Re: [Xen-devel] [RFC] PVFB: Add refresh period to XenStore parameters?

[Markus Armbruster <armbru@xxxxxxxxxx>]

Samuel Thibault <samuel.thibault@xxxxxxxxxxxxx> writes:

> Markus Armbruster, le Mon 05 May 2008 10:26:14 +0200, a Ãcrit :
>> Any particular reason for going through xenstore instead of the event
>> rings?
>
> Just because it exists.

Sufficient reason for climbing mountains, if you're so inclined, but
is it sane guidance for software development?

So far, PVFB uses xenstore *only* for device initialization and
shutdown.  For normal operations, we communicate through the event
rings.  Therefore, there's just one place dealing with events related
to device initialization and shutdown (in the fb frontend:
xenfb_backend_changed()), and just one place dealing with all other
events (fb frontend: xenfb_event_handler()).

I fear mixing in more communication channels without a technical need
just adds complexity for no real gain.

>> If I read your patch correctly, xenfb_read_frontend_fb_config() sets
>> request-update in xenstore depending on xenfb->ds->idle, during
>> startup.  The only place that changes it is xenfb_update(), which is
>> #ifdef CONFIG_STUBDOM.
>
> Err, no, it is not #ifdef CONFIG_STUBDOM.

Oops, you're right.  You'll have to update its comment, by the way.

>> What if ds->idle is true when xenfb_read_frontend_fb_config() runs
>> during startup of the ordinary (!CONFIG_STUBDOM) backend?  Can this
>> happen?
>
> Yes it can, when running a vnc server for instance.  And when later on a
> client connects, on next iteration of xenfb_update() the frontend will
> be notified.
>
>> Other than that, the change seems to affect only Mini-OS and
>> CONFIG_STUBDOM, which are both outside the area of my expertise (and
>> frankly, interest).
>
> Well, that could be used in the linux frontend to disable the timer too.
> That may interest e.g. laptop users.
>
> Samuel

The frontend should not send any events unless something's drawing on
the display.  And when something's drawing on the display, the guest,
and therefore the laptop, is already burning processor cycles, and
whether disabling sending updates saves any battery power is not
obvious.  Of course, if you can demonstrate it does...

Anyway, I'm not opposed to the feature.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel