Hi, with tail recursion and accumulation it is quite common case that the profile is unrealistic and the recursive call is triggered more often then the entry block. This patch prevents tailcall from dropping entry block profile to 0 (and making it very cold) in this case.
Bootstrapped/regtested x86_64-linux, comitted. * tree-tailcall.c (eliminate_tail_call): Be more careful about not disturbin profile of entry block. Index: tree-tailcall.c =================================================================== --- tree-tailcall.c (revision 254884) +++ tree-tailcall.c (working copy) @@ -889,10 +889,18 @@ eliminate_tail_call (struct tailcall *t) /* Number of executions of function has reduced by the tailcall. */ e = single_succ_edge (gsi_bb (t->call_gsi)); - decrease_profile (EXIT_BLOCK_PTR_FOR_FN (cfun), e->count ()); - decrease_profile (ENTRY_BLOCK_PTR_FOR_FN (cfun), e->count ()); + + profile_count count = e->count (); + + /* When profile is inconsistent and the recursion edge is more frequent + than number of executions of functions, scale it down, so we do not end + up with 0 executions of entry block. */ + if (count >= ENTRY_BLOCK_PTR_FOR_FN (cfun)->count) + count = ENTRY_BLOCK_PTR_FOR_FN (cfun)->count.apply_scale (7, 8); + decrease_profile (EXIT_BLOCK_PTR_FOR_FN (cfun), count); + decrease_profile (ENTRY_BLOCK_PTR_FOR_FN (cfun), count); if (e->dest != EXIT_BLOCK_PTR_FOR_FN (cfun)) - decrease_profile (e->dest, e->count ()); + decrease_profile (e->dest, count); /* Replace the call by a jump to the start of function. */ e = redirect_edge_and_branch (single_succ_edge (gsi_bb (t->call_gsi)),