This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: SSA usage question
- From: Richard Henderson <rth at redhat dot com>
- To: Richard Kenner <kenner at vlsi1 dot ultra dot nyu dot edu>
- Cc: law at redhat dot com, gcc at gcc dot gnu dot org
- Date: Mon, 7 Jun 2004 09:39:15 -0700
- Subject: Re: SSA usage question
- References: <10406071413.AA09370@vlsi1.ultra.nyu.edu>
On Mon, Jun 07, 2004 at 10:13:02AM -0400, Richard Kenner wrote:
> I guess I should make my proposal more explicit:
>
> (1) Add two operands to the ARRAY_REF node. One is a copy of TYPE_MIN_VALUE
> of the index and the second is a copy of TYPE_SIZE_UNIT of the element type.
> (2) Add an operand to COMPONENT_REF that's a copy of DECL_FIELD_POSITION of
> the second operand.
>
> In both cases, any PLACEHOLDER_EXPRs will be expanded, the copy will be
> gimplified, and the operand will not be present if it would be a constant.
>
> Does this solve the problem?
Likely yes.
> Then I propose that we allow VIEW_CONVERT_EXPR in a nested tree of
> references if it doesn't change TYPE_MODE. It will be treated as a NOP,
> just there for type-correctness purposes.
TYPE_MODE shouldn't matter. Let me think about this one.
r~