[PATCH] Fix __cxa_exception ABI breakage in 4.4 (PR libstdc++/38732)

Richard Guenther richard.guenther@gmail.com
Wed Jan 7 21:12:00 GMT 2009


On Wed, Jan 7, 2009 at 7:16 PM, Paolo Carlini <paolo.carlini@oracle.com> wrote:
> Hi,
>>> I see, if you could experiment a bit with that...
>>>
>> What I wrote is what we do.
> Of course, sorry.
>>   So I don't know what should I experiment with.
>> I think we never access fields in __cxa_refcounted_exception,
>> __cxa_exception and _Unwind_Exception through 2 different pointers
>> in one function, but even if we would, this IMHO boils down to:
>>
>> struct A { int i; struct B { short s; long l; } b; };
>>
>> int foo (struct A *a, struct B *b)
>> {
>>   b->s = 3;
>>   a->b.s = 4;
>>   a->b.l = 7;
>>   b->l = 6;
>>   return b->s + a->b.l;
>> }
>> which is IMHO valid.
>>
> Agreed. Therefore, I propose waiting a bit for Richard's opinion and, in
> any case, decouple the spurious warning issue and 38732, thus commit
> your work on the latter and separately try to figure out if the false
> positive is avoidable and / or related to the warning in 38477.

It's certainly unrelated to PR38477.  PR38477 looks more like a
frontend/middle-end
issue with respect to the representation of types.

Richard.

> Paolo.
>



More information about the Gcc-patches mailing list