On Thu, 2003-02-06 at 21:46, rahul shetty wrote: > We are writing a usb device driver for a webcam(twinkle). We are > encountering a host of problems in writing the read entry point. > We have opted for isochronous transfer
Unless you designed the webcam yourself and are in control of its behavior, you can not "opt" for this or that protocol. A device usually has very rigid access scheme, and the driver designer has not much of freedom to choose this or that method. > but are now not in a > position to proceed with it. We would appretiate any help which > will guide us through as to "How does one perform isochronous > transfer?". You can find a lot of existing Linux drivers that deal with Isoc transfers. Just look under linux/drivers/usb/ in your kernel tree. I don't think anyone can guide you through the development of your driver, though, because there are just too many reasons for failure. Driver development is quite demanding. You probably will be better off asking very specific questions. > Our code utilizes the URB structure for isochronous > transfer. But once the URB is submitted it gives us a pending > message and does not proceed from thereon. Probably the device does not send anything? Or you read from a wrong pipe? Or maybe there is a bug in your code? Hard to tell. It is quite common to use specialized hardware (CATC and friends) in such situations, to see what really happens on the wire. Dmitri
signature.asc
Description: This is a digitally signed message part
