[mlir] Make GpuAsyncRegion pass depend on async dialect.

Do not cache gpu.async.token type so that the pass can be created before the GPU dialect is registered.

Reviewed By: ftynse

Differential Revision: https://reviews.llvm.org/D94397

Author: chsigg

mlir/include/mlir/Dialect/GPU/Passes.td

@@ -19,6 +19,7 @@ def GpuKernelOutlining : Pass<"gpu-kernel-outlining", "ModuleOp"> {
 def GpuAsyncRegionPass : FunctionPass<"gpu-async-region"> {
   let summary = "Make GPU ops async";
   let constructor = "mlir::createGpuAsyncRegionPass()";
+  let dependentDialects = ["async::AsyncDialect"];
 }
 
 #endif // MLIR_DIALECT_GPU_PASSES

mlir/lib/Dialect/GPU/Transforms/AsyncRegionRewriter.cpp

@@ -78,6 +78,8 @@ struct GpuAsyncRegionPass::ThreadTokenCallback {
     if (op->getNumRegions() > 0)
       return op->emitOpError("regions are not supported");
 
+    auto tokenType = builder.getType<gpu::AsyncTokenType>();
+
     // If there is no current token, insert a `gpu.wait async` without
     // dependencies to create one.
     if (!currentToken)
@@ -108,7 +110,7 @@ struct GpuAsyncRegionPass::ThreadTokenCallback {
   }
 
   OpBuilder builder;
-  const Type tokenType = builder.getType<gpu::AsyncTokenType>();
+
   // The token that represents the current asynchronous dependency. It's valid
   // range starts with a `gpu.wait async` op, and ends with a `gpu.wait` op.
   // In between, each gpu::AsyncOpInterface depends on the current token and

mlir/lib/Dialect/GPU/Transforms/PassDetail.h

@@ -9,6 +9,7 @@
 #ifndef DIALECT_GPU_TRANSFORMS_PASSDETAIL_H_
 #define DIALECT_GPU_TRANSFORMS_PASSDETAIL_H_
 
+#include "mlir/Dialect/Async/IR/Async.h"
 #include "mlir/Pass/Pass.h"
 
 namespace mlir {