Re: [PATCH 2/2] i7300_idle driver v1.55

[Andi Kleen]

Pallipadi, Venkatesh wrote:
> > > +#if CONFIG_I7300_IDLE_IOAT_CHANNEL
> > > +     device->common.chancnt--;
> > > +#endif
> > I still think this lone decrement looks fishy. Can there please be some
> > explanation how it exactly relates to the i7300 idle driver, where the
> > matching increment is, etc.?
>
> No. This is not a increment/decrement thing. It is basically telling other
> Users of IOAT that they have one IOAT channel less that they can use.
> The last IOAT channel is used by i7300 idle driver to get the throttling to
> work.

Ok then it should be made conditional on the i7300 actually be available
in the system?  It looks like you do it always no matter what chipset is in there.


> udelay may be more power efficient than polling. 

Isn't that function a one off anyways?

> > > +/* Check for known platforms with I/O-AT */
> > > +static int __init i7300_idle_platform_probe(void)
> > > +{
> > > +     int i;
> > > +
> > > +     fbd_dev = pci_get_bus_and_slot(MEMCTL_BUS, MEMCTL_DEVFN)
> > Is there a specific reason you cannot match this by pci vendor/devid
> > like all standard drivers do?
> >
> > If there is a good reason add a comment.
>
> They have to match vendor device and function. Its not unique to
> just vendor and device id.

It might be still better then to use the standard callback style and just ignore
devices that don't match?


> > ;
> > > +static void __exit i7300_idle_exit(void)
> > > +{
> > > +     idle_notifier_unregister(&i7300_idle_nb);
> > I still think this needs some kind of idle synchronization.
>
> This unregister uses atomic_notifier_chain_unregister() which
> uses RCU and handles the race conditions that way. 

I didn't do a full analysis, but I didn't think RCU protects
against the idle handler itself.  At least not standard RCU.
You would need to go through at least two quiescent periods,
not one as standard code.

There used to be an older
idle synchronization interface for this case (which was removed
at some point), perhaps it needs to be readded.

Or just use stop_machine(). That protects against idle too
and while it's a big hammer for unloading it should be ok.

-Andi
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/