Skip to content

[RISCV][GISel] Use maskedValueIsZero in RISCVInstructionSelector::selectZExtBits. #115244

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Nov 7, 2024
Merged

[RISCV][GISel] Use maskedValueIsZero in RISCVInstructionSelector::selectZExtBits. #115244

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
a6f5080
[RISCV][GISel] Use maskedValueIsZero in RISCVInstructionSelector::sel…
topperc Nov 7, 2024
a0fb8db
fixup! getScalarSizeInBits.
topperc Nov 7, 2024
eaea0cb
Merge remote-tracking branch 'origin/main' into pr/zext-knownbits
topperc Nov 7, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 3 additions & 1 deletion llvm/lib/Target/RISCV/GISel/RISCVInstructionSelector.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -261,7 +261,9 @@ RISCVInstructionSelector::selectZExtBits(MachineOperand &Root,
return {{[=](MachineInstrBuilder &MIB) { MIB.addReg(RegX); }}};
}

// TODO: Use computeKnownBits.
unsigned Size = MRI->getType(RootReg).getScalarSizeInBits();
if (KB->maskedValueIsZero(RootReg, APInt::getBitsSetFrom(Size, Bits)))
return {{[=](MachineInstrBuilder &MIB) { MIB.add(Root); }}};

return std::nullopt;
}
Expand Down
94 changes: 61 additions & 33 deletions llvm/test/CodeGen/RISCV/GlobalISel/rv32zbkb.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,11 +25,16 @@ define i32 @pack_i32(i32 %a, i32 %b) nounwind {
}

define i32 @pack_i32_2(i16 zeroext %a, i16 zeroext %b) nounwind {
; CHECK-LABEL: pack_i32_2:
; CHECK: # %bb.0:
; CHECK-NEXT: slli a1, a1, 16
; CHECK-NEXT: or a0, a1, a0
; CHECK-NEXT: ret
; RV32I-LABEL: pack_i32_2:
; RV32I: # %bb.0:
; RV32I-NEXT: slli a1, a1, 16
; RV32I-NEXT: or a0, a1, a0
; RV32I-NEXT: ret
;
; RV32ZBKB-LABEL: pack_i32_2:
; RV32ZBKB: # %bb.0:
; RV32ZBKB-NEXT: pack a0, a0, a1
; RV32ZBKB-NEXT: ret
%zexta = zext i16 %a to i32
%zextb = zext i16 %b to i32
%shl1 = shl i32 %zextb, 16
Expand All @@ -38,12 +43,18 @@ define i32 @pack_i32_2(i16 zeroext %a, i16 zeroext %b) nounwind {
}

define i32 @pack_i32_3(i16 zeroext %0, i16 zeroext %1, i32 %2) {
; CHECK-LABEL: pack_i32_3:
; CHECK: # %bb.0:
; CHECK-NEXT: slli a0, a0, 16
; CHECK-NEXT: or a0, a0, a1
; CHECK-NEXT: add a0, a0, a2
; CHECK-NEXT: ret
; RV32I-LABEL: pack_i32_3:
; RV32I: # %bb.0:
; RV32I-NEXT: slli a0, a0, 16
; RV32I-NEXT: or a0, a0, a1
; RV32I-NEXT: add a0, a0, a2
; RV32I-NEXT: ret
;
; RV32ZBKB-LABEL: pack_i32_3:
; RV32ZBKB: # %bb.0:
; RV32ZBKB-NEXT: pack a0, a1, a0
; RV32ZBKB-NEXT: add a0, a0, a2
; RV32ZBKB-NEXT: ret
%4 = zext i16 %0 to i32
%5 = shl nuw i32 %4, 16
%6 = zext i16 %1 to i32
Expand Down Expand Up @@ -165,9 +176,8 @@ define i64 @packh_i64_2(i64 %a, i64 %b) nounwind {
; RV32ZBKB-LABEL: packh_i64_2:
; RV32ZBKB: # %bb.0:
; RV32ZBKB-NEXT: andi a1, a2, 255
; RV32ZBKB-NEXT: slli a3, zero, 8
; RV32ZBKB-NEXT: srli a1, a1, 24
; RV32ZBKB-NEXT: or a1, a3, a1
; RV32ZBKB-NEXT: packh a1, a1, zero
; RV32ZBKB-NEXT: packh a0, a0, a2
; RV32ZBKB-NEXT: ret
%and = and i64 %a, 255
Expand All @@ -179,11 +189,16 @@ define i64 @packh_i64_2(i64 %a, i64 %b) nounwind {


