aboutsummaryrefslogtreecommitdiff
path: root/lib/Target/X86/X86InstrSSE.td
diff options
context:
space:
mode:
Diffstat (limited to 'lib/Target/X86/X86InstrSSE.td')
-rw-r--r--lib/Target/X86/X86InstrSSE.td162
1 files changed, 104 insertions, 58 deletions
diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td
index 212958025bde..5580ba74e64e 100644
--- a/lib/Target/X86/X86InstrSSE.td
+++ b/lib/Target/X86/X86InstrSSE.td
@@ -117,17 +117,17 @@ def alignedload : PatFrag<(ops node:$ptr), (load node:$ptr), [{
return cast<LoadSDNode>(N)->getAlignment() >= 16;
}]>;
-def alignedloadfsf32 : PatFrag<(ops node:$ptr),
+def alignedloadfsf32 : PatFrag<(ops node:$ptr),
(f32 (alignedload node:$ptr))>;
-def alignedloadfsf64 : PatFrag<(ops node:$ptr),
+def alignedloadfsf64 : PatFrag<(ops node:$ptr),
(f64 (alignedload node:$ptr))>;
-def alignedloadv4f32 : PatFrag<(ops node:$ptr),
+def alignedloadv4f32 : PatFrag<(ops node:$ptr),
(v4f32 (alignedload node:$ptr))>;
-def alignedloadv2f64 : PatFrag<(ops node:$ptr),
+def alignedloadv2f64 : PatFrag<(ops node:$ptr),
(v2f64 (alignedload node:$ptr))>;
-def alignedloadv4i32 : PatFrag<(ops node:$ptr),
+def alignedloadv4i32 : PatFrag<(ops node:$ptr),
(v4i32 (alignedload node:$ptr))>;
-def alignedloadv2i64 : PatFrag<(ops node:$ptr),
+def alignedloadv2i64 : PatFrag<(ops node:$ptr),
(v2i64 (alignedload node:$ptr))>;
// Like 'load', but uses special alignment checks suitable for use in
@@ -387,11 +387,11 @@ def MOVSSrr : SSI<0x10, MRMSrcReg,
let AddedComplexity = 15 in
def : Pat<(v4f32 (movl VR128:$src1, VR128:$src2)),
(MOVSSrr (v4f32 VR128:$src1),
- (EXTRACT_SUBREG (v4f32 VR128:$src2), x86_subreg_ss))>;
+ (EXTRACT_SUBREG (v4f32 VR128:$src2), sub_ss))>;
// Implicitly promote a 32-bit scalar to a vector.
def : Pat<(v4f32 (scalar_to_vector FR32:$src)),
- (INSERT_SUBREG (v4f32 (IMPLICIT_DEF)), FR32:$src, x86_subreg_ss)>;
+ (INSERT_SUBREG (v4f32 (IMPLICIT_DEF)), FR32:$src, sub_ss)>;
// Loading from memory automatically zeroing upper bits.
let canFoldAsLoad = 1, isReMaterializable = 1 in
@@ -403,11 +403,11 @@ def MOVSSrm : SSI<0x10, MRMSrcMem, (outs FR32:$dst), (ins f32mem:$src),
// with SUBREG_TO_REG.
let AddedComplexity = 20 in {
def : Pat<(v4f32 (X86vzmovl (v4f32 (scalar_to_vector (loadf32 addr:$src))))),
- (SUBREG_TO_REG (i32 0), (MOVSSrm addr:$src), x86_subreg_ss)>;
+ (SUBREG_TO_REG (i32 0), (MOVSSrm addr:$src), sub_ss)>;
def : Pat<(v4f32 (scalar_to_vector (loadf32 addr:$src))),
- (SUBREG_TO_REG (i32 0), (MOVSSrm addr:$src), x86_subreg_ss)>;
+ (SUBREG_TO_REG (i32 0), (MOVSSrm addr:$src), sub_ss)>;
def : Pat<(v4f32 (X86vzmovl (loadv4f32 addr:$src))),
- (SUBREG_TO_REG (i32 0), (MOVSSrm addr:$src), x86_subreg_ss)>;
+ (SUBREG_TO_REG (i32 0), (MOVSSrm addr:$src), sub_ss)>;
}
// Store scalar value to memory.
@@ -419,7 +419,7 @@ def MOVSSmr : SSI<0x11, MRMDestMem, (outs), (ins f32mem:$dst, FR32:$src),
def : Pat<(store (f32 (vector_extract (v4f32 VR128:$src), (iPTR 0))),
addr:$dst),
(MOVSSmr addr:$dst,
- (EXTRACT_SUBREG (v4f32 VR128:$src), x86_subreg_ss))>;
+ (EXTRACT_SUBREG (v4f32 VR128:$src), sub_ss))>;
// Conversion instructions
def CVTTSS2SIrr : SSI<0x2C, MRMSrcReg, (outs GR32:$dst), (ins FR32:$src),
@@ -449,7 +449,7 @@ def Int_CVTSS2SIrm : SSI<0x2D, MRMSrcMem, (outs GR32:$dst), (ins f32mem:$src),
[(set GR32:$dst, (int_x86_sse_cvtss2si
(load addr:$src)))]>;
-// Match intrinisics which expect MM and XMM operand(s).
+// Match intrinsics which expect MM and XMM operand(s).
def Int_CVTPS2PIrr : PSI<0x2D, MRMSrcReg, (outs VR64:$dst), (ins VR128:$src),
"cvtps2pi\t{$src, $dst|$dst, $src}",
[(set VR64:$dst, (int_x86_sse_cvtps2pi VR128:$src))]>;
@@ -509,6 +509,17 @@ let mayLoad = 1 in
def CMPSSrm : SSIi8<0xC2, MRMSrcMem,
(outs FR32:$dst), (ins FR32:$src1, f32mem:$src, SSECC:$cc),
"cmp${cc}ss\t{$src, $dst|$dst, $src}", []>;
+
+ // Accept explicit immediate argument form instead of comparison code.
+let isAsmParserOnly = 1 in {
+ def CMPSSrr_alt : SSIi8<0xC2, MRMSrcReg,
+ (outs FR32:$dst), (ins FR32:$src1, FR32:$src, i8imm:$src2),
+ "cmpss\t{$src2, $src, $dst|$dst, $src, $src2}", []>;
+let mayLoad = 1 in
+ def CMPSSrm_alt : SSIi8<0xC2, MRMSrcMem,
+ (outs FR32:$dst), (ins FR32:$src1, f32mem:$src, i8imm:$src2),
+ "cmpss\t{$src2, $src, $dst|$dst, $src, $src2}", []>;
+}
}
let Defs = [EFLAGS] in {
@@ -518,25 +529,25 @@ def UCOMISSrr: PSI<0x2E, MRMSrcReg, (outs), (ins FR32:$src1, FR32:$src2),
def UCOMISSrm: PSI<0x2E, MRMSrcMem, (outs), (ins FR32:$src1, f32mem:$src2),
"ucomiss\t{$src2, $src1|$src1, $src2}",
[(set EFLAGS, (X86cmp FR32:$src1, (loadf32 addr:$src2)))]>;
-
+
def COMISSrr: PSI<0x2F, MRMSrcReg, (outs), (ins VR128:$src1, VR128:$src2),
"comiss\t{$src2, $src1|$src1, $src2}", []>;
def COMISSrm: PSI<0x2F, MRMSrcMem, (outs), (ins VR128:$src1, f128mem:$src2),
"comiss\t{$src2, $src1|$src1, $src2}", []>;
-
+
} // Defs = [EFLAGS]
// Aliases to match intrinsics which expect XMM operand(s).
