Re: Bogous trees in ivopts again!

2005-06-02 Thread Richard Guenther
On 6/1/05, Richard Guenther <[EMAIL PROTECTED]> wrote: > rewrite_address_base is trying at > > *op = build1 (INDIRECT_REF, TREE_TYPE (*op), with); To see the failure in various places during a build of gcc apply the attached patch with places gcc_assert (TREE_TYPE (*op) == TREE_TYPE (TREE_TY

Re: Bogous trees in ivopts again!

2005-06-02 Thread Richard Guenther
The problem is (?) that get_base_address for state_inD.6032_16->savedD.6026[regD.6037_3] (which is of type ) returns *state_inD.6032_16 (which is of type ). The documentation of get_base_address does not talk about any type guarantees - but I guess IVOPTs rather wanted to have state_inD.60

Bogous trees in ivopts again!

2005-06-01 Thread Richard Guenther
rewrite_address_base is trying at *op = build1 (INDIRECT_REF, TREE_TYPE (*op), with); to substitute *op state_inD.7206_16->savedD.7200[regD.7211_3] which is of type integer_type 0x4014a3cc char with *state_inD.7206_38 where state_inD.7206_38 is of type pointer_type 0x4047b21c type . Th