From 52bda948465582df2d7bb1b0ba71758a7ff2b1e4 Mon Sep 17 00:00:00 2001
From: David Majnemer <david.majnemer@gmail.com>
Date: Wed, 6 Apr 2016 07:04:53 +0000
Subject: [PATCH] [SLPVectorizer] Vectorizing the libm sqrt to llvm's sqrt
 intrinsic requires nnan

To quote the langref "Unlike sqrt in libm, however, llvm.sqrt has
undefined behavior for negative numbers other than -0.0 (which allows
for better optimization, because there is no need to worry about errno
being set). llvm.sqrt(-0.0) is defined to return -0.0 like IEEE sqrt."

This means that it's unsafe to replace sqrt with llvm.sqrt unless the
call is annotated with nnan.

Thanks to Hal Finkel for pointing this out!

llvm-svn: 265521
---
 lib/Analysis/VectorUtils.cpp                  |  4 ++-
 .../LoopVectorize/X86/veclib-calls.ll         | 26 ++++++++++++++++++-
 test/Transforms/SLPVectorizer/X86/call.ll     |  4 +--
 3 files changed, 30 insertions(+), 4 deletions(-)

diff --git a/lib/Analysis/VectorUtils.cpp b/lib/Analysis/VectorUtils.cpp
index aa5bb7a610c..4a6cdf973e2 100644
--- a/lib/Analysis/VectorUtils.cpp
+++ b/lib/Analysis/VectorUtils.cpp
@@ -223,7 +223,9 @@ Intrinsic::ID llvm::getIntrinsicIDForCall(CallInst *CI,
   case LibFunc::sqrt:
   case LibFunc::sqrtf:
   case LibFunc::sqrtl:
-    return checkUnaryFloatSignature(*CI, Intrinsic::sqrt);
+    if (CI->hasNoNaNs())
+      return checkUnaryFloatSignature(*CI, Intrinsic::sqrt);
+    return Intrinsic::not_intrinsic;
   }
 
   return Intrinsic::not_intrinsic;
diff --git a/test/Transforms/LoopVectorize/X86/veclib-calls.ll b/test/Transforms/LoopVectorize/X86/veclib-calls.ll
index e7e36850f4e..6f8f5223ce4 100644
--- a/test/Transforms/LoopVectorize/X86/veclib-calls.ll
+++ b/test/Transforms/LoopVectorize/X86/veclib-calls.ll
@@ -3,6 +3,31 @@
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
 target triple = "x86_64-unknown-linux-gnu"
 
+;CHECK-LABEL: @sqrt_f32(
+;CHECK: vsqrtf{{.*}}<4 x float>
+;CHECK: ret void
+declare float @sqrtf(float) nounwind readnone
+define void @sqrt_f32(i32 %n, float* noalias %y, float* noalias %x) nounwind uwtable {
+entry:
+  %cmp6 = icmp sgt i32 %n, 0
+  br i1 %cmp6, label %for.body, label %for.end
+
+for.body:                                         ; preds = %entry, %for.body
+  %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
+  %arrayidx = getelementptr inbounds float, float* %y, i64 %indvars.iv
+  %0 = load float, float* %arrayidx, align 4
+  %call = tail call float @sqrtf(float %0) nounwind readnone
+  %arrayidx2 = getelementptr inbounds float, float* %x, i64 %indvars.iv
+  store float %call, float* %arrayidx2, align 4
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+  %exitcond = icmp eq i32 %lftr.wideiv, %n
+  br i1 %exitcond, label %for.end, label %for.body
+
+for.end:                                          ; preds = %for.body, %entry
+  ret void
+}
+
 ;CHECK-LABEL: @exp_f32(
 ;CHECK: vexpf{{.*}}<4 x float>
 ;CHECK: ret void
@@ -135,7 +160,6 @@ for.end:                                          ; preds = %for.body, %entry
 ;CHECK-LABEL: @sqrt_f32_nobuiltin(
 ;CHECK-NOT: vsqrtf{{.*}}<4 x float>
 ;CHECK: ret void
-declare float @sqrtf(float) nounwind readnone
 define void @sqrt_f32_nobuiltin(i32 %n, float* noalias %y, float* noalias %x) nounwind uwtable {
 entry:
   %cmp6 = icmp sgt i32 %n, 0
diff --git a/test/Transforms/SLPVectorizer/X86/call.ll b/test/Transforms/SLPVectorizer/X86/call.ll
index 79f01106dc9..d6c0ebd6b07 100644
--- a/test/Transforms/SLPVectorizer/X86/call.ll
+++ b/test/Transforms/SLPVectorizer/X86/call.ll
@@ -105,13 +105,13 @@ entry:
   %i0 = load double, double* %a, align 8
   %i1 = load double, double* %b, align 8
   %mul = fmul double %i0, %i1
-  %call = tail call double @sqrt(double %mul) nounwind readnone
+  %call = tail call nnan double @sqrt(double %mul) nounwind readnone
   %arrayidx3 = getelementptr inbounds double, double* %a, i64 1
   %i3 = load double, double* %arrayidx3, align 8
   %arrayidx4 = getelementptr inbounds double, double* %b, i64 1
   %i4 = load double, double* %arrayidx4, align 8
   %mul5 = fmul double %i3, %i4
-  %call5 = tail call double @sqrt(double %mul5) nounwind readnone
+  %call5 = tail call nnan double @sqrt(double %mul5) nounwind readnone
   store double %call, double* %c, align 8
   %arrayidx5 = getelementptr inbounds double, double* %c, i64 1
   store double %call5, double* %arrayidx5, align 8