Lines Matching full:predecessors
63 /// multiple predecessors and multiple successors. If one or more of the
64 /// predecessors of the block can be proven to always jump to one of the
321 /// in any of our predecessors. If so, return the known list of value and pred
340 // If V is a constant, then it is known in all predecessors.
354 // of any of our predecessors.
515 // live-in value on any predecessors.
608 /// fewest predecessors. This should reduce the in-degree of the others.
614 // Compute the successor with the minimum number of predecessors.
638 /// ProcessBlock - If there are any predecessors whose control can be threaded
650 // predecessors of our predecessor block.
805 // a PHI node in the current block. If we can prove that any predecessors
806 // compute a predictable value based on a PHI node, thread those predecessors.
919 // predecessors. If the value is unavailable in more than one unique
920 // predecessor, we want to insert a merge block for those common predecessors.
932 // Otherwise, we had multiple unavailable predecessors or we had a critical
940 // Add all the unavailable predecessors to the PredsToSplit list.
957 // If the value isn't available in all predecessors, then there will be
1092 // predecessors and keeps track of the undefined inputs (which are represented
1148 // predecessors that will jump to it into a single predecessor.
1164 // the destination that these predecessors should get to.
1254 // predecessors can be of the set true, false, or undef.
1284 // If we inferred a value for all of the predecessors, then duplication won't
1362 // And finally, do it! Start by factoring the predecessors is needed.
1368 << " common predecessors.\n");
1461 // NewBB instead of BB. This eliminates predecessors from BB, which requires
1506 // And finally, do it! Start by factoring the predecessors is needed.
1512 << " common predecessors.\n");
