[llvm] r220933 - Fix incorrect invariant check in DAG Combine

[Hal Finkel]

----- Original Message -----
> From: "Louis Gerbarg" <lgg at apple.com>
> To: llvm-commits at cs.uiuc.edu
> Sent: Thursday, October 30, 2014 5:21:04 PM
> Subject: [llvm] r220933 - Fix incorrect invariant check in DAG Combine
> 
> Author: louis
> Date: Thu Oct 30 17:21:03 2014
> New Revision: 220933
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=220933&view=rev
> Log:
> Fix incorrect invariant check in DAG Combine
> 
> Earlier this summer I fixed an issue where we were incorrectly
> combining
> multiple loads that had different constraints such alignment,
> invariance,
> temporality, etc. Apparently in one case I made copt paste error and
> swapped
> alignment and invariance.
> 
> Tests included.
> 
> rdar://18816719
> 
> Added:
>     llvm/trunk/test/CodeGen/AArch64/dag-combine-invaraints.ll
> Modified:
>     llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
> 
> Modified: llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
> URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp?rev=220933&r1=220932&r2=220933&view=diff
> ==============================================================================
> --- llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp (original)
> +++ llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp Thu Oct 30
> 17:21:03 2014
> @@ -11542,7 +11542,7 @@ bool DAGCombiner::SimplifySelectOps(SDNo
>      // It is safe to replace the two loads if they have different
>      alignments,
>      // but the new load must be the minimum (most restrictive)
>      alignment of the
>      // inputs.
> -    bool isInvariant = LLD->getAlignment() & RLD->getAlignment();
> +    bool isInvariant = LLD->isInvariant() & RLD->isInvariant();
>      unsigned Alignment = std::min(LLD->getAlignment(),
>      RLD->getAlignment());
>      if (LLD->getExtensionType() == ISD::NON_EXTLOAD) {
>        Load = DAG.getLoad(TheSelect->getValueType(0),
> 
> Added: llvm/trunk/test/CodeGen/AArch64/dag-combine-invaraints.ll
> URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/AArch64/dag-combine-invaraints.ll?rev=220933&view=auto
> ==============================================================================
> --- llvm/trunk/test/CodeGen/AArch64/dag-combine-invaraints.ll (added)
> +++ llvm/trunk/test/CodeGen/AArch64/dag-combine-invaraints.ll Thu Oct
> 30 17:21:03 2014
> @@ -0,0 +1,36 @@
> +; RUN: llc -mtriple=arm64-apple-darwin8.0 -relocation-model=pic -O1
> < %s | FileCheck %s
> +
> + at .str2 = private unnamed_addr constant [9 x i8] c"_%d____\0A\00",
> align 1
> +
> +; Function Attrs: nounwind ssp
> +define i32 @main(i32 %argc, i8** %argv) #0 {
> +main_:
> +  %tmp = alloca i32, align 4
> +  %i32T = alloca i32, align 4
> +  %i32F = alloca i32, align 4
> +  %i32X = alloca i32, align 4
> +  store i32 0, i32* %tmp
> +  store i32 15, i32* %i32T, align 4
> +  store i32 5, i32* %i32F, align 4
> +  %tmp6 = load i32* %tmp, align 4
> +  %tmp7 = icmp ne i32 %tmp6, 0
> +  %tmp8 = xor i1 %tmp7, true
> +  %tmp9 = load i32* %i32T, align 4
> +  %tmp10 = load i32* %i32F, align 4
> +  %DHSelect = select i1 %tmp8, i32 %tmp9, i32 %tmp10
> +  store i32 %DHSelect, i32* %i32X, align 4
> +  %tmp15 = load i32* %i32X, align 4
> +  %tmp17 = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds
> ([9 x i8]* @.str2, i32 0, i32 0), i32 %tmp15)
> +  ret i32 0
> +
> +; CHECK: main:
> +; CHECK-DAG: movz
> +; CHECK-DAG: orr
> +; CHECK: csel
> +}
> +
> +
> +declare i32 @printf(i8*, ...) #1
> +
> +attributes #0 = { nounwind ssp "less-precise-fpmad"="false"
> "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf"
> "no-infs-fp-math"="false" "no-nans-fp-math"="false"
> "stack-protector-buffer-size"="8" "unsafe-fp-math"="false"
> "use-soft-float"="false" }
> +attributes #1 = { "less-precise-fpmad"="false"
> "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf"
> "no-infs-fp-math"="false" "no-nans-fp-math"="false"
> "stack-protector-buffer-size"="8" "unsafe-fp-math"="false"
> "use-soft-float"="false" }

Please don't include unnecessary attributes in regression tests.

Thanks again,
Hal

> 
> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
> 

-- 
Hal Finkel
Assistant Computational Scientist
Leadership Computing Facility
Argonne National Laboratory