[Bug tree-optimization/102192] Curious '-O2'-only '-Wmaybe-uninitialized' diagnostics for 'libgomp.oacc-fortran/routine-10.f90'
msebor at gcc dot gnu.org
gcc-bugzilla@gcc.gnu.org
Wed Oct 6 16:44:50 GMT 2021
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102192
Martin Sebor <msebor at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |WAITING
Ever confirmed|0 |1
CC| |msebor at gcc dot gnu.org
Last reconfirmed| |2021-10-06
--- Comment #1 from Martin Sebor <msebor at gcc dot gnu.org> ---
There are lots of reports about -Wmaybe-uninitialized triggering when it's not
expected to. Some are bugs/limitations in the warning but others are inherent
in the IL the warning works with and can't be avoided given its design.
Without a small test case to easily reproduce the suspected problem, given the
large number of reports with test cases, it's unlikely that anyone will go try
to extract one from a historical Git revision of a test, especially one written
in FORTRAN and involving complex control flow, to determine were this one
falls. I would suggest to include the test in the bug report, along with the
output (as we ask users to do). Better yet would be reducing the suspected
false positive to a smaller test case, and ideally converting it to C.
More information about the Gcc-bugs
mailing list