https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83398
Jakub Jelinek <jakub at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |aoliva at gcc dot gnu.org, | |jakub at gcc dot gnu.org --- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> --- Started with r255569, got fixed with r255609. I think it just went latent though, because the testcase doesn't have any debug bind stmts. --- pr83398.c.230t.optimized_ 2017-12-14 07:54:28.000000000 -0500 +++ pr83398.c.230t.optimized 2017-12-14 07:56:34.000000000 -0500 @@ -28,9 +28,9 @@ foo () goto <bb 3>; [66.00%] <bb 3> [local count: 708669604]: - # DEBUG BEGIN_STMT c: # DEBUG BEGIN_STMT + # DEBUG BEGIN_STMT a = 0; <bb 4> [local count: 1073741825]: is the difference between r255607 and r255609, but # DEBUG BEGIN_STMT may appear before labels even with current trunk. Perhaps dup of the https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83396#c41 Alex is looking into?