https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84478
Martin Sebor <msebor at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|REOPENED |RESOLVED
Resolution|--- |FIXED
--- Comment #8 from Martin Sebor <msebor at gcc dot gnu.org> ---
The test started failing only today:
https://gcc.gnu.org/ml/gcc-regression/2019-01/msg00039.html
so the failures are a regression in GCC 9 but not in GCC 8 as the bug summary
indicates. Since GCC 8 is not affected, neither is the patch referenced in
comment #3, and so tracking any new failures in this area under a separate bug
would be appropriate.
r267503 is also part of a bigger patch kit with at least one change after it
(r267531) so it's possible that the regression was only transient. As Jeff
mentions in his post to gcc-patches below, he is aware of a few outstanding
issues on a few 32-bit targets as a result of some of the patches in this
series and is in the process of validating the full kit:
https://gcc.gnu.org/ml/gcc-patches/2019-01/msg00064.html
I have just built the top of trunk and verified the test passes in both 32-bit
and 64-bit modes (with a 64-bit host GCC) so unless the original bug was also
specific to an ILP32 host compiler the new failures are, making the root cause
also likely different. In any case, if the failures do persist on any targets
let's open a new bug and make both the host and target clear. As the author of
the changes behind both regressions I prefer it that way.
