Re: fyi: gengtype patches checked in

[Paolo Bonzini <paolo dot bonzini at lu dot unisi dot ch>]

Steve Ellcey wrote:
> gengtype is seg faulting on IA64 HP-UX.  It looks like oprintf in
> gcc/gengtype.c is calling vsnprintf when o->buf is NULL and expecting it
> to return zero and then fix things up later.  On IA64 HP-UX the call to
> vsnprintf is giving a seg fault when o->buf is NULL.
> 
> I tried putting that chunk of code in side "if (o->buf)" and
> initializing slength to zero but now I hit the "gcc_assert (slen2 ==
> slength)" and wind up stopping there.

You just have to create the buffer with a fixed length if it is
NULL, like this (untested):

  size_t slength;
  va_list ap;

  va_start (ap, format);
  if (!o->buf)
    {
      o->buf = XNEWVEC (char, 1024);
      o->buflength = 1024;
    }

  /* Try first with the assumption that there is enough space.  */
  {
    va_list ap;
    va_start (ap, format);
    slength = vsnprintf (o->buf + o->bufused, o->buflength - o->bufused,
			 format, ap);
    va_end (ap);
  }
 
  if (o->bufused + slength >= o->buflength)
    {
      /* There wasn't enough space.  */
      size_t new_len = o->buflength;
      do
        new_len *= 2;
      while (o->bufused + slength >= new_len);
      o->buf = XRESIZEVEC (char, o->buf, new_len);
      o->buflength = new_len;

      /* We now know that there is enough space. */
      {
	size_t slen2;
	va_list ap;
	va_start (ap, format);
	slen2 = vsnprintf (o->buf + o->bufused, o->buflength - o->bufused,
			   format, ap);
	va_end (ap);

	gcc_assert (slen2 == slength);
	gcc_assert (o->bufused + slen2 < o->buflength);
      }
    }

  o->bufused += slength;