Separate the Registration from Loading dialects in the Context

This changes the behavior of constructing MLIRContext to no longer load globally
registered dialects on construction. Instead Dialects are only loaded explicitly
on demand:
- the Parser is lazily loading Dialects in the context as it encounters them
during parsing. This is the only purpose for registering dialects and not load
them in the context.
- Passes are expected to declare the dialects they will create entity from
(Operations, Attributes, or Types), and the PassManager is loading Dialects into
the Context when starting a pipeline.

This changes simplifies the configuration of the registration: a compiler only
need to load the dialect for the IR it will emit, and the optimizer is
self-contained and load the required Dialects. For example in the Toy tutorial,
the compiler only needs to load the Toy dialect in the Context, all the others
(linalg, affine, std, LLVM, ...) are automatically loaded depending on the
optimization pipeline enabled.

To adjust to this change, stop using the existing dialect registration: the
global registry will be removed soon.

1) For passes, you need to override the method:

virtual void getDependentDialects(DialectRegistry &registry) const {}

and registery on the provided registry any dialect that this pass can produce.
Passes defined in TableGen can provide this list in the dependentDialects list
field.

2) For dialects, on construction you can register dependent dialects using the
provided MLIRContext: `context.getOrLoadDialect<DialectName>()`
This is useful if a dialect may canonicalize or have interfaces involving
another dialect.

3) For loading IR, dialect that can be in the input file must be explicitly
registered with the context. `MlirOptMain()` is taking an explicit registry for
this purpose. See how the standalone-opt.cpp example is setup:

  mlir::DialectRegistry registry;
  registry.insert<mlir::standalone::StandaloneDialect>();
  registry.insert<mlir::StandardOpsDialect>();

Only operations from these two dialects can be in the input file. To include all
of the dialects in MLIR Core, you can populate the registry this way:

  mlir::registerAllDialects(registry);

4) For `mlir-translate` callback, as well as frontend, Dialects can be loaded in
the context before emitting the IR: context.getOrLoadDialect<ToyDialect>()

Differential Revision: https://reviews.llvm.org/D85622

mlir/examples/standalone/standalone-opt/standalone-opt.cpp

@@ -24,9 +24,16 @@
 int main(int argc, char **argv) {
   mlir::registerAllDialects();
   mlir::registerAllPasses();
-
-  mlir::registerDialect<mlir::standalone::StandaloneDialect>();
   // TODO: Register standalone passes here.
 
-  return failed(mlir::MlirOptMain(argc, argv, "Standalone optimizer driver\n"));
+  mlir::DialectRegistry registry;
+  registry.insert<mlir::standalone::StandaloneDialect>();
+  registry.insert<mlir::StandardOpsDialect>();
+  // Add the following to include *all* MLIR Core dialects, or selectively
+  // include what you need like above. You only need to register dialects that
+  // will be *parsed* by the tool, not the one generated
+  // registerAllDialects(registry);
+
+  return failed(
+      mlir::MlirOptMain(argc, argv, "Standalone optimizer driver\n", registry));
 }

mlir/examples/standalone/test/Standalone/standalone-opt.mlir

@@ -1,3 +1,3 @@
 // RUN: standalone-opt --show-dialects | FileCheck %s
-// CHECK: Registered Dialects:
+// CHECK: Available Dialects:
 // CHECK: standalone

mlir/examples/toy/Ch2/toyc.cpp

@@ -68,10 +68,9 @@ std::unique_ptr<toy::ModuleAST> parseInputFile(llvm::StringRef filename) {
 }
 
 int dumpMLIR() {
-  // Register our Dialect with MLIR.
-  mlir::registerDialect<mlir::toy::ToyDialect>();
-
-  mlir::MLIRContext context;
+  mlir::MLIRContext context(/*loadAllDialects=*/false);
+  // Load our Dialect in this MLIR Context.
+  context.getOrLoadDialect<mlir::toy::ToyDialect>();
 
