[llvm-branch-commits] [llvm-branch] r252480 - Merging r246900:
Tom Stellard via llvm-branch-commits
llvm-branch-commits at lists.llvm.org
Mon Nov 9 08:25:09 PST 2015
Author: tstellar
Date: Mon Nov 9 10:25:09 2015
New Revision: 252480
URL: http://llvm.org/viewvc/llvm-project?rev=252480&view=rev
Log:
Merging r246900:
------------------------------------------------------------------------
r246900 | hfinkel | 2015-09-04 20:02:59 -0400 (Fri, 04 Sep 2015) | 14 lines
[PowerPC] Fix and(or(x, c1), c2) -> rlwimi generation
PPCISelDAGToDAG has a transformation that generates a rlwimi instruction from
an input pattern that looks like this:
and(or(x, c1), c2)
but the associated logic does not work if there are bits that are 1 in c1 but 0
in c2 (these are normally canonicalized away, but that can't happen if the 'or'
has other users. Make sure we abort the transformation if such bits are
discovered.
Fixes PR24704.
------------------------------------------------------------------------
Added:
llvm/branches/release_37/test/CodeGen/PowerPC/rlwimi-and-or-bits.ll
Modified:
llvm/branches/release_37/lib/Target/PowerPC/PPCISelDAGToDAG.cpp
Modified: llvm/branches/release_37/lib/Target/PowerPC/PPCISelDAGToDAG.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/branches/release_37/lib/Target/PowerPC/PPCISelDAGToDAG.cpp?rev=252480&r1=252479&r2=252480&view=diff
==============================================================================
--- llvm/branches/release_37/lib/Target/PowerPC/PPCISelDAGToDAG.cpp (original)
+++ llvm/branches/release_37/lib/Target/PowerPC/PPCISelDAGToDAG.cpp Mon Nov 9 10:25:09 2015
@@ -2570,13 +2570,25 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *
return nullptr;
}
// ISD::OR doesn't get all the bitfield insertion fun.
- // (and (or x, c1), c2) where isRunOfOnes(~(c1^c2)) is a bitfield insert
+ // (and (or x, c1), c2) where isRunOfOnes(~(c1^c2)) might be a
+ // bitfield insert.
if (isInt32Immediate(N->getOperand(1), Imm) &&
N->getOperand(0).getOpcode() == ISD::OR &&
isInt32Immediate(N->getOperand(0).getOperand(1), Imm2)) {
+ // The idea here is to check whether this is equivalent to:
+ // (c1 & m) | (x & ~m)
+ // where m is a run-of-ones mask. The logic here is that, for each bit in
+ // c1 and c2:
+ // - if both are 1, then the output will be 1.
+ // - if both are 0, then the output will be 0.
+ // - if the bit in c1 is 0, and the bit in c2 is 1, then the output will
+ // come from x.
+ // - if the bit in c1 is 1, and the bit in c2 is 0, then the output will
+ // be 0.
+ // If that last condition is never the case, then we can form m from the
+ // bits that are the same between c1 and c2.
unsigned MB, ME;
- Imm = ~(Imm^Imm2);
- if (isRunOfOnes(Imm, MB, ME)) {
+ if (isRunOfOnes(~(Imm^Imm2), MB, ME) && !(~Imm & Imm2)) {
SDValue Ops[] = { N->getOperand(0).getOperand(0),
N->getOperand(0).getOperand(1),
getI32Imm(0, dl), getI32Imm(MB, dl),
Added: llvm/branches/release_37/test/CodeGen/PowerPC/rlwimi-and-or-bits.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/branches/release_37/test/CodeGen/PowerPC/rlwimi-and-or-bits.ll?rev=252480&view=auto
==============================================================================
--- llvm/branches/release_37/test/CodeGen/PowerPC/rlwimi-and-or-bits.ll (added)
+++ llvm/branches/release_37/test/CodeGen/PowerPC/rlwimi-and-or-bits.ll Mon Nov 9 10:25:09 2015
@@ -0,0 +1,27 @@
+; RUN: llc < %s | FileCheck %s
+target datalayout = "E-m:e-i64:64-n32:64"
+target triple = "powerpc64-unknown-linux-gnu"
+
+ at m = external global i32, align 4
+
+; Function Attrs: nounwind
+define signext i32 @main() #0 {
+entry:
+
+; CHECK-LABEL: @main
+; CHECK-NOT: rlwimi
+; CHECK: andi
+
+ %0 = load i32, i32* @m, align 4
+ %or = or i32 %0, 250
+ store i32 %or, i32* @m, align 4
+ %and = and i32 %or, 249
+ %sub.i = sub i32 %and, 0
+ %sext = shl i32 %sub.i, 24
+ %conv = ashr exact i32 %sext, 24
+ ret i32 %conv
+}
+
+attributes #0 = { nounwind "target-cpu"="pwr7" }
+attributes #1 = { nounwind }
+
More information about the llvm-branch-commits
mailing list