https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111049
Sam James <sjames at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|--- |11.0
CC| |sjames at gcc dot gnu.org
Status|UNCONFIRMED |RESOLVED
Resolution|--- |FIXED
--- Comment #3 from Sam James <sjames at gcc dot gnu.org> ---
10.x branch is (long) closed.
(In reply to Mark Pizzolato from comment #2)
> > Looks like this is fixed for GCC 11.
>
> The latest Raspberry Pi OS is still packaging gcc 10.2.1.
>
> How was this tested with gcc 11?
He was retrying the testcase and noticed it now works.
>
> Was there a specific change that addressed the problem? If so, can we get a
> pointer to that?
>
> Did the testing include running with a wide range of structure sizes?
> 10.2.1 demonstrates the issue with some structure sizes and not others. It
> is not just sizes above a certain size which reliably demonstrate the
> problem.
This class of warning is "special" (middle-end warnings) and subject to noise.
You can see some details at
https://developers.redhat.com/blog/2019/03/13/understanding-gcc-warnings and
https://developers.redhat.com/blog/2019/03/13/understanding-gcc-warnings-part-2.
That is, they're very dependent on the optimisation which is, or isn't, done.
Issues like this are usually because of a missed optimisation. Much has changed
since GCC 10, including the integration of ranger.
