[Mlir-commits] [mlir] [MLIR][GPU] Fix async.yield gpu.async.token lowering race (PR #190717)

[Jared Hoberock]

================
@@ -30,9 +30,15 @@ class RewritePatternSet;
 /// the corresponding async.yield ops need to update their types accordingly to
 /// the TypeConverter, but otherwise don't care what type conversions are
 /// happening.
+///
+/// `addStructuralYieldPattern` controls whether the structural pattern for
+/// `async.yield` is registered. Callers that install their own (more
+/// specialized) `async.yield` rewriter should pass `false` to avoid pattern
+/// competition with that rewriter. The dynamic legality marker for
----------------
jaredhoberock wrote:

Yes, I tried bumping the benefit first, and it seemed to work. If that's the conventional way to deal with pattern competition, I can just do that. Is it?

https://github.com/llvm/llvm-project/pull/190717