Hi Roman,
> > That readl() stuff may, the theory goes, also handle some
> > flakey hardware cases ... maybe fewer could be used, but
> > in this case I thought "better safe than sorry".
>
> ....
>
> Note: It is expected that writes to the Host Controller will
> be posted and have minimal impact on CPU performance.
>
> I think they had another design goal in mind ...;-)
There shouldn't be much; the critical paths shouldn't involve
hosts synchronizing with controllers about important state
like "is this hc/schedule/reqest/... running".
Though anyone with performance tools is welcome to report
any measured differences!
- Dave
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
http://lists.sourceforge.net/lists/listinfo/linux-usb-devel