   // Handle '.toy' input to the compiler.
   if (inputType != InputType::MLIR &&

mlir/examples/toy/Ch3/toyc.cpp

@@ -102,10 +102,10 @@ int loadMLIR(llvm::SourceMgr &sourceMgr, mlir::MLIRContext &context,
 }
 
 int dumpMLIR() {
-  // Register our Dialect with MLIR.
-  mlir::registerDialect<mlir::toy::ToyDialect>();
+  mlir::MLIRContext context(/*loadAllDialects=*/false);
+  // Load our Dialect in this MLIR Context.
+  context.getOrLoadDialect<mlir::toy::ToyDialect>();
 
-  mlir::MLIRContext context;
   mlir::OwningModuleRef module;
   llvm::SourceMgr sourceMgr;
   mlir::SourceMgrDiagnosticHandler sourceMgrHandler(sourceMgr, &context);

mlir/examples/toy/Ch4/toyc.cpp

@@ -103,10 +103,10 @@ int loadMLIR(llvm::SourceMgr &sourceMgr, mlir::MLIRContext &context,
 }
 
 int dumpMLIR() {
-  // Register our Dialect with MLIR.
-  mlir::registerDialect<mlir::toy::ToyDialect>();
+  mlir::MLIRContext context(/*loadAllDialects=*/false);
+  // Load our Dialect in this MLIR Context.
+  context.getOrLoadDialect<mlir::toy::ToyDialect>();
 
-  mlir::MLIRContext context;
   mlir::OwningModuleRef module;
   llvm::SourceMgr sourceMgr;
   mlir::SourceMgrDiagnosticHandler sourceMgrHandler(sourceMgr, &context);

mlir/examples/toy/Ch5/mlir/LowerToAffineLoops.cpp

@@ -256,6 +256,9 @@ struct TransposeOpLowering : public ConversionPattern {
 namespace {
 struct ToyToAffineLoweringPass
     : public PassWrapper<ToyToAffineLoweringPass, FunctionPass> {
+  void getDependentDialects(DialectRegistry &registry) const override {
+    registry.insert<AffineDialect, StandardOpsDialect>();
+  }
   void runOnFunction() final;
 };
 } // end anonymous namespace.

mlir/examples/toy/Ch5/toyc.cpp

@@ -106,10 +106,10 @@ int loadMLIR(llvm::SourceMgr &sourceMgr, mlir::MLIRContext &context,
 }
 
 int dumpMLIR() {
-  // Register our Dialect with MLIR.
-  mlir::registerDialect<mlir::toy::ToyDialect>();
+  mlir::MLIRContext context(/*loadAllDialects=*/false);
+  // Load our Dialect in this MLIR Context.
+  context.getOrLoadDialect<mlir::toy::ToyDialect>();
 
-  mlir::MLIRContext context;
   mlir::OwningModuleRef module;
   llvm::SourceMgr sourceMgr;
   mlir::SourceMgrDiagnosticHandler sourceMgrHandler(sourceMgr, &context);

mlir/examples/toy/Ch6/mlir/LowerToAffineLoops.cpp

@@ -255,6 +255,9 @@ struct TransposeOpLowering : public ConversionPattern {
 namespace {
 struct ToyToAffineLoweringPass
     : public PassWrapper<ToyToAffineLoweringPass, FunctionPass> {
+  void getDependentDialects(DialectRegistry &registry) const override {
+    registry.insert<AffineDialect, StandardOpsDialect>();
+  }
   void runOnFunction() final;
 };
 } // end anonymous namespace.

mlir/examples/toy/Ch6/mlir/LowerToLLVM.cpp

@@ -159,6 +159,9 @@ class PrintOpLowering : public ConversionPattern {
 namespace {
 struct ToyToLLVMLoweringPass
     : public PassWrapper<ToyToLLVMLoweringPass, OperationPass<ModuleOp>> {
+  void getDependentDialects(DialectRegistry &registry) const override {
+    registry.insert<LLVM::LLVMDialect, scf::SCFDialect>();
+  }
   void runOnOperation() final;
 };
 } // end anonymous namespace

mlir/examples/toy/Ch6/toyc.cpp

@@ -255,10 +255,10 @@ int main(int argc, char **argv) {
 
   // If we aren't dumping the AST, then we are compiling with/to MLIR.
 
