Re: [PATCH]: update to df for mainline.

[Bernd Schmidt <bernds_cb1 at t-online dot de>]

Kenneth Zadeck wrote:
> All of the little comments will be done as you ask.
> > > +  /* Kill this register if it is not a subreg store.  */
> > > +  if (!(DF_REF_FLAGS (def) & DF_REF_PARTIAL))
> > > +    bitmap_clear_bit (live, dregno);
> > Looks like this depends on another part.  I have to admit that I'm not
> > convinced yet by the subreg handling in DF as a whole.  A normal
> > (non-multiword non-strict-low) subreg store _does_ kill the whole reg.
> I find this somewhat confusing.  Are there other cases of subregs that
> cover the whole register?

Paradoxical subregs might also count.  As I said, the whole situation is 
a bit of a mess, the subreg rtx code is overloaded to mean a whole set 
of different things.

> a<- ...
>
> (part of a) <- ...
>
> use of a
>
> If we assume that second store complete, the first store looks dead.

The first store is dead if A is a single register and the partial access 
is smaller and not using strict_low_part.  The second store is 
documented to have undefined effects on the rest of the register in that 
case.

> If there is a set of test that we can apply when setting the df_ref
> _partial flag to ignore the cases where it really is to the entire word,
> we will add this in the next flip of the patch.

There's some code in flow.c's mark_set_1 which may be usable.

> > Do you think we need this patch in stage3, or can it wait until 4.3
> > stage1?
>
> As danny said this patch has been in for a long time and fixes or at
> least partially fixes several bugs.

Okay, let's sort out the subreg situation and get it in.


Bernd