Re: [Tinycc-devel] Broken commit e460f7dbb216

[Vincent Lefevre]

On 2022-07-30 19:48:29 +0200, grischka wrote:
> On 30.07.2022 12:45, Vincent Lefevre wrote:
> > I changed the & to && because it is more correct and makes more sense,
> > but note that this is not really redundant because n is signed and
> > n & (n - 1) is not portable when n = 0 or 1 if two's complement isn't
> > necessarily used, because 0 may have several representations (well,
> > tcc is not designed to work on such exotic platforms, but who knows
> > in the future... and moreover, because of this general portability
> > issue, compilers may issue warnings).
> 
> Sorry, what?  Are you saying that (n & (n - 1)) with n == 1 -> (1 & 0)
> and with n == 0 -> (0 & -1) does not evaluate to 0 in both cases always
> necessarily, in C?

Yes, if the 0 happens to be a negative 0, 1 & 0 will give 1
in ones' complement, and 0 & -1 will also give non-zero in
both ones' complement and "sign and magnitude".

ISO C17 says:

  It is unspecified whether these cases actually generate a
  negative zero or a normal zero, and whether a negative zero
  becomes a normal zero when stored in an object.

> > > As to the "n < 0 ||" clause by the way. the C-standards seem to say:
> > > 
> > >    "Alignments are represented as values of the type size_t. Valid 
> > > alignments
> > >     include only fundamental alignments, plus an additional 
> > > implementation-
> > >     defined set of values, which may be empty.  Every valid alignment 
> > > value
> > >     shall be a nonnegative integral power of two."
> > > 
> > > Well, when "size_t" means unsigned, how could it be not "nonnegative"
> > > then ?!?
> > 
> > I don't understand what you mean here. The fact is that in the code,
> > n is of type int, not size_t. So the case n < 0 needs to be taken
> > into account. Note that n comes from
> > 
> >                    n = expr_const();
> > 
> > thus may be negative.
> 
> Seems I have misunderstood what they are trying to say.  It seems that by
>    "nonnegative integral power of two"
> they mean any powers 0..n of two, as in
>     2 ^ 0..n
> as opposed to
>     2 ^ -1.2
> for example.
> 
> That hasn't to do much with "int n = expr_const();" in tcc though because
> the 'n' there does not mean "2 ^ n", just as _Alignas(1) does not mean
> to align at 2 ^ 1.  It means to align at 2 ^ 0 = 1.

The point is that _Alignas(-2) is invalid.

> Therfor the combination
> 
>     if (n < 0 || ...
>         tcc_error("alignment must be a positive power of two");
> 
> still seems misleading.  "_Alignas(0x80000000) int x;" is not an
> invalid declaration, per se.

Well, this depends on the implementation. If supported by tcc,
this is fine. But this should be documented, as this is an
implementation-defined feature.

-- 
Vincent Lefèvre <vincent@vinc17.net> - Web: <https://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)