[PATCH] hidclass.sys: fixed size passed IOCTL_HID_GET_INPUT_REPORT (Coverity)

[Aric Stewart]

On 10/19/15 9:08 AM, Sebastian Lackner wrote:
> On 19.10.2015 15:04, Aric Stewart wrote:
>> On 10/17/15 8:34 AM, Marcus Meissner wrote:
>>> On Sat, Oct 17, 2015 at 03:31:24PM +0200, Sebastian Lackner wrote:
>>>> On 17.10.2015 14:52, Marcus Meissner wrote:
>>>>> 1327477 Wrong sizeof argument
>>>>>
>>>>> Signed-off-by: Marcus Meissner <marcus at jet.franken.de>
>>>>> ---
>>>>>    dlls/hidclass.sys/device.c | 2 +-
>>>>>    1 file changed, 1 insertion(+), 1 deletion(-)
>>>>>
>>>>> diff --git a/dlls/hidclass.sys/device.c b/dlls/hidclass.sys/device.c
>>>>> index e7e7e11..dcc7d3c 100644
>>>>> --- a/dlls/hidclass.sys/device.c
>>>>> +++ b/dlls/hidclass.sys/device.c
>>>>> @@ -277,7 +277,7 @@ static DWORD CALLBACK hid_device_thread(void *args)
>>>>>                packet->reportId = 0;
>>>>>    
>>>>>                irp = IoBuildDeviceIoControlRequest(IOCTL_HID_GET_INPUT_REPORT,
>>>>> -                device, NULL, 0, packet, sizeof(packet), TRUE, events[0],
>>>>> +                device, NULL, 0, packet, sizeof(*packet)+ext->preparseData->caps.InputReportByteLength, TRUE, events[0],
>>>>>                    &irp_status);
>>>>>    
>>>>>                irpsp = IoGetNextIrpStackLocation(irp);
>>>>>
>>>>
>>>> This looks wrong, you have to allocate a separate buffer, and then copy it (like in the code below).
>>>> I don't know why some of these buffers are HEAP_ZERO_MEMORY though, and others not? Adding Aric, he
>>>> might want to review this part again, especially since Coverity detected more issues in this code
>>>> (unnecessary assignment of "rc" for example).
>>>
>>> something seems wrong though here, yes.
>>>
>>> rc issue is "CID 1327478 Unused value"
>>>
>>> Ciao, Marcus
>>>
>>
>> What is happening here is that the HID_XFER_PACKET is not a variable length structure, it takes a pointer to the buffer.  I am taking a bit of a shortcut and instead of allocating the HID_XFER_PACKET structure and a separate buffer, I am allocating 1 block of memory that is both, the buffer just comes after the HID_XFER_PACKET structure.
>>
>> I am pretty sure Marcus' original patch is unnecessary.  I am not surprised that there are issues in this code area though. It when through a lot of revision and tweaking as I was getting things working and that often results in cruft.
>>
>> -aric
>>
> 
> I still don't think its correct. The IOCTL_HID_GET_INPUT_REPORT expects a raw data buffer,
> without any header in front of it, right? Then you would have to pass packet->reportBuffer
> (and the correct size) to IoBuildDeviceIoControlRequest, but that doesn't work because
> the buffer will be deallocated automatically after the request is done.
> 


The confusion here, and I remember it well, is that IOCTL_HID_GET_INPUT_REPORT is used twice, one from user->hidclass IOCTL and again from hidclass->minidriver IOCTL, and the IOCTL parameters are different depending on usage.

This case is the minidriver IOCTL: https://msdn.microsoft.com/en-us/library/windows/hardware/ff541126%28v=vs.85%29.aspx

The minidriver IOCTL uses the HID_XFER_PACKET structure.

-aric