[RISCV] Add a PseudoVSETVLIX0X0 pseudo for the x0,x0 vsetvli. NFC #141875
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Strengthen the register class on PseudoVSETVLIX0 to disallow X0 as a destination. This allows removal of an opcode check from RISCVDeadRegisterDefinitions. Now the register class will prevent the conversion to X0. I'm considering removing the explicit X0 from the operands and adding them during PseudoExpansion, but it complicates finding the vtype operand.
|
@llvm/pr-subscribers-backend-risc-v Author: Craig Topper (topperc) ChangesStrengthen the register class on PseudoVSETVLIX0 to disallow X0 as a destination. This allows removal of an opcode check from RISCVDeadRegisterDefinitions. Now the register class will prevent the conversion to X0. I'm considering removing the explicit X0 operands and adding them during PseudoExpansion, but it complicates finding the vtype operand. Patch is 27.91 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/141875.diff 9 Files Affected:
diff --git a/llvm/lib/Target/RISCV/RISCVDeadRegisterDefinitions.cpp b/llvm/lib/Target/RISCV/RISCVDeadRegisterDefinitions.cpp
index 47510ec6b0ea8..e842b07255af2 100644
--- a/llvm/lib/Target/RISCV/RISCVDeadRegisterDefinitions.cpp
+++ b/llvm/lib/Target/RISCV/RISCVDeadRegisterDefinitions.cpp
@@ -74,9 +74,6 @@ bool RISCVDeadRegisterDefinitions::runOnMachineFunction(MachineFunction &MF) {
MI.getOpcode() != RISCV::PseudoVSETVLI &&
MI.getOpcode() != RISCV::PseudoVSETIVLI)
continue;
- // For PseudoVSETVLIX0, Rd = X0 has special meaning.
- if (MI.getOpcode() == RISCV::PseudoVSETVLIX0)
- continue;
for (int I = 0, E = Desc.getNumDefs(); I != E; ++I) {
MachineOperand &MO = MI.getOperand(I);
if (!MO.isReg() || !MO.isDef() || MO.isEarlyClobber())
diff --git a/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp b/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp
index 8fe8dfabee297..0deeae5bbb817 100644
--- a/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp
+++ b/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp
@@ -1056,7 +1056,7 @@ void RISCVInsertVSETVLI::insertVSETVLI(MachineBasicBlock &MBB,
// Use X0, X0 form if the AVL is the same and the SEW+LMUL gives the same
// VLMAX.
if (Info.hasSameAVL(PrevInfo) && Info.hasSameVLMAX(PrevInfo)) {
- auto MI = BuildMI(MBB, InsertPt, DL, TII->get(RISCV::PseudoVSETVLIX0))
+ auto MI = BuildMI(MBB, InsertPt, DL, TII->get(RISCV::PseudoVSETVLIX0X0))
.addReg(RISCV::X0, RegState::Define | RegState::Dead)
.addReg(RISCV::X0, RegState::Kill)
.addImm(Info.encodeVTYPE())
@@ -1074,11 +1074,12 @@ void RISCVInsertVSETVLI::insertVSETVLI(MachineBasicBlock &MBB,
DefMI && RISCVInstrInfo::isVectorConfigInstr(*DefMI)) {
VSETVLIInfo DefInfo = getInfoForVSETVLI(*DefMI);
if (DefInfo.hasSameAVL(PrevInfo) && DefInfo.hasSameVLMAX(PrevInfo)) {
- auto MI = BuildMI(MBB, InsertPt, DL, TII->get(RISCV::PseudoVSETVLIX0))
- .addReg(RISCV::X0, RegState::Define | RegState::Dead)
- .addReg(RISCV::X0, RegState::Kill)
- .addImm(Info.encodeVTYPE())
- .addReg(RISCV::VL, RegState::Implicit);
+ auto MI =
+ BuildMI(MBB, InsertPt, DL, TII->get(RISCV::PseudoVSETVLIX0X0))
+ .addReg(RISCV::X0, RegState::Define | RegState::Dead)
+ .addReg(RISCV::X0, RegState::Kill)
+ .addImm(Info.encodeVTYPE())
+ .addReg(RISCV::VL, RegState::Implicit);
if (LIS)
LIS->InsertMachineInstrInMaps(*MI);
return;
@@ -1098,7 +1099,7 @@ void RISCVInsertVSETVLI::insertVSETVLI(MachineBasicBlock &MBB,
}
if (Info.hasAVLVLMAX()) {
- Register DestReg = MRI->createVirtualRegister(&RISCV::GPRRegClass);
+ Register DestReg = MRI->createVirtualRegister(&RISCV::GPRNoX0RegClass);
auto MI = BuildMI(MBB, InsertPt, DL, TII->get(RISCV::PseudoVSETVLIX0))
.addReg(DestReg, RegState::Define | RegState::Dead)
.addReg(RISCV::X0, RegState::Kill)
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfo.cpp b/llvm/lib/Target/RISCV/RISCVInstrInfo.cpp
index 818dedac08dde..840300acb2737 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfo.cpp
@@ -275,6 +275,7 @@ static bool isConvertibleToVMV_V_V(const RISCVSubtarget &STI,
if (MBBI->getOpcode() == RISCV::PseudoVSETVLI ||
MBBI->getOpcode() == RISCV::PseudoVSETVLIX0 ||
+ MBBI->getOpcode() == RISCV::PseudoVSETVLIX0X0 ||
MBBI->getOpcode() == RISCV::PseudoVSETIVLI) {
// There is a vsetvli between COPY and source define instruction.
// vy = def_vop ... (producing instruction)
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td b/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
index e35c98e95705e..281f8d55932b9 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
@@ -6056,7 +6056,7 @@ let hasSideEffects = 0, mayLoad = 0, mayStore = 0, isCodeGenOnly = 1 in {
PseudoInstExpansion<(CSRRS GPR:$rd, SysRegVLENB.Encoding, X0)>,
Sched<[WriteRdVLENB]>;
let Defs = [VL, VTYPE] in {
- def PseudoReadVLENBViaVSETVLIX0 : Pseudo<(outs GPR:$rd), (ins uimm5:$shamt),
+ def PseudoReadVLENBViaVSETVLIX0 : Pseudo<(outs GPRNoX0:$rd), (ins uimm5:$shamt),
[]>,
Sched<[WriteVSETVLI, ReadVSETVLI]>;
}
@@ -6096,9 +6096,12 @@ let hasSideEffects = 0, mayLoad = 0, mayStore = 0, Defs = [VL, VTYPE] in {
def PseudoVSETVLI : Pseudo<(outs GPR:$rd), (ins GPRNoX0:$rs1, VTypeIOp11:$vtypei), []>,
PseudoInstExpansion<(VSETVLI GPR:$rd, GPR:$rs1, VTypeIOp11:$vtypei)>,
Sched<[WriteVSETVLI, ReadVSETVLI]>;
-def PseudoVSETVLIX0 : Pseudo<(outs GPR:$rd), (ins GPRX0:$rs1, VTypeIOp11:$vtypei), []>,
+def PseudoVSETVLIX0 : Pseudo<(outs GPRNoX0:$rd), (ins GPRX0:$rs1, VTypeIOp11:$vtypei), []>,
PseudoInstExpansion<(VSETVLI GPR:$rd, GPR:$rs1, VTypeIOp11:$vtypei)>,
Sched<[WriteVSETVLI, ReadVSETVLI]>;
+def PseudoVSETVLIX0X0 : Pseudo<(outs GPRX0:$rd), (ins GPRX0:$rs1, VTypeIOp11:$vtypei), []>,
+ PseudoInstExpansion<(VSETVLI GPR:$rd, GPR:$rs1, VTypeIOp11:$vtypei)>,
+ Sched<[WriteVSETVLI, ReadVSETVLI]>;
def PseudoVSETIVLI : Pseudo<(outs GPR:$rd), (ins uimm5:$rs1, VTypeIOp10:$vtypei), []>,
PseudoInstExpansion<(VSETIVLI GPR:$rd, uimm5:$rs1, VTypeIOp10:$vtypei)>,
Sched<[WriteVSETIVLI]>;
diff --git a/llvm/lib/Target/RISCV/RISCVInstrPredicates.td b/llvm/lib/Target/RISCV/RISCVInstrPredicates.td
index ca193559cc7f6..59025bfe4efb1 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrPredicates.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrPredicates.td
@@ -70,6 +70,7 @@ def isVectorConfigInstr
CheckOpcode<[
PseudoVSETVLI,
PseudoVSETVLIX0,
+ PseudoVSETVLIX0X0,
PseudoVSETIVLI
]>>>;
@@ -79,8 +80,7 @@ def isVLPreservingConfig
: TIIPredicate<"isVLPreservingConfig",
MCReturnStatement<
CheckAll<[
- CheckOpcode<[PseudoVSETVLIX0]>,
- CheckRegOperand<0, X0>
+ CheckOpcode<[PseudoVSETVLIX0X0]>
]>>>;
def isFloatScalarMoveOrScalarSplatInstr
diff --git a/llvm/test/CodeGen/RISCV/rvv/vmv-copy.mir b/llvm/test/CodeGen/RISCV/rvv/vmv-copy.mir