-  // Register our Dialect with MLIR.
-  mlir::registerDialect<mlir::toy::ToyDialect>();
+  mlir::MLIRContext context(/*loadAllDialects=*/false);
+  // Load our Dialect in this MLIR Context.
+  context.getOrLoadDialect<mlir::toy::ToyDialect>();
 
-  mlir::MLIRContext context;
   mlir::OwningModuleRef module;
   if (int error = loadAndProcessMLIR(context, module))
     return error;

mlir/examples/toy/Ch7/mlir/LowerToAffineLoops.cpp

@@ -256,6 +256,9 @@ struct TransposeOpLowering : public ConversionPattern {
 namespace {
 struct ToyToAffineLoweringPass
     : public PassWrapper<ToyToAffineLoweringPass, FunctionPass> {
+  void getDependentDialects(DialectRegistry &registry) const override {
+    registry.insert<AffineDialect, StandardOpsDialect>();
+  }
   void runOnFunction() final;
 };
 } // end anonymous namespace.

mlir/examples/toy/Ch7/mlir/LowerToLLVM.cpp

@@ -159,6 +159,9 @@ class PrintOpLowering : public ConversionPattern {
 namespace {
 struct ToyToLLVMLoweringPass
     : public PassWrapper<ToyToLLVMLoweringPass, OperationPass<ModuleOp>> {
+  void getDependentDialects(DialectRegistry &registry) const override {
+    registry.insert<LLVM::LLVMDialect, scf::SCFDialect>();
+  }
   void runOnOperation() final;
 };
 } // end anonymous namespace

mlir/examples/toy/Ch7/toyc.cpp

@@ -256,10 +256,10 @@ int main(int argc, char **argv) {
 
   // If we aren't dumping the AST, then we are compiling with/to MLIR.
 
-  // Register our Dialect with MLIR.
-  mlir::registerDialect<mlir::toy::ToyDialect>();
+  mlir::MLIRContext context(/*loadAllDialects=*/false);
+  // Load our Dialect in this MLIR Context.
+  context.getOrLoadDialect<mlir::toy::ToyDialect>();
 
-  mlir::MLIRContext context;
   mlir::OwningModuleRef module;
   if (int error = loadAndProcessMLIR(context, module))
     return error;

mlir/include/mlir-c/Registration.h

@@ -10,14 +10,16 @@
 #ifndef MLIR_C_REGISTRATION_H
 #define MLIR_C_REGISTRATION_H
 
+#include "mlir-c/IR.h"
+
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-/** Registers all dialects known to core MLIR with the system. This must be
- * called before creating an MlirContext if it needs access to the registered
- * dialects. */
-void mlirRegisterAllDialects();
+/** Registers all dialects known to core MLIR with the provided Context.
+ * This is needed before creating IR for these Dialects.
+ */
+void mlirRegisterAllDialects(MlirContext context);
 
 #ifdef __cplusplus
 }

mlir/include/mlir/Conversion/Passes.td

@@ -66,6 +66,11 @@ def ConvertAffineToStandard : Pass<"lower-affine"> {
         `affine.apply`.
   }];
   let constructor = "mlir::createLowerAffinePass()";
+  let dependentDialects = [
+    "scf::SCFDialect",
+    "StandardOpsDialect",
+    "vector::VectorDialect"
+  ];
 }
 
 //===----------------------------------------------------------------------===//
@@ -76,6 +81,7 @@ def ConvertAVX512ToLLVM : Pass<"convert-avx512-to-llvm", "ModuleOp"> {
   let summary = "Convert the operations from the avx512 dialect into the LLVM "
                 "dialect";
   let constructor = "mlir::createConvertAVX512ToLLVMPass()";
+  let dependentDialects = ["LLVM::LLVMDialect", "LLVM::LLVMAVX512Dialect"];
 }
 
