///////////////////////////////////////////////////////////////////////// // $Id$ ///////////////////////////////////////////////////////////////////////// // // Copyright (c) 2023 Stanislav Shwartsman // Written by Stanislav Shwartsman [sshwarts at sourceforge net] // // This library is free software; you can redistribute it and/or // modify it under the terms of the GNU Lesser General Public // License as published by the Free Software Foundation; either // version 2 of the License, or (at your option) any later version. // // This library 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 // Lesser General Public License for more details. // // You should have received a copy of the GNU Lesser General Public // License along with this library; if not, write to the Free Software // Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA B 02110-1301 USA // ///////////////////////////////////////////////////////////////////////// #define NEED_CPU_REG_SHORTCUTS 1 #include "bochs.h" #include "cpu.h" #define LOG_THIS BX_CPU_THIS_PTR #if BX_SUPPORT_EVEX extern softfloat_status_t mxcsr_to_softfloat_status_word(bx_mxcsr_t mxcsr); extern void mxcsr_to_softfloat_status_word_imm_override(softfloat_status_t &status, Bit8u immb); #include "softfloat3e/include/softfloat-compare.h" #include "simd_int.h" #include "simd_pfp.h" #define EVEX_OP_SCALAR_HALF(HANDLER, func) \ void BX_CPP_AttrRegparmN(1) BX_CPU_C:: HANDLER (bxInstruction_c *i) \ { \ BxPackedXmmRegister op1 = BX_READ_XMM_REG(i->src1()); \ \ float16 op2 = BX_READ_XMM_REG_LO_WORD(i->src2()); \ softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); \ softfloat_status_word_rc_override(status, i); \ op1.xmm16u(0) = (func)(op1.xmm16u(0), op2, &status); \ check_exceptionsSSE(softfloat_getExceptionFlags(&status)); \ \ BX_WRITE_XMM_REG_CLEAR_HIGH(i->dst(), op1); \ BX_NEXT_INSTR(i); \ } EVEX_OP_SCALAR_HALF(VADDSH_VshHphWshR, f16_add) EVEX_OP_SCALAR_HALF(VSUBSH_VshHphWshR, f16_sub) EVEX_OP_SCALAR_HALF(VMULSH_VshHphWshR, f16_mul) EVEX_OP_SCALAR_HALF(VDIVSH_VshHphWshR, f16_div) EVEX_OP_SCALAR_HALF(VMINSH_VshHphWshR, f16_min) EVEX_OP_SCALAR_HALF(VMAXSH_VshHphWshR, f16_max) EVEX_OP_SCALAR_HALF(VSCALEFSH_VshHphWshR, f16_scalef) #define EVEX_OP_SCALAR_HALF_MASK(HANDLER, func) \ void BX_CPP_AttrRegparmN(1) BX_CPU_C:: HANDLER (bxInstruction_c *i) \ { \ BxPackedXmmRegister op1 = BX_READ_XMM_REG(i->src1()); \ \ if (BX_SCALAR_ELEMENT_MASK(i->opmask())) { \ float16 op2 = BX_READ_XMM_REG_LO_WORD(i->src2()); \ softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); \ softfloat_status_word_rc_override(status, i); \ op1.xmm16u(0) = (func)(op1.xmm16u(0), op2, &status); \ check_exceptionsSSE(softfloat_getExceptionFlags(&status)); \ } \ else { \ if (i->isZeroMasking()) \ op1.xmm16u(0) = 0; \ else \ op1.xmm16u(0) = BX_READ_XMM_REG_LO_WORD(i->dst()); \ } \ \ BX_WRITE_XMM_REG_CLEAR_HIGH(i->dst(), op1); \ BX_NEXT_INSTR(i); \ } EVEX_OP_SCALAR_HALF_MASK(VADDSH_MASK_VshHphWshR, f16_add) EVEX_OP_SCALAR_HALF_MASK(VSUBSH_MASK_VshHphWshR, f16_sub) EVEX_OP_SCALAR_HALF_MASK(VMULSH_MASK_VshHphWshR, f16_mul) EVEX_OP_SCALAR_HALF_MASK(VDIVSH_MASK_VshHphWshR, f16_div) EVEX_OP_SCALAR_HALF_MASK(VMINSH_MASK_VshHphWshR, f16_min) EVEX_OP_SCALAR_HALF_MASK(VMAXSH_MASK_VshHphWshR, f16_max) EVEX_OP_SCALAR_HALF_MASK(VSCALEFSH_MASK_VshHphWshR, f16_scalef) extern float16_compare_method avx_compare16[32]; void BX_CPP_AttrRegparmN(1) BX_CPU_C::VCMPSH_MASK_KGbHshWshIbR(bxInstruction_c *i) { Bit32u result = 0; if (! i->opmask() || BX_SCALAR_ELEMENT_MASK(i->opmask())) { float16 op1 = BX_READ_XMM_REG_LO_WORD(i->src1()); float16 op2 = BX_READ_XMM_REG_LO_WORD(i->src2()); softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); if (avx_compare16[i->Ib() & 0x1F](op1, op2, &status)) result = 1; check_exceptionsSSE(softfloat_getExceptionFlags(&status)); } BX_WRITE_OPMASK(i->dst(), result); BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VCMPPH_MASK_KGdHphWphIbR(bxInstruction_c *i) { BxPackedAvxRegister op1 = BX_READ_AVX_REG(i->src1()), op2 = BX_READ_AVX_REG(i->src2()); unsigned num_elements = WORD_ELEMENTS(i->getVL()); Bit32u opmask = i->opmask() ? BX_READ_32BIT_OPMASK(i->opmask()) : (Bit32u) -1; Bit32u result = 0; softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); int ib = i->Ib() & 0x1F; for (unsigned n=0, mask = 0x1; n < num_elements; n++, mask <<= 1) { if (opmask & mask) { if (avx_compare16[ib](op1.vmm16u(n), op2.vmm16u(n), &status)) result |= mask; } } check_exceptionsSSE(softfloat_getExceptionFlags(&status)); BX_WRITE_OPMASK(i->dst(), result); BX_NEXT_INSTR(i); } #include "cpu/decoder/ia_opcodes.h" void BX_CPP_AttrRegparmN(1) BX_CPU_C::VCOMISH_VshWshR(bxInstruction_c *i) { float16 op1 = BX_READ_XMM_REG_LO_WORD(i->dst()), op2 = BX_READ_XMM_REG_LO_WORD(i->src()); softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); bool quiet = (i->getIaOpcode() == BX_IA_EVEX_VUCOMISH_VshWsh); int rc = f16_compare(op1, op2, quiet, &status); check_exceptionsSSE(softfloat_getExceptionFlags(&status)); BX_CPU_THIS_PTR write_eflags_fpu_compare(rc); BX_NEXT_INSTR(i); } #include "fpu/softfloat-specialize.h" static BX_CPP_INLINE int f16_fpclass(float16 op, int selector, int daz) { extern int fpclass(softfloat_class_t op_class, int sign, int selector); if (daz) op = f16_denormal_to_zero(op); return fpclass(f16_class(op), f16_sign(op), selector); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VFPCLASSPH_MASK_KGdWphIbR(bxInstruction_c *i) { BxPackedAvxRegister op = BX_READ_AVX_REG(i->src()); unsigned num_elements = WORD_ELEMENTS(i->getVL()); Bit32u opmask = i->opmask() ? BX_READ_32BIT_OPMASK(i->opmask()) : (Bit32u) -1; Bit32u result = 0; int selector = i->Ib(), daz = MXCSR.get_DAZ(); for (unsigned n=0, mask = 0x1; n < num_elements; n++, mask <<= 1) { if (opmask & mask) { if (f16_fpclass(op.vmm16u(n), selector, daz)) result |= mask; } } BX_WRITE_OPMASK(i->dst(), result); BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VFPCLASSSH_MASK_KGbWshIbR(bxInstruction_c *i) { if (! i->opmask() || BX_SCALAR_ELEMENT_MASK(i->opmask())) { BX_WRITE_OPMASK(i->dst(), f16_fpclass(BX_READ_XMM_REG_LO_WORD(i->src()), i->Ib(), MXCSR.get_DAZ())); } else { BX_WRITE_OPMASK(i->dst(), 0); } BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VSQRTSH_VshHphWshR(bxInstruction_c *i) { BxPackedXmmRegister op1 = BX_READ_XMM_REG(i->src1()); float16 op2 = BX_READ_XMM_REG_LO_WORD(i->src2()); softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); op1.xmm16u(0) = f16_sqrt(op2, &status); check_exceptionsSSE(softfloat_getExceptionFlags(&status)); BX_WRITE_XMM_REG_CLEAR_HIGH(i->dst(), op1); BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VSQRTSH_MASK_VshHphWshR(bxInstruction_c *i) { BxPackedXmmRegister op1 = BX_READ_XMM_REG(i->src1()); if (BX_SCALAR_ELEMENT_MASK(i->opmask())) { float16 op2 = BX_READ_XMM_REG_LO_WORD(i->src2()); softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); op1.xmm16u(0) = f16_sqrt(op2, &status); check_exceptionsSSE(softfloat_getExceptionFlags(&status)); } else { if (i->isZeroMasking()) op1.xmm16u(0) = 0; else op1.xmm16u(0) = BX_READ_XMM_REG_LO_WORD(i->dst()); } BX_WRITE_XMM_REG_CLEAR_HIGH(i->dst(), op1); BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VGETEXPSH_VshHphWshR(bxInstruction_c *i) { BxPackedXmmRegister op1 = BX_READ_XMM_REG(i->src1()); float16 op2 = BX_READ_XMM_REG_LO_WORD(i->src2()); softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); op1.xmm16u(0) = f16_getExp(op2, &status); check_exceptionsSSE(softfloat_getExceptionFlags(&status)); BX_WRITE_XMM_REG_CLEAR_HIGH(i->dst(), op1); BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VGETEXPSH_MASK_VshHphWshR(bxInstruction_c *i) { BxPackedXmmRegister op1 = BX_READ_XMM_REG(i->src1()); if (BX_SCALAR_ELEMENT_MASK(i->opmask())) { float16 op2 = BX_READ_XMM_REG_LO_WORD(i->src2()); softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); op1.xmm16u(0) = f16_getExp(op2, &status); check_exceptionsSSE(softfloat_getExceptionFlags(&status)); } else { if (i->isZeroMasking()) op1.xmm16u(0) = 0; else op1.xmm16u(0) = BX_READ_XMM_REG_LO_WORD(i->dst()); } BX_WRITE_XMM_REG_CLEAR_HIGH(i->dst(), op1); BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VGETMANTSH_MASK_VshHphWshIbR(bxInstruction_c *i) { BxPackedXmmRegister op1 = BX_READ_XMM_REG(i->src1()); if (! i->opmask() || BX_SCALAR_ELEMENT_MASK(i->opmask())) { float16 op2 = BX_READ_XMM_REG_LO_WORD(i->src2()); int sign_ctrl = (i->Ib() >> 2) & 0x3; int interv = i->Ib() & 0x3; softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); op1.xmm16u(0) = f16_getMant(op2, &status, sign_ctrl, interv); check_exceptionsSSE(softfloat_getExceptionFlags(&status)); } else { if (i->isZeroMasking()) op1.xmm16u(0) = 0; else op1.xmm16u(0) = BX_READ_XMM_REG_LO_WORD(i->dst()); } BX_WRITE_XMM_REG_CLEAR_HIGH(i->dst(), op1); BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VGETMANTPH_MASK_VphWphIbR(bxInstruction_c *i) { BxPackedAvxRegister op = BX_READ_AVX_REG(i->src()); Bit32u opmask = i->opmask() ? BX_READ_32BIT_OPMASK(i->opmask()) : (Bit32u) -1; unsigned len = i->getVL(); unsigned num_elements = WORD_ELEMENTS(len); softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); int sign_ctrl = (i->Ib() >> 2) & 0x3; int interv = i->Ib() & 0x3; for (unsigned n=0, mask = 0x1; n < num_elements; n++, mask <<= 1) { if (opmask & mask) op.vmm16u(n) = f16_getMant(op.vmm16u(n), &status, sign_ctrl, interv); else op.vmm16u(n) = 0; } check_exceptionsSSE(softfloat_getExceptionFlags(&status)); if (! i->isZeroMasking()) { simd_pblendw(&BX_READ_AVX_REG(i->dst()), &op, opmask, num_elements); BX_CLEAR_AVX_REGZ(i->dst(), len); } else { BX_WRITE_AVX_REGZ(i->dst(), op, len); } BX_NEXT_INSTR(i); } static BX_CPP_INLINE float16 float16_reduce(float16 