Rr */ + tcg_gen_andi_tl(R, R, 0xff); /* make it 8 bits */ + + /* update status register */ + gen_sub_CHf(R, Rd, Rr); + gen_sub_Vf(R, Rd, Rr); + gen_ZNSf(R); + + tcg_temp_free_i32(R); + + return true; +} + +/* + * This instruction performs a compare between two registers Rd and Rr and + * also takes into account the previous carry. None of the registers are + * changed. All conditional branches can be used after this instruction. + */ +static bool trans_CPC(DisasContext *ctx, arg_CPC *a) +{ + TCGv Rd = cpu_r[a->rd]; + TCGv Rr = cpu_r[a->rr]; + TCGv R = tcg_temp_new_i32(); + TCGv zero = tcg_const_i32(0); + + tcg_gen_sub_tl(R, Rd, Rr); /* R = Rd - Rr - Cf */ + tcg_gen_sub_tl(R, R, cpu_Cf); + tcg_gen_andi_tl(R, R, 0xff); /* make it 8 bits */ + /* update status register */ + gen_sub_CHf(R, Rd, Rr); + gen_sub_Vf(R, Rd, Rr); + gen_NSf(R); + + /* + * Previous value remains unchanged when the result is zero; + * cleared otherwise. + */ + tcg_gen_movcond_tl(TCG_COND_EQ, cpu_Zf, R, zero, cpu_Zf, zero); + + tcg_temp_free_i32(zero); + tcg_temp_free_i32(R); + + return true; +} + +/* + * This instruction performs a compare between register Rd and a constant. + * The register is not changed. All conditional branches can be used after this + * instruction. + */ +static bool trans_CPI(DisasContext *ctx, arg_CPI *a) +{ + TCGv Rd = cpu_r[a->rd]; + int Imm = a->imm; + TCGv Rr = tcg_const_i32(Imm); + TCGv R = tcg_temp_new_i32(); + + tcg_gen_sub_tl(R, Rd, Rr); /* R = Rd - Rr */ + tcg_gen_andi_tl(R, R, 0xff); /* make it 8 bits */ + + /* update status register */ + gen_sub_CHf(R, Rd, Rr); + gen_sub_Vf(R, Rd, Rr); + gen_ZNSf(R); + + tcg_temp_free_i32(R); + tcg_temp_free_i32(Rr); + + return true; +} + +/* + * This instruction tests a single bit in a register and skips the next + * instruction if the bit is cleared. + */ +static bool trans_SBRC(DisasContext *ctx, arg_SBRC *a) +{ + TCGv Rr = cpu_r[a->rr]; + + ctx->skip_cond = TCG_COND_EQ; + ctx->skip_var0 = tcg_temp_new(); + ctx->free_skip_var0 = true; + + tcg_gen_andi_tl(ctx->skip_var0, Rr, 1 << a->bit); + return true; +} + +/* + * This instruction tests a single bit in a register and skips the next + * instruction if the bit is set. + */ +static bool trans_SBRS(DisasContext *ctx, arg_SBRS *a) +{ + TCGv Rr = cpu_r[a->rr]; + + ctx->skip_cond = TCG_COND_NE; + ctx->skip_var0 = tcg_temp_new(); + ctx->free_skip_var0 = true; + + tcg_gen_andi_tl(ctx->skip_var0, Rr, 1 << a->bit); + return true; +} + +/* + * This instruction tests a single bit in an I/O Register and skips the + * next instruction if the bit is cleared. This instruction operates on the + * lower 32 I/O Registers -- addresses 0-31. + */ +static bool trans_SBIC(DisasContext *ctx, arg_SBIC *a) +{ + TCGv temp = tcg_const_i32(a->reg); + + gen_helper_inb(temp, cpu_env, temp); + tcg_gen_andi_tl(temp, temp, 1 << a->bit); + ctx->skip_cond = TCG_COND_EQ; + ctx->skip_var0 = temp; + ctx->free_skip_var0 = true; + + return true; +} + +/* + * This instruction tests a single bit in an I/O Register and skips the + * next instruction if the bit is set. This instruction operates on the lower + * 32 I/O Registers -- addresses 0-31. + */ +static bool trans_SBIS(DisasContext *ctx, arg_SBIS *a) +{ + TCGv temp = tcg_const_i32(a->reg); + + gen_helper_inb(temp, cpu_env, temp); + tcg_gen_andi_tl(temp, temp, 1 << a->bit); + ctx->skip_cond = TCG_COND_NE; + ctx->skip_var0 = temp; + ctx->free_skip_var0 = true; + + return true; +} + +/* + * Conditional relative branch. Tests a single bit in SREG and branches + * relatively to PC if the bit is cleared. This instruction branches relatively + * to PC in either direction (PC - 63 < = destination <= PC + 64). The + * parameter k is the offset from PC and is represented in two's complement + * form. + */ +static bool trans_BRBC(DisasContext *ctx, arg_BRBC *a) +{ + TCGLabel *not_taken = gen_new_label(); + + TCGv var; + + switch (a->bit) { + case 0x00: + var = cpu_Cf; + break; + case 0x01: + var = cpu_Zf; + break; + case 0x02: + var = cpu_Nf; + break; + case 0x03: + var = cpu_Vf; + break; + case 0x04: + var = cpu_Sf; + break; + case 0x05: + var = cpu_Hf; + break; + case 0x06: + var = cpu_Tf; + break; + case 0x07: + var = cpu_If; + break; + default: + g_assert_not_reached(); + } + + tcg_gen_brcondi_i32(TCG_COND_NE, var, 0, not_taken); + gen_goto_tb(ctx, 0, ctx->npc + a->imm); + gen_set_label(not_taken); + + ctx->bstate = DISAS_CHAIN; + return true; +} + +/* + * Conditional relative branch. Tests a single bit in SREG and branches + * relatively to PC if the bit is set. This instruction branches relatively to + * PC in either direction (PC - 63 < = destination <= PC + 64). The parameter k + * is the offset from PC and is represented in two's complement form. + */ +static bool trans_BRBS(DisasContext *ctx, arg_BRBS *a) +{ + TCGLabel *not_taken = gen_new_label(); + + TCGv var; + + switch (a->bit) { + case 0x00: + var = cpu_Cf; + break; + case 0x01: + var = cpu_Zf; + break; + case 0x02: + var = cpu_Nf; + break; + case 0x03: + var = cpu_Vf; + break; + case 0x04: + var = cpu_Sf; + break; + case 0x05: + var = cpu_Hf; + break; + case 0x06: + var = cpu_Tf; + break; + case 0x07: + var = cpu_If; + break; + default: + g_assert_not_reached(); + } + + tcg_gen_brcondi_i32(TCG_COND_EQ, var, 0, not_taken); + gen_goto_tb(ctx, 0, ctx->npc + a->imm); + gen_set_label(not_taken); + + ctx->bstate = DISAS_CHAIN; + return true; +} -- 2.26.2