 //===----------------------------------------------------------------------===//
@@ -98,6 +104,7 @@ def GpuToLLVMConversionPass : Pass<"gpu-to-llvm", "ModuleOp"> {
 def ConvertGpuOpsToNVVMOps : Pass<"convert-gpu-to-nvvm", "gpu::GPUModuleOp"> {
   let summary = "Generate NVVM operations for gpu operations";
   let constructor = "mlir::createLowerGpuOpsToNVVMOpsPass()";
+  let dependentDialects = ["NVVM::NVVMDialect"];
   let options = [
     Option<"indexBitwidth", "index-bitwidth", "unsigned",
            /*default=kDeriveIndexBitwidthFromDataLayout*/"0",
@@ -112,6 +119,7 @@ def ConvertGpuOpsToNVVMOps : Pass<"convert-gpu-to-nvvm", "gpu::GPUModuleOp"> {
 def ConvertGpuOpsToROCDLOps : Pass<"convert-gpu-to-rocdl", "gpu::GPUModuleOp"> {
   let summary = "Generate ROCDL operations for gpu operations";
   let constructor = "mlir::createLowerGpuOpsToROCDLOpsPass()";
+  let dependentDialects = ["ROCDL::ROCDLDialect"];
   let options = [
     Option<"indexBitwidth", "index-bitwidth", "unsigned",
            /*default=kDeriveIndexBitwidthFromDataLayout*/"0",
@@ -126,6 +134,7 @@ def ConvertGpuOpsToROCDLOps : Pass<"convert-gpu-to-rocdl", "gpu::GPUModuleOp"> {
 def ConvertGPUToSPIRV : Pass<"convert-gpu-to-spirv", "ModuleOp"> {
   let summary = "Convert GPU dialect to SPIR-V dialect";
   let constructor = "mlir::createConvertGPUToSPIRVPass()";
+  let dependentDialects = ["spirv::SPIRVDialect"];
 }
 
 //===----------------------------------------------------------------------===//
@@ -136,13 +145,15 @@ def ConvertGpuLaunchFuncToVulkanLaunchFunc
     : Pass<"convert-gpu-launch-to-vulkan-launch", "ModuleOp"> {
   let summary = "Convert gpu.launch_func to vulkanLaunch external call";
   let constructor = "mlir::createConvertGpuLaunchFuncToVulkanLaunchFuncPass()";
+  let dependentDialects = ["spirv::SPIRVDialect"];
 }
 
 def ConvertVulkanLaunchFuncToVulkanCalls
     : Pass<"launch-func-to-vulkan", "ModuleOp"> {
   let summary = "Convert vulkanLaunch external call to Vulkan runtime external "
                 "calls";
   let constructor = "mlir::createConvertVulkanLaunchFuncToVulkanCallsPass()";
+  let dependentDialects = ["LLVM::LLVMDialect"];
 }
 
 //===----------------------------------------------------------------------===//
@@ -153,6 +164,7 @@ def ConvertLinalgToLLVM : Pass<"convert-linalg-to-llvm", "ModuleOp"> {
   let summary = "Convert the operations from the linalg dialect into the LLVM "
                 "dialect";
   let constructor = "mlir::createConvertLinalgToLLVMPass()";
+  let dependentDialects = ["scf::SCFDialect", "LLVM::LLVMDialect"];
 }
 
 //===----------------------------------------------------------------------===//
@@ -163,6 +175,7 @@ def ConvertLinalgToStandard : Pass<"convert-linalg-to-std", "ModuleOp"> {
   let summary = "Convert the operations from the linalg dialect into the "
                 "Standard dialect";
   let constructor = "mlir::createConvertLinalgToStandardPass()";
+  let dependentDialects = ["StandardOpsDialect"];
 }
 
 //===----------------------------------------------------------------------===//
@@ -172,6 +185,7 @@ def ConvertLinalgToStandard : Pass<"convert-linalg-to-std", "ModuleOp"> {
 def ConvertLinalgToSPIRV : Pass<"convert-linalg-to-spirv", "ModuleOp"> {
   let summary = "Convert Linalg ops to SPIR-V ops";
   let constructor = "mlir::createLinalgToSPIRVPass()";
+  let dependentDialects = ["spirv::SPIRVDialect"];
 }
 