let Constraints = "$src1 = $dst" in {
def Int_CMPSSrr : SSIi8<0xC2, MRMSrcReg,
- (outs VR128:$dst),
+ (outs VR128:$dst),
(ins VR128:$src1, VR128:$src, SSECC:$cc),
"cmp${cc}ss\t{$src, $dst|$dst, $src}",
- [(set VR128:$dst, (int_x86_sse_cmp_ss
+ [(set VR128:$dst, (int_x86_sse_cmp_ss
VR128:$src1,
VR128:$src, imm:$cc))]>;
def Int_CMPSSrm : SSIi8<0xC2, MRMSrcMem,
- (outs VR128:$dst),
+ (outs VR128:$dst),
(ins VR128:$src1, f32mem:$src, SSECC:$cc),
"cmp${cc}ss\t{$src, $dst|$dst, $src}",
[(set VR128:$dst, (int_x86_sse_cmp_ss VR128:$src1,
@@ -1009,6 +1020,16 @@ let Constraints = "$src1 = $dst" in {
"cmp${cc}ps\t{$src, $dst|$dst, $src}",
[(set VR128:$dst, (int_x86_sse_cmp_ps VR128:$src1,
(memop addr:$src), imm:$cc))]>;
+
+ // Accept explicit immediate argument form instead of comparison code.
+let isAsmParserOnly = 1 in {
+ def CMPPSrri_alt : PSIi8<0xC2, MRMSrcReg,
+ (outs VR128:$dst), (ins VR128:$src1, VR128:$src, i8imm:$src2),
+ "cmpps\t{$src2, $src, $dst|$dst, $src, $src}", []>;
+ def CMPPSrmi_alt : PSIi8<0xC2, MRMSrcMem,
+ (outs VR128:$dst), (ins VR128:$src1, f128mem:$src, i8imm:$src2),
+ "cmpps\t{$src2, $src, $dst|$dst, $src, $src}", []>;
+}
}
def : Pat<(v4i32 (X86cmpps (v4f32 VR128:$src1), VR128:$src2, imm:$cc)),
(CMPPSrri (v4f32 VR128:$src1), (v4f32 VR128:$src2), imm:$cc)>;
@@ -1102,7 +1123,8 @@ def MOVNTI_64mr : RI<0xC3, MRMDestMem, (outs), (ins i64mem:$dst, GR64:$src),
}
// Load, store, and memory fence
-def SFENCE : PSI<0xAE, MRM7r, (outs), (ins), "sfence", [(int_x86_sse_sfence)]>;
+def SFENCE : I<0xAE, MRM_F8, (outs), (ins), "sfence", [(int_x86_sse_sfence)]>,
+ TB, Requires<[HasSSE1]>;
// MXCSR register
def LDMXCSR : PSI<0xAE, MRM2m, (outs), (ins i32mem:$src),
@@ -1130,7 +1152,7 @@ def : Pat<(v8i16 immAllZerosV), (V_SET0PI)>;
def : Pat<(v16i8 immAllZerosV), (V_SET0PI)>;
def : Pat<(f32 (vector_extract (v4f32 VR128:$src), (iPTR 0))),
- (f32 (EXTRACT_SUBREG (v4f32 VR128:$src), x86_subreg_ss))>;
+ (f32 (EXTRACT_SUBREG (v4f32 VR128:$src), sub_ss))>;
//===---------------------------------------------------------------------===//
// SSE2 Instructions
@@ -1152,11 +1174,11 @@ def MOVSDrr : SDI<0x10, MRMSrcReg,
let AddedComplexity = 15 in
def : Pat<(v2f64 (movl VR128:$src1, VR128:$src2)),
(MOVSDrr (v2f64 VR128:$src1),
- (EXTRACT_SUBREG (v2f64 VR128:$src2), x86_subreg_sd))>;
+ (EXTRACT_SUBREG (v2f64 VR128:$src2), sub_sd))>;
// Implicitly promote a 64-bit scalar to a vector.