index 8a489bd7598b4..2786f58826b91 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vmv-copy.mir
+++ b/llvm/test/CodeGen/RISCV/rvv/vmv-copy.mir
@@ -98,14 +98,14 @@ body: |
; CHECK-NEXT: $v28m4 = PseudoVLE32_V_M4 undef $v28m4, killed $x16, $noreg, 5 /* e32 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
; CHECK-NEXT: $x15 = PseudoVSETVLI $x17, 73 /* e16, m2, ta, mu */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: $v0m2 = PseudoVLE32_V_M2 undef $v0m2, $x18, $noreg, 4 /* e16 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
- ; CHECK-NEXT: $x0 = PseudoVSETVLIX0 $x0, 82 /* e32, m4, ta, mu */, implicit-def $vl, implicit-def $vtype
+ ; CHECK-NEXT: $x0 = PseudoVSETVLIX0X0 $x0, 82 /* e32, m4, ta, mu */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: $v4m4 = PseudoVLE32_V_M4 undef $v4m4, killed $x18, $noreg, 5 /* e32 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
; CHECK-NEXT: $v12m4 = VMV4R_V $v28m4, implicit $vtype
$x15 = PseudoVSETVLI $x14, 82, implicit-def $vl, implicit-def $vtype
$v28m4 = PseudoVLE32_V_M4 undef $v28m4, killed $x16, $noreg, 5, 0, implicit $vl, implicit $vtype
$x15 = PseudoVSETVLI $x17, 73, implicit-def $vl, implicit-def $vtype
$v0m2 = PseudoVLE32_V_M2 undef $v0m2, $x18, $noreg, 4, 0, implicit $vl, implicit $vtype
- $x0 = PseudoVSETVLIX0 $x0, 82, implicit-def $vl, implicit-def $vtype
+ $x0 = PseudoVSETVLIX0X0 $x0, 82, implicit-def $vl, implicit-def $vtype
$v4m4 = PseudoVLE32_V_M4 undef $v4m4, killed $x18, $noreg, 5, 0, implicit $vl, implicit $vtype
$v12m4 = COPY $v28m4
...
@@ -120,16 +120,16 @@ body: |
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $x15 = PseudoVSETVLI $x14, 82 /* e32, m4, ta, mu */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: $v28m4 = PseudoVLE32_V_M4 undef $v28m4, killed $x16, $noreg, 5 /* e32 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
- ; CHECK-NEXT: $x0 = PseudoVSETVLIX0 $x0, 73 /* e16, m2, ta, mu */, implicit-def $vl, implicit-def $vtype
+ ; CHECK-NEXT: $x0 = PseudoVSETVLIX0X0 $x0, 73 /* e16, m2, ta, mu */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: $v0m2 = PseudoVLE32_V_M2 undef $v0m2, $x18, $noreg, 4 /* e16 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
- ; CHECK-NEXT: $x0 = PseudoVSETVLIX0 $x0, 82 /* e32, m4, ta, mu */, implicit-def $vl, implicit-def $vtype
+ ; CHECK-NEXT: $x0 = PseudoVSETVLIX0X0 $x0, 82 /* e32, m4, ta, mu */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: $v4m4 = PseudoVLE32_V_M4 undef $v4m4, killed $x18, $noreg, 5 /* e32 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
; CHECK-NEXT: $v12m4 = PseudoVMV_V_V_M4 undef $v12m4, $v28m4, $noreg, 5 /* e32 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
$x15 = PseudoVSETVLI $x14, 82, implicit-def $vl, implicit-def $vtype
$v28m4 = PseudoVLE32_V_M4 undef $v28m4, killed $x16, $noreg, 5, 0, implicit $vl, implicit $vtype
- $x0 = PseudoVSETVLIX0 $x0, 73, implicit-def $vl, implicit-def $vtype
+ $x0 = PseudoVSETVLIX0X0 $x0, 73, implicit-def $vl, implicit-def $vtype
$v0m2 = PseudoVLE32_V_M2 undef $v0m2, $x18, $noreg, 4, 0, implicit $vl, implicit $vtype
- $x0 = PseudoVSETVLIX0 $x0, 82, implicit-def $vl, implicit-def $vtype
+ $x0 = PseudoVSETVLIX0X0 $x0, 82, implicit-def $vl, implicit-def $vtype
$v4m4 = PseudoVLE32_V_M4 undef $v4m4, killed $x18, $noreg, 5, 0, implicit $vl, implicit $vtype
$v12m4 = COPY $v28m4
...
