On Thu, Aug 18, 2016 at 9:39 AM, Jakub Jelinek <ja...@redhat.com> wrote:
> Hi!
>
> While looking if something doesn't need to be updated for the
> i386-builtin.def addition (I think we are ok, this file doesn't need to be
> shipped for plugins, as it is only included in i386.c, the first time
> i386.c is built no matter what and next time we'll have it listed in
> .deps/i386.Po), I've noticed IMHO incorrect dependency.
> i386-builtin-types.inc is included only by i386.c, not by i386-c.c, and
> while the second time it will be listed in .deps/i386.Po, I think if
> somebody is unlucky enough that i386.o is compiled before i386-c.o, then
> i386-builtin-types.inc might not be compiled yet.
>
> As an experiment,
> rm -f i386{,-c}.o .deps/i386{,-c}.Po s-i386-bt i386-builtin-types.inc
> make i386.o
> fails to compile without this patch. i386-c.o apparently comes before
> i386.o in dependencies of cc1, so even with make -j1 it is built before
> that. If I add sleep 4s into the s-i386-bt: rules before move-if-change
> then even make -j16 build fails after the above rm, so it is all about good
> timing.
>
> Ok for trunk if testing passes?
Uh, yes. This is also OK for backports.
Thanks,
Uros.
> 2016-08-18 Jakub Jelinek <ja...@redhat.com>
>
> * config/i386/t-i386 (i386-c.o): Don't depend on
> i386-builtin-types.inc.
> (i386.o): Depend on i386-builtin-types.inc.
>
> --- gcc/config/i386/t-i386.jj 2016-01-04 14:55:56.000000000 +0100
> +++ gcc/config/i386/t-i386 2016-08-18 09:23:28.523058356 +0200
> @@ -19,10 +19,12 @@
> OPTIONS_H_EXTRA += $(srcdir)/config/i386/stringop.def
> TM_H += $(srcdir)/config/i386/x86-tune.def
>
> -i386-c.o: $(srcdir)/config/i386/i386-c.c i386-builtin-types.inc
> +i386-c.o: $(srcdir)/config/i386/i386-c.c
> $(COMPILE) $<
> $(POSTCOMPILE)
>
> +i386.o: i386-builtin-types.inc
> +
> i386-builtin-types.inc: s-i386-bt ; @true
> s-i386-bt: $(srcdir)/config/i386/i386-builtin-types.awk \
> $(srcdir)/config/i386/i386-builtin-types.def
>
> Jakub
