This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [patch] Fix pr23046


On Mon, Sep 19, 2005 at 03:53:29PM -0400, Richard Kenner wrote:
> I don't think so because the point would be to *validate* the data in the
> object.  Except for the bitfield case, the actual data is 8 bits wide
> and the idea is to verify that the data is valid.  So you don't want somebody
> assuming that the data is already known to be 0-7, since it isn't.

No, the actual data is 3 bits wide.  That's what expand_expr is
supposed to be enforcing.  Thus the data *is* in 0-7, by definition.

But it's probably less efficient than just using the 8 bit data type,
and so isn't worth considering further.  Which begs the question of why
the subtype has TYPE_PRECISION 3 instead of 8...

> We do have to be careful that the removal of "unnecessary" conversions doesn't
> remove the V_C_E.  I think what it means is that we never consider a V_C_E
> unnecessary if it's between two scalar types.

We'll have to document this carefully.


r~


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]