Re: [PATCH] powerpc/boot: Use address-of operator on section symbols
From: Geert Uytterhoeven
Date: Mon Aug 03 2020 - 07:32:43 EST
Hi Michael,
On Mon, Aug 3, 2020 at 1:09 PM Michael Ellerman <mpe@xxxxxxxxxxxxxx> wrote:
> Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> writes:
> > On Mon, Jul 20, 2020 at 11:03 PM Segher Boessenkool
> > <segher@xxxxxxxxxxxxxxxxxxx> wrote:
> >> On Sat, Jul 18, 2020 at 09:50:50AM +0200, Geert Uytterhoeven wrote:
> >> > On Wed, Jun 24, 2020 at 6:02 AM Nathan Chancellor
> >> > <natechancellor@xxxxxxxxx> wrote:
> >> > > /* If we have an image attached to us, it overrides anything
> >> > > * supplied by the loader. */
> >> > > - if (_initrd_end > _initrd_start) {
> >> > > + if (&_initrd_end > &_initrd_start) {
> >> >
> >> > Are you sure that fix is correct?
> >> >
> >> > extern char _initrd_start[];
> >> > extern char _initrd_end[];
> >> > extern char _esm_blob_start[];
> >> > extern char _esm_blob_end[];
> >> >
> >> > Of course the result of their comparison is a constant, as the addresses
> >> > are constant. If clangs warns about it, perhaps that warning should be moved
> >> > to W=1?
> >> >
> >> > But adding "&" is not correct, according to C.
> >>
> >> Why not?
> >>
> >> 6.5.3.2/3
> >> The unary & operator yields the address of its operand. [...]
> >> Otherwise, the result is a pointer to the object or function designated
> >> by its operand.
> >>
> >> This is the same as using the name of an array without anything else,
> >> yes. It is a bit clearer if it would not be declared as array, perhaps,
> >> but it is correct just fine like this.
> >
> > Thanks, I stand corrected.
> >
> > Regardless, the comparison is still a comparison between two constant
> > addresses, so my fear is that the compiler will start generating
> > warnings for that in the near or distant future, making this change
> > futile.
>
> They're not constant at compile time though. So I don't think the
> compiler could (sensibly) warn about that? (surely!)
They're constant, but the compiler doesn't know their value.
That doesn't change by (not) using the address-of operator.
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds