Re: [PATCH/RFC v6 05/36] leds: Improve setting brightness in a non sleeping way

From: Jacek Anaszewski
Date: Mon Aug 24 2015 - 04:11:57 EST

On 08/21/2015 10:42 PM, Andrew Lunn wrote:
This work queue from led-class.c is used for setting brightness, when
blink timer is on. Blinking is the functionality from the LED core,
so the work queue should also belong to the core. It should be moved
there along with led_timer_function, for consistency reasons. In view
of the above, using it for setting brightness by blocking drivers
would be correct from the architectural point of view.

So lets assume we are not doing all the changes in a single patch
set. Its too complex. You say the work queue should be in the
core. Lets put a work queue in the core to handle the blocking op, and
strip it out of the drivers. That allows you to get a lot of driver
patches merged.

Moving the work queue to the core is rather a cosmetic change and it is
not required for adapting it to setting brightness for blocking drivers.
I think that we could postpone the cosmetic changes to the moment when
other essential patches are merged.

Like I explained in the previous message, patches from 1 to 4 are
the minimum of the changes required for removing work queues from
drivers. I am aware that the changes may be hard to analyze, but the
things are entangled together and changes must be done in a few places
simultaneously. Those crucial changes are gathered in the patch [1].
I tried to explain them in the commit messages. Maybe my explanations
weren't comprehensive enough, if so, I'd be glad if you could indicate
what could be added/modified in the description to make it more clear.

You can then work on moving led_timer_function into the core, etc,
with a much smaller patch set, and hopefully in a number of small
steps which are easy to review.


Best Regards,
Jacek Anaszewski
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at