def : Pat<(v2f64 (scalar_to_vector FR64:$src)),
- (INSERT_SUBREG (v2f64 (IMPLICIT_DEF)), FR64:$src, x86_subreg_sd)>;
+ (INSERT_SUBREG (v2f64 (IMPLICIT_DEF)), FR64:$src, sub_sd)>;
// Loading from memory automatically zeroing upper bits.
let canFoldAsLoad = 1, isReMaterializable = 1, AddedComplexity = 20 in
@@ -1168,15 +1190,15 @@ def MOVSDrm : SDI<0x10, MRMSrcMem, (outs FR64:$dst), (ins f64mem:$src),
// with SUBREG_TO_REG.
let AddedComplexity = 20 in {
def : Pat<(v2f64 (X86vzmovl (v2f64 (scalar_to_vector (loadf64 addr:$src))))),
- (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), x86_subreg_sd)>;
+ (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), sub_sd)>;
def : Pat<(v2f64 (scalar_to_vector (loadf64 addr:$src))),
- (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), x86_subreg_sd)>;
+ (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), sub_sd)>;
def : Pat<(v2f64 (X86vzmovl (loadv2f64 addr:$src))),
- (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), x86_subreg_sd)>;
+ (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), sub_sd)>;
def : Pat<(v2f64 (X86vzmovl (bc_v2f64 (loadv4f32 addr:$src)))),
- (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), x86_subreg_sd)>;
+ (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), sub_sd)>;
def : Pat<(v2f64 (X86vzload addr:$src)),
- (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), x86_subreg_sd)>;
+ (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), sub_sd)>;
}
// Store scalar value to memory.
@@ -1188,7 +1210,7 @@ def MOVSDmr : SDI<0x11, MRMDestMem, (outs), (ins f64mem:$dst, FR64:$src),
def : Pat<(store (f64 (vector_extract (v2f64 VR128:$src), (iPTR 0))),
addr:$dst),
(MOVSDmr addr:$dst,
- (EXTRACT_SUBREG (v2f64 VR128:$src), x86_subreg_sd))>;
+ (EXTRACT_SUBREG (v2f64 VR128:$src), sub_sd))>;
// Conversion instructions
def CVTTSD2SIrr : SDI<0x2C, MRMSrcReg, (outs GR32:$dst), (ins FR64:$src),
@@ -1255,7 +1277,7 @@ def Int_CVTSD2SIrm : SDI<0x2D, MRMSrcMem, (outs GR32:$dst), (ins f128mem:$src),
[(set GR32:$dst, (int_x86_sse2_cvtsd2si
(load addr:$src)))]>;
-// Match intrinisics which expect MM and XMM operand(s).
+// Match intrinsics which expect MM and XMM operand(s).
def Int_CVTPD2PIrr : PDI<0x2D, MRMSrcReg, (outs VR64:$dst), (ins VR128:$src),
"cvtpd2pi\t{$src, $dst|$dst, $src}",
[(set VR64:$dst, (int_x86_sse_cvtpd2pi VR128:$src))]>;
@@ -1297,6 +1319,17 @@ let mayLoad = 1 in
def CMPSDrm : SDIi8<0xC2, MRMSrcMem,
(outs FR64:$dst), (ins FR64:$src1, f64mem:$src, SSECC:$cc),
"cmp${cc}sd\t{$src, $dst|$dst, $src}", []>;
+
+ // Accept explicit immediate argument form instead of comparison code.
+let isAsmParserOnly = 1 in {
+ def CMPSDrr_alt : SDIi8<0xC2, MRMSrcReg,
+ (outs FR64:$dst), (ins FR64:$src1, FR64:$src, i8imm:$src2),
+ "cmpsd\t{$src2, $src, $dst|$dst, $src, $src2}", []>;
+let mayLoad = 1 in
+ def CMPSDrm_alt : SDIi8<0xC2, MRMSrcMem,
+ (outs FR64:$dst), (ins FR64:$src1, f64mem:$src, i8imm:$src2),
+ "cmpsd\t{$src2, $src, $dst|$dst, $src, $src2}", []>;
+}
}
let Defs = [EFLAGS] in {
@@ -1311,13 +1344,13 @@ def UCOMISDrm: PDI<0x2E, MRMSrcMem, (outs), (ins FR64:$src1, f64mem:$src2),
// Aliases to match intrinsics which expect XMM operand(s).
let Constraints = "$src1 = $dst" in {
def Int_CMPSDrr : SDIi8<0xC2, MRMSrcReg,
- (outs VR128:$dst),
+ (outs VR128:$dst),
(ins VR128:$src1, VR128:$src, SSECC:$cc),
"cmp${cc}sd\t{$src, $dst|$dst, $src}",
[(set VR128:$dst, (int_x86_sse2_cmp_sd VR128:$src1,
VR128:$src, imm:$cc))]>;
def Int_CMPSDrm : SDIi8<0xC2, MRMSrcMem,
- (outs VR128:$dst),
+ (outs VR128:$dst),
(ins VR128:$src1, f64mem:$src, SSECC:$cc),
"cmp${cc}sd\t{$src, $dst|$dst, $src}",
[(set VR128:$dst, (int_x86_sse2_cmp_sd VR128:$src1,
@@ -1655,7 +1688,7 @@ def CVTTPS2DQrm : SSI<0x5B, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
def Int_CVTTPS2DQrr : I<0x5B, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
"cvttps2dq\t{$src, $dst|$dst, $src}",
- [(set VR128:$dst,
+ [(set VR128:$dst,
(int_x86_sse2_cvttps2dq VR128:$src))]>,
XS, Requires<[HasSSE2]>;
def Int_CVTTPS2DQrm : I<0x5B, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
@@ -1890,6 +1923,16 @@ let Constraints = "$src1 = $dst" in {
"cmp${cc}pd\t{$src, $dst|$dst, $src}",
[(set VR128:$dst, (int_x86_sse2_cmp_pd VR128:$src1,
(memop addr:$src), imm:$cc))]>;
+
+ // Accept explicit immediate argument form instead of comparison code.
+let isAsmParserOnly = 1 in {
+ def CMPPDrri_alt : PDIi8<0xC2, MRMSrcReg,
+ (outs VR128:$dst), (ins VR128:$src1, VR128:$src, i8imm:$src2),
+ "cmppd\t{$src2, $src, $dst|$dst, $src, $src2}", []>;
+ def CMPPDrmi_alt : PDIi8<0xC2, MRMSrcMem,
+ (outs VR128:$dst), (ins VR128:$src1, f128mem:$src, i8imm:$src2),
+ "cmppd\t{$src2, $src, $dst|$dst, $src, $src2}", []>;
+}
}
def : Pat<(v2i64 (X86cmppd (v2f64 VR128:$src1), VR128:$src2, imm:$cc)),
(CMPPDrri VR128:$src1, VR128:$src2, imm:$cc)>;
@@ -1980,24 +2023,24 @@ let Constraints = "$src1 = $dst" in {
multiclass PDI_binop_rm_int<bits<8> opc, string OpcodeStr, Intrinsic IntId,
bit Commutable = 0> {
- def rr : PDI<opc, MRMSrcReg, (outs VR128:$dst),
