Re: Using strlcpy if target OS allows it

[Joe Buck <Joe dot Buck at synopsys dot COM>]

On Thu, Apr 21, 2005 at 02:31:28AM +0200, Marc Espie wrote:
> On Wed, Apr 20, 2005 at 06:18:31PM -0500, Aaron W. LaFramboise wrote:
> > Marc Espie wrote:
> > 
> > > The linker now warns about uses of strcpy, which means that the whole
> > > libstdc++ testsuite fails on OpenBSD now.
> > 
> > If you're just trying to fix the testsuite failures, perhaps you could
> > use the gcc-dg-prune callback to ignore this warning?
> > 
> 
> As far as the testsuite goes, yep.
> In general, having the library warn is a major annoyance...
> well, it's fixed on my box now by sprinkling strlcpy all over the place,
> but having yet another patch to maintain internally is stuff I could live
> without...

There appear to be only four calls of strcpy in the library itself
(as opposed to the testsuite).  All four take this exact form:

      char* __sav = new char[std::strlen(__old) + 1];
      std::strcpy(__sav, __old);

Perhaps the V3 maintainers could be persuaded to accept a patch that
replaces all four with a call to

inline char* __cstring_duplicate(const char* __old) {
  unsigned __size = std::strlen(__old) + 1;
  char* __sav = new char[__size];
  return std::strcpy(__sav, __old);
}

This would have maintainance advantages, as four identical code sequences
become one.

Then for OpenBSD you only need a simple patch:

 -  return std::strcpy(__sav, __old);
 +  strlcpy(__sav, __old, __size);
 +  return __sav;