speedtch usbatm.c,1.20,1.21 usbatm.h,1.13,1.14

Roman Kagan rkagan at mail.ru
Wed May 4 10:14:30 EDT 2005

On Wed, May 04, 2005 at 10:09:08AM +0200, Duncan Sands wrote:
> I'm kind of confused here - what's wrong with having the usbatm core
> detect the endpoint type?

The problem is that I had implemented it, but it looked so ugly that I
decided to abandon it.

Traditionally the usb drivers were assumed to know what they were doing,
hence the pipe has its own notion of the transfer type, not directly
coupled to that stored in the endpoint descriptor.

Besides, that was your "do nothing" argument :)  Endpoint type detection
is a few tens of lines of code, plus some amount of code in the
minidriver to allow for bind-time selection of the endpoint, vs. two lines
assigning the bulk pipes to each channel.endpoint plus one more
assignment in the subdriver to override the rx pipe.


More information about the Usbatm mailing list