Bug#362422: FTBFS with GCC 4.2: the address of 'hwparams' will always evaluate as 'true'

[Ben Elliston]

> Hmm.  This is a complex one.  The addresses of local variables are
> always non-NULL, as required by the C/C++ standards (and common
> sense).  And its often the case explicitly testing this in code
> indicates a potential logic error, "assert (&hwparams)" instead of
> "assert (hwparams)" for example.  However, in this particular case,
> because of the use of a macro, it does appear to be an annoying
> false negative.

There are always the potential for false negatives like this. :-(
However, the warning has already proved its worth -- it would have
caught the recent X11 security hole!

Ben