+ def rr : PDI<opc, MRMSrcReg, (outs VR128:$dst),
(ins VR128:$src1, VR128:$src2),
!strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
[(set VR128:$dst, (IntId VR128:$src1, VR128:$src2))]> {
let isCommutable = Commutable;
}
- def rm : PDI<opc, MRMSrcMem, (outs VR128:$dst),
+ def rm : PDI<opc, MRMSrcMem, (outs VR128:$dst),
(ins VR128:$src1, i128mem:$src2),
!strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
[(set VR128:$dst, (IntId VR128:$src1,
- (bitconvert (memopv2i64
+ (bitconvert (memopv2i64
addr:$src2))))]>;
}
multiclass PDI_binop_rmi_int<bits<8> opc, bits<8> opc2, Format ImmForm,
string OpcodeStr,
Intrinsic IntId, Intrinsic IntId2> {
- def rr : PDI<opc, MRMSrcReg, (outs VR128:$dst),
+ def rr : PDI<opc, MRMSrcReg, (outs VR128:$dst),
(ins VR128:$src1, VR128:$src2),
!strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
[(set VR128:$dst, (IntId VR128:$src1, VR128:$src2))]>;
@@ -2006,7 +2049,7 @@ multiclass PDI_binop_rmi_int<bits<8> opc, bits<8> opc2, Format ImmForm,
!strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
[(set VR128:$dst, (IntId VR128:$src1,
(bitconvert (memopv2i64 addr:$src2))))]>;
- def ri : PDIi8<opc2, ImmForm, (outs VR128:$dst),
+ def ri : PDIi8<opc2, ImmForm, (outs VR128:$dst),
(ins VR128:$src1, i32i8imm:$src2),
!strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
[(set VR128:$dst, (IntId2 VR128:$src1, (i32 imm:$src2)))]>;
@@ -2015,13 +2058,13 @@ multiclass PDI_binop_rmi_int<bits<8> opc, bits<8> opc2, Format ImmForm,
/// PDI_binop_rm - Simple SSE2 binary operator.
multiclass PDI_binop_rm<bits<8> opc, string OpcodeStr, SDNode OpNode,
ValueType OpVT, bit Commutable = 0> {
- def rr : PDI<opc, MRMSrcReg, (outs VR128:$dst),
+ def rr : PDI<opc, MRMSrcReg, (outs VR128:$dst),
(ins VR128:$src1, VR128:$src2),
!strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
[(set VR128:$dst, (OpVT (OpNode VR128:$src1, VR128:$src2)))]> {
let isCommutable = Commutable;
}
- def rm : PDI<opc, MRMSrcMem, (outs VR128:$dst),
+ def rm : PDI<opc, MRMSrcMem, (outs VR128:$dst),
(ins VR128:$src1, i128mem:$src2),
!strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
[(set VR128:$dst, (OpVT (OpNode VR128:$src1,
@@ -2416,6 +2459,10 @@ def LFENCE : I<0xAE, MRM_E8, (outs), (ins),
def MFENCE : I<0xAE, MRM_F0, (outs), (ins),
"mfence", [(int_x86_sse2_mfence)]>, TB, Requires<[HasSSE2]>;
+// Pause. This "instruction" is encoded as "rep; nop", so even though it
+// was introduced with SSE2, it's backward compatible.
