On Thursday 28 February 2002 01:33, David Brownell wrote:
> > I was thinking that this would be a good idea too (a usb subsystem wide
> > urb pool). ...
>
> A driver-wide cache of pre-initialized URBs could make some
> sense, but a USB-wide one would just duplicate kmem_cache
> functionality so far as I can tell.
So it would seems. But that does not mean that using the slab
cache is a bad idea, does it ?
Do you know how to use the slab cache ?
> > (right now everyone usually just creates them once, on
> > device init.)
>
> I wouldn't say "usually" there. Any driver using the synchronous
> calls (for bulk or control messages) does dynamic allocation.
> So do quite a few others ... as a rule, drivers spend more time
> idle than busy, so pre-allocation can just be a waste.
Even if it's only a small fixed number of URBs ?
Perhaps allocating urbs on open() instead of probe()
would be a good compromise.
Regards
Oliver
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