@@ -144,12 +144,12 @@ body: |
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $x15 = PseudoVSETVLI $x14, 82 /* e32, m4, ta, mu */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: $v28m4 = PseudoVLE32_V_M4 undef $v28m4, killed $x16, $noreg, 5 /* e32 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
- ; CHECK-NEXT: $x0 = PseudoVSETVLIX0 $x0, 73 /* e16, m2, ta, mu */, implicit-def $vl, implicit-def $vtype
+ ; CHECK-NEXT: $x0 = PseudoVSETVLIX0X0 $x0, 73 /* e16, m2, ta, mu */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: $v0m2 = PseudoVLE32_V_M2 undef $v0m2, $x18, $noreg, 4 /* e16 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
; CHECK-NEXT: $v12m4 = VMV4R_V $v28m4, implicit $vtype
$x15 = PseudoVSETVLI $x14, 82, implicit-def $vl, implicit-def $vtype
$v28m4 = PseudoVLE32_V_M4 undef $v28m4, killed $x16, $noreg, 5, 0, implicit $vl, implicit $vtype
- $x0 = PseudoVSETVLIX0 $x0, 73, implicit-def $vl, implicit-def $vtype
+ $x0 = PseudoVSETVLIX0X0 $x0, 73, implicit-def $vl, implicit-def $vtype
$v0m2 = PseudoVLE32_V_M2 undef $v0m2, $x18, $noreg, 4, 0, implicit $vl, implicit $vtype
$v12m4 = COPY $v28m4
...
@@ -185,11 +185,11 @@ body: |
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $x15 = PseudoVSETVLI $x14, 82 /* e32, m4, ta, mu */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: $v28m4 = PseudoVLE32_V_M4 undef $v28m4, killed $x16, $noreg, 5 /* e32 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
- ; CHECK-NEXT: $x0 = PseudoVSETVLIX0 $x0, 74 /* e16, m4, ta, mu */, implicit-def $vl, implicit-def $vtype
+ ; CHECK-NEXT: $x0 = PseudoVSETVLIX0X0 $x0, 74 /* e16, m4, ta, mu */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: $v12m4 = VMV4R_V $v28m4, implicit $vtype
$x15 = PseudoVSETVLI $x14, 82, implicit-def $vl, implicit-def $vtype
$v28m4 = PseudoVLE32_V_M4 undef $v28m4, killed $x16, $noreg, 5, 0, implicit $vl, implicit $vtype
- $x0 = PseudoVSETVLIX0 $x0, 74, implicit-def $vl, implicit-def $vtype
+ $x0 = PseudoVSETVLIX0X0 $x0, 74, implicit-def $vl, implicit-def $vtype
$v12m4 = COPY $v28m4
...
