On 11/11/2014 03:57 PM, Terje Bergström wrote:
On 11.11.2014 06:29, Alexandre Courbot wrote:
I think (after a quick look at devfreq's source) that you can avoid
polling altogether if you set polling_ms to 0 in your
devfreq_dev_profile instance. Then it is up to you to call
update_devfreq() from your custom governor whenever it sees fit.
ACTMON support seems to overlap between being a governor (which reacts
to ACTMON interrupts and calls update_devfreq() when needed) and part of
a devfreq_dev_profile (get_dev_status() needs to use the actmon
counters). If we keep your current design where the driver simply
controls a clock, you could have the ACTMON driver obtain that clock,
register its governor, create a non-polling devfreq_dev_profile that
controls that clock, and just call devfreq_add_device() with both. Then
we will have the benefit of being able to use ACTMON as well as the
performance and powersave governors on EMC, and switch policies
dynamically.
Another way to use it is that governor is just a governor. It takes in
load values and generates new target frequencies, and doesn't know
anything about HW.
Device profile is the one that enables threshold interrupts and disables
polling. Device profile receives the interrupt, retrieves new load
number from hardware, and calls update_devfreq().
This way we keep all HW specific code in device profile, and there's
potential to use a generic governor instead of writing your own.
I see several obstacles with this approach:
1) update_devfreq() is a governor private function (defined in
drivers/devfreq/governor.h) and it would need to be called from the
device profile.
2) devfreq events (start monitoring, stop monitoring, suspend, resume,
...) are processed by the governor. So it would still need access to the
actmon registers to honor these requests.
3) simple monitors like performance and powersave set the frequency (max
or min) once and for all when they are started, and don't need to be
called again afterwards. But this is probably what will happen if you
let the devfreq_dev_profile handle the ACTMON registers, since it can
make no assumption on when the governor expects to be invoked.
It would be good to have the feedback of devfreq's maintainers about
this (added them). How could an IP capable of monitoring activity
through counters and firing interrupts when these counters reach a
certain threshold be integrated nicely into devfreq? The functions seem
to overlap between the governor (reacting to specific events, in this
case ACTMON interrupts) and dev_profile (querying current status through
the counters), and it seems difficult to come with a design where the
governor and dev_profile are not tightly intertwined.
--
To unsubscribe from this list: send the line "unsubscribe linux-tegra" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
