diff options
-rw-r--r-- | target-sh4/helper.h | 2 | ||||
-rw-r--r-- | target-sh4/op.c | 46 | ||||
-rw-r--r-- | target-sh4/op_helper.c | 9 | ||||
-rw-r--r-- | target-sh4/translate.c | 94 |
4 files changed, 69 insertions, 82 deletions
diff --git a/target-sh4/helper.h b/target-sh4/helper.h index ca6496873e..4dca7ceb8c 100644 --- a/target-sh4/helper.h +++ b/target-sh4/helper.h @@ -16,3 +16,5 @@ DEF_HELPER(uint32_t, helper_subc, (uint32_t, uint32_t)) DEF_HELPER(uint32_t, helper_negc, (uint32_t)) DEF_HELPER(void, helper_macl, (uint32_t, uint32_t)) DEF_HELPER(void, helper_macw, (uint32_t, uint32_t)) + +DEF_HELPER(void, helper_ld_fpscr, (uint32_t)) diff --git a/target-sh4/op.c b/target-sh4/op.c index a2c602aca6..8313070232 100644 --- a/target-sh4/op.c +++ b/target-sh4/op.c @@ -88,52 +88,6 @@ void OPPROTO op_shld_T0_T1(void) RETURN(); } -void OPPROTO op_ldc_T0_sr(void) -{ - env->sr = T0 & 0x700083f3; - RETURN(); -} - -void OPPROTO op_stc_sr_T0(void) -{ - T0 = env->sr; - RETURN(); -} - -#define LDSTOPS(target,load,store) \ -void OPPROTO op_##load##_T0_##target (void) \ -{ env ->target = T0; RETURN(); \ -} \ -void OPPROTO op_##store##_##target##_T0 (void) \ -{ T0 = env->target; RETURN(); \ -} \ - - LDSTOPS(gbr, ldc, stc) - LDSTOPS(vbr, ldc, stc) - LDSTOPS(ssr, ldc, stc) - LDSTOPS(spc, ldc, stc) - LDSTOPS(sgr, ldc, stc) - LDSTOPS(dbr, ldc, stc) - LDSTOPS(mach, lds, sts) - LDSTOPS(macl, lds, sts) - LDSTOPS(pr, lds, sts) - LDSTOPS(fpul, lds, sts) - -void OPPROTO op_lds_T0_fpscr(void) -{ - env->fpscr = T0 & 0x003fffff; - env->fp_status.float_rounding_mode = T0 & 0x01 ? - float_round_to_zero : float_round_nearest_even; - - RETURN(); -} - -void OPPROTO op_sts_fpscr_T0(void) -{ - T0 = env->fpscr & 0x003fffff; - RETURN(); -} - void OPPROTO op_rotcl_Rn(void) { helper_rotcl(&env->gregs[PARAM1]); diff --git a/target-sh4/op_helper.c b/target-sh4/op_helper.c index e3f9527d24..a4498ce5d0 100644 --- a/target-sh4/op_helper.c +++ b/target-sh4/op_helper.c @@ -388,3 +388,12 @@ void helper_rotcr(uint32_t * addr) env->sr &= ~SR_T; *addr = new; } + +void helper_ld_fpscr(uint32_t val) +{ + env->fpscr = val & 0x003fffff; + if (val & 0x01) + set_float_rounding_mode(float_round_to_zero, &env->fp_status); + else + set_float_rounding_mode(float_round_nearest_even, &env->fp_status); +} diff --git a/target-sh4/translate.c b/target-sh4/translate.c index 8e01fad93e..985675bfa7 100644 --- a/target-sh4/translate.c +++ b/target-sh4/translate.c @@ -959,40 +959,34 @@ void _decode_opc(DisasContext * ctx) gen_cmp_imm(TCG_COND_EQ, cpu_T[0], B7_0s); return; case 0xc400: /* mov.b @(disp,GBR),R0 */ - gen_op_stc_gbr_T0(); - tcg_gen_addi_i32(cpu_T[0], cpu_T[0], B7_0); + tcg_gen_addi_i32(cpu_T[0], cpu_gbr, B7_0); tcg_gen_qemu_ld8s(cpu_T[0], cpu_T[0], ctx->memidx); tcg_gen_mov_i32(cpu_gregs[REG(0)], cpu_T[0]); return; case 0xc500: /* mov.w @(disp,GBR),R0 */ - gen_op_stc_gbr_T0(); - tcg_gen_addi_i32(cpu_T[0], cpu_T[0], B7_0 * 2); + tcg_gen_addi_i32(cpu_T[0], cpu_gbr, B7_0 * 2); tcg_gen_qemu_ld16s(cpu_T[0], cpu_T[0], ctx->memidx); tcg_gen_mov_i32(cpu_gregs[REG(0)], cpu_T[0]); return; case 0xc600: /* mov.l @(disp,GBR),R0 */ - gen_op_stc_gbr_T0(); - tcg_gen_addi_i32(cpu_T[0], cpu_T[0], B7_0 * 4); + tcg_gen_addi_i32(cpu_T[0], cpu_gbr, B7_0 * 4); tcg_gen_qemu_ld32s(cpu_T[0], cpu_T[0], ctx->memidx); tcg_gen_mov_i32(cpu_gregs[REG(0)], cpu_T[0]); return; case 0xc000: /* mov.b R0,@(disp,GBR) */ - gen_op_stc_gbr_T0(); - tcg_gen_addi_i32(cpu_T[0], cpu_T[0], B7_0); + tcg_gen_addi_i32(cpu_T[0], cpu_gbr, B7_0); tcg_gen_mov_i32(cpu_T[1], cpu_T[0]); tcg_gen_mov_i32(cpu_T[0], cpu_gregs[REG(0)]); tcg_gen_qemu_st8(cpu_T[0], cpu_T[1], ctx->memidx); return; case 0xc100: /* mov.w R0,@(disp,GBR) */ - gen_op_stc_gbr_T0(); - tcg_gen_addi_i32(cpu_T[0], cpu_T[0], B7_0 * 2); + tcg_gen_addi_i32(cpu_T[0], cpu_gbr, B7_0 * 2); tcg_gen_mov_i32(cpu_T[1], cpu_T[0]); tcg_gen_mov_i32(cpu_T[0], cpu_gregs[REG(0)]); tcg_gen_qemu_st16(cpu_T[0], cpu_T[1], ctx->memidx); return; case 0xc200: /* mov.l R0,@(disp,GBR) */ - gen_op_stc_gbr_T0(); - tcg_gen_addi_i32(cpu_T[0], cpu_T[0], B7_0 * 4); + tcg_gen_addi_i32(cpu_T[0], cpu_gbr, B7_0 * 4); tcg_gen_mov_i32(cpu_T[1], cpu_T[0]); tcg_gen_mov_i32(cpu_T[0], cpu_gregs[REG(0)]); tcg_gen_qemu_st32(cpu_T[0], cpu_T[1], ctx->memidx); @@ -1131,43 +1125,71 @@ void _decode_opc(DisasContext * ctx) ctx->flags |= DELAY_SLOT; ctx->delayed_pc = (uint32_t) - 1; return; -#define LDST(reg,ldnum,ldpnum,ldop,stnum,stpnum,stop,extrald) \ + case 0x400e: /* lds Rm,SR */ + tcg_gen_andi_i32(cpu_sr, cpu_gregs[REG(B11_8)], 0x700083f3); + ctx->bstate = BS_STOP; + return; + case 0x4007: /* lds.l @Rm+,SR */ + tcg_gen_qemu_ld32s(cpu_T[0], cpu_gregs[REG(B11_8)], ctx->memidx); + tcg_gen_addi_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], 4); + tcg_gen_andi_i32(cpu_sr, cpu_T[0], 0x700083f3); + ctx->bstate = BS_STOP; + return; + case 0x0002: /* sts SR,Rn */ + tcg_gen_mov_i32(cpu_gregs[REG(B11_8)], cpu_sr); + return; + case 0x4003: /* sts SR,@-Rn */ + tcg_gen_subi_i32(cpu_T[0], cpu_gregs[REG(B11_8)], 4); + tcg_gen_qemu_st32(cpu_sr, cpu_T[0], ctx->memidx); + tcg_gen_subi_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], 4); + return; +#define LDST(reg,ldnum,ldpnum,stnum,stpnum) \ case ldnum: \ - tcg_gen_mov_i32 (cpu_T[0], cpu_gregs[REG(B11_8)]); \ - gen_op_##ldop##_T0_##reg (); \ - extrald \ + tcg_gen_mov_i32 (cpu_##reg, cpu_gregs[REG(B11_8)]); \ return; \ case ldpnum: \ - tcg_gen_qemu_ld32s (cpu_T[0], cpu_gregs[REG(B11_8)], ctx->memidx); \ + tcg_gen_qemu_ld32s (cpu_##reg, cpu_gregs[REG(B11_8)], ctx->memidx); \ tcg_gen_addi_i32(cpu_gregs[REG(B11_8)], \ cpu_gregs[REG(B11_8)], 4); \ - gen_op_##ldop##_T0_##reg (); \ - extrald \ return; \ case stnum: \ - gen_op_##stop##_##reg##_T0 (); \ - tcg_gen_mov_i32 (cpu_gregs[REG(B11_8)], cpu_T[0]); \ + tcg_gen_mov_i32 (cpu_gregs[REG(B11_8)], cpu_##reg); \ return; \ case stpnum: \ - gen_op_##stop##_##reg##_T0 (); \ tcg_gen_subi_i32(cpu_T[1], cpu_gregs[REG(B11_8)], 4); \ - tcg_gen_qemu_st32 (cpu_T[0], cpu_T[1], ctx->memidx); \ + tcg_gen_qemu_st32 (cpu_##reg, cpu_T[1], ctx->memidx); \ tcg_gen_subi_i32(cpu_gregs[REG(B11_8)], \ cpu_gregs[REG(B11_8)], 4); \ return; - LDST(sr, 0x400e, 0x4007, ldc, 0x0002, 0x4003, stc, ctx->bstate = - BS_STOP;) - LDST(gbr, 0x401e, 0x4017, ldc, 0x0012, 0x4013, stc,) - LDST(vbr, 0x402e, 0x4027, ldc, 0x0022, 0x4023, stc,) - LDST(ssr, 0x403e, 0x4037, ldc, 0x0032, 0x4033, stc,) - LDST(spc, 0x404e, 0x4047, ldc, 0x0042, 0x4043, stc,) - LDST(dbr, 0x40fa, 0x40f6, ldc, 0x00fa, 0x40f2, stc,) - LDST(mach, 0x400a, 0x4006, lds, 0x000a, 0x4002, sts,) - LDST(macl, 0x401a, 0x4016, lds, 0x001a, 0x4012, sts,) - LDST(pr, 0x402a, 0x4026, lds, 0x002a, 0x4022, sts,) - LDST(fpul, 0x405a, 0x4056, lds, 0x005a, 0x4052, sts,) - LDST(fpscr, 0x406a, 0x4066, lds, 0x006a, 0x4062, sts, ctx->bstate = - BS_STOP;) + LDST(gbr, 0x401e, 0x4017, 0x0012, 0x4013) + LDST(vbr, 0x402e, 0x4027, 0x0022, 0x4023) + LDST(ssr, 0x403e, 0x4037, 0x0032, 0x4033) + LDST(spc, 0x404e, 0x4047, 0x0042, 0x4043) + LDST(dbr, 0x40fa, 0x40f6, 0x00fa, 0x40f2) + LDST(mach, 0x400a, 0x4006, 0x000a, 0x4002) + LDST(macl, 0x401a, 0x4016, 0x001a, 0x4012) + LDST(pr, 0x402a, 0x4026, 0x002a, 0x4022) + LDST(fpul, 0x405a, 0x4056, 0x005a, 0x4052) + case 0x406a: /* lds Rm,FPSCR */ + tcg_gen_helper_0_1(helper_ld_fpscr, cpu_gregs[REG(B11_8)]); + ctx->bstate = BS_STOP; + return; + case 0x4066: /* lds.l @Rm+,FPSCR */ + tcg_gen_qemu_ld32s(cpu_T[0], cpu_gregs[REG(B11_8)], ctx->memidx); + tcg_gen_addi_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], 4); + tcg_gen_helper_0_1(helper_ld_fpscr, cpu_T[0]); + ctx->bstate = BS_STOP; + return; + case 0x006a: /* sts FPSCR,Rn */ + tcg_gen_andi_i32(cpu_T[0], cpu_fpscr, 0x003fffff); + tcg_gen_mov_i32(cpu_gregs[REG(B11_8)], cpu_T[0]); + return; + case 0x4062: /* sts FPSCR,@-Rn */ + tcg_gen_andi_i32(cpu_T[0], cpu_fpscr, 0x003fffff); + tcg_gen_subi_i32(cpu_T[1], cpu_gregs[REG(B11_8)], 4); + tcg_gen_qemu_st32(cpu_T[0], cpu_T[1], ctx->memidx); + tcg_gen_subi_i32(cpu_gregs[REG(B11_8)], cpu_gregs[REG(B11_8)], 4); + return; case 0x00c3: /* movca.l R0,@Rm */ tcg_gen_mov_i32(cpu_T[0], cpu_gregs[REG(0)]); tcg_gen_mov_i32(cpu_T[1], cpu_gregs[REG(B11_8)]); |