---
diff --git a/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-coalesce.mir b/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-coalesce.mir
index 6f052d6377cf6..5e81b832381b6 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-coalesce.mir
+++ b/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-coalesce.mir
@@ -15,7 +15,7 @@ body: |
; CHECK-NEXT: bb.1:
; CHECK-NEXT: successors: %bb.2(0x80000000)
; CHECK-NEXT: {{ $}}
- ; CHECK-NEXT: dead [[PseudoVSETVLIX0_:%[0-9]+]]:gpr = PseudoVSETVLIX0 killed $x0, 209 /* e32, m2, ta, ma */, implicit-def $vl, implicit-def $vtype
+ ; CHECK-NEXT: dead [[PseudoVSETVLIX0_:%[0-9]+]]:gprnox0 = PseudoVSETVLIX0 killed $x0, 209 /* e32, m2, ta, ma */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: renamable $v10m2 = PseudoVMV_V_I_M2 undef renamable $v10m2, 0, -1, 5 /* e32 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.2:
diff --git a/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-crossbb.mir b/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-crossbb.mir
index 449ee44d5dc5e..140875c4b24ad 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-crossbb.mir
+++ b/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-crossbb.mir
@@ -282,14 +282,14 @@ body: |
; CHECK-NEXT: bb.1.if.then:
; CHECK-NEXT: successors: %bb.3(0x80000000)
; CHECK-NEXT: {{ $}}
- ; CHECK-NEXT: dead $x0 = PseudoVSETVLIX0 killed $x0, 216 /* e64, m1, ta, ma */, implicit-def $vl, implicit-def $vtype, implicit $vl
+ ; CHECK-NEXT: dead $x0 = PseudoVSETVLIX0X0 killed $x0, 216 /* e64, m1, ta, ma */, implicit-def $vl, implicit-def $vtype, implicit $vl
; CHECK-NEXT: early-clobber %9:vr = PseudoVZEXT_VF2_M1 undef $noreg, [[PseudoVLE32_V_MF2_]], $noreg, 6 /* e64 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
; CHECK-NEXT: PseudoBR %bb.3
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.2.if.else:
; CHECK-NEXT: successors: %bb.3(0x80000000)
; CHECK-NEXT: {{ $}}
- ; CHECK-NEXT: dead $x0 = PseudoVSETVLIX0 killed $x0, 216 /* e64, m1, ta, ma */, implicit-def $vl, implicit-def $vtype, implicit $vl
+ ; CHECK-NEXT: dead $x0 = PseudoVSETVLIX0X0 killed $x0, 216 /* e64, m1, ta, ma */, implicit-def $vl, implicit-def $vtype, implicit $vl
; CHECK-NEXT: early-clobber %9:vr = PseudoVSEXT_VF2_M1 undef $noreg, [[PseudoVLE32_V_MF2_]], $noreg, 6 /* e64 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3.if.end:
@@ -499,9 +499,9 @@ body: |
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:gpr = COPY $x11
; CHECK-NEXT: [[COPY1:%[0-9]+]]:gpr = COPY $x10
- ; CHECK-NEXT: dead [[PseudoVSETVLIX0_:%[0-9]+]]:gpr = PseudoVSETVLIX0 killed $x0, 223 /* e64, mf2, ta, ma */, implicit-def $vl, implicit-def $vtype
+ ; CHECK-NEXT: dead [[PseudoVSETVLIX0_:%[0-9]+]]:gprnox0 = PseudoVSETVLIX0 killed $x0, 223 /* e64, mf2, ta, ma */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: [[PseudoVID_V_MF2_:%[0-9]+]]:vr = PseudoVID_V_MF2 undef $noreg, -1, 6 /* e64 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
- ; CHECK-NEXT: dead [[PseudoVSETVLIX0_1:%[0-9]+]]:gpr = PseudoVSETVLIX0 killed $x0, 215 /* e32, mf2, ta, ma */, implicit-def $vl, implicit-def $vtype
