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

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

Daniel Berlin wrote:
> Bernd Schmidt wrote:
> > Daniel Berlin wrote:
> >
> > > Based on this discussion, I also have a hard time believing any part
> > > outside of flow (or possibly including flow) actually gets subregs right
> > > all the time, since nobody can seem to articulate the entire complete
> > > set of dataflow distinctions that must be made for them. :)
> > Ian and I seem to be giving you the same set of rules.
>
> But you aren't, AFAICT.
>
> You said it's not a partial set if it's a subreg of a pseudo not inside
> a strict low part, and it's a set of a smaller than the entire size of
> the register.

I may have been imprecise, for which I apologize - it is hard to 
describe this verbally with the necessary amount of precision.  I said 
"if A is a single register", which is ambiguous - I intended it to mean, 
not a multiword one.  For pseudos, that was a poor choice of wording, 
given that you have pseudos that'll end up covering more than one hard 
register.  My mental model of the situation is that there are 
essentially two cases, multiword values, and single-word values, and 
SUBREG has different meanings for the two, but (ignoring 
strict_low_part) it always affects an entire machine register (but not 
necessarily an entire pseudo).

I recommend that you try to read the code in flow.c and follow its 
behaviour, rather than any English-language description of it (the 
latter usually not being tested by a full bootstrap of the compiler).


Bernd