This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Fix ICE due to commit for PR88834
- From: Jeff Law <law at redhat dot com>
- To: Segher Boessenkool <segher at kernel dot crashing dot org>
- Cc: Kugan Vivekanandarajah <kugan dot vivekanandarajah at linaro dot org>, GCC Patches <gcc-patches at gcc dot gnu dot org>, richard dot sandiford at arm dot com
- Date: Tue, 2 Jul 2019 14:18:42 -0600
- Subject: Re: Fix ICE due to commit for PR88834
- References: <CAELXzTOYeSyR=O02ZwhwbUJnUJp8w1C7_ETr-aCxFiW3A17UaQ@mail.gmail.com> <mptk1dky3xz.fsf@arm.com> <386e2ec9-1356-dd4d-acca-c3c4aea48f00@redhat.com> <20190629155136.GA17570@gate.crashing.org>
On 6/29/19 9:51 AM, Segher Boessenkool wrote:
> On Mon, Jun 17, 2019 at 01:13:43PM -0600, Jeff Law wrote:
>>> (Hopefully one day GET_MODE_SIZE & co. will assert on BLKmode and VOIDmode.)
>> Yea. Not sure why. There's probably a compile-time hit, but I suspect
>> it'd turn up some interesting bugs.
>
> There turn out to be at least twenty or so places in GCC that rely on the
> size of BLKmode and VOIDmode being 0, currently. After those I gave up.
>
> Maybe it's easier to first only handle BLKmode? Or have a
> GET_MODE_SIZE_INTERNAL for all places that currently handle size 0 fine?
I guess we could split them up, but there's something to be said for
getting away from this special behavior. But it's clearly not the most
important thing on our list right now.
jeff