This is the mail archive of the
mailing list for the GCC project.
Re: [ubsan] Introduce pointer_sized_int_node
- From: Richard Biener <richard dot guenther at gmail dot com>
- To: Marek Polacek <polacek at redhat dot com>
- Cc: GCC Patches <gcc-patches at gcc dot gnu dot org>
- Date: Wed, 28 Aug 2013 12:48:41 +0200
- Subject: Re: [ubsan] Introduce pointer_sized_int_node
- Authentication-results: sourceware.org; auth=none
- References: <20130826101509 dot GH4968 at redhat dot com> <CAFiYyc2O5N91=kDF-aJOTUz67_m7RpF9b7m6zi9apw0bZECkrA at mail dot gmail dot com> <20130827125616 dot GB574 at redhat dot com>
On Tue, Aug 27, 2013 at 2:56 PM, Marek Polacek <email@example.com> wrote:
> On Tue, Aug 27, 2013 at 01:48:29PM +0200, Richard Biener wrote:
>> > + TI_POINTER_SIZED_TYPE,
>> I'd rather see TI_UINTPTR_TYPE and TI_INTPTR_TYPE (note they might
>> not be exactly of POINTER_SIZE but larger).
> We already have [u]intptr_type_node -- but only in c-family/, thus
> ubsan.c/asan.c cannot use those nodes. I can create both
> TI_SIGNED_POINTER_SIZED_TYPE and TI_UNSIGNED_POINTER_SIZED_TYPE,
> but we currently need only the latter...
So simply move them to the middle-end. The set of C language types that
define the target ABI should be constructed and maintained by the middle-end
(you need it for various builtins anyway)
>> TI_POINTER_SIZED_TYPE is ambiguous, too - what's its signedness?
> Unsigned. But yeah, one can't tell by just looking at the name.
>> All around the compiler we use sizetype and ssizetype to munge pointers
>> (well, not strictly correct as targets may define sizetype to be larger/smaller
>> than actual pointers).
> I see.