a, Bit8u scale, softfloat_status_t &status) { if (a == float16_negative_inf || a == float16_positive_inf) return 0; float16 tmp = f16_roundToInt(a, scale, &status); return f16_sub(a, tmp, &status); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VREDUCEPH_MASK_VphWphIbR(bxInstruction_c *i) { BxPackedAvxRegister op = BX_READ_AVX_REG(i->src()); Bit32u opmask = i->opmask() ? BX_READ_32BIT_OPMASK(i->opmask()) : (Bit32u) -1; unsigned len = i->getVL(); unsigned num_elements = WORD_ELEMENTS(len); Bit8u control = i->Ib(), scale = control >> 4; softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); mxcsr_to_softfloat_status_word_imm_override(status, control); status.softfloat_suppressException |= softfloat_flag_denormal | softfloat_flag_underflow | softfloat_flag_overflow; for (unsigned n=0, mask = 0x1; n < num_elements; n++, mask <<= 1) { if (opmask & mask) op.vmm16u(n) = float16_reduce(op.vmm16u(n), scale, status); else op.vmm16u(n) = 0; } check_exceptionsSSE(softfloat_getExceptionFlags(&status)); if (! i->isZeroMasking()) { simd_pblendw(&BX_READ_AVX_REG(i->dst()), &op, opmask, num_elements); BX_CLEAR_AVX_REGZ(i->dst(), len); } else { BX_WRITE_AVX_REGZ(i->dst(), op, len); } BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VREDUCESH_MASK_VshHphWshIbR(bxInstruction_c *i) { BxPackedXmmRegister op1 = BX_READ_XMM_REG(i->src1()); if (! i->opmask() || BX_SCALAR_ELEMENT_MASK(i->opmask())) { float16 op2 = BX_READ_XMM_REG_LO_WORD(i->src2()); Bit8u control = i->Ib(), scale = control >> 4; softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); mxcsr_to_softfloat_status_word_imm_override(status, control); status.softfloat_suppressException |= softfloat_flag_denormal | softfloat_flag_underflow | softfloat_flag_overflow; op1.xmm16u(0) = float16_reduce(op2, scale, status); check_exceptionsSSE(softfloat_getExceptionFlags(&status)); } else { if (i->isZeroMasking()) op1.xmm16u(0) = 0; else op1.xmm16u(0) = BX_READ_XMM_REG_LO_WORD(i->dst()); } BX_WRITE_XMM_REG_CLEAR_HIGH(i->dst(), op1); BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VRNDSCALEPH_MASK_VphWphIbR(bxInstruction_c *i) { BxPackedAvxRegister op = BX_READ_AVX_REG(i->src()); Bit32u opmask = i->opmask() ? BX_READ_32BIT_OPMASK(i->opmask()) : (Bit32u) -1; unsigned len = i->getVL(); unsigned num_elements = WORD_ELEMENTS(len); Bit8u control = i->Ib(), scale = control >> 4; softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); mxcsr_to_softfloat_status_word_imm_override(status, control); for (unsigned n=0, mask = 0x1; n < num_elements; n++, mask <<= 1) { if (opmask & mask) op.vmm16u(n) = f16_roundToInt(op.vmm16u(n), scale, &status); else op.vmm16u(n) = 0; } check_exceptionsSSE(softfloat_getExceptionFlags(&status)); if (! i->isZeroMasking()) { simd_pblendw(&BX_READ_AVX_REG(i->dst()), &op, opmask, num_elements); BX_CLEAR_AVX_REGZ(i->dst(), len); } else { BX_WRITE_AVX_REGZ(i->dst(), op, len); } BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VRNDSCALESH_MASK_VshHphWshIbR(bxInstruction_c *i) { BxPackedXmmRegister op1 = BX_READ_XMM_REG(i->src1()); if (! i->opmask() || BX_SCALAR_ELEMENT_MASK(i->opmask())) { float16 op2 = BX_READ_XMM_REG_LO_WORD(i->src2()); Bit8u control = i->Ib(), scale = control >> 4; softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); mxcsr_to_softfloat_status_word_imm_override(status, control); op1.xmm16u(0) = f16_roundToInt(op2, scale, &status); check_exceptionsSSE(softfloat_getExceptionFlags(&status)); } else { if (i->isZeroMasking()) op1.xmm16u(0) = 0; else op1.xmm16u(0) = BX_READ_XMM_REG_LO_WORD(i->dst()); } BX_WRITE_XMM_REG_CLEAR_HIGH(i->dst(), op1); BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VFCMULCSH_MASK_VshHphWshR(bxInstruction_c *i) { if (i->dst() == i->src1() || i->dst() == i->src2()) { BX_DEBUG(("%s: dst register must be different from sources", i->getIaOpcodeNameShort())); exception(BX_UD_EXCEPTION, 0); } BxPackedXmmRegister *op1 = &BX_READ_XMM_REG(i->src1()), *op2 = &BX_READ_XMM_REG(i->src2()), dst = BX_READ_XMM_REG(i->src1()); dst.xmm32u(0) = BX_READ_XMM_REG_LO_DWORD(i->dst()); softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); status.softfloat_exceptionMasks = softfloat_all_exceptions_mask; int fma_control1 = (i->getIaOpcode() == BX_IA_EVEX_VFMULCSH_VshHphWsh_Kmask) ? softfloat_muladd_negate_product : 0; int fma_control2 = (i->getIaOpcode() == BX_IA_EVEX_VFMULCSH_VshHphWsh_Kmask) ? 0 : softfloat_muladd_negate_product; if (! i->opmask() || BX_SCALAR_ELEMENT_MASK(i->opmask())) { dst.xmm16u(0) = f16_mul(op1->xmm16u(0), op2->xmm16u(0), &status); dst.xmm16u(1) = f16_mul(op1->xmm16u(1), op2->xmm16u(0), &status); dst.xmm16u(0) = f16_mulAdd(op1->xmm16u(1), op2->xmm16u(1), dst.xmm16u(0), fma_control1, &status); dst.xmm16u(1) = f16_mulAdd(op1->xmm16u(0), op2->xmm16u(1), dst.xmm16u(1), fma_control2, &status); } else if (i->isZeroMasking()) { dst.xmm32u(0) = 0; } check_exceptionsSSE(softfloat_getExceptionFlags(&status)); BX_WRITE_XMM_REG_CLEAR_HIGH(i->dst(), dst); BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VFCMULCPH_MASK_VphHphWphR(bxInstruction_c *i) { if (i->dst() == i->src1() || i->dst() == i->src2()) { BX_DEBUG(("%s: dst register must be different from sources", i->getIaOpcodeNameShort())); exception(BX_UD_EXCEPTION, 0); } BxPackedAvxRegister *op1 = &BX_READ_AVX_REG(i->src1()), *op2 = &BX_READ_AVX_REG(i->src2()), dst = BX_READ_AVX_REG(i->dst()); unsigned len = i->getVL(), n; Bit32u mask = (i->opmask() != 0) ? BX_READ_16BIT_OPMASK(i->opmask()) : 0xffff, tmp_mask; softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); status.softfloat_exceptionMasks = softfloat_all_exceptions_mask; for (n=0, tmp_mask = mask; n < DWORD_ELEMENTS(len); n++, tmp_mask >>= 1) { if (tmp_mask & 0x1) { dst.vmm16u(2*n) = f16_mul(op1->vmm16u(2*n), op2->vmm16u(2*n), &status); dst.vmm16u(2*n+1) = f16_mul(op1->vmm16u(2*n+1), op2->vmm16u(2*n), &status); } } int fma_control1 = (i->getIaOpcode() == BX_IA_EVEX_VFMULCPH_VphHphWph_Kmask) ? softfloat_muladd_negate_product : 0; int fma_control2 = (i->getIaOpcode() == BX_IA_EVEX_VFMULCPH_VphHphWph_Kmask) ? 