/* 66 0F 3A 62 */
void BX_CPP_AttrRegparmN(1) BX_CPU_C::PCMPISTRM_VdqWdqIb(bxInstruction_c *i)
{
#if (BX_SUPPORT_SSE >= 5) || (BX_SUPPORT_SSE >= 4 && BX_SUPPORT_SSE_EXTENSION > 0)
BX_CPU_THIS_PTR prepareSSE();
BxPackedXmmRegister op1 = BX_READ_XMM_REG(i->nnn()), op2, result;
Bit8u imm8 = i->Ib();
/* op2 is a register or memory reference */
if (i->modC0()) {
op2 = BX_READ_XMM_REG(i->rm());
}
else {
BX_CPU_CALL_METHODR(i->ResolveModrm, (i));
/* pointer, segment address pair */
readVirtualDQwordAligned(i->seg(), RMAddr(i), (Bit8u *) &op2);
}
// compare all pairs of Ai, Bj
bx_bool BoolRes[16][16];
compare_strings(BoolRes, op1, op2, imm8);
unsigned num_elements = (imm8 & 0x1) ? 8 : 16;
unsigned len1 = find_eos(op1, imm8);
unsigned len2 = find_eos(op2, imm8);
Bit16u result2 = aggregate(BoolRes, len1, len2, imm8);
// As defined by imm8[6], result2 is then either stored to the least
// significant bits of XMM0 (zero extended to 128 bits) or expanded
// into a byte/word-mask and then stored to XMM0
if (imm8 & 0x40) {
if (num_elements == 8) {
for (int index = 0; index < 8; index++)
result.xmm16u(index) = (result2 & (1<<index)) ? 0xffff : 0;
}
else { // num_elements = 16
for (int index = 0; index < 16; index++)
result.xmmubyte(index) = (result2 & (1<<index)) ? 0xff : 0;
}
}
else {
result.xmm64u(1) = 0;
result.xmm64u(0) = (Bit64u) result2;
}
Bit32u flags = 0;
if (result2 != 0) flags |= EFlagsCFMask;
if (len1 < num_elements) flags |= EFlagsSFMask;
if (len2 < num_elements) flags |= EFlagsZFMask;
if (result2 & 0x1)
flags |= EFlagsOFMask;
setEFlagsOSZAPC(flags);
BX_WRITE_XMM_REG(0, result); /* store result XMM0 */
#else
BX_INFO(("PCMPISTRM_VdqWdqIb: required SSE4.2, use --enable-sse and --enable-sse-extension options"));
UndefinedOpcode(i);
#endif
}
/* 66 0F 3A 63 */
void BX_CPP_AttrRegparmN(1) BX_CPU_C::PCMPISTRI_VdqWdqIb(bxInstruction_c *i)
{
#if (BX_SUPPORT_SSE >= 5) || (BX_SUPPORT_SSE >= 4 && BX_SUPPORT_SSE_EXTENSION > 0)
BX_CPU_THIS_PTR prepareSSE();
BxPackedXmmRegister op1 = BX_READ_XMM_REG(i->nnn()), op2;
Bit8u imm8 = i->Ib();
/* op2 is a register or memory reference */
if (i->modC0()) {
op2 = BX_READ_XMM_REG(i->rm());
}
else {
BX_CPU_CALL_METHODR(i->ResolveModrm, (i));
/* pointer, segment address pair */
readVirtualDQwordAligned(i->seg(), RMAddr(i), (Bit8u *) &op2);
}
// compare all pairs of Ai, Bj
bx_bool BoolRes[16][16];
compare_strings(BoolRes, op1, op2, imm8);
unsigned num_elements = (imm8 & 0x1) ? 8 : 16;
int index;
unsigned len1 = find_eos(op1, imm8);
unsigned len2 = find_eos(op2, imm8);
Bit16u result2 = aggregate(BoolRes, len1, len2, imm8);
// The index of the first (or last, according to imm8[6]) set bit of result2
// is returned to ECX. If no bits are set in IntRes2, ECX is set to 16 (8)
if (imm8 & 0x40) {
// The index returned to ECX is of the MSB in result2
for (index=num_elements-1; index>=0; index--)
if (result2 & (1<<index)) break;
if (index < 0) index = num_elements;
}
else {
// The index returned to ECX is of the LSB in result2
for (index=0; index<(int)num_elements; index++)
if (result2 & (1<<index)) break;
}
RCX = index;
Bit32u flags = 0;
if (result2 != 0) flags |= EFlagsCFMask;
if (len1 < num_elements) flags |= EFlagsSFMask;
if (len2 < num_elements) flags |= EFlagsZFMask;
if (result2 & 0x1)
flags |= EFlagsOFMask;
setEFlagsOSZAPC(flags);
#else
BX_INFO(("PCMPISTRI_VdqWdqIb: required SSE4.2, use --enable-sse and --enable-sse-extension options"));
UndefinedOpcode(i);
#endif
