[patch PR libstdc++/51135]: Fix [4.7 Regression] SIGSEGV during exception cleanup on win32

[Kai Tietz]

2011/12/12 Paolo Carlini <paolo.carlini@oracle.com>:
> On 12/12/2011 07:17 PM, Kai Tietz wrote:
>>
>> 2011/12/12 Paolo Carlini<paolo.carlini@oracle.com>:
>>>
>>> On 12/12/2011 06:45 PM, Marc Glisse wrote:
>>>>
>>>> Actually, g++ currently simply ignores the linkage as part of function
>>>> types, so this shouldn't have any effect.
>>>
>>> Also note that other systems have been using libsupc++ (without using the
>>> whole C++ lib), and I don't think we should play dirty games here and
>>> make
>>> the life unnecessarily hard to compilers which in this specific respect
>>> are
>>> more conforming than GCC.
>>>
>>> Paolo.
>>
>> Well, could you be please more specific, what in the unwind-cxx.h is
>> causing especially your questioning here?
>
> Very simply, I don't want we fiddle with delicate things in Stage 3. Thus,
> for anything != mingw **nothing** should be changed, signatures, linkages,
> etc. Do whatever you prefer in the config files, but make sure that by
> default nothing changes.
>
> Thanks,
> Paolo.

Well, so we are at the variant I've shown at beginning, which is using
a config-macro to add optional the thiscall.

Kai