Am Dienstag, 21. September 2004 20:37 schrieb Alan Stern:
> On Tue, 21 Sep 2004, Oliver Neukum wrote:
>
> > Is there enough contention that a global lock would
> > be impractical?
>
> That's hard to answer without actual measurement. But remember, a global
> lock would have to be acquired whenever a probe, disconnect, power change,
> or tree traversal occurs on any device in the system. Even things like
> probing for devices on different buses wouldn't be able to run in
> parallel.
Tree transversals could be kept happy with a rwsem.
The rest is really rare. And where it is not rare, I am not convinced
that uncontrolled paralellism is a good idea. IMHO you would soon
block on higher order allocations for hotplug anyway. Maybe you'd
even need rate limiting for hotplug outcalls.
I am suspecting overengineering here.
Regards
Oliver
-------------------------------------------------------
This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
Project Admins to receive an Apple iPod Mini FREE for your judgement on
who ports your project to Linux PPC the best. Sponsored by IBM.
Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
