https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125873
Sam James <sjames at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |sjames at gcc dot gnu.org
--- Comment #8 from Sam James <sjames at gcc dot gnu.org> ---
(In reply to Saksham Gupta from comment #7)
> Hi Richard,
>
> I'd like to take a crack at this bug.
>
> Since the previous bisection was thrown off by modifying the test case to
> use return, I am going to set up a fresh automated git bisect using the
> original small.c with __builtin_abort() to pinpoint the actual offending
> commit.
>
If it has 'needs-bisection' in the keywords, you should just feel free to do it
;)
> Once I have the correct revision, my plan is to reduce the test case with
> creduce and compare the GIMPLE dumps to see where
> -fno-aggressive-loop-optimizations is being ignored.
I don't think it is being ignored, more likely that with that argument, a
transformation is done that inhibits a later bad one.
>
> Please let me know if anyone is already actively working on a fix for this,
> otherwise I will share my bisection results here shortly.
I don't think anybody is.
