[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: perl_5.32.0~rc1-1 FTBFS on m68k (experimental)



Hi Adrian,

On Fri, Jun 19, 2020 at 10:20 AM John Paul Adrian Glaubitz
<glaubitz@physik.fu-berlin.de> wrote:
> > On Jun 19, 2020, at 10:02 AM, Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> >> On Thu, Jun 18, 2020 at 10:46 PM John Paul Adrian Glaubitz
> >> <glaubitz@physik.fu-berlin.de> wrote:
> >>> On 6/18/20 12:10 PM, John Paul Adrian Glaubitz wrote:
> >>> Reported upstream [1].
> >>
> >> It's an alignment issue and can be trivially fixed with this patch:
> >>
> >> diff --git a/op.h b/op.h
> >> index fc21f03cda..480c95245b 100644
> >> --- a/op.h
> >> +++ b/op.h
> >> @@ -698,7 +698,7 @@ struct opslot {
> >>     U16         opslot_size;        /* size of this slot (in pointers) */
> >>     U16         opslot_offset;      /* offset from start of slab (in ptr units) */
> >>     OP         opslot_op;              /* the op itself */
> >> -};
> >> +} __attribute__ ((aligned (4)));
> >>
> >> struct opslab {
> >>     OPSLAB *   opslab_next;            /* next slab */
> >
> > In the mean time, you changed this to add explicit padding instead:
> >
> > https://github.com/Perl/perl5/issues/17871
> >
> >> diff --git a/op.h b/op.h
> >> index fc21f03cda..fb9f538e23 100644
> >> --- a/op.h
> >> +++ b/op.h
> >> @@ -714,6 +714,7 @@ struct opslab {
> >> # ifdef PERL_DEBUG_READONLY_OPS
> >>     bool       opslab_readonly;
> >> # endif
> >> +    U16         opslab_padding;                /* padding to ensure proper alignment */
> >>     OPSLOT     opslab_slots;           /* slots begin here */
> >> };
> >
> > I take it PERL_DEBUG_READONLY_OPS is enabled?
> > Hence the padding should be moved inside the #ifdef,
> > Furthermore, sizeof(bool) = 1, right? So you still have an implicit
> > hole, and it would be better to add 3 bytes of explicit padding
> > instead one 16-bit quantity.
>
> No, I didn’t take that #ifdef into consideration.
>
> And I’m confused now.
>
> Add the three bytes how? Inside the #ifdef makes no sense as that wouldn’t fix release builds.

So release builds fail, too?
And they don't set PERL_DEBUG_READONLY_OPS?

If PERL_DEBUG_READONLY_OPS is not set, I see no reason why
the padding would be needed.

Disclaimer: looking at the 5.26 version from Ubuntu, as github seems
to be down.

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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


Reply to: