On Sun, Oct 13, 2002, David Brownell <[EMAIL PROTECTED]> wrote: > > > Ok, I see your distinction between endpoint and URB. > > > > It unfortunately still doesn't work. > > Oh yes it does ... :)
I give up. > >>When that second driver submits its urb, assuming that needs > >>to reserve some bandwidth, the "is it available" logic will be > >>accounting for the first endpoint's bandwidth consumption. > >>Because it would never have been released. > > > > Yes it will be released. There are no URB's outstanding for that > > endpoint. > > But the endpoint was still busy, so bandwidth wasn't released. > Go back and reread what I wrote. > > Soon I'll get _really_ tired of repeating that part of what > I've been saying all along. When that happens, you can just > assume it still stays true. Same here. Same shit as the reference counting thread. You have some screwed up way of doing something in your head and you're too stubborn to realize it's dumb. > > Regardless, explaining this race condition to you is moot, because from > > a conceptual level, your implementation is still not reserving the > > bandwidth. > > It most certainly is, although you don't want to admit it. > > What you're talking about is splitting apart two aspects > of a reservation: making it, and using it. I've been > talking about the latter aspect. (A "wildlife reserve" > surely had wildlife before it some planning commission > labeled it as a reserve.) Stop abusing definitions. Yes, reserve has multiple definitions, but you know what it means in this case. Using it is moot. It's completely unneeded. All you need to do is reserve it. All of the information to later use it has already be figured out. > > I'll give you a *completely* seperate example: > > > > HID devices. They don't use their interrupt out pipe often, but the > > bandwidth needs to be reserved so they are never denied it. > > Such a device _could_ easily reserve bandwidth by keeping a transfer > queued at all times -- with no new APIs needed -- if it ever notices > bandwidth shortage as a problem. Huh? You can't. It's an out endpoint. You can't queue it and have it do nothing, it sends data. And don't tell me you're going to add some new method to queue an URB to an out endpoint where it doesn't send data. More importantly, drivers shouldn't need to jump through hoops like that. JE ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
