diff --git a/target/hexagon/arch.c b/target/hexagon/arch.c new file mode 100644 index 0000000000..09de124818 --- /dev/null +++ b/target/hexagon/arch.c @@ -0,0 +1,300 @@ +/* + * Copyright(c) 2019-2021 Qualcomm Innovation Center, Inc. All Rights Reserved. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, see . + */ + +#include "qemu/osdep.h" +#include "fpu/softfloat.h" +#include "cpu.h" +#include "fma_emu.h" +#include "arch.h" +#include "macros.h" + +#define SF_BIAS 127 +#define SF_MAXEXP 254 +#define SF_MANTBITS 23 +#define float32_nan make_float32(0xffffffff) + +#define BITS_MASK_8 0x5555555555555555ULL +#define PAIR_MASK_8 0x3333333333333333ULL +#define NYBL_MASK_8 0x0f0f0f0f0f0f0f0fULL +#define BYTE_MASK_8 0x00ff00ff00ff00ffULL +#define HALF_MASK_8 0x0000ffff0000ffffULL +#define WORD_MASK_8 0x00000000ffffffffULL + +uint64_t interleave(uint32_t odd, uint32_t even) +{ + /* Convert to long long */ + uint64_t myodd = odd; + uint64_t myeven = even; + /* First, spread bits out */ + myodd = (myodd | (myodd << 16)) & HALF_MASK_8; + myeven = (myeven | (myeven << 16)) & HALF_MASK_8; + myodd = (myodd | (myodd << 8)) & BYTE_MASK_8; + myeven = (myeven | (myeven << 8)) & BYTE_MASK_8; + myodd = (myodd | (myodd << 4)) & NYBL_MASK_8; + myeven = (myeven | (myeven << 4)) & NYBL_MASK_8; + myodd = (myodd | (myodd << 2)) & PAIR_MASK_8; + myeven = (myeven | (myeven << 2)) & PAIR_MASK_8; + myodd = (myodd | (myodd << 1)) & BITS_MASK_8; + myeven = (myeven | (myeven << 1)) & BITS_MASK_8; + /* Now OR together */ + return myeven | (myodd << 1); +} + +uint64_t deinterleave(uint64_t src) +{ + /* Get odd and even bits */ + uint64_t myodd = ((src >> 1) & BITS_MASK_8); + uint64_t myeven = (src & BITS_MASK_8); + + /* Unspread bits */ + myeven = (myeven | (myeven >> 1)) & PAIR_MASK_8; + myodd = (myodd | (myodd >> 1)) & PAIR_MASK_8; + myeven = (myeven | (myeven >> 2)) & NYBL_MASK_8; + myodd = (myodd | (myodd >> 2)) & NYBL_MASK_8; + myeven = (myeven | (myeven >> 4)) & BYTE_MASK_8; + myodd = (myodd | (myodd >> 4)) & BYTE_MASK_8; + myeven = (myeven | (myeven >> 8)) & HALF_MASK_8; + myodd = (myodd | (myodd >> 8)) & HALF_MASK_8; + myeven = (myeven | (myeven >> 16)) & WORD_MASK_8; + myodd = (myodd | (myodd >> 16)) & WORD_MASK_8; + + /* Return odd bits in upper half */ + return myeven | (myodd << 32); +} + +uint32_t carry_from_add64(uint64_t a, uint64_t b, uint32_t c) +{ + uint64_t tmpa, tmpb, tmpc; + tmpa = fGETUWORD(0, a); + tmpb = fGETUWORD(0, b); + tmpc = tmpa + tmpb + c; + tmpa = fGETUWORD(1, a); + tmpb = fGETUWORD(1, b); + tmpc = tmpa + tmpb + fGETUWORD(1, tmpc); + tmpc = fGETUWORD(1, tmpc); + return tmpc; +} + +int32_t conv_round(int32_t a, int n) +{ + int64_t val; + + if (n == 0) { + val = a; + } else if ((a & ((1 << (n - 1)) - 1)) == 0) { /* N-1..0 all zero? */ + /* Add LSB from int part */ + val = ((fSE32_64(a)) + (int64_t) (((uint32_t) ((1 << n) & a)) >> 1)); + } else { + val = ((fSE32_64(a)) + (1 << (n - 1))); + } + + val = val >> n; + return (int32_t)val; +} + +/* Floating Point Stuff */ + +static const int softfloat_roundingmodes[] = { + float_round_nearest_even, + float_round_to_zero, + float_round_down, + float_round_up, +}; + +void arch_fpop_start(CPUHexagonState *env) +{ + set_float_exception_flags(0, &env->fp_status); + set_float_rounding_mode( + softfloat_roundingmodes[fREAD_REG_FIELD(USR, USR_FPRND)], + &env->fp_status); +} + +#ifdef CONFIG_USER_ONLY +/* + * Hexagon Linux kernel only sets the relevant bits in USR (user status + * register). The exception isn't raised to user mode, so we don't + * model it in qemu user mode. + */ +#define RAISE_FP_EXCEPTION do {} while (0) +#endif + +#define SOFTFLOAT_TEST_FLAG(FLAG, MYF, MYE) \ + do { \ + if (flags & FLAG) { \ + if (GET_USR_FIELD(USR_##MYF) == 0) { \ + SET_USR_FIELD(USR_##MYF, 1); \ + if (GET_USR_FIELD(USR_##MYE)) { \ + RAISE_FP_EXCEPTION; \ + } \ + } \ + } \ + } while (0) + +void arch_fpop_end(CPUHexagonState *env) +{ + int flags = get_float_exception_flags(&env->fp_status); + if (flags != 0) { + SOFTFLOAT_TEST_FLAG(float_flag_inexact, FPINPF, FPINPE); + SOFTFLOAT_TEST_FLAG(float_flag_divbyzero, FPDBZF, FPDBZE); + SOFTFLOAT_TEST_FLAG(float_flag_invalid, FPINVF, FPINVE); + SOFTFLOAT_TEST_FLAG(float_flag_overflow, FPOVFF, FPOVFE); + SOFTFLOAT_TEST_FLAG(float_flag_underflow, FPUNFF, FPUNFE); + } +} + +static float32 float32_mul_pow2(float32 a, uint32_t p, float_status *fp_status) +{ + float32 b = make_float32((SF_BIAS + p) << SF_MANTBITS); + return float32_mul(a, b, fp_status); +} + +int arch_sf_recip_common(float32 *Rs, float32 *Rt, float32 *Rd, int *adjust, + float_status *fp_status) +{ + int n_exp; + int d_exp; + int ret = 0; + float32 RsV, RtV, RdV; + int PeV = 0; + RsV = *Rs; + RtV = *Rt; + if (float32_is_any_nan(RsV) && float32_is_any_nan(RtV)) { + if (extract32(RsV & RtV, 22, 1) == 0) { + float_raise(float_flag_invalid, fp_status); + } + RdV = RsV = RtV = float32_nan; + } else if (float32_is_any_nan(RsV)) { + if (extract32(RsV, 22, 1) == 0) { + float_raise(float_flag_invalid, fp_status); + } + RdV = RsV = RtV = float32_nan; + } else if (float32_is_any_nan(RtV)) { + /* or put NaN in num/den fixup? */ + if (extract32(RtV, 22, 1) == 0) { + float_raise(float_flag_invalid, fp_status); + } + RdV = RsV = RtV = float32_nan; + } else if (float32_is_infinity(RsV) && float32_is_infinity(RtV)) { + /* or put Inf in num fixup? */ + RdV = RsV = RtV = float32_nan; + float_raise(float_flag_invalid, fp_status); + } else if (float32_is_zero(RsV) && float32_is_zero(RtV)) { + /* or put zero in num fixup? */ + RdV = RsV = RtV = float32_nan; + float_raise(float_flag_invalid, fp_status); + } else if (float32_is_zero(RtV)) { + /* or put Inf in num fixup? */ + uint8_t RsV_sign = float32_is_neg(RsV); + uint8_t RtV_sign = float32_is_neg(RtV); + RsV = infinite_float32(RsV_sign ^ RtV_sign); + RtV = float32_one; + RdV = float32_one; + if (float32_is_infinity(RsV)) { + float_raise(float_flag_divbyzero, fp_status); + } + } else if (float32_is_infinity(RtV)) { + RsV = make_float32(0x80000000 & (RsV ^ RtV)); + RtV = float32_one; + RdV = float32_one; + } else if (float32_is_zero(RsV)) { + /* Does this just work itself out? */ + /* No, 0/Inf causes problems. */ + RsV = make_float32(0x80000000 & (RsV ^ RtV)); + RtV = float32_one; + RdV = float32_one; + } else if (float32_is_infinity(RsV)) { + uint8_t RsV_sign = float32_is_neg(RsV); + uint8_t RtV_sign = float32_is_neg(RtV); + RsV = infinite_float32(RsV_sign ^ RtV_sign); + RtV = float32_one; + RdV = float32_one; + } else { + PeV = 0x00; + /* Basic checks passed */ + n_exp = float32_getexp(RsV); + d_exp = float32_getexp(RtV); + if ((n_exp - d_exp + SF_BIAS) <= SF_MANTBITS) { + /* Near quotient underflow / inexact Q */ + PeV = 0x80; + RtV = float32_mul_pow2(RtV, -64, fp_status); + RsV = float32_mul_pow2(RsV, 64, fp_status); + } else if ((n_exp - d_exp + SF_BIAS) > (SF_MAXEXP - 24)) { + /* Near quotient overflow */ + PeV = 0x40; + RtV = float32_mul_pow2(RtV, 32, fp_status); + RsV = float32_mul_pow2(RsV, -32, fp_status); + } else if (n_exp <= SF_MANTBITS + 2) { + RtV = float32_mul_pow2(RtV, 64, fp_status); + RsV = float32_mul_pow2(RsV, 64, fp_status); + } else if (d_exp <= 1) { + RtV = float32_mul_pow2(RtV, 32, fp_status); + RsV = float32_mul_pow2(RsV, 32, fp_status); + } else if (d_exp > 252) { + RtV = float32_mul_pow2(RtV, -32, fp_status); + RsV = float32_mul_pow2(RsV, -32, fp_status); + } + RdV = 0; + ret = 1; + } + *Rs = RsV; + *Rt = RtV; + *Rd = RdV; + *adjust = PeV; + return ret; +} + +int arch_sf_invsqrt_common(float32 *Rs, float32 *Rd, int *adjust, + float_status *fp_status) +{ + float32 RsV, RdV; + int PeV = 0; + int r_exp; + int ret = 0; + RsV = *Rs; + if (float32_is_infinity(RsV)) { + if (extract32(RsV, 22, 1) == 0) { + float_raise(float_flag_invalid, fp_status); + } + RdV = RsV = float32_nan; + } else if (float32_lt(RsV, float32_zero, fp_status)) { + /* Negative nonzero values are NaN */ + float_raise(float_flag_invalid, fp_status); + RsV = float32_nan; + RdV = float32_nan; + } else if (float32_is_infinity(RsV)) { + /* or put Inf in num fixup? */ + RsV = infinite_float32(1); + RdV = infinite_float32(1); + } else if (float32_is_zero(RsV)) { + /* or put zero in num fixup? */ + RdV = float32_one; + } else { + PeV = 0x00; + /* Basic checks passed */ + r_exp = float32_getexp(RsV); + if (r_exp <= 24) { + RsV = float32_mul_pow2(RsV, 64, fp_status); + PeV = 0xe0; + } + RdV = 0; + ret = 1; + } + *Rs = RsV; + *Rd = RdV; + *adjust = PeV; + return ret; +} diff --git a/target/hexagon/arch.h b/target/hexagon/arch.h new file mode 100644 index 0000000000..1f7f03693a --- /dev/null +++ b/target/hexagon/arch.h @@ -0,0 +1,34 @@ +/* + * Copyright(c) 2019-2021 Qualcomm Innovation Center, Inc. All Rights Reserved. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, see . + */ + +#ifndef HEXAGON_ARCH_H +#define HEXAGON_ARCH_H + +#include "qemu/int128.h" + +uint64_t interleave(uint32_t odd, uint32_t even); +uint64_t deinterleave(uint64_t src); +uint32_t carry_from_add64(uint64_t a, uint64_t b, uint32_t c); +int32_t conv_round(int32_t a, int n); +void arch_fpop_start(CPUHexagonState *env); +void arch_fpop_end(CPUHexagonState *env); +int arch_sf_recip_common(float32 *Rs, float32 *Rt, float32 *Rd, + int *adjust, float_status *fp_status); +int arch_sf_invsqrt_common(float32 *Rs, float32 *Rd, int *adjust, + float_status *fp_status); + +#endif