| Message ID | 1577023823-9615-2-git-send-email-akinobu.mita@gmail.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | hwmon: (pwm-fan) add option to stop fan on shutdown | expand |
On Sun, Dec 22, 2019 at 11:10:22PM +0900, Akinobu Mita wrote: > The pwm-fan driver leaves the fun running when shutting down the system. > (On the other hand the gpio-fan driver stops it.) Seemms like we should have consistent behavior independent of what the underlying implementation uses. Is there actually a case you'd want to leave the fan on? It seems strange to disable in suspend and leave on in shutdown. Wouldn't the shutdown state depend if the PWM off state is high or low? IIRC, i.MX PWM has a quirk that the PWM disabled state is high. Doesn't it also depend on what the PWM driver does in shutdown? Rob > > This adds an optional property "disable-state-shutdown" in case someone > wants to stop the fun running when shutting down the system. > > Cc: Rob Herring <robh+dt@kernel.org> > Cc: Mark Rutland <mark.rutland@arm.com> > Cc: Kamil Debski <kamil@wypas.org> > Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> > Cc: Guenter Roeck <linux@roeck-us.net> > Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> > --- > drivers/hwmon/pwm-fan.c | 24 ++++++++++++++++++++++-- > 1 file changed, 22 insertions(+), 2 deletions(-) > > diff --git a/drivers/hwmon/pwm-fan.c b/drivers/hwmon/pwm-fan.c > index 42ffd2e..8775d37 100644 > --- a/drivers/hwmon/pwm-fan.c > +++ b/drivers/hwmon/pwm-fan.c > @@ -33,6 +33,7 @@ struct pwm_fan_ctx { > u8 pulses_per_revolution; > ktime_t sample_start; > struct timer_list rpm_timer; > + bool disable_state_shutdown; > > unsigned int pwm_value; > unsigned int pwm_fan_state; > @@ -292,6 +293,9 @@ static int pwm_fan_probe(struct platform_device *pdev) > > mutex_init(&ctx->lock); > > + ctx->disable_state_shutdown = > + of_property_read_bool(dev->of_node, "disable-state-shutdown"); > + > ctx->pwm = devm_of_pwm_get(dev, dev->of_node, NULL); > if (IS_ERR(ctx->pwm)) { > ret = PTR_ERR(ctx->pwm); > @@ -390,8 +394,7 @@ static int pwm_fan_probe(struct platform_device *pdev) > return 0; > } > > -#ifdef CONFIG_PM_SLEEP > -static int pwm_fan_suspend(struct device *dev) > +static int pwm_fan_disable(struct device *dev) > { > struct pwm_fan_ctx *ctx = dev_get_drvdata(dev); > struct pwm_args args; > @@ -418,6 +421,22 @@ static int pwm_fan_suspend(struct device *dev) > return 0; > } > > +static void pwm_fan_shutdown(struct platform_device *pdev) > +{ > + struct pwm_fan_ctx *ctx = platform_get_drvdata(pdev); > + > + if (!ctx->disable_state_shutdown) > + return; > + > + pwm_fan_disable(&pdev->dev); > +} > + > +#ifdef CONFIG_PM_SLEEP > +static int pwm_fan_suspend(struct device *dev) > +{ > + return pwm_fan_disable(dev); > +} > + > static int pwm_fan_resume(struct device *dev) > { > struct pwm_fan_ctx *ctx = dev_get_drvdata(dev); > @@ -455,6 +474,7 @@ MODULE_DEVICE_TABLE(of, of_pwm_fan_match); > > static struct platform_driver pwm_fan_driver = { > .probe = pwm_fan_probe, > + .shutdown = pwm_fan_shutdown, > .driver = { > .name = "pwm-fan", > .pm = &pwm_fan_pm, > -- > 2.7.4 >
2020年1月9日(木) 1:13 Rob Herring <robh@kernel.org>: > > On Sun, Dec 22, 2019 at 11:10:22PM +0900, Akinobu Mita wrote: > > The pwm-fan driver leaves the fun running when shutting down the system. > > (On the other hand the gpio-fan driver stops it.) > > Seemms like we should have consistent behavior independent of what the > underlying implementation uses. Is there actually a case you'd want to > leave the fan on? It seems strange to disable in suspend and leave on in > shutdown. I agree. I was trying to keep the current behavior unchanged, so I added "disable-state-shutdown" property. But I can't think of any case we want to leave the fun on in shutdown. So it's better to change the shutdown behavior and remove the option completely or add "retain-state-shutdown" property instead. (The "retain-state-shutdown" property is inspired by gpio-leds) > Wouldn't the shutdown state depend if the PWM off state is high or low? > IIRC, i.MX PWM has a quirk that the PWM disabled state is high. Doesn't It could be possible to affect the shutdown behavior for pwm-fan. There are three i.MX PWM drivers (pwm-imx1, pwm-imx27, and pwm-tpm). Do you remember which one actually have such limitation? Maybe it should be handled by the PWM controller/chip driver and PWM core. From the perspective of PWM user drivers for now, we have nothing to do other than setting duty cycle zero and then disable PWM for stopping the pwm-fan. > it also depend on what the PWM driver does in shutdown? I think so. But as far as I can see, no PWM drivers implement shutdown callback.
