Alan,

> > There is only one device connected to the host and it has only control
> > endpoint. This is not to intermix different control messages, see there
> > are some 10 TDs containing data from one control message, If the UHCI is
> > scheduling them vertically, every second TD in the frame causes NAK hence
> > there is only one TD getting trasacted per frame. If there is some delay
> > introduced between 2 TDs (may be by using a dummy TD), the device may be
> > able to finish the first one before it gets the second.
> 
> I'm not sure how 2.4 works.  Under 2.6 the UHCI driver uses horizontal
> scheduling, not vertical.  It also uses full-speed bandwidth 
> reclamation, so it would make many attempts to send the TDs during each 
> frame if your device was full-speed.  Since the device is low-speed, that 
> explains your difficulty.

In 2.4, horizontal/vertical scheduling is being controlled through a flag
in the code. By default it is breadth first in 2.4 also.
 
> If your device is so slow that it can't handle two TDs in a row, what 
> makes you think it will be able to handle two TDs that are separated by 
> some sort of dummy?

I am really not sure that it will speed up, but its sort of
experimentation to see if a dummy TD can be added and it can help in 
displaying things faster.
 
> Modifying the driver to create such dummy TDs would be a very non-trivial 
> task, and the result probably wouldn't be accepted into the kernel.

That's true, its just to see if this approach is feasible.


        Thanks
        Manoj



-------------------------------------------------------
This SF.net email is sponsored by: The SF.net Donation Program.
Do you like what SourceForge.net is doing for the Open
Source Community?  Make a contribution, and help us add new
features and functionality. Click here: http://sourceforge.net/donate/
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to