https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69850
Manuel López-Ibáñez <manu at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |manu at gcc dot gnu.org
--- Comment #3 from Manuel López-Ibáñez <manu at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #2)
> Created attachment 37723 [details]
> gcc6-pr69850.patch
>
> Untested fix. Another possibility is to STRIP_NOPS and if it is this, set
> ifexp to integer_one_node. But we need the TREE_NO_WARNING stuff anyway,
> because e.g. in function with nonnull parameter other than this, that at
> some point is changed, the comparison would be still desirable and valid.
Aren't compiler-generated expressions marked with ARTIFICIAL or some such? In
any case, setting TREE_NO_WARNING for compiler-generated expressions seems the
right thing to do.
I'd suggest to add "-Wnonnul-compare" explicitly to the testcase, otherwise it
may start silently failing if -Wall stops enabling the warning.