 //===----------------------------------------------------------------------===//
@@ -182,6 +196,7 @@ def SCFToStandard : Pass<"convert-scf-to-std"> {
   let summary = "Convert SCF dialect to Standard dialect, replacing structured"
                 " control flow with a CFG";
   let constructor = "mlir::createLowerToCFGPass()";
+  let dependentDialects = ["StandardOpsDialect"];
 }
 
 //===----------------------------------------------------------------------===//
@@ -191,6 +206,7 @@ def SCFToStandard : Pass<"convert-scf-to-std"> {
 def ConvertAffineForToGPU : FunctionPass<"convert-affine-for-to-gpu"> {
   let summary = "Convert top-level AffineFor Ops to GPU kernels";
   let constructor = "mlir::createAffineForToGPUPass()";
+  let dependentDialects = ["gpu::GPUDialect"];
   let options = [
     Option<"numBlockDims", "gpu-block-dims", "unsigned", /*default=*/"1u",
            "Number of GPU block dimensions for mapping">,
@@ -202,6 +218,7 @@ def ConvertAffineForToGPU : FunctionPass<"convert-affine-for-to-gpu"> {
 def ConvertParallelLoopToGpu : Pass<"convert-parallel-loops-to-gpu"> {
   let summary = "Convert mapped scf.parallel ops to gpu launch operations";
   let constructor = "mlir::createParallelLoopToGpuPass()";
+  let dependentDialects = ["AffineDialect", "gpu::GPUDialect"];
 }
 
 //===----------------------------------------------------------------------===//
@@ -212,6 +229,7 @@ def ConvertShapeToStandard : Pass<"convert-shape-to-std", "ModuleOp"> {
   let summary = "Convert operations from the shape dialect into the standard "
                 "dialect";
   let constructor = "mlir::createConvertShapeToStandardPass()";
+  let dependentDialects = ["StandardOpsDialect"];
 }
 
 //===----------------------------------------------------------------------===//
@@ -221,6 +239,7 @@ def ConvertShapeToStandard : Pass<"convert-shape-to-std", "ModuleOp"> {
 def ConvertShapeToSCF : FunctionPass<"convert-shape-to-scf"> {
   let summary = "Convert operations from the shape dialect to the SCF dialect";
   let constructor = "mlir::createConvertShapeToSCFPass()";
+  let dependentDialects = ["scf::SCFDialect"];
 }
 
 //===----------------------------------------------------------------------===//
@@ -230,6 +249,7 @@ def ConvertShapeToSCF : FunctionPass<"convert-shape-to-scf"> {
 def ConvertSPIRVToLLVM : Pass<"convert-spirv-to-llvm", "ModuleOp"> {
   let summary = "Convert SPIR-V dialect to LLVM dialect";
   let constructor = "mlir::createConvertSPIRVToLLVMPass()";
+  let dependentDialects = ["LLVM::LLVMDialect"];
 }
 
 //===----------------------------------------------------------------------===//
@@ -264,6 +284,7 @@ def ConvertStandardToLLVM : Pass<"convert-std-to-llvm", "ModuleOp"> {
     LLVM IR types.
   }];
   let constructor = "mlir::createLowerToLLVMPass()";
+  let dependentDialects = ["LLVM::LLVMDialect"];
   let options = [
     Option<"useAlignedAlloc", "use-aligned-alloc", "bool", /*default=*/"false",
            "Use aligned_alloc in place of malloc for heap allocations">,
@@ -291,11 +312,13 @@ def ConvertStandardToLLVM : Pass<"convert-std-to-llvm", "ModuleOp"> {
 def LegalizeStandardForSPIRV : Pass<"legalize-std-for-spirv"> {
   let summary = "Legalize standard ops for SPIR-V lowering";
   let constructor = "mlir::createLegalizeStdOpsForSPIRVLoweringPass()";
+  let dependentDialects = ["spirv::SPIRVDialect"];
 }
 
 def ConvertStandardToSPIRV : Pass<"convert-std-to-spirv", "ModuleOp"> {
   let summary = "Convert Standard Ops to SPIR-V dialect";
   let constructor = "mlir::createConvertStandardToSPIRVPass()";
+  let dependentDialects = ["spirv::SPIRVDialect"];
 }
 