On Thu, Jan 9, 2020 at 11:07 PM Akinobu Mita <akinobu.mita@gmail.com> wrote: > > 2020年1月9日(木) 1:13 Rob Herring <robh@kernel.org>: > > > > On Sun, Dec 22, 2019 at 11:10:22PM +0900, Akinobu Mita wrote: > > > The pwm-fan driver leaves the fun running when shutting down the system. > > > (On the other hand the gpio-fan driver stops it.) > > > > Seemms like we should have consistent behavior independent of what the > > underlying implementation uses. Is there actually a case you'd want to > > leave the fan on? It seems strange to disable in suspend and leave on in > > shutdown. > > I agree. I was trying to keep the current behavior unchanged, so I added > "disable-state-shutdown" property. But I can't think of any case we want > to leave the fun on in shutdown. > > So it's better to change the shutdown behavior and remove the option > completely or add "retain-state-shutdown" property instead. > (The "retain-state-shutdown" property is inspired by gpio-leds) I would just turn off the fan in shutdown and see if anyone complains. > > Wouldn't the shutdown state depend if the PWM off state is high or low? > > IIRC, i.MX PWM has a quirk that the PWM disabled state is high. Doesn't > > It could be possible to affect the shutdown behavior for pwm-fan. > There are three i.MX PWM drivers (pwm-imx1, pwm-imx27, and pwm-tpm). > Do you remember which one actually have such limitation? No. I believe the fix was to use pinctrl modes to force the state. And I think the issue was for suspend rather than shutdown (but it seems unlikely you'd want the fan off in suspend and on in shutdown). > > Maybe it should be handled by the PWM controller/chip driver and PWM core. > From the perspective of PWM user drivers for now, we have nothing to do > other than setting duty cycle zero and then disable PWM for stopping the > pwm-fan. > > > it also depend on what the PWM driver does in shutdown? > > I think so. But as far as I can see, no PWM drivers implement shutdown > callback. Do they need to? I'd assume most SoC's are powered off or put in reset which makes the PWN pin go low. Rob
diff --git a/drivers/hwmon/pwm-fan.c b/drivers/hwmon/pwm-fan.c index 42ffd2e..8775d37 100644 --- a/drivers/hwmon/pwm-fan.c +++ b/drivers/hwmon/pwm-fan.c @@ -33,6 +33,7 @@ struct pwm_fan_ctx { u8 pulses_per_revolution; ktime_t sample_start; struct timer_list rpm_timer; + bool disable_state_shutdown; unsigned int pwm_value; unsigned int pwm_fan_state; @@ -292,6 +293,9 @@ static int pwm_fan_probe(struct platform_device *pdev) mutex_init(&ctx->lock); + ctx->disable_state_shutdown = + of_property_read_bool(dev->of_node, "disable-state-shutdown"); + ctx->pwm = devm_of_pwm_get(dev, dev->of_node, NULL); if (IS_ERR(ctx->pwm)) { ret = PTR_ERR(ctx->pwm); @@ -390,8 +394,7 @@ static int pwm_fan_probe(struct platform_device *pdev) return 0; } -#ifdef CONFIG_PM_SLEEP -static int pwm_fan_suspend(struct device *dev) +static int pwm_fan_disable(struct device *dev) { struct pwm_fan_ctx *ctx = dev_get_drvdata(dev); struct pwm_args args; @@ -418,6 +421,22 @@ static int pwm_fan_suspend(struct device *dev) return 0; } +static void pwm_fan_shutdown(struct platform_device *pdev) +{ + struct pwm_fan_ctx *ctx = platform_get_drvdata(pdev); + + if (!ctx->disable_state_shutdown) + return; + + pwm_fan_disable(&pdev->dev); +} + +#ifdef CONFIG_PM_SLEEP +static int pwm_fan_suspend(struct device *dev) +{ + return pwm_fan_disable(dev); +} + static int pwm_fan_resume(struct device *dev) { struct pwm_fan_ctx *ctx = dev_get_drvdata(dev); @@ -455,6 +474,7 @@ MODULE_DEVICE_TABLE(of, of_pwm_fan_match); static struct platform_driver pwm_fan_driver = { .probe = pwm_fan_probe, + .shutdown = pwm_fan_shutdown, .driver = { .name = "pwm-fan", .pm = &pwm_fan_pm,
The pwm-fan driver leaves the fun running when shutting down the system. (On the other hand the gpio-fan driver stops it.) This adds an optional property "disable-state-shutdown" in case someone wants to stop the fun running when shutting down the system. Cc: Rob Herring <robh+dt@kernel.org> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Kamil Debski <kamil@wypas.org> Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Cc: Guenter Roeck <linux@roeck-us.net> Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> --- drivers/hwmon/pwm-fan.c | 24 ++++++++++++++++++++++-- 1 file changed, 22 insertions(+), 2 deletions(-)