[PATCH] Fix may_alias canonical types regression

[Doug Gregor]

On Wed, Sep 24, 2008 at 2:15 AM, Eric Botcazou <ebotcazou@adacore.com> wrote:
>> Tested i386-apple-darwin8.11.1 with no regressions. Assuming bootstrap
>> runs through, okay for mainline and 4.3?
>
> Is the type_hash_eq change necessary to fix the regression?  If no, I think
> that it's too dangerous for the branch.

Yes. Fixing the canonical-types issue in
build_type_attribute_qual_variant exposes the wchar_t bug in
g++.dg/ext/alias-canon.C.

>> 2008-09-24  Douglas Gregor  <doug.gregor@gmail.com>
>>
>>       PR c++/37553
>>       * tree.c (build_type_attribute_qual_variant): Hash on the
>>       unqualified type, and don't overwrite an existing
>>       (type_hash_eq): For INTEGER_TYPE nodes, make the TYPE_NAME of the
>>       types significant, to allow distinguishing between wchar_t and its
>>       underlying type. This also means that we'll retain a little more
>>       typedef information.
>
> Please bootstrap and test Ada as well for this kind of change.

Will do.

  - Doug