This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [C++ PATCH] Fix constexpr OBJ_TYPE_REF handling on array elts (PR c++/87398)
- From: Jason Merrill <jason at redhat dot com>
- To: Jakub Jelinek <jakub at redhat dot com>
- Cc: Marek Polacek <polacek at redhat dot com>, gcc-patches List <gcc-patches at gcc dot gnu dot org>
- Date: Tue, 25 Sep 2018 11:07:19 -0400
- Subject: Re: [C++ PATCH] Fix constexpr OBJ_TYPE_REF handling on array elts (PR c++/87398)
- References: <20180925071754.GX8250@tucnak> <CADzB+2nc+62ioqFcPNJY7SKxsfQBMuXC=Xi-45V_P6zoWDBHbQ@mail.gmail.com> <20180925145723.GE8250@tucnak>
On Tue, Sep 25, 2018 at 10:57 AM, Jakub Jelinek <jakub@redhat.com> wrote:
> On Tue, Sep 25, 2018 at 10:42:08AM -0400, Jason Merrill wrote:
>> On Tue, Sep 25, 2018 at 3:17 AM, Jakub Jelinek <jakub@redhat.com> wrote:
>> > ARRAY_REF is a handled component, so when a virtual call is on an array
>> > element (or some component thereof), the loop to look through handled
>> > components will look through the ARRAY_REFs too and then TREE_TYPE (obj)
>> > might be an ARRAY_TYPE. The code wants to look at the class type instead
>> > though.
>>
>> Yeah, I think looping through handled components is wrong here; given
>> something like
>>
>> struct A
>> {
>> virtual void f();
>> };
>>
>> struct B
>> {
>> A a;
>> } b;
>>
>> int main()
>> {
>> b.a.f();
>> }
>>
>> Looking through all handled components would give us "b", which has no
>> virtual functions.
>
> You're right. So like this (only look through COMPONENT_REFs where the
> FIELD_DECL is DECL_FIELD_IS_BASE)?
> Only checked
> make check-c++-all RUNTESTFLAGS="dg.exp='constexpr-virt* pr87398*'"
> so far, but it also fixes the PR87425 testcase and doesn't regress anything.
Looks good.
Jason