Met vriendelijke groet / kind regards,
Mike Looijmans
System Expert
TOPIC Embedded Products B.V.
Materiaalweg 4, 5681 RJ Best
The Netherlands
T: +31 (0) 499 33 69 69
E: mike.looijmans@xxxxxxxxxxxxxxxxx
W: www.topicproducts.com
Please consider the environment before printing this e-mail
On 07-09-2020 09:44, Felipe Balbi wrote:
Hi,
Mike Looijmans <mike.looijmans@xxxxxxxx> writes:
Met vriendelijke groet / kind regards,
Mike Looijmans
System Expert
TOPIC Embedded Products B.V.
Materiaalweg 4, 5681 RJ Best
The Netherlands
T: +31 (0) 499 33 69 69
E: mike.looijmans@xxxxxxxxxxxxxxxxx
W: www.topicproducts.com
Please consider the environment before printing this e-mail
On 27-07-2020 12:23, Mark Brown wrote:
On Sun, Jul 26, 2020 at 09:10:39AM +0200, Mike Looijmans wrote:
On 23-07-2020 13:05, Mark Brown wrote:
Does the device actually support running without power so that's a thing
that can happen? _get_optional() should only ever be used for supplies
that may be physically absent.
It's the 5V VBUS power for the USB "plug" that's being controlled here. It
must turned on when the controller is in "host" mode. Some boards arrange
this in hardware through the PHY, and some just don't have any control at
all and have it permanently on or off. On a board where the 5V is controlled
using a GPIO line or an I2C chip, this patch is required to make it work.
That sounds like the driver should not be using _get_optional() then.
Making it mandatory would break most (read: all except Topic's) existing
boards as they won't have it in their devicetree. I'm perfectly okay with
that, but others might disagree.
you're perfectly okay with break all existing users of the driver?
That's a bit harsh
It turned out that "optional" when used for regulators means the
opposite of when used in clk context. For regulators, "optional" means
"don't supply a dummy regulator if there's none provided". So
get_optional will just fail when the supply isn't defined, while
get_regulator will just return a dummy in that case.
So the v4 patch which doesn't use "_get_optional" works for both cases
and doesn't break existing use(r)s.
--
Mike Looijmans
