egcs segfaults on rh51 (example)

[Jeffrey A Law]

  In message < 9808012100.aa04098@poker.informatik.uni-kl.de >you write:
  > the following programm segfaults with the -O6 optimisation.
  > The strange thing is: if you uncomment : //pow(6.0,3.0);
  > it even wotks with -O6.
  > 
  > Can someone tell me if its a bug or I'm doing something wrong?
  > Please reply by mail, because im not on the mailing list.
  > 
  > I'm using RH 5.1. 
There is a known code generation bug in the compiler which exhibits
itself when using the inlined pow function from certain versions of
glibc on the x86.

This bug has been fixed for the upcoming egcs-1.1 release.
jeff