define zeroext i16 @packh_i16(i8 zeroext %a, i8 zeroext %b) nounwind {
; CHECK-LABEL: packh_i16:
; CHECK: # %bb.0:
; CHECK-NEXT: slli a1, a1, 8
; CHECK-NEXT: or a0, a1, a0
; CHECK-NEXT: ret
; RV32I-LABEL: packh_i16:
; RV32I: # %bb.0:
; RV32I-NEXT: slli a1, a1, 8
; RV32I-NEXT: or a0, a1, a0
; RV32I-NEXT: ret
;
; RV32ZBKB-LABEL: packh_i16:
; RV32ZBKB: # %bb.0:
; RV32ZBKB-NEXT: packh a0, a0, a1
; RV32ZBKB-NEXT: ret
%zext = zext i8 %a to i16
%zext1 = zext i8 %b to i16
%shl = shl i16 %zext1, 8
Expand All @@ -193,13 +208,19 @@ define zeroext i16 @packh_i16(i8 zeroext %a, i8 zeroext %b) nounwind {


define zeroext i16 @packh_i16_2(i8 zeroext %0, i8 zeroext %1, i8 zeroext %2) {
; CHECK-LABEL: packh_i16_2:
; CHECK: # %bb.0:
; CHECK-NEXT: add a0, a1, a0
; CHECK-NEXT: andi a0, a0, 255
; CHECK-NEXT: slli a0, a0, 8
; CHECK-NEXT: or a0, a0, a2
; CHECK-NEXT: ret
; RV32I-LABEL: packh_i16_2:
; RV32I: # %bb.0:
; RV32I-NEXT: add a0, a1, a0
; RV32I-NEXT: andi a0, a0, 255
; RV32I-NEXT: slli a0, a0, 8
; RV32I-NEXT: or a0, a0, a2
; RV32I-NEXT: ret
;
; RV32ZBKB-LABEL: packh_i16_2:
; RV32ZBKB: # %bb.0:
; RV32ZBKB-NEXT: add a0, a1, a0
; RV32ZBKB-NEXT: packh a0, a2, a0
; RV32ZBKB-NEXT: ret
%4 = add i8 %1, %0
%5 = zext i8 %4 to i16
%6 = shl i16 %5, 8
Expand All @@ -209,14 +230,21 @@ define zeroext i16 @packh_i16_2(i8 zeroext %0, i8 zeroext %1, i8 zeroext %2) {
}

define void @packh_i16_3(i8 zeroext %0, i8 zeroext %1, i8 zeroext %2, ptr %p) {
; CHECK-LABEL: packh_i16_3:
; CHECK: # %bb.0:
; CHECK-NEXT: add a0, a1, a0
; CHECK-NEXT: andi a0, a0, 255
; CHECK-NEXT: slli a0, a0, 8
; CHECK-NEXT: or a0, a0, a2
; CHECK-NEXT: sh a0, 0(a3)
; CHECK-NEXT: ret
; RV32I-LABEL: packh_i16_3:
; RV32I: # %bb.0:
; RV32I-NEXT: add a0, a1, a0
; RV32I-NEXT: andi a0, a0, 255
; RV32I-NEXT: slli a0, a0, 8
; RV32I-NEXT: or a0, a0, a2
; RV32I-NEXT: sh a0, 0(a3)
; RV32I-NEXT: ret
;
; RV32ZBKB-LABEL: packh_i16_3:
; RV32ZBKB: # %bb.0:
; RV32ZBKB-NEXT: add a0, a1, a0
; RV32ZBKB-NEXT: packh a0, a2, a0
; RV32ZBKB-NEXT: sh a0, 0(a3)
; RV32ZBKB-NEXT: ret
%4 = add i8 %1, %0
%5 = zext i8 %4 to i16
%6 = shl i16 %5, 8
Expand Down
3 changes: 1 addition & 2 deletions llvm/test/CodeGen/RISCV/GlobalISel/rv64zbkb.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -130,8 +130,7 @@ define i64 @pack_i64_3(ptr %0, ptr %1) {
; RV64ZBKB: # %bb.0:
; RV64ZBKB-NEXT: lwu a0, 0(a0)
; RV64ZBKB-NEXT: lwu a1, 0(a1)
; RV64ZBKB-NEXT: slli a0, a0, 32
; RV64ZBKB-NEXT: or a0, a0, a1
; RV64ZBKB-NEXT: pack a0, a1, a0
; RV64ZBKB-NEXT: ret
%3 = load i32, ptr %0, align 4
%4 = zext i32 %3 to i64
Expand Down
Loading