No space left on device

Hi
I am developing a project about stereo vision with my mini2440, use 2 USB
webcams  and OpenCV-2.0.
I can successfully cross compile my program and my libraries and moved them
on my board.
I have a problem with my webcams.
With one webcam (and a program using single camera) it works fine. But when
I connect two webcams via hub this error appears :

VIDIOC_STREAMON : No space left on device.
VIDIOC_QBUF : Invalid argument.

I searched a little about it and I found that it can be solved with PCI USB
Host.

My question is that is it possible to connect it to board or not?
I mean we have to install its driver?

thank you

You have to install the driver for the webcam. 
For no space left on device, 
post your df -lh output

I am also trying the same (but in a laptop, for opencv). I need 3 cameras
to work simultaneously. I am using integrated webcam of laptop and 2
external cameras.I was able to get all 3 cameras working if I change the
camera capture property resolution to 320 x 240, using
cvSetCaptureProperty() api. But this resolution is not enough as I need
minimum 640 x 480, which gives the error, 
libv4l2: error turning on stream: No space left on device
VIDIOC_STREAMON: No space left on device
output of dl -h is
Filesystem            Size  Used Avail Use% Mounted on
/dev/loop0             29G   16G   12G  57% /
udev                  1.9G  4.0K  1.9G   1% /dev
tmpfs                 765M  876K  765M   1% /run
none                  5.0M     0  5.0M   0% /run/lock
none                  1.9G  104K  1.9G   1% /run/shm
/dev/sda3             457G  392G   65G  86% /host

So how to get all 3 cameras working at more resolution? How to increase the
space?

I believe in this context "No space left on device" means that there is
insufficient USB bus bandwidth to stream from 3 cameras simultaneously, it
isn't referring to disk space.

I am encountering the same problem, does anyone have a hardware or software
solution for this? I'm working on a multi-camera input for my Cubieboard
2.0 ARM based board. Any recommendations would be welcome. Thanks.

I think imaging from three cameras will take a lot of data moving about. IS
this real-time or still images? If you can sequence the reads from the
cameras one at a time you can avoid the bottle neck. There is plenty of
room in RAM even on a Mini2440. I use the standard kernel, Debian, Python,
and X with tKinter and have 24 MBytes of RAM left for application space.

You can use a Tiny210SDK with the USB hub chip on board and 3 USB host
connectors. The 512M RAM and 1 GHZ might solve the problem. Plus there is
video/image transcoding hardware that could be accessed with some custom
routines.

If not, the new Tiny4412 is a quad core 1.5 GHz. Basically a Samsung Series
3 Chromebook with two more cores.

Other alternative that only costs time: Bare metal no OS C program. Or use
the sample uC/OSII real-time OS. Small with very low overhead but they have
some libs that could make reading the cameras a lot easier.

The restriction is the USB bus and its available bandwidth, *not* CPU
power. Three USB HC would solve the problem (one HC per camera), but maybe
not one USB HC with a HUB to attach three cameras to.

Hello Concern Person,
Greetings!!!

We are connecting usb web camera to linux centos 6.3 system but not
connecting we are getting following error:

libv4l2: error turning on stream: No space left on device

Please let us know your feedbcak for same.