kern/68232: [patch] ugen(4) isochronous handling correction and tx support

[Matthew Gream]

The following reply was made to PR kern/68232; it has been noted by GNATS.

From: "Matthew Gream" <matthew.gream at pobox.com>
To: <hselasky at c2i.net>, <bug-followup at freebsd.org>
Cc:  
Subject: RE: kern/68232: [patch] ugen(4) isochronous handling correction and tx support
Date: Fri, 10 Jun 2005 17:43:07 +0100

 Hans,
 
 Thanks for the feedback. I think I mentioned in the defect report that the
 USB UGEN code looked like it needed a really good dose of attention to
 provide flexible and generic ISOC handling, my patch was admittedly a little
 limited, but I hoped something that could feed into anyone who had
 resource/time to take a good look at the USB code. I wasn't really happy
 with the ioctl's that I specified, they definitely do need improvement. I've
 moved on from that project, so am out of touch with what's been happening
 across BSD USB, and not able to give it any attention due to other
 priorities. You sound like you've got the issues covered, and hope you have
 the time/space to work your solutions in. If these comments, and the others
 you sent me, were attached to the defect report, they'd help inform anyone
 else working in the area.
 
 Best of luck,
 
 Matthew
 
  
 
 -----Original Message-----
 From: Hans Petter Selasky [mailto:hselasky at c2i.net] 
 Sent: 08 June 2005 16:55
 To: bug-followup at freebsd.org; matthew.gream at pobox.com
 Subject: Re: kern/68232: [patch] ugen(4) isochronous handling correction and
 tx support
 
 Hi,
 
 sce->isorate must be initialized somewhere. 
 Also it must be range checked.
 
 Is USBD_SET_RATE or USBD_GET_RATE documented somewhere.
 I think it is not so smart to pass the total number of 
 bytes per second. What is important is the frame size, 
 to synchronize data. Therefore USBD_SET_FRAME_SIZE and 
 USBD_GET_FRAME_SIZE would be better. And more generic! 
 Can you change this?
 
 Remember there is USB2.0 with 1000*8 frames per second.
 So USBD_SET_RATE might require two different values for 
 the same application, which is not very good.
 
 --HPS