[DAGCombine] Skip re-visiting EntryToken to avoid compile time explosion

During the main DAGCombine loop, whenever a node gets replaced, the new
node and all its users are pushed onto the worklist.  Omit this if the
new node is the EntryToken (e.g. if a store managed to get optimized
out), because re-visiting the EntryToken and its users will not uncover
any additional opportunities, but there may be a large number of such
users, potentially causing compile time explosion.

This compile time explosion showed up in particular when building the
SingleSource/UnitTests/matrix-types-spec.cpp test-suite case on any
platform without SIMD vector support.

Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D86963
This commit is contained in:
Ulrich Weigand 2020-09-09 19:09:52 +02:00
parent fc4bff0cd3
commit 1a25133bcd
1 changed files with 9 additions and 3 deletions

View File

@ -1558,9 +1558,15 @@ void DAGCombiner::Run(CombineLevel AtLevel) {
DAG.ReplaceAllUsesWith(N, &RV); DAG.ReplaceAllUsesWith(N, &RV);
} }
// Push the new node and any users onto the worklist // Push the new node and any users onto the worklist. Omit this if the
AddToWorklist(RV.getNode()); // new node is the EntryToken (e.g. if a store managed to get optimized
AddUsersToWorklist(RV.getNode()); // out), because re-visiting the EntryToken and its users will not uncover
// any additional opportunities, but there may be a large number of such
// users, potentially causing compile time explosion.
if (RV.getOpcode() != ISD::EntryToken) {
AddToWorklist(RV.getNode());
AddUsersToWorklist(RV.getNode());
}
// Finally, if the node is now dead, remove it from the graph. The node // Finally, if the node is now dead, remove it from the graph. The node
// may not be dead if the replacement process recursively simplified to // may not be dead if the replacement process recursively simplified to