+def PAUSE : I<0x90, RawFrm, (outs), (ins), "pause", []>, REP;
+
//TODO: custom lower this so as to never even generate the noop
def : Pat<(membarrier (i8 imm), (i8 imm), (i8 imm), (i8 imm),
(i8 0)), (NOOP)>;
@@ -2462,7 +2509,7 @@ def MOVPQI2QImr : PDI<0xD6, MRMDestMem, (outs), (ins i64mem:$dst, VR128:$src),
(iPTR 0))), addr:$dst)]>;
def : Pat<(f64 (vector_extract (v2f64 VR128:$src), (iPTR 0))),
- (f64 (EXTRACT_SUBREG (v2f64 VR128:$src), x86_subreg_sd))>;
+ (f64 (EXTRACT_SUBREG (v2f64 VR128:$src), sub_sd))>;
def MOVPDI2DIrr : PDI<0x7E, MRMDestReg, (outs GR32:$dst), (ins VR128:$src),
"movd\t{$src, $dst|$dst, $src}",
@@ -2903,7 +2950,7 @@ defm PMADDUBSW : SS3I_binop_rm_int_8 <0x04, "pmaddubsw",
defm PMULHRSW : SS3I_binop_rm_int_16<0x0B, "pmulhrsw",
int_x86_ssse3_pmul_hr_sw,
int_x86_ssse3_pmul_hr_sw_128, 1>;
-
+
defm PSHUFB : SS3I_binop_rm_int_8 <0x00, "pshufb",
int_x86_ssse3_pshuf_b,
int_x86_ssse3_pshuf_b_128>;
@@ -3042,10 +3089,10 @@ def : Pat<(v4f32 (X86vzmovl (v4f32 (scalar_to_vector FR32:$src)))),
(MOVSSrr (v4f32 (V_SET0PS)), FR32:$src)>;
def : Pat<(v4f32 (X86vzmovl (v4f32 VR128:$src))),
(MOVSSrr (v4f32 (V_SET0PS)),
- (f32 (EXTRACT_SUBREG (v4f32 VR128:$src), x86_subreg_ss)))>;
+ (f32 (EXTRACT_SUBREG (v4f32 VR128:$src), sub_ss)))>;
def : Pat<(v4i32 (X86vzmovl (v4i32 VR128:$src))),
(MOVSSrr (v4i32 (V_SET0PI)),
- (EXTRACT_SUBREG (v4i32 VR128:$src), x86_subreg_ss))>;
+ (EXTRACT_SUBREG (v4i32 VR128:$src), sub_ss))>;
}
// Splat v2f64 / v2i64
@@ -3181,17 +3228,17 @@ let AddedComplexity = 15 in {
// Setting the lowest element in the vector.
def : Pat<(v4i32 (movl VR128:$src1, VR128:$src2)),
(MOVSSrr (v4i32 VR128:$src1),
- (EXTRACT_SUBREG (v4i32 VR128:$src2), x86_subreg_ss))>;
+ (EXTRACT_SUBREG (v4i32 VR128:$src2), sub_ss))>;
def : Pat<(v2i64 (movl VR128:$src1, VR128:$src2)),
(MOVSDrr (v2i64 VR128:$src1),
- (EXTRACT_SUBREG (v2i64 VR128:$src2), x86_subreg_sd))>;
+ (EXTRACT_SUBREG (v2i64 VR128:$src2), sub_sd))>;
// vector_shuffle v1, v2 <4, 5, 2, 3> using movsd
def : Pat<(v4f32 (movlp VR128:$src1, VR128:$src2)),
- (MOVSDrr VR128:$src1, (EXTRACT_SUBREG VR128:$src2, x86_subreg_sd))>,
+ (MOVSDrr VR128:$src1, (EXTRACT_SUBREG VR128:$src2, sub_sd))>,
Requires<[HasSSE2]>;
def : Pat<(v4i32 (movlp VR128:$src1, VR128:$src2)),
- (MOVSDrr VR128:$src1, (EXTRACT_SUBREG VR128:$src2, x86_subreg_sd))>,
+ (MOVSDrr VR128:$src1, (EXTRACT_SUBREG VR128:$src2, sub_sd))>,
Requires<[HasSSE2]>;
}
@@ -3464,14 +3511,14 @@ let Constraints = "$src1 = $dst" in {
let Constraints = "$src1 = $dst" in {
multiclass SS48I_binop_rm<bits<8> opc, string OpcodeStr, SDNode OpNode,
ValueType OpVT, bit Commutable = 0> {
- def rr : SS48I<opc, MRMSrcReg, (outs VR128:$dst),
