6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2025-08-09 11:25:55 +00:00
Code Issues Projects Releases Wiki Activity

Legalizer: Use the scalar bit width when promoting bit counting instrs on

Browse Source 
vectors.

e.g. when promoting ctlz from <2 x i32> to <2 x i64> we have to fixup
the result by 32 bits, not 64. PR20917.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@217671 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Benjamin Kramer
2014-09-12 12:50:27 +00:00
parent 4a93e8dd02
commit 66feb63c3c
2 changed files with 56 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp
View File

@@ -342,9 +342,10 @@ SDValue DAGTypeLegalizer::PromoteIntRes_CTLZ(SDNode *N) {
EVT NVT = Op.getValueType(); EVT NVT = Op.getValueType();
Op = DAG.getNode(N->getOpcode(), dl, NVT, Op); Op = DAG.getNode(N->getOpcode(), dl, NVT, Op);
// Subtract off the extra leading bits in the bigger type. // Subtract off the extra leading bits in the bigger type.
return DAG.getNode(ISD::SUB, dl, NVT, Op, return DAG.getNode(
DAG.getConstant(NVT.getSizeInBits() - ISD::SUB, dl, NVT, Op,
OVT.getSizeInBits(), NVT)); DAG.getConstant(NVT.getScalarSizeInBits() - OVT.getScalarSizeInBits(),
NVT));
} }
SDValue DAGTypeLegalizer::PromoteIntRes_CTPOP(SDNode *N) { SDValue DAGTypeLegalizer::PromoteIntRes_CTPOP(SDNode *N) {
@@ -362,8 +363,8 @@ SDValue DAGTypeLegalizer::PromoteIntRes_CTTZ(SDNode *N) {
// The count is the same in the promoted type except if the original // The count is the same in the promoted type except if the original
// value was zero. This can be handled by setting the bit just off // value was zero. This can be handled by setting the bit just off
// the top of the original type. // the top of the original type.
APInt TopBit(NVT.getSizeInBits(), 0); auto TopBit = APInt::getOneBitSet(NVT.getScalarSizeInBits(),
TopBit.setBit(OVT.getSizeInBits()); OVT.getScalarSizeInBits());
Op = DAG.getNode(ISD::OR, dl, NVT, Op, DAG.getConstant(TopBit, NVT)); Op = DAG.getNode(ISD::OR, dl, NVT, Op, DAG.getConstant(TopBit, NVT));
} }
return DAG.getNode(N->getOpcode(), dl, NVT, Op); return DAG.getNode(N->getOpcode(), dl, NVT, Op);

51
test/CodeGen/X86/vec_ctbits.ll
View File

@@ -1,4 +1,4 @@
; RUN: llc < %s -march=x86-64 ; RUN: llc < %s -march=x86-64 -mattr=+sse2 | FileCheck %s
declare <2 x i64> @llvm.cttz.v2i64(<2 x i64>, i1) declare <2 x i64> @llvm.cttz.v2i64(<2 x i64>, i1)
declare <2 x i64> @llvm.ctlz.v2i64(<2 x i64>, i1) declare <2 x i64> @llvm.ctlz.v2i64(<2 x i64>, i1)
@@ -7,12 +7,61 @@ declare <2 x i64> @llvm.ctpop.v2i64(<2 x i64>)
define <2 x i64> @footz(<2 x i64> %a) nounwind { define <2 x i64> @footz(<2 x i64> %a) nounwind {
%c = call <2 x i64> @llvm.cttz.v2i64(<2 x i64> %a, i1 true) %c = call <2 x i64> @llvm.cttz.v2i64(<2 x i64> %a, i1 true)
ret <2 x i64> %c ret <2 x i64> %c
; CHECK-LABEL: footz
; CHECK: bsfq
; CHECK: bsfq
} }
define <2 x i64> @foolz(<2 x i64> %a) nounwind { define <2 x i64> @foolz(<2 x i64> %a) nounwind {
%c = call <2 x i64> @llvm.ctlz.v2i64(<2 x i64> %a, i1 true) %c = call <2 x i64> @llvm.ctlz.v2i64(<2 x i64> %a, i1 true)
ret <2 x i64> %c ret <2 x i64> %c
; CHECK-LABEL: foolz
; CHECK: bsrq
; CHECK: xorq $63
; CHECK: bsrq
; CHECK: xorq $63
} }
define <2 x i64> @foopop(<2 x i64> %a) nounwind { define <2 x i64> @foopop(<2 x i64> %a) nounwind {
%c = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> %a) %c = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> %a)
ret <2 x i64> %c ret <2 x i64> %c
} }
declare <2 x i32> @llvm.cttz.v2i32(<2 x i32>, i1)
declare <2 x i32> @llvm.ctlz.v2i32(<2 x i32>, i1)
declare <2 x i32> @llvm.ctpop.v2i32(<2 x i32>)
define <2 x i32> @promtz(<2 x i32> %a) nounwind {
%c = call <2 x i32> @llvm.cttz.v2i32(<2 x i32> %a, i1 false)
ret <2 x i32> %c
; CHECK: .quad 4294967296
; CHECK: .quad 4294967296
; CHECK-LABEL: promtz
; CHECK: bsfq
; CHECK: cmov
; CHECK: bsfq
; CHECK: cmov
}
define <2 x i32> @promlz(<2 x i32> %a) nounwind {
%c = call <2 x i32> @llvm.ctlz.v2i32(<2 x i32> %a, i1 false)
ret <2 x i32> %c
; CHECK: .quad 4294967295
; CHECK: .quad 4294967295
; CHECK: .quad 32
; CHECK: .quad 32
; CHECK-LABEL: promlz
; CHECK: pand
; CHECK: bsrq
; CHECK: xorq $63
; CHECK: bsrq
; CHECK: xorq $63
; CHECK: psub
}
define <2 x i32> @prompop(<2 x i32> %a) nounwind {
%c = call <2 x i32> @llvm.ctpop.v2i32(<2 x i32> %a)
ret <2 x i32> %c
}