“Shashidhar” wrote in message
news:xxxxx@ntdev…
> The maximum transfer size (AFAIK) same as maximum packet size of the end
> point descriptors(Correct me if I am wrong).
You’re wrong.
> But this is not the limit I
> am encountering, if I submit an URB with data greater than 4k then the
> request is failed by lower driver.
Read my previous post. The default limit is 4K, you are exceeding that.
> This has nothing to do with packet size from the descriptor.
Right. That is why you are failing.
> As of now I am splitting the request into 1k chunks and I feel that is
> safe too.
That is fine. But you could also split them into 4K chunks and that would
work.
Marc Reinig
System Solutions
> >-----Original Message-----
> >From: Marc Reinig [mailto:xxxxx@nowwhere.com]
> >Sent: Friday, October 04, 2002 9:42 PM
> >To: NT Developers Interest List
> >Subject: [ntdev] Re: [ntdev]:To change page size
> >
> >MaximumTransferSize is used to specify the maximum transfer request
> > size for
> >pipe. It defaults to 4096. It can be changed and is not directly
> >related to page size.
> >
> >Search for MaximumTransferSize in the DDK to see how to change it.
> >
> >Why do you want to circumvent this behavior? Very little benefit will
> >be gained if any, and if the size is increased by a large amount, the
> >performance of other devices on the USB may suffer. Just divide up
> >your
> >desired size into blocks of MaximumTransferSize and feed them to the
> USB.
> >> This was to verify the behavior of the USB host control driver
> >> which does not allow read/write above 4096K per URB. I suspected
> >> this to be dependent on the page size and if it is so what is
> >> the reason behind it. And is there anyway to override or
> >> circumvent this behavior of USBHCD.sys