+ ; CHECK-NEXT: dead [[PseudoVSETVLIX0_1:%[0-9]+]]:gprnox0 = PseudoVSETVLIX0 killed $x0, 215 /* e32, mf2, ta, ma */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: [[PseudoVMV_V_I_MF2_:%[0-9]+]]:vrnov0 = PseudoVMV_V_I_MF2 undef $noreg, 0, -1, 5 /* e32 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.1:
@@ -509,9 +509,9 @@ body: |
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[PseudoVMSEQ_VI_MF2_:%[0-9]+]]:vmv0 = PseudoVMSEQ_VI_MF2 [[PseudoVID_V_MF2_]], 0, -1, 5 /* e32 */, implicit $vl, implicit $vtype
; CHECK-NEXT: $v0 = COPY [[PseudoVMSEQ_VI_MF2_]]
- ; CHECK-NEXT: dead $x0 = PseudoVSETVLIX0 killed $x0, 23 /* e32, mf2, tu, mu */, implicit-def $vl, implicit-def $vtype, implicit $vl
+ ; CHECK-NEXT: dead $x0 = PseudoVSETVLIX0X0 killed $x0, 23 /* e32, mf2, tu, mu */, implicit-def $vl, implicit-def $vtype, implicit $vl
; CHECK-NEXT: [[PseudoVLE32_V_MF2_MASK:%[0-9]+]]:vrnov0 = PseudoVLE32_V_MF2_MASK [[PseudoVMV_V_I_MF2_]], [[COPY]], $v0, -1, 5 /* e32 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
- ; CHECK-NEXT: dead $x0 = PseudoVSETVLIX0 killed $x0, 197 /* e8, mf8, ta, ma */, implicit-def $vl, implicit-def $vtype, implicit $vl
+ ; CHECK-NEXT: dead $x0 = PseudoVSETVLIX0X0 killed $x0, 197 /* e8, mf8, ta, ma */, implicit-def $vl, implicit-def $vtype, implicit $vl
; CHECK-NEXT: [[PseudoVCPOP_M_B64_:%[0-9]+]]:gpr = PseudoVCPOP_M_B64 [[PseudoVMSEQ_VI_MF2_]], -1, 0 /* e8 */, implicit $vl, implicit $vtype
; CHECK-NEXT: [[DEF:%[0-9]+]]:gpr = IMPLICIT_DEF
; CHECK-NEXT: BEQ [[PseudoVCPOP_M_B64_]], $x0, %bb.3
@@ -523,7 +523,7 @@ body: |
; CHECK-NEXT: [[DEF:%[0-9]+]]:gpr = LWU [[COPY1]], 0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3:
- ; CHECK-NEXT: dead $x0 = PseudoVSETVLIX0 killed $x0, 215 /* e32, mf2, ta, ma */, implicit-def $vl, implicit-def $vtype, implicit $vl
+ ; CHECK-NEXT: dead $x0 = PseudoVSETVLIX0X0 killed $x0, 215 /* e32, mf2, ta, ma */, implicit-def $vl, implicit-def $vtype, implicit $vl
; CHECK-NEXT: [[PseudoVADD_VX_MF2_:%[0-9]+]]:vr = nsw PseudoVADD_VX_MF2 undef $noreg, [[PseudoVLE32_V_MF2_MASK]], [[DEF]], -1, 5 /* e32 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
; CHECK-NEXT: $v0 = COPY [[PseudoVADD_VX_MF2_]]
; CHECK-NEXT: PseudoRET implicit $v0
@@ -584,7 +584,7 @@ body: |
; CHECK-NEXT: [[PseudoReadVLENB:%[0-9]+]]:gpr = PseudoReadVLENB
; CHECK-NEXT: [[SRLI:%[0-9]+]]:gpr = SRLI [[PseudoReadVLENB]], 3
; CHECK-NEXT: [[COPY1:%[0-9]+]]:gpr = COPY $x11
- ; CHECK-NEXT: dead [[PseudoVSETVLIX0_:%[0-9]+]]:gpr = PseudoVSETVLIX0 killed $x0, 216 /* e64, m1, ta, ma */, implicit-def $vl, implicit-def $vtype
+ ; CHECK-NEXT: dead [[PseudoVSETVLIX0_:%[0-9]+]]:gprnox0 = PseudoVSETVLIX0 killed $x0, 216 /* e64, m1, ta, ma */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: [[PseudoVID_V_M1_:%[0-9]+]]:vr = PseudoVID_V_M1 undef $noreg, -1, 6 /* e64 */, 0 /* tu, mu */, implicit $vl, implicit $vtype
; CHECK-NEXT: [[COPY2:%[0-9]+]]:gpr = COPY $x0
; CHECK-NEXT: {{ $}}
@@ -651,7 +651,7 @@ body: |
; CHECK-NEXT: [[PseudoReadVLENB:%[0-9]+]]:gpr = PseudoReadVLENB
; CHECK-NEXT: [[SRLI:%[0-9]+]]:gpr = SRLI [[PseudoReadVLENB]], 3
; CHECK-NEXT: [[COPY1:%[0-9]+]]:gpr = COPY $x11
- ; CHECK-NEXT: dead [[PseudoVSETVLIX0_:%[0-9]+]]:gpr = PseudoVSETVLIX0 killed $x0, 216 /* e64, m1, ta, ma */, implicit-def $vl, implicit-def $vtype
