Add support for scalarizing/splitting vector bswap.

Summary:
  SLP Vectorization of intrinsics (r203707) has exposed cases where the
  expansion of vector bswap is failing (PR19151).

Reviewers: hfinkel

CC: chandlerc

Differential Revision: http://llvm-reviews.chandlerc.com/D3104

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204163 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Raul E. Silvera 2014-03-18 17:49:12 +00:00
parent e1ea4faca1
commit 370981ad17
2 changed files with 21 additions and 0 deletions

View File

@ -65,6 +65,7 @@ void DAGTypeLegalizer::ScalarizeVectorResult(SDNode *N, unsigned ResNo) {
case ISD::UNDEF: R = ScalarizeVecRes_UNDEF(N); break; case ISD::UNDEF: R = ScalarizeVecRes_UNDEF(N); break;
case ISD::VECTOR_SHUFFLE: R = ScalarizeVecRes_VECTOR_SHUFFLE(N); break; case ISD::VECTOR_SHUFFLE: R = ScalarizeVecRes_VECTOR_SHUFFLE(N); break;
case ISD::ANY_EXTEND: case ISD::ANY_EXTEND:
case ISD::BSWAP:
case ISD::CTLZ: case ISD::CTLZ:
case ISD::CTPOP: case ISD::CTPOP:
case ISD::CTTZ: case ISD::CTTZ:
@ -533,6 +534,7 @@ void DAGTypeLegalizer::SplitVectorResult(SDNode *N, unsigned ResNo) {
SplitVecRes_VECTOR_SHUFFLE(cast<ShuffleVectorSDNode>(N), Lo, Hi); SplitVecRes_VECTOR_SHUFFLE(cast<ShuffleVectorSDNode>(N), Lo, Hi);
break; break;
case ISD::BSWAP:
case ISD::CONVERT_RNDSAT: case ISD::CONVERT_RNDSAT:
case ISD::CTLZ: case ISD::CTLZ:
case ISD::CTTZ: case ISD::CTTZ:

View File

@ -22,7 +22,26 @@ entry:
ret i64 %or.7 ret i64 %or.7
} }
define <4 x i32> @bswapv4i32(<4 x i32> %x) nounwind readnone {
entry:
; MIPS32-LABEL: bswapv4i32:
; MIPS32: wsbh $[[R0:[0-9]+]]
; MIPS32: rotr ${{[0-9]+}}, $[[R0]], 16
; MIPS32: wsbh $[[R0:[0-9]+]]
; MIPS32: rotr ${{[0-9]+}}, $[[R0]], 16
; MIPS32: wsbh $[[R0:[0-9]+]]
; MIPS32: rotr ${{[0-9]+}}, $[[R0]], 16
; MIPS32: wsbh $[[R0:[0-9]+]]
; MIPS32: rotr ${{[0-9]+}}, $[[R0]], 16
; mips16: .ent bswapv4i32
%ret = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %x)
ret <4 x i32> %ret
}
declare i32 @llvm.bswap.i32(i32) nounwind readnone declare i32 @llvm.bswap.i32(i32) nounwind readnone
declare i64 @llvm.bswap.i64(i64) nounwind readnone declare i64 @llvm.bswap.i64(i64) nounwind readnone
declare <4 x i32> @llvm.bswap.v4i32(<4 x i32>) nounwind readnone