https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121344
--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> --- (In reply to Andrew Pinski from comment #2) > The problem is the same issue as PR 111280. In both cases > number_of_iterations_cltz_complement produces the condition. > > *** This bug has been marked as a duplicate of bug 111280 *** Note the only difference between this and the other one is CLZ ifn is used for the other one since there is a CLZ optab for the 64bit size while there is none for 8bit. If we do s/char/long long/ then you get the CLZ ifn instead of the builtin. This is why I said this and other ones are exactly the same issue.
