This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: PATCH: Reguire __cxa_atexit for ext/{new, throw}_allocator/deallocate_global.cc
- From: Dave Korn <dave dot korn dot cygwin at googlemail dot com>
- To: Steve Ellcey <sje at cup dot hp dot com>
- Cc: Dave Korn <dave dot korn dot cygwin at googlemail dot com>, Paolo Carlini <paolo dot carlini at oracle dot com>, gcc-patches at gcc dot gnu dot org, libstdc++ at gcc dot gnu dot org
- Date: Wed, 03 Mar 2010 00:48:16 +0000
- Subject: Re: PATCH: Reguire __cxa_atexit for ext/{new, throw}_allocator/deallocate_global.cc
- References: <201003030025.o230PxT02210@lucas.cup.hp.com>
On 03/03/2010 00:25, Steve Ellcey wrote:
> Dave Korn <dave.korn.cygwin@googlemail> writes:
>
>> As far as I can see, this is a real genuine bug in the platform support for
>> one specific target only.
>
> These tests fail on HP-UX platforms too (hppa and ia64) and for the same
> reason, so now we have two platforms.
Ok; it's still not actually *contingent* on "do not have cxa_atexit".
> I think having the list of targets
> that 'screw up the destructor ordering' centralized in a feature test
> instead of listed them in each test makes sense. I think how the feature
> test is implemented (it could just be a list of targets) is a secondary
> issue to having it in one place.
If problems with static ?tor ordering exists on more than one platform, I
see why it makes sense to consider it as a category. But it really ought to
be true by default, other targets that aren't affected by this bug shouldn't
be excluded from the test on the basis of an overly-broad filter.
cheers,
DaveK