[MLIR][NVVM] Add support for nvvm.breakpoint Op (#107193)

This commit adds support for `nvvm.breakpoint` Op which lowers to the
PTX brkpt instruction. Also, added the respective tests in `nvvmir.mlir`

Author: schwarzschild-radius

mlir/include/mlir/Dialect/LLVMIR/NVVMOps.td

@@ -2101,6 +2101,23 @@ def NVVM_WgmmaMmaAsyncOp : NVVM_Op<"wgmma.mma_async",
   }];
 }
 
+//===----------------------------------------------------------------------===//
+// NVVM breakpoint Op
+//===----------------------------------------------------------------------===//
+
+def NVVM_Breakpoint : NVVM_Op<"breakpoint"> {
+  let summary = "Breakpoint Op";
+  let description = [{
+    Breakpoint suspends execution of the program for debugging.
+    [For more information, see PTX ISA](https://docs.nvidia.com/cuda/parallel-thread-execution/index.html#miscellaneous-instructions-brkpt)
+  }];
+  string llvmBuilder = [{
+    createIntrinsicCall(builder, llvm::Intrinsic::debugtrap);
+  }];
+
+  let assemblyFormat = "attr-dict";
+}
+
 //===----------------------------------------------------------------------===//
 // NVVM target attribute.
 //===----------------------------------------------------------------------===//

mlir/test/Target/LLVMIR/nvvmir.mlir

@@ -610,4 +610,12 @@ llvm.func @nvvm_fence_proxy_tensormap_generic_acquire(%addr : !llvm.ptr) {
   // CHECK: call void @llvm.nvvm.fence.proxy.tensormap_generic.acquire.sys(ptr {{%[0-9]+}}, i32 128)
   nvvm.fence.proxy.acquire #nvvm.mem_scope<sys> %addr, %c128
   llvm.return
-}
+}
+
+// -----
+// CHECK-LABEL: @nvvm_breakpoint
+llvm.func @nvvm_breakpoint() {
+  // CHECK: call void @llvm.debugtrap()
+  nvvm.breakpoint
+  llvm.return
+}