+ ; CHECK-NEXT: dead [[PseudoVSETVLIX0_:%[0-9]+]]:gprnox0 = PseudoVSETVLIX0 killed $x0, 216 /* e64, m1, ta, ma */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: [[PseudoVID_V_M1_:%[0-9]+]]:vr = PseudoVID_V_M1 undef $noreg, -1, 6 /* e64 */, 3 /* ta, ma */, implicit $vl, implicit $vtype
; CHECK-NEXT: [[COPY2:%[0-9]+]]:gpr = COPY $x0
; CHECK-NEXT: {{ $}}
@@ -815,7 +815,7 @@ body: |
; CHECK-NEXT: %t3:vr = COPY $v2
; CHECK-NEXT: [[COPY:%[0-9]+]]:vrnov0 = COPY $v3
; CHECK-NEXT: %t5:vrnov0 = COPY $v1
- ; CHECK-NEXT: dead [[PseudoVSETVLIX0_:%[0-9]+]]:gpr = PseudoVSETVLIX0 killed $x0, 216 /* e64, m1, ta, ma */, implicit-def $vl, implicit-def $vtype
+ ; CHECK-NEXT: dead [[PseudoVSETVLIX0_:%[0-9]+]]:gprnox0 = PseudoVSETVLIX0 killed $x0, 216 /* e64, m1, ta, ma */, implicit-def $vl, implicit-def $vtype
; CHECK-NEXT: %t6:vr = PseudoVMSEQ_VI_M1 %t1, 0, -1, 6 /* e64 */, implicit $vl, implicit $vtype
; CHECK-NEXT: PseudoBR %bb.1
; CHECK-NEXT: {{ $}}
@@ -830,12 +830,12 @@ body: |
; CHECK-NEXT: successors: %bb.3(0x80000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $v0 = COPY %mask
- ; CHECK-NEXT: dead $x0 = PseudoVSETVLIX0 killed $x0, 69 /* e8, mf8, ta, mu */, implicit-def $vl, implicit-def $vtype, implicit $vl
+ ; CHECK-NEXT: dead $x0 = PseudoVSETVLIX0X0 k...
[truncated]
|
lukel97
reviewed
May 29, 2025
lukel97
approved these changes
May 29, 2025
svkeerthy
pushed a commit
that referenced
this pull request
May 29, 2025
…41875) Strengthen the register class on PseudoVSETVLIX0 to disallow X0 as a destination. This allows removal of an opcode check from RISCVDeadRegisterDefinitions. Now the register class will prevent the conversion to X0. I'm considering removing the explicit X0 operands and adding them during PseudoExpansion, but it complicates finding the vtype operand.
google-yfyang
pushed a commit
to google-yfyang/llvm-project
that referenced
this pull request
May 29, 2025
…vm#141875) Strengthen the register class on PseudoVSETVLIX0 to disallow X0 as a destination. This allows removal of an opcode check from RISCVDeadRegisterDefinitions. Now the register class will prevent the conversion to X0. I'm considering removing the explicit X0 operands and adding them during PseudoExpansion, but it complicates finding the vtype operand.
sivan-shani
pushed a commit
to sivan-shani/llvm-project
that referenced
this pull request
Jun 3, 2025
…vm#141875) Strengthen the register class on PseudoVSETVLIX0 to disallow X0 as a destination. This allows removal of an opcode check from RISCVDeadRegisterDefinitions. Now the register class will prevent the conversion to X0. I'm considering removing the explicit X0 operands and adding them during PseudoExpansion, but it complicates finding the vtype operand.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Strengthen the register class on PseudoVSETVLIX0 to disallow X0 as a destination. This allows removal of an opcode check from RISCVDeadRegisterDefinitions. Now the register class will prevent the conversion to X0.
I'm considering removing the explicit X0 operands and adding them during PseudoExpansion, but it complicates finding the vtype operand.