https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123848
--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> --- >the issue seems to start off in first gimple pass itself Huh? Considering inet_csk/1 with 10 size to be inlined into inet_csk/1 in /app/example.c:130 Estimated badness is -0.000098, frequency 0.90. Performing recursive inlining on inet_csk/1 Inlining call of depth 1 called approx. 0.90 times per call Parm map:
