pwm: Read initial hardware state at request time

Drivers that support reading the hardware state (using ->get_state())
may want to rely on per-PWM data to do so. Defer reading the hardware
state for the first time until the PWM has been requested and after
drivers have had a chance to allocate per-PWM data.

Conceptually this is also a more natural place to read the hardware
state because the PWM core doesn't need to know the hardware state of a
PWM unless there is a user for it. This also ensures that the state is
read everytime a user requests a PWM. If the PWM changes between users
for some reason, the PWM core will reload the state from hardware and
keep its copy of the state up-to-date.

Tested-by: Enric Balletbo i Serra <enric.balletbo@collabora.com>
Tested-by: Michal Vokáč <michal.vokac@ysoft.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
This commit is contained in:
Thierry Reding 2019-10-21 12:51:56 +02:00
parent e42617b825
commit cfc4c189bc

View File

@ -114,6 +114,9 @@ static int pwm_device_request(struct pwm_device *pwm, const char *label)
}
}
if (pwm->chip->ops->get_state)
pwm->chip->ops->get_state(pwm->chip, pwm, &pwm->state);
set_bit(PWMF_REQUESTED, &pwm->flags);
pwm->label = label;
@ -283,9 +286,6 @@ int pwmchip_add_with_polarity(struct pwm_chip *chip,
pwm->hwpwm = i;
pwm->state.polarity = polarity;
if (chip->ops->get_state)
chip->ops->get_state(chip, pwm, &pwm->state);
radix_tree_insert(&pwm_tree, pwm->pwm, pwm);
}