+ def rr : SS48I<opc, MRMSrcReg, (outs VR128:$dst),
(ins VR128:$src1, VR128:$src2),
!strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
[(set VR128:$dst, (OpVT (OpNode VR128:$src1, VR128:$src2)))]>,
OpSize {
let isCommutable = Commutable;
}
- def rm : SS48I<opc, MRMSrcMem, (outs VR128:$dst),
+ def rm : SS48I<opc, MRMSrcMem, (outs VR128:$dst),
(ins VR128:$src1, i128mem:$src2),
!strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
[(set VR128:$dst, (OpNode VR128:$src1,
@@ -3949,15 +3996,15 @@ let Defs = [EFLAGS], Uses = [EAX, EDX], usesCustomInserter = 1 in {
def PCMPESTRM128REG : SS42AI<0, Pseudo, (outs VR128:$dst),
(ins VR128:$src1, VR128:$src3, i8imm:$src5),
"#PCMPESTRM128rr PSEUDO!",
- [(set VR128:$dst,
- (int_x86_sse42_pcmpestrm128
+ [(set VR128:$dst,
+ (int_x86_sse42_pcmpestrm128
VR128:$src1, EAX, VR128:$src3, EDX, imm:$src5))]>, OpSize;
def PCMPESTRM128MEM : SS42AI<0, Pseudo, (outs VR128:$dst),
(ins VR128:$src1, i128mem:$src3, i8imm:$src5),
"#PCMPESTRM128rm PSEUDO!",
- [(set VR128:$dst, (int_x86_sse42_pcmpestrm128
- VR128:$src1, EAX, (load addr:$src3), EDX, imm:$src5))]>,
+ [(set VR128:$dst, (int_x86_sse42_pcmpestrm128
+ VR128:$src1, EAX, (load addr:$src3), EDX, imm:$src5))]>,
OpSize;
}
@@ -3972,7 +4019,7 @@ def PCMPESTRM128rm : SS42AI<0x60, MRMSrcMem, (outs),
let Defs = [ECX, EFLAGS] in {
multiclass SS42AI_pcmpistri<Intrinsic IntId128> {
- def rr : SS42AI<0x63, MRMSrcReg, (outs),
+ def rr : SS42AI<0x63, MRMSrcReg, (outs),
(ins VR128:$src1, VR128:$src2, i8imm:$src3),
"pcmpistri\t{$src3, $src2, $src1|$src1, $src2, $src3}",
[(set ECX, (IntId128 VR128:$src1, VR128:$src2, imm:$src3)),
@@ -4003,7 +4050,7 @@ let Uses = [EAX, EDX] in {
def rm : SS42AI<0x61, MRMSrcMem, (outs),
(ins VR128:$src1, i128mem:$src3, i8imm:$src5),
"pcmpestri\t{$src5, $src3, $src1|$src1, $src3, $src5}",
- [(set ECX,
+ [(set ECX,
(IntId128 VR128:$src1, EAX, (load addr:$src3), EDX, imm:$src5)),
(implicit EFLAGS)]>, OpSize;
}
@@ -4081,16 +4128,15 @@ def AESIMCrm : AES8I<0xDB, MRMSrcMem, (outs VR128:$dst),
OpSize;
def AESKEYGENASSIST128rr : AESAI<0xDF, MRMSrcReg, (outs VR128:$dst),
- (ins VR128:$src1, i32i8imm:$src2),
+ (ins VR128:$src1, i8imm:$src2),
"aeskeygenassist\t{$src2, $src1, $dst|$dst, $src1, $src2}",
[(set VR128:$dst,
(int_x86_aesni_aeskeygenassist VR128:$src1, imm:$src2))]>,
OpSize;
def AESKEYGENASSIST128rm : AESAI<0xDF, MRMSrcMem, (outs VR128:$dst),
- (ins i128mem:$src1, i32i8imm:$src2),
+ (ins i128mem:$src1, i8imm:$src2),
"aeskeygenassist\t{$src2, $src1, $dst|$dst, $src1, $src2}",
[(set VR128:$dst,
(int_x86_aesni_aeskeygenassist (bitconvert (memopv2i64 addr:$src1)),
imm:$src2))]>,
OpSize;
-
generated by cgit v1.2.3 (git 2.25.1) at 2025-04-13 14:51:30 +0000