[Dwarf-discuss] Re: Interpretation of FBREG: implied deref of frame_base?

[Todd Allen]

On Mon, Apr 18, 2005 at 06:03:23PM -0700, Chris Quenelle wrote:
> 
> >
> >
> >Since DW_OP_fbreg -20 is unambiguously computing a location
> >(not a value) I suspect Daniel is right, and an automatic dereference
> >is in order.
> >
> 
> I'm not following this discussion 100%, but in the future I'll want
> to use dwarf expressions as value expressions as well as location
> expressions.  So it would be good if the definitions of the operators
> were compatible with that use.  (Sorry I can't offer more explicit
> feedback on the subject under discussion)
> 

I don't think anything said precludes this.  I believe David was talking
about the case of a location description, that subset of dwarf expressions
which compute a location, i.e. either an address or a register.  For the more
general dwarf expression case, it would compute whatever it computes, as
defined by the particular attribute.  Of course, if it computes a value,
DW_OP_reg{n} and DW_OP_regx alone would be meaningless.

In fact, my first instinct with the DW_AT_frame_base attribute was that a
DW_OP_reg{n} or DW_OP_regx alone was meaningless.  But I suppose it's mostly
harmless to say that they're equivalent to DW_OP_breg{n} or DW_OP_bregx,
respectively.  Especially if some producers already are producing them and
meaning that, and some consumers already are interpreting them that way.

-- 
Todd Allen
Concurrent Computer Corporation