[InstCombine] pow(x, +/- 0.0) -> 1.0

[jfb]

Summary

This isn't an important optimization at all... We're already doing:

pow(x, 0.0) -> 1.0

My patch merely teaches instcombine that -0.0 does the same.

However, doing this fixes an AMAZING bug! Compile this program:

extern "C" double pow(double, double);
double boom(double base) {
  return pow(base, -0.0);
}

With:

clang++ ~/Desktop/fast-math.cpp -ffast-math -O2 -S

And clang will crash with a signal. Wow, fast math is so fast it ICEs the
compiler! Arguably, the generated math is infinitely fast.

What's actually happening is that we recurse infinitely in getPow. In debug we
hit its assertion:

assert(Exp != 0 && "Incorrect exponent 0 not handled");

We avoid this entire mess if we instead recognize that an exponent of positive
and negative zero yield 1.0.

rdar://problem/54598300

[arsenm]

llvm/lib/Transforms/Utils/SimplifyLibCalls.cpp

1566–1567

You can use m_AnyZeroFP

1566–1567

Also gives free vector support

llvm/test/Transforms/InstCombine/pow-0.ll

31

Can you add some vectors?

[jfb]

• Address arsenm's comments

[arsenm]

LGTM

[scanon]

"pow(x, ±0) returns 1 for any x, even a NaN"

LGTM.

[xbolva00]

lg too

[xbolva00]

@spatel fixed it

https://reviews.llvm.org/rG4f0e429acca3b224b7b58fb89899c2de08a2e4c6

[jfb]

In D67248#1661067, @xbolva00 wrote:

@spatel fixed it

https://reviews.llvm.org/rG4f0e429acca3b224b7b58fb89899c2de08a2e4c6

🤷‍♂️

I just committed the added tests.

[spatel]

Sorry - I didn't see this review, only the bug report that came in today:
https://bugs.llvm.org/show_bug.cgi?id=43233

I made an identical code change, but I didn't include as many tests. Feel free to commit those or let me know if you want me to.

[jfb]

In D67248#1661110, @spatel wrote:

Sorry - I didn't see this review, only the bug report that came in today:
https://bugs.llvm.org/show_bug.cgi?id=43233

I made an identical code change, but I didn't include as many tests. Feel free to commit those or let me know if you want me to.

Already committed the tests :)

llvm/lib/Transforms/Utils/SimplifyLibCalls.cpp

if (Value *Exp = replacePowWithExp(Pow, B))
     return Exp;
 
   // Evaluate special cases related to the exponent.
 
   // pow(x, -1.0) -> 1.0 / x
   if (match(Expo, m_SpecificFP(-1.0)))
     return B.CreateFDiv(ConstantFP::get(Ty, 1.0), Base, "reciprocal");
 
-  // pow(x, 0.0) -> 1.0
-  if (match(Expo, m_SpecificFP(0.0)))
+  // pow(x, +/- 0.0) -> 1.0
+  if (match(Expo, m_AnyZeroFP()))
     return ConstantFP::get(Ty, 1.0);
 
   // pow(x, 1.0) -> x
   if (match(Expo, m_FPOne()))
     return Base;
 
   // pow(x, 2.0) -> x * x
   if (match(Expo, m_SpecificFP(2.0)))
     return B.CreateFMul(Base, Base, "square");

llvm/test/Transforms/InstCombine/pow-0.ll

+; RUN: opt < %s -instcombine -S | FileCheck %s
+
+; CHECK-LABEL: @zero(
+; CHECK-NEXT:  ret double 1.000000e+00
+define double @zero(double %value) {
+  %res = call double @llvm.pow.f64(double %value, double 0.000000e+00)
+  ret double %res
+}
+
+; CHECK-LABEL: @minus_zero(
+; CHECK-NEXT:  ret double 1.000000e+00
+define double @minus_zero(double %value) {
+  %res = call double @llvm.pow.f64(double %value, double -0.000000e+00)
+  ret double %res
+}
+
+; CHECK-LABEL: @fast_zero(
+; CHECK-NEXT:  ret double 1.000000e+00
+define double @fast_zero(double %value) {
+  %res = call fast double @llvm.pow.f64(double %value, double 0.000000e+00)
+  ret double %res
+}
+
+; CHECK-LABEL: @fast_minus_zero(
+; CHECK-NEXT:  ret double 1.000000e+00
+define double @fast_minus_zero(double %value) {
+  %res = call fast double @llvm.pow.f64(double %value, double -0.000000e+00)
+  ret double %res
+}
+
+; CHECK-LABEL: @vec_zero(
+; CHECK-NEXT:  ret <2 x double> <double 1.000000e+00, double 1.000000e+00>
+define <2 x double> @vec_zero(<2 x double> %value) {
+  %res = call <2 x double> @llvm.pow.v2f64(<2 x double> %value, <2 x double> <double 0.000000e+00, double 0.000000e+00>)
+  ret <2 x double> %res
+}
+
+; CHECK-LABEL: @vec_minus_zero(
+; CHECK-NEXT:  ret <2 x double> <double 1.000000e+00, double 1.000000e+00>
+define <2 x double> @vec_minus_zero(<2 x double> %value) {
+  %res = call <2 x double> @llvm.pow.v2f64(<2 x double> %value, <2 x double> <double -0.000000e+00, double -0.000000e+00>)
+  ret <2 x double> %res
+}
+
+; CHECK-LABEL: @vec_fast_zero(
+; CHECK-NEXT:  ret <2 x double> <double 1.000000e+00, double 1.000000e+00>
+define <2 x double> @vec_fast_zero(<2 x double> %value) {
+  %res = call fast <2 x double> @llvm.pow.v2f64(<2 x double> %value, <2 x double> <double 0.000000e+00, double 0.000000e+00>)
+  ret <2 x double> %res
+}
+
+; CHECK-LABEL: @vec_fast_minus_zero(
+; CHECK-NEXT:  ret <2 x double> <double 1.000000e+00, double 1.000000e+00>
+define <2 x double> @vec_fast_minus_zero(<2 x double> %value) {
+  %res = call fast <2 x double> @llvm.pow.v2f64(<2 x double> %value, <2 x double> <double -0.000000e+00, double -0.000000e+00>)
+  ret <2 x double> %res
+}
+
+declare double @llvm.pow.f64(double, double)
+declare <2 x double> @llvm.pow.v2f64(<2 x double>, <2 x double>)