[8/13](resend)hidclass.sys: Add a processing thread for HID devices
Aric Stewart
aric at codeweavers.com
Thu Sep 3 07:22:40 CDT 2015
>> +
>> +static DWORD CALLBACK hid_device_thread(void *args)
>> +{
>> + DEVICE_OBJECT *device = (DEVICE_OBJECT*)args;
>> +
>> + IRP *irp;
>> + IO_STATUS_BLOCK irp_status;
>> + IO_STACK_LOCATION *irpsp;
>> + DWORD rc;
>> + HANDLE events[2];
>> + NTSTATUS ntrc;
>> +
>> + BASE_DEVICE_EXTENSION *ext = device->DeviceExtension;
>> + events[0] = CreateEventA(NULL, FALSE, FALSE, NULL);
>> + events[1] = ext->halt_event;
>> +
>> + if (ext->information.Polled)
>> + {
>> + while(1)
>> + {
>> + HID_XFER_PACKET *packet;
>> + ResetEvent(events[0]);
>> +
>> + packet = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(*packet) + ext->preparseData->caps.InputReportByteLength);
>> + packet->reportBufferLen = ext->preparseData->caps.InputReportByteLength;
>> + packet->reportBuffer = ((BYTE*)packet) + sizeof(*packet);
>> + packet->reportId = 0;
>> +
>> + irp = IoBuildDeviceIoControlRequest(IOCTL_HID_GET_INPUT_REPORT,
>> + device, NULL, 0, packet, sizeof(packet), TRUE, events[0],
>> + &irp_status);
>> +
>> + irpsp = IoGetNextIrpStackLocation(irp);
>> + irpsp->CompletionRoutine = read_Completion;
>> + irpsp->Control = SL_INVOKE_ON_SUCCESS | SL_INVOKE_ON_ERROR;
>> +
>> + ntrc = IoCallDriver(device, irp);
>> +
>> + if (ntrc == STATUS_PENDING)
>> + rc = WaitForMultipleObjects(2, events, FALSE, INFINITE);
>> +
>> + if (irp->IoStatus.u.Status == STATUS_SUCCESS)
>> + {
>> + RingBuffer_Write(ext->ring_buffer, packet);
>
> I'm not opposed to what you have here, but a get-buffer/commit-buffer
> model would avoid an allocation and a memcpy. Although this would
> complicate simultaneous writes, if that's a concern.
>
Part of the trickiness here is that the the packet gets 'consumed' by the IoCallDriver/IoCompleteRequest process. It needs to be a stack pointer. So I end up having to allocate it on the stack and then copy the result out of it.
-aric
More information about the wine-devel
mailing list