0 : softfloat_muladd_negate_product; for (n=0, tmp_mask = mask; n < DWORD_ELEMENTS(len); n++, tmp_mask >>= 1) { if (tmp_mask & 0x1) { dst.vmm16u(2*n) = f16_mulAdd(op1->vmm16u(2*n+1), op2->vmm16u(2*n+1), dst.vmm16u(2*n), fma_control1, &status); dst.vmm16u(2*n+1) = f16_mulAdd(op1->vmm16u(2*n), op2->vmm16u(2*n+1), dst.vmm16u(2*n+1), fma_control2, &status); } else if (i->isZeroMasking()) { dst.vmm32u(n) = 0; } } check_exceptionsSSE(softfloat_getExceptionFlags(&status)); BX_WRITE_AVX_REGZ(i->dst(), dst, len); BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VFCMADDCSH_MASK_VshHphWshR(bxInstruction_c *i) { if (i->dst() == i->src1() || i->dst() == i->src2()) { BX_DEBUG(("%s: dst register must be different from sources", i->getIaOpcodeNameShort())); exception(BX_UD_EXCEPTION, 0); } BxPackedXmmRegister *op1 = &BX_READ_XMM_REG(i->src1()), *op2 = &BX_READ_XMM_REG(i->src2()), dst = BX_READ_XMM_REG(i->src1()); dst.xmm32u(0) = BX_READ_XMM_REG_LO_DWORD(i->dst()); softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); status.softfloat_exceptionMasks = softfloat_all_exceptions_mask; int fma_control1 = (i->getIaOpcode() == BX_IA_EVEX_VFMADDCSH_VshHphWsh_Kmask) ? softfloat_muladd_negate_product : 0; int fma_control2 = (i->getIaOpcode() == BX_IA_EVEX_VFMADDCSH_VshHphWsh_Kmask) ? 0 : softfloat_muladd_negate_product; if (! i->opmask() || BX_SCALAR_ELEMENT_MASK(i->opmask())) { dst.xmm16u(0) = f16_mulAdd(op1->xmm16u(0), op2->xmm16u(0), dst.xmm16u(0), 0, &status); dst.xmm16u(1) = f16_mulAdd(op1->xmm16u(1), op2->xmm16u(0), dst.xmm16u(1), 0, &status); dst.xmm16u(0) = f16_mulAdd(op1->xmm16u(1), op2->xmm16u(1), dst.xmm16u(0), fma_control1, &status); dst.xmm16u(1) = f16_mulAdd(op1->xmm16u(0), op2->xmm16u(1), dst.xmm16u(1), fma_control2, &status); } else if (i->isZeroMasking()) { dst.xmm32u(0) = 0; } check_exceptionsSSE(softfloat_getExceptionFlags(&status)); BX_WRITE_XMM_REG_CLEAR_HIGH(i->dst(), dst); BX_NEXT_INSTR(i); } void BX_CPP_AttrRegparmN(1) BX_CPU_C::VFCMADDCPH_MASK_VphHphWphR(bxInstruction_c *i) { if (i->dst() == i->src1() || i->dst() == i->src2()) { BX_DEBUG(("%s: dst register must be different from sources", i->getIaOpcodeNameShort())); exception(BX_UD_EXCEPTION, 0); } BxPackedAvxRegister *op1 = &BX_READ_AVX_REG(i->src1()), *op2 = &BX_READ_AVX_REG(i->src2()), dst = BX_READ_AVX_REG(i->dst()); unsigned len = i->getVL(), n; Bit32u mask = (i->opmask() != 0) ? BX_READ_16BIT_OPMASK(i->opmask()) : 0xffff, tmp_mask; softfloat_status_t status = mxcsr_to_softfloat_status_word(MXCSR); softfloat_status_word_rc_override(status, i); status.softfloat_exceptionMasks = softfloat_all_exceptions_mask; for (n=0, tmp_mask = mask; n < DWORD_ELEMENTS(len); n++, tmp_mask >>= 1) { if (tmp_mask & 0x1) { dst.vmm16u(2*n) = f16_mulAdd(op1->vmm16u(2*n), op2->vmm16u(2*n), dst.vmm16u(2*n), 0, &status); dst.vmm16u(2*n+1) = f16_mulAdd(op1->vmm16u(2*n+1), op2->vmm16u(2*n), dst.vmm16u(2*n+1), 0, &status); } } int fma_control1 = (i->getIaOpcode() == BX_IA_EVEX_VFMADDCPH_VphHphWph_Kmask) ? softfloat_muladd_negate_product : 0; int fma_control2 = (i->getIaOpcode() == BX_IA_EVEX_VFMADDCPH_VphHphWph_Kmask) ? 0 : softfloat_muladd_negate_product; for (n=0, tmp_mask = mask; n < DWORD_ELEMENTS(len); n++, tmp_mask >>= 1) { if (tmp_mask & 0x1) { dst.vmm16u(2*n) = f16_mulAdd(op1->vmm16u(2*n+1), op2->vmm16u(2*n+1), dst.vmm16u(2*n), fma_control1, &status); dst.vmm16u(2*n+1) = f16_mulAdd(op1->vmm16u(2*n), op2->vmm16u(2*n+1), dst.vmm16u(2*n+1), fma_control2, &status); } else if (i->isZeroMasking()) { dst.vmm32u(n) = 0; } } check_exceptionsSSE(softfloat_getExceptionFlags(&status)); BX_WRITE_AVX_REGZ(i->dst(), dst, len); BX_NEXT_INSTR(i); } #endif