 //===----------------------------------------------------------------------===//
@@ -306,6 +329,7 @@ def ConvertVectorToSCF : FunctionPass<"convert-vector-to-scf"> {
   let summary = "Lower the operations from the vector dialect into the SCF "
                 "dialect";
   let constructor = "mlir::createConvertVectorToSCFPass()";
+  let dependentDialects = ["AffineDialect", "scf::SCFDialect"];
   let options = [
     Option<"fullUnroll", "full-unroll", "bool", /*default=*/"false",
            "Perform full unrolling when converting vector transfers to SCF">,
@@ -320,6 +344,7 @@ def ConvertVectorToLLVM : Pass<"convert-vector-to-llvm", "ModuleOp"> {
   let summary = "Lower the operations from the vector dialect into the LLVM "
                 "dialect";
   let constructor = "mlir::createConvertVectorToLLVMPass()";
+  let dependentDialects = ["LLVM::LLVMDialect"];
   let options = [
     Option<"reassociateFPReductions", "reassociate-fp-reductions",
            "bool", /*default=*/"false",
@@ -335,6 +360,7 @@ def ConvertVectorToROCDL : Pass<"convert-vector-to-rocdl", "ModuleOp"> {
   let summary = "Lower the operations from the vector dialect into the ROCDL "
                 "dialect";
   let constructor = "mlir::createConvertVectorToROCDLPass()";
+  let dependentDialects = ["ROCDL::ROCDLDialect"];
 }
 
 #endif // MLIR_CONVERSION_PASSES

mlir/include/mlir/Dialect/Affine/Passes.td

@@ -94,6 +94,7 @@ def AffineLoopUnrollAndJam : FunctionPass<"affine-loop-unroll-jam"> {
 def AffineVectorize : FunctionPass<"affine-super-vectorize"> {
   let summary = "Vectorize to a target independent n-D vector abstraction";
   let constructor = "mlir::createSuperVectorizePass()";
+  let dependentDialects = ["vector::VectorDialect"];
   let options = [
     ListOption<"vectorSizes", "virtual-vector-size", "int64_t",
                "Specify an n-D virtual vector size for vectorization",

mlir/include/mlir/Dialect/LLVMIR/LLVMDialect.h

@@ -15,6 +15,7 @@
 #define MLIR_DIALECT_LLVMIR_LLVMDIALECT_H_
 
 #include "mlir/Dialect/LLVMIR/LLVMTypes.h"
+#include "mlir/Dialect/OpenMP/OpenMPDialect.h"
 #include "mlir/IR/Dialect.h"
 #include "mlir/IR/Function.h"
 #include "mlir/IR/OpDefinition.h"

mlir/include/mlir/Dialect/LLVMIR/LLVMOpBase.td

@@ -19,6 +19,11 @@ include "mlir/IR/OpBase.td"
 def LLVM_Dialect : Dialect {
   let name = "llvm";
   let cppNamespace = "LLVM";
+
+  /// FIXME: at the moment this is a dependency of the translation to LLVM IR,
+  /// not really one of this dialect per-se.
+  let dependentDialects = ["omp::OpenMPDialect"];
+
   let hasRegionArgAttrVerify = 1;
   let hasOperationAttrVerify = 1;
   let extraClassDeclaration = [{

mlir/include/mlir/Dialect/LLVMIR/NVVMDialect.h

@@ -14,6 +14,7 @@
 #ifndef MLIR_DIALECT_LLVMIR_NVVMDIALECT_H_
 #define MLIR_DIALECT_LLVMIR_NVVMDIALECT_H_
 
+#include "mlir/Dialect/LLVMIR/LLVMDialect.h"
 #include "mlir/IR/Dialect.h"
 #include "mlir/IR/OpDefinition.h"
 #include "mlir/Interfaces/SideEffectInterfaces.h"