Re: [PATCH] configure: Improve alias attribute check

[Gavin Shan]

Hi Paolo and Richard,

On 3/21/21 9:33 AM, Richard Henderson wrote:
> On 3/20/21 11:52 AM, Paolo Bonzini wrote:
> > > +int main(void)
> > > +{
> > > +    return read_y();
> > > +}
> >
> > I think this should be "read_y() == 1 ? 0 : 1".
>
> As a testcase returning 0 on success, yes.

Ok. I will include the changes in v2. Also, I will
wrap the lines, for example:

int main(void) { return (read_y() == 1) ? 0 : 1; }

if compile_prog "" "" && $TMPE >/dev/null 2>/dev/null; then
   attralias=yes
fi

> > I can reproduce it with -flto -O2 but not without -flto, do you agree?
>
> Agreed.  Replicated with a random recent gcc 11 snapshot.
> This is really annoying of lto.  It's clear something needs to change though.

The command I used is:

gcc -O2 -flto=auto config-temp.c config-temp-b.c -o config-temp.exe.

Removing "-O2" or "-flto=auto" can make the gcc alias attribute workable again.

> > Perhaps we can obtain the same optimization by wrapping reads of the page size 
> > in an inline __attribute__((const)) function.  Richard, what do you think?
>
> I'll give it a shot and see what happens.

Thanks,
Gavin