Josh Stone <[email protected]> writes:

> With CFLAGS='-Og -g', F21 gcc hits -Werror=maybe-uninitialized on jmp in
> the call "prepare_thread (pid2, jmp)".  It's fine with -O2/-O0.
>
> The only way this could be unset is if the loop before ran to
> completion, and there's already an assert against this case.

GCC is sort of right in that that assert could in theory be a nop.  It's
not appropirate in the first place, for handling what amounts to user
input, but this is a test case, so who cares.

Looks good to me.

Thanks,
Petr

Reply via email to