Hello Laurent,
I can't use your patch since I am not in a Linux OS system, but I do
believe your patch is going to be able to work. In my system, I use
the following method to patch these alignment problems:
static uint8_t get_byte(uint32_t dword, int offset)
{
return (0x000000FF & (dword >> (offset * 8)));
}
:
//*(__le32*)&data[18] = cpu_to_le32(ctrl->dwMaxVideoFrameSize);
data[18] = get_byte(ctrl->dwMaxVideoFrameSize,0);
data[19] = get_byte(ctrl->dwMaxVideoFrameSize,1);
data[20] = get_byte(ctrl->dwMaxVideoFrameSize,2);
data[21] = get_byte(ctrl->dwMaxVideoFrameSize,3);
This is working solution and I carefully inspect these data on my USB
bus, it is fixed.
After I have done fixing the alignment issue, I found my MicroSoft
LifeCam NX-6000 still working on a very slow framerate. Fortunately by
examine the alignment issue I mentioned previously, I had accidentally
found that it is my system that caused the cam to work slowly. (I
added a breakpoint at the code that trying to set the format of the
cam, and that makes the hub_events(..) finished its execution by
stopping at my video streaming thread which has try_set_format
function, in this condition the video starts streaming at 30fps)
So, that's what I said, its my system that caused the streaming slow.
My system currently is a test system that once detect a UVC device, it
will start streaming for a while. There are no typical Linux device
driver architecture on my OS.
I just simulate important things on my system.
Hope that helps
-Brian
On Feb 13, 2008 3:45 PM, Laurent Pinchart <[EMAIL PROTECTED]> wrote:
> Hi Brian,
>
> On Friday 01 February 2008, Brian Pin wrote:
> > Hi Laurent, I found the answer why I get slow framerate on Microsoft CAM
> > NX600. See the following three lines: (uvc_set_video_ctrl)
> > *(__le16*)&data[16] = cpu_to_le16(ctrl->wDelay);
> > *(__le32*)&data[18] = cpu_to_le32(ctrl->dwMaxVideoFrameSize);
> > *(__le32*)&data[22] = cpu_to_le32(ctrl->dwMaxPayloadTransferSize);
> > When I write dwMaxVideoFrameSize to data[18], it over writes the
> > wDelay value and makes me like asking for delays in Microsoft CAM, so
> > the CAM do what I said, and it gives me slow frame rate, actually the
> > delay.
> >
> > The reason seems like when I try to cast &data[18] to a 32bit address,
> > my system transfers it to a 32bit aligned address, which covers the
> > address of wDelay. and that's why I get abnormal device behavior.
> >
> > Is it happening on other ARM platforms too?
>
> Have you been able to solve your problem ? Did the patch I sent you help ?
>
> Best regards,
>
> Laurent Pinchart
>
--
Best Regards
Brian
_______________________________________________
Linux-uvc-devel mailing list
Linux-uvc-devel@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/linux-uvc-devel
