[CLANG_X] use of infinity is undefined behavior due to the currently enabled floating-point options

[iarspider]

Splitting off from cms-sw/cmsdist#9200 for visibility

In CLANG_X IBs, a lot of Wnan-infinity-disabled warnings are generated, e.g.

  src/DataFormats/EgammaReco/interface/ElectronSeed.h:115:46: warning: use of infinity is undefined behavior due to the currently enabled floating-point options [-Wnan-infinity-disabled]
   115 |     void setNegAttributes(const float dRZ2 = std::numeric_limits<float>::infinity(),
      |                                              ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  src/DataFormats/EgammaReco/interface/ElectronSeed.h:116:47: warning: use of infinity is undefined behavior due to the currently enabled floating-point options [-Wnan-infinity-disabled]
   116 |                           const float dPhi2 = std::numeric_limits<float>::infinity(),
      |                                               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  src/DataFormats/EgammaReco/interface/ElectronSeed.h:117:46: warning: use of infinity is undefined behavior due to the currently enabled floating-point options [-Wnan-infinity-disabled]
   117 |                           const float dRZ1 = std::numeric_limits<float>::infinity(),
      |                                              ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  src/DataFormats/EgammaReco/interface/ElectronSeed.h:118:47: warning: use of infinity is undefined behavior due to the currently enabled floating-point options [-Wnan-infinity-disabled]
   118 |                           const float dPhi1 = std::numeric_limits<float>::infinity());
      |                                               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  src/DataFormats/EgammaReco/interface/ElectronSeed.h:119:46: warning: use of infinity is undefined behavior due to the currently enabled floating-point options [-Wnan-infinity-disabled]
   119 |     void setPosAttributes(const float dRZ2 = std::numeric_limits<float>::infinity(),
      |                                              ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  src/DataFormats/EgammaReco/interface/ElectronSeed.h:120:47: warning: use of infinity is undefined behavior due to the currently enabled floating-point options [-Wnan-infinity-disabled]
   120 |                           const float dPhi2 = std::numeric_limits<float>::infinity(),
      |                                               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  src/DataFormats/EgammaReco/interface/ElectronSeed.h:121:46: warning: use of infinity is undefined behavior due to the currently enabled floating-point options [-Wnan-infinity-disabled]
   121 |                           const float dRZ1 = std::numeric_limits<float>::infinity(),
      |                                              ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  src/DataFormats/EgammaReco/interface/ElectronSeed.h:122:47: warning: use of infinity is undefined behavior due to the currently enabled floating-point options [-Wnan-infinity-disabled]
   122 |                           const float dPhi1 = std::numeric_limits<float>::infinity());
      |                                               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  src/DataFormats/EgammaReco/interface/ElectronSeed.h:143:63: warning: use of infinity is undefined behavior due to the currently enabled floating-point options [-Wnan-infinity-disabled]
   143 |       return hitNr < hitInfo_.size() ? hitInfo_[hitNr].*val : std::numeric_limits<T>::infinity();
      |                                                               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
src/DataFormats/EgammaReco/interface/ElectronSeed.h:102:48: note: in instantiation of function template specialization 'reco::ElectronSeed::getVal<float>' requested here
  102 |     float dPhiNeg(size_t hitNr) const { return getVal(hitNr, &PMVars::dPhiNeg); }
      |                                                ^
  src/DataFormats/EgammaReco/interface/ElectronSeed.h:143:63: warning: use of infinity is undefined behavior due to the currently enabled floating-point options [-Wnan-infinity-disabled]
   143 |       return hitNr < hitInfo_.size() ? hitInfo_[hitNr].*val : std::numeric_limits<T>::infinity();
      |                                                               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
src/DataFormats/EgammaReco/interface/ElectronSeed.h:110:52: note: in instantiation of function template specialization 'reco::ElectronSeed::getVal<int>' requested here
  110 |     int layerOrDiskNr(size_t hitNr) const { return getVal(hitNr, &PMVars::layerOrDiskNr); }
      |                                                    ^

We are building cmssw with -ffast-math, which sets (among other flags) -ffinite-math-only, allowing compiler to assume all floating-point numbers will be finite (no NaNs or infinities), but some functions explicitly return std::numeric_limits<float>::infinity and std::numeric_limits<float>::quiet_NaN.

If I'm reading @makortel's comment right, we check for NaN/infinity ourselves (instead of relying on compiler to do it), so this warning can be disabled.

[iarspider]

assign core

[cmsbuild]

New categories assigned: core

@Dr15Jones,@makortel,@smuzaffar you have been requested to review this Pull request/Issue and eventually sign? Thanks

[cmsbuild]

cms-bot internal usage

[cmsbuild]

A new Issue was created by @iarspider.

@smuzaffar, @rappoccio, @sextonkennedy, @Dr15Jones, @antoniovilela, @makortel can you please review it and eventually sign/assign? Thanks.

cms-bot commands are listed here

[makortel]

If I'm reading @makortel's comment right, we check for NaN/infinity ourselves (instead of relying on compiler to do it), so this warning can be disabled.

I meant we need to use edm::isFinite etc to check for infinity/nan in presence of -ffast-math (although one could also ask if such code relying on IEEE infinity behavior should really be compiled with -ffast-math...).

This warning (and the similar static analyzer check) are very useful to catch the problems.

[VinInn]

Just use std::numeric_limits<float>::max()

[makortel]

+core

[cmsbuild]

This issue is fully signed and ready to be closed.