On Thu, May 25, 2017 at 04:11:19PM +0200, Jakub Jelinek wrote:
> On Thu, May 25, 2017 at 03:02:42PM +0100, Jonathan Wakely wrote:
> > On 25/05/17 11:07 +0100, Jonathan Wakely wrote:
> > > On 25/05/17 10:05 +0200, Andreas Schwab wrote:
> > > > ../../gcc/ada/gcc-interface/utils2.c: In function 'int
> > > > compare_elmt_bitpos(const void*, const void*)':
> > > > ../../gcc/ada/gcc-interface/utils2.c:1937:73: error: type qualifiers
> > > > ignored on cast result type [-Werror=ignored-qualifiers]
> > > > const constructor_elt * const elmt1 = (const constructor_elt * const)
> > > > rt1;
> > > >
> > > > ^~~
> > > > ../../gcc/ada/gcc-interface/utils2.c:1938:73: error: type qualifiers
> > > > ignored on cast result type [-Werror=ignored-qualifiers]
> > > > const constructor_elt * const elmt2 = (const constructor_elt * const)
> > > > rt2;
> > >
> > > I'm testing this obvious fix.
> >
> > Committed as r248458 because it gets bootstrap past the error above,
> > although now Ada fails for me with:
> >
> > /home/jwakely/src/gcc/bootstrap/./gcc/xgcc
> > -B/home/jwakely/src/gcc/bootstrap/./gcc/
> > -B/usr/local/x86_64-pc-linux-gnu/bin/ -B/usr/local/x86_64-pc-linux-gnu/lib/
> > -isystem /usr/local/x86_64-pc-linux-gnu/include -isystem
> > /usr/local/x86_64-pc-linux-gnu/sys-include -c -g -O2 -m32 -fpic -W
> > -Wall -gnatpg -nostdinc -m32 s-regpat.adb -o s-regpat.o
> >
> > raised STORAGE_ERROR : stack overflow or erroneous memory access
> > ../gcc-interface/Makefile:296: recipe for target 's-regpat.o' failed
> >
> >
> > > diff --git a/gcc/ada/gcc-interface/utils2.c
> > > b/gcc/ada/gcc-interface/utils2.c
> > > index fc6f1b8..cd37791 100644
> > > --- a/gcc/ada/gcc-interface/utils2.c
> > > +++ b/gcc/ada/gcc-interface/utils2.c
> > > @@ -1934,8 +1934,8 @@ build_call_raise_range (int msg, Node_Id gnat_node,
> > > char kind,
> > > static int
> > > compare_elmt_bitpos (const PTR rt1, const PTR rt2)
> > > {
> > > - const constructor_elt * const elmt1 = (const constructor_elt * const)
> > > rt1;
> > > - const constructor_elt * const elmt2 = (const constructor_elt * const)
> > > rt2;
> > > + const constructor_elt * const elmt1 = (const constructor_elt *) rt1;
> > > + const constructor_elt * const elmt2 = (const constructor_elt *) rt2;
> > > const_tree const field1 = elmt1->index;
> > > const_tree const field2 = elmt2->index;
> > > const int ret
>
> So, what can one do with typeof or similar to avoid the warning?
>
> void
> foo (const void *p)
> {
> const int *const q = (const int *const) p;
> typeof (q) r = (typeof (q)) p;
> (void) q;
> (void) r;
> }
>
> AFAIK typeof doesn't strip the toplevel qualifiers and I see current trunk
> warns even about the cast in r initialization. Similarly to what has been
> noted recently in another (C) PR, it would be nice if we had toplevel cv
> stripping variant of typeof or some special builtin that could wrap
> typeof or some type and could be used in places where typeof can,
> __strip_cv (typeof (q)) = (__strip_cv (typeof (q))) p;
> or
> typeof (q) = (__strip_cv (typeof (q))) p;
> or
> __strip_cv (const int *const) z;
> where the last one would be effectively
> const int *z;
I remember trying to implement the stripping version of __typeof; I even had
a prototype patch that I'm of course not finding right now, but I'd be happy to
work on this again.
Ok, I at least found the PR:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65455
and
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=39985
Also there's
https://gcc.gnu.org/ml/gcc-patches/2016-02/msg00268.html
Let me know if I should get back to it.
Marek
