On Tue, 11 Sep 2007, Philip Rakity wrote: > Alan, > > Agree that ehci.txt is not the right place. However submit_urb is a > linux function and it is not at all obvious that each buffer (in iso > mode) corresponds to a SEPARATE transaction. I have read the usb > spec (for echi) and all it talks about are alignment of buffers to 4K > pages.
Well, since we are discussing specifications let's start using the correct words. The "buffers" you have been talking about are really just parts of a single large buffer; they are described in the kernel by usb_iso_packet_descriptor structures. (In earlier versions of the kernel the same structures were called "usb_iso_frame_descriptor".) The name makes it clear that each structure refers to a single packet -- or more precisely, a single transaction. The older name was not appropriate for high-speed devices since they can have more than one transaction per frame. Furthermore, when operating in high-bandwidth mode they can have up to 3 transactions per microframe, all of which need to be described in a single usb_iso_packet_descriptor -- so in that respect the name still isn't perfect. If you would like to submit a patch adding some comments to include/linux/usb.h to explain all of this, please feel free to do so. You can send it to me first for suggestions or corrections, if you like. When it's ready to be submitted, post it on linux-usb-devel and CC: Greg KH <[EMAIL PROTECTED]>. Don't forget to follow the instructions in Documentation/SubmittingPatches. Alan Stern ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Linux-usb-users@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-users
