> It is a formal predicate of the GIMPLE grammar. But there is an
> inconsistency now. I don't think we want to allow min-invariant as
> ARRAY_REF indexes because this would allow things like
>
> &a[&b[3]]
>
> which is too gross for words. So, the grammar should be updated to
> include invariant addresses in 'val',
Like so?
Index: doc/tree-ssa.texi
===================================================================
--- doc/tree-ssa.texi (revision 126300)
+++ doc/tree-ssa.texi (working copy)
@@ -1,4 +1,4 @@
[EMAIL PROTECTED] Copyright (c) 2004, 2005 Free Software Foundation, Inc.
[EMAIL PROTECTED] Copyright (c) 2004, 2005, 2007 Free Software Foundation, Inc.
@c Free Software Foundation, Inc.
@c This is part of the GCC manual.
@c For copying conditions, see the file gcc.texi.
@@ -680,7 +680,7 @@ void f()
bitfieldref : BIT_FIELD_REF
op0 -> inner-compref
op1 -> CONST
- op2 -> var
+ op2 -> val
compref : inner-compref
| TARGET_MEM_REF
@@ -718,6 +718,8 @@ void f()
op1 -> val
val : ID
+ | invariant ADDR_EXPR
+ op0 -> addr-expr-arg
| CONST
rhs : lhs
> but it should have a new rule so that no invariant addresses are allowed as
> array indices.
This would require more substantial changes in the gimplifier than what is
required to fix the is_gimple_min_invariant problem. Can't we live with it?
--
Eric Botcazou
