[Bug tree-optimization/25737] [4.1/4.2 Regression] ACATS tests c974001 and c974013 do not terminate with struct aliasing

["dberlin at dberlin dot org" <gcc-bugzilla at gcc dot gnu dot org>]

------- Comment #21 from dberlin at gcc dot gnu dot org  2006-03-24 04:04 -------
Subject: Re:  [4.1/4.2 Regression] ACATS tests
        c974001 and c974013 do not terminate with struct aliasing

On Thu, 2006-03-23 at 22:33 +0000, ebotcazou at gcc dot gnu dot org
wrote:
> 
> ------- Comment #19 from ebotcazou at gcc dot gnu dot org  2006-03-23 22:33 -------
> > How can the addressability of Succ make a difference for the testcase?
> 
> How do we determine that the SMT of the structure type and the SFT alias?

It depends.
In the case i've fixed, when we don't have a valid points-to set (which
is the case here), we walk the *addressable* variables, and see if they
alias with the other *addresable* variables.  These addressable
variables include the SFT's with TREE_ADDRESSABLE set.

However, it is non-sensible for an SFT of a DECL_NONADDRESSABLE_P field
to have TREE_ADDRESSABLE set, since it can't be addressed.

That said *do not* currently make use of this DECL_NONADRESSABLE_P
information, so this cannot be the problem, as it will be marked
TREE_ADDRESSABLE if it is addressed in some way, and that flag will not
be cleared as long as something is taking what appears to be the address
of it.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25737