}
/* 66 0F 3A 60 */
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::PCMPESTRM_VdqWdqIbR(bxInstruction_c *i)
{
BxPackedXmmRegister op1 = BX_READ_XMM_REG(i->nnn());
BxPackedXmmRegister op2 = BX_READ_XMM_REG(i->rm()), result;
Bit8u imm8 = i->Ib();
// compare all pairs of Ai, Bj
Bit8u BoolRes[16][16];
compare_strings(BoolRes, op1, op2, imm8);
unsigned len1, len2, num_elements = (imm8 & 0x1) ? 8 : 16;
#if BX_SUPPORT_X86_64
if (i->os64L()) {
len1 = find_eos64(RAX, imm8);
len2 = find_eos64(RDX, imm8);
}
else
#endif
{
len1 = find_eos32(EAX, imm8);
len2 = find_eos32(EDX, imm8);
}
Bit16u result2 = aggregate(BoolRes, len1, len2, imm8);
// As defined by imm8[6], result2 is then either stored to the least
// significant bits of XMM0 (zero extended to 128 bits) or expanded
// into a byte/word-mask and then stored to XMM0
if (imm8 & 0x40) {
if (num_elements == 8) {
for (int index = 0; index < 8; index++)
result.xmm16u(index) = (result2 & (1<<index)) ? 0xffff : 0;
}
else { // num_elements = 16
for (int index = 0; index < 16; index++)
result.xmmubyte(index) = (result2 & (1<<index)) ? 0xff : 0;
}
}
else {
result.xmm64u(1) = 0;
result.xmm64u(0) = (Bit64u) result2;
}
Bit32u flags = 0;
if (result2 != 0) flags |= EFlagsCFMask;
if (len1 < num_elements) flags |= EFlagsSFMask;
if (len2 < num_elements) flags |= EFlagsZFMask;
if (result2 & 0x1)
flags |= EFlagsOFMask;
setEFlagsOSZAPC(flags);
BX_WRITE_XMM_REGZ(0, result, i->getVL()); /* store result XMM0 */
BX_NEXT_INSTR(i);
}
/* 66 0F 3A 61 */
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::PCMPESTRI_VdqWdqIbR(bxInstruction_c *i)
{
BxPackedXmmRegister op1 = BX_READ_XMM_REG(i->nnn()), op2 = BX_READ_XMM_REG(i->rm());
Bit8u imm8 = i->Ib();
// compare all pairs of Ai, Bj
Bit8u BoolRes[16][16];
compare_strings(BoolRes, op1, op2, imm8);
unsigned len1, len2, num_elements = (imm8 & 0x1) ? 8 : 16;
int index;
#if BX_SUPPORT_X86_64
if (i->os64L()) {
len1 = find_eos64(RAX, imm8);
len2 = find_eos64(RDX, imm8);
}
else
#endif
{
len1 = find_eos32(EAX, imm8);
len2 = find_eos32(EDX, imm8);
}
Bit16u result2 = aggregate(BoolRes, len1, len2, imm8);
// The index of the first (or last, according to imm8[6]) set bit of result2
// is returned to ECX. If no bits are set in IntRes2, ECX is set to 16 (8)
if (imm8 & 0x40) {
// The index returned to ECX is of the MSB in result2
for (index=num_elements-1; index>=0; index--)
if (result2 & (1<<index)) break;
if (index < 0) index = num_elements;
}
else {
// The index returned to ECX is of the LSB in result2
for (index=0; index<(int)num_elements; index++)
if (result2 & (1<<index)) break;
}
RCX = index;
Bit32u flags = 0;
if (result2 != 0) flags |= EFlagsCFMask;
if (len1 < num_elements) flags |= EFlagsSFMask;
if (len2 < num_elements) flags |= EFlagsZFMask;
if (result2 & 0x1)
flags |= EFlagsOFMask;
setEFlagsOSZAPC(flags);
BX_NEXT_INSTR(i);
}
/* VMOVSS: VEX.F3.0F 10 (VEX.W ignore) */
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VMOVSS_VssHpsWssR(bxInstruction_c *i)
{
BxPackedXmmRegister op = BX_READ_XMM_REG(i->vvv());
op.xmm32u(0) = BX_READ_XMM_REG_LO_DWORD(i->rm());
BX_WRITE_XMM_REG_CLEAR_HIGH(i->nnn(), op);
BX_NEXT_INSTR(i);
}
/* Opcode: VEX.66.0F.3A 18 (VEX.W=0) */
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VINSERTF128_VdqHdqWdqIbR(bxInstruction_c *i)
{
BxPackedAvxRegister op1 = BX_READ_AVX_REG(i->vvv());
op1.avx128(i->Ib() & 1) = BX_READ_XMM_REG(i->rm());
BX_WRITE_AVX_REG(i->nnn(), op1);
BX_NEXT_INSTR(i);
}
