2017-06-21 07:05:05

by Chun-Yi Lee

Subject: [PATCH v2] acpi: handle the acpi hotplug schedule error

Kernel should decrements the reference count of acpi device
when the scheduling of acpi hotplug work is failed, and
evaluates _OST to notify BIOS the failure.

v2:
To simplify the code. (Andy Shevchenko)

Cc: "Rafael J. Wysocki" <[email protected]>
Cc: Len Brown <[email protected]>
Cc: Michal Hocko <[email protected]>
Reviewed-by: Andy Shevchenko <[email protected]>
Signed-off-by: "Lee, Chun-Yi" <[email protected]>
---
drivers/acpi/bus.c | 9 +++++++--
1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/drivers/acpi/bus.c b/drivers/acpi/bus.c
index 34fbe02..91adb71 100644
--- a/drivers/acpi/bus.c
+++ b/drivers/acpi/bus.c
@@ -427,12 +427,17 @@ static void acpi_bus_notify(acpi_handle handle, u32 type, void *data)
(driver->flags & ACPI_DRIVER_ALL_NOTIFY_EVENTS))
driver->ops.notify(adev, type);

- if (hotplug_event && ACPI_SUCCESS(acpi_hotplug_schedule(adev, type)))
- return;
+ if (hotplug_event) {
+ if (ACPI_SUCCESS(acpi_hotplug_schedule(adev, type)))
+ return;
+ goto err_put_device;
+ }

acpi_bus_put_acpi_device(adev);
return;

+ err_put_device:
+ acpi_bus_put_acpi_device(adev);
err:
acpi_evaluate_ost(handle, type, ost_code, NULL);
}
--
2.10.2

2017-06-28 22:13:54

by Rafael J. Wysocki

Subject: Re: [PATCH v2] acpi: handle the acpi hotplug schedule error

On Wednesday, June 21, 2017 03:04:34 PM Lee, Chun-Yi wrote:
> Kernel should decrements the reference count of acpi device
> when the scheduling of acpi hotplug work is failed, and
> evaluates _OST to notify BIOS the failure.
>
> v2:
> To simplify the code. (Andy Shevchenko)
>
> Cc: "Rafael J. Wysocki" <[email protected]>
> Cc: Len Brown <[email protected]>
> Cc: Michal Hocko <[email protected]>
> Reviewed-by: Andy Shevchenko <[email protected]>
> Signed-off-by: "Lee, Chun-Yi" <[email protected]>

Well, what about:

if (!hotplug_event) {
acpi_bus_put_acpi_device(adev);
return;
}

if (ACPI_SUCCESS(acpi_hotplug_schedule(adev, type)))
return;

acpi_bus_put_acpi_device(adev);

err:
acpi_evaluate_ost(handle, type, ost_code, NULL);

This achieves the same as the patch below to my eyes, but looks
less convoluted IMO.


> ---
> drivers/acpi/bus.c | 9 +++++++--
> 1 file changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/acpi/bus.c b/drivers/acpi/bus.c
> index 34fbe02..91adb71 100644
> --- a/drivers/acpi/bus.c
> +++ b/drivers/acpi/bus.c
> @@ -427,12 +427,17 @@ static void acpi_bus_notify(acpi_handle handle, u32 type, void *data)
> (driver->flags & ACPI_DRIVER_ALL_NOTIFY_EVENTS))
> driver->ops.notify(adev, type);
>
> - if (hotplug_event && ACPI_SUCCESS(acpi_hotplug_schedule(adev, type)))
> - return;
> + if (hotplug_event) {
> + if (ACPI_SUCCESS(acpi_hotplug_schedule(adev, type)))
> + return;
> + goto err_put_device;
> + }
>
> acpi_bus_put_acpi_device(adev);
> return;
>
> + err_put_device:
> + acpi_bus_put_acpi_device(adev);
> err:
> acpi_evaluate_ost(handle, type, ost_code, NULL);
> }
>

2017-06-29 23:17:31

by joeyli

Subject: Re: [PATCH v2] acpi: handle the acpi hotplug schedule error

Hi Rafael,

Thanks for your review.

On Thu, Jun 29, 2017 at 12:06:20AM +0200, Rafael J. Wysocki wrote:
> On Wednesday, June 21, 2017 03:04:34 PM Lee, Chun-Yi wrote:
> > Kernel should decrements the reference count of acpi device
> > when the scheduling of acpi hotplug work is failed, and
> > evaluates _OST to notify BIOS the failure.
> >
> > v2:
> > To simplify the code. (Andy Shevchenko)
> >
> > Cc: "Rafael J. Wysocki" <[email protected]>
> > Cc: Len Brown <[email protected]>
> > Cc: Michal Hocko <[email protected]>
> > Reviewed-by: Andy Shevchenko <[email protected]>
> > Signed-off-by: "Lee, Chun-Yi" <[email protected]>
>
> Well, what about:
>
> if (!hotplug_event) {
> acpi_bus_put_acpi_device(adev);
> return;
> }
>
> if (ACPI_SUCCESS(acpi_hotplug_schedule(adev, type)))
> return;
>
> acpi_bus_put_acpi_device(adev);
>
> err:
> acpi_evaluate_ost(handle, type, ost_code, NULL);
>
> This achieves the same as the patch below to my eyes, but looks
> less convoluted IMO.
>

Yes, that's better. I will send v3 patch.

Joey Lee

>
> > ---
> > drivers/acpi/bus.c | 9 +++++++--
> > 1 file changed, 7 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/acpi/bus.c b/drivers/acpi/bus.c
> > index 34fbe02..91adb71 100644
> > --- a/drivers/acpi/bus.c
> > +++ b/drivers/acpi/bus.c
> > @@ -427,12 +427,17 @@ static void acpi_bus_notify(acpi_handle handle, u32 type, void *data)
> > (driver->flags & ACPI_DRIVER_ALL_NOTIFY_EVENTS))
> > driver->ops.notify(adev, type);
> >
> > - if (hotplug_event && ACPI_SUCCESS(acpi_hotplug_schedule(adev, type)))
> > - return;
> > + if (hotplug_event) {
> > + if (ACPI_SUCCESS(acpi_hotplug_schedule(adev, type)))
> > + return;
> > + goto err_put_device;
> > + }
> >
> > acpi_bus_put_acpi_device(adev);
> > return;
> >
> > + err_put_device:
> > + acpi_bus_put_acpi_device(adev);
> > err:
> > acpi_evaluate_ost(handle, type, ost_code, NULL);
> > }
> >
>