egcs-970901, sparc-sun-solaris2.5 problem with sys_siglist ...

[Joe Buck]

> 	BTW, solaris2.5 does have strsignal() which could be checked for
> and used instead.  In fact, we should write all gcc code in terms of
> strsignal() (whether or not it exists,) and provide a replacement if it
> is absent.  Then if the replacement was necessary, we'd only have to get
> it working in one place.

This is basically the libiberty approach.  Now that gcc is being packaged
with libiberty, why not use it?  [ Or does this create a circular
dependency? ]