[MLIR] Fix typo and expand gpu.host_register description.

[csigg]

Summary

See comments in https://reviews.llvm.org/D85631.

[herhut]

Thanks!

mlir/include/mlir/Dialect/GPU/GPUOps.td

]> {
   let parser = [{ return success(); }];
   let printer = [{ p << getOperationName(); }];
 }
 
 def GPU_HostRegisterOp : GPU_Op<"host_register">,
     Arguments<(ins AnyUnrankedMemRef:$value)> {
   let summary = "Registers a memref for access from device.";
   let description = [{
-    This op registers the host memory pointed to by a memref to be accessed from
-    a device.
+    This op maps the provided host buffer into the device address space.
+
+    This operation may not be supported in every environment, there is not yet a
+    way to check at runtime whether this feature is supported.
+
+    Writes from the host are guaranteed to be visible to device kernels that are
+    launched afterwards. Writes from the device are guaranteed to be visible on
+    the host after synchronizing with the device kernel completion.
   }];
 
   let assemblyFormat = "$value attr-dict `:` type($value)";
   let verifier = [{ return success(); }];
 }
 
 #endif // GPU_OPS

mlir/lib/Conversion/GPUCommon/ConvertLaunchFuncToRuntimeCalls.cpp

protected:
   FunctionCallBuilder hostRegisterCallBuilder = {
       "mgpuMemHostRegisterMemRef",
       llvmVoidType,
       {llvmIntPtrType /* intptr_t rank */,
        llvmPointerType /* void *memrefDesc */,
        llvmIntPtrType /* intptr_t elementSizeBytes */}};
 };
 
-/// A rewrite patter to convert gpu.host_register operations into a GPU runtime
+/// A rewrite pattern to convert gpu.host_register operations into a GPU runtime
 /// call. Currently it supports CUDA and ROCm (HIP).
 class ConvertHostRegisterOpToGpuRuntimeCallPattern
     : public ConvertOpToGpuRuntimeCallPattern<gpu::HostRegisterOp> {
 public:
   ConvertHostRegisterOpToGpuRuntimeCallPattern(LLVMTypeConverter &typeConverter)
       : ConvertOpToGpuRuntimeCallPattern<gpu::HostRegisterOp>(typeConverter) {}
 
 private: