The code mistakenly thinks any cond_jump has two successors. This is not true if both destinations are the same, as can happen with weird patterns as in the PR.
Bootstrapped and tested on powerpc64-linux; also tested the simplified test in the PR on an x86_64-linux cross. Sorry for the breakage. Is this okay for trunk? Segher 2015-11-09 Segher Boessenkool <seg...@kernel.crashing.org> * gcc/bb-reorder.c (reorder_basic_blocks_simple): Treat a conditional branch with only one successor just like unconditional branches. --- gcc/bb-reorder.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/gcc/bb-reorder.c b/gcc/bb-reorder.c index 5f1c2cc..950b1a1 100644 --- a/gcc/bb-reorder.c +++ b/gcc/bb-reorder.c @@ -2304,7 +2304,9 @@ reorder_basic_blocks_simple (void) if (JUMP_P (end) && extract_asm_operands (end)) continue; - if (any_condjump_p (end)) + if (single_succ_p (bb)) + edges[n++] = EDGE_SUCC (bb, 0); + else if (any_condjump_p (end)) { edge e0 = EDGE_SUCC (bb, 0); edge e1 = EDGE_SUCC (bb, 1); @@ -2315,8 +2317,6 @@ reorder_basic_blocks_simple (void) edges[n++] = e0; edges[n++] = e1; } - else if (single_succ_p (bb)) - edges[n++] = EDGE_SUCC (bb, 0); } /* Sort the edges, the most desirable first. When optimizing for size -- 1.9.3