This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [C++ PATCH] Fix -fsanitize=vptr (PR c++/70147)
- From: Jason Merrill <jason at redhat dot com>
- To: Jakub Jelinek <jakub at redhat dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Wed, 16 Mar 2016 00:11:43 -0400
- Subject: Re: [C++ PATCH] Fix -fsanitize=vptr (PR c++/70147)
- Authentication-results: sourceware.org; auth=none
- References: <20160314214329 dot GQ3017 at tucnak dot redhat dot com> <20160315162428 dot GY3017 at tucnak dot redhat dot com>
On 03/15/2016 12:24 PM, Jakub Jelinek wrote:
Bernd E. mentioned in the PR the problem that if some subobject ctor throws,
if for -fsanitize=vptr we clear again the vtable pointers even for virtual
bases then they won't be properly destructed.
So, here is an incremental patch to the earlier patch, which will clear
the virtual base vtbl pointers only in the in-charge ctor.
Right. If we aren't in charge of constructing the base, we shouldn't
mess with its vptr either. Both patches are OK.
Though, this brings a non-sanitizer issue, for -flifetime-dse=2
we emit a clobber of the whole subobject even in a ctor with _vtt_parm
argument, and the virtual bases at that point might live inside of the
area that is clobbered by the ctor {CLOBBER}
(both data and vtable pointers).
See my comment in the PR.
Jason