[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [bluetooth-dev] Stack design and some anoucements



On Fri, May 04, 2001 at 10:29:56AM -0700, Maksim Krasnyanskiy wrote:
> Greg,
> 
> >The one current advantage of using a tty device as a low level driver
> >is that the current usb bluetooth driver works with about 4 different
> >Linux Bluetooth stacks (yes there are more out there.)  
> Agree. However my goal was to design and develop general hardware
> independent interface for HCI core. People are working on HCI over PCI
> and stuff like that, so emulating TTY functionality would completely
> inefficient.  Current interface is clean and simple. 
> 
> Stack uses SKBs on all layers. Somebody in the list said that they are
> no different from kmem_alloc which is _not_ true.  SKBs are different
> because they are SLABified which means that there is a cache for SKB
> heads and allocations are optimized. Also SKBs provide lots of nice
> optimized functions and queues, so reinventing the wheel is not
> necessary :) 

I agree, skbs are very nice and fast.  And I like the design.  It's just
the compatibility with other stacks that I'm worried about.

> >I'm all for increased speed, and code sharing, but because of this
> >compatibility, I can't justify removing the current bluetooth.c driver
> >from the usb driver tree.
> Let's keep them both :).

We'll run into problems with hotplugging, which driver should get
loaded when a bluetooth device is plugged in?

> >I have a few minor usb driver suggestions that I'll send on later based
> >on skimming hci_usb.c.
> Great. Would be happy to discuss them.

Just sent a patch (actually 2) to the bluez-users list.

Crud, should have used bluez-devel, sorry.

greg k-h
-
To unsubscribe from this list: send the line "unsubscribe bluetooth-dev" in
the body of a message to majordomo@xxxxxxx.com