2020-04-30 15:45:24

by Alain Volmat

[permalink] [raw]
Subject: [PATCH] i2c: fix missing pm_runtime_put_sync in i2c_device_probe

In case of the I2C client exposes the flag I2C_CLIENT_HOST_NOTIFY,
pm_runtime_get_sync is called in order to always keep active the
adapter. However later on, pm_runtime_put_sync is never called
within the function in case of an error. This commit add this
error handling.

Fixes: 72bfcee11cf8 ("i2c: Prevent runtime suspend of adapter when Host Notify is required")
Signed-off-by: Alain Volmat <[email protected]>
---
drivers/i2c/i2c-core-base.c | 22 ++++++++++++++++------
1 file changed, 16 insertions(+), 6 deletions(-)

diff --git a/drivers/i2c/i2c-core-base.c b/drivers/i2c/i2c-core-base.c
index 139aea351ffb..2e4560671183 100644
--- a/drivers/i2c/i2c-core-base.c
+++ b/drivers/i2c/i2c-core-base.c
@@ -338,8 +338,10 @@ static int i2c_device_probe(struct device *dev)
} else if (ACPI_COMPANION(dev)) {
irq = i2c_acpi_get_irq(client);
}
- if (irq == -EPROBE_DEFER)
- return irq;
+ if (irq == -EPROBE_DEFER) {
+ status = irq;
+ goto put_sync_adapter;
+ }

if (irq < 0)
irq = 0;
@@ -353,15 +355,19 @@ static int i2c_device_probe(struct device *dev)
*/
if (!driver->id_table &&
!i2c_acpi_match_device(dev->driver->acpi_match_table, client) &&
- !i2c_of_match_device(dev->driver->of_match_table, client))
- return -ENODEV;
+ !i2c_of_match_device(dev->driver->of_match_table, client)) {
+ status = -ENODEV;
+ goto put_sync_adapter;
+ }

if (client->flags & I2C_CLIENT_WAKE) {
int wakeirq;

wakeirq = of_irq_get_byname(dev->of_node, "wakeup");
- if (wakeirq == -EPROBE_DEFER)
- return wakeirq;
+ if (wakeirq == -EPROBE_DEFER) {
+ status = wakeirq;
+ goto put_sync_adapter;
+ }

device_init_wakeup(&client->dev, true);

@@ -408,6 +414,10 @@ static int i2c_device_probe(struct device *dev)
err_clear_wakeup_irq:
dev_pm_clear_wake_irq(&client->dev);
device_init_wakeup(&client->dev, false);
+put_sync_adapter:
+ if (client->flags & I2C_CLIENT_HOST_NOTIFY)
+ pm_runtime_put_sync(&client->adapter->dev);
+
return status;
}

--
2.17.1


2020-04-30 16:36:56

by Wolfram Sang

[permalink] [raw]
Subject: Re: [PATCH] i2c: fix missing pm_runtime_put_sync in i2c_device_probe

On Thu, Apr 30, 2020 at 05:43:21PM +0200, Alain Volmat wrote:
> In case of the I2C client exposes the flag I2C_CLIENT_HOST_NOTIFY,
> pm_runtime_get_sync is called in order to always keep active the
> adapter. However later on, pm_runtime_put_sync is never called
> within the function in case of an error. This commit add this
> error handling.
>
> Fixes: 72bfcee11cf8 ("i2c: Prevent runtime suspend of adapter when Host Notify is required")

Adding the patch author to CC.

> Signed-off-by: Alain Volmat <[email protected]>
> ---
> drivers/i2c/i2c-core-base.c | 22 ++++++++++++++++------
> 1 file changed, 16 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/i2c/i2c-core-base.c b/drivers/i2c/i2c-core-base.c
> index 139aea351ffb..2e4560671183 100644
> --- a/drivers/i2c/i2c-core-base.c
> +++ b/drivers/i2c/i2c-core-base.c
> @@ -338,8 +338,10 @@ static int i2c_device_probe(struct device *dev)
> } else if (ACPI_COMPANION(dev)) {
> irq = i2c_acpi_get_irq(client);
> }
> - if (irq == -EPROBE_DEFER)
> - return irq;
> + if (irq == -EPROBE_DEFER) {
> + status = irq;
> + goto put_sync_adapter;
> + }
>
> if (irq < 0)
> irq = 0;
> @@ -353,15 +355,19 @@ static int i2c_device_probe(struct device *dev)
> */
> if (!driver->id_table &&
> !i2c_acpi_match_device(dev->driver->acpi_match_table, client) &&
> - !i2c_of_match_device(dev->driver->of_match_table, client))
> - return -ENODEV;
> + !i2c_of_match_device(dev->driver->of_match_table, client)) {
> + status = -ENODEV;
> + goto put_sync_adapter;
> + }
>
> if (client->flags & I2C_CLIENT_WAKE) {
> int wakeirq;
>
> wakeirq = of_irq_get_byname(dev->of_node, "wakeup");
> - if (wakeirq == -EPROBE_DEFER)
> - return wakeirq;
> + if (wakeirq == -EPROBE_DEFER) {
> + status = wakeirq;
> + goto put_sync_adapter;
> + }
>
> device_init_wakeup(&client->dev, true);
>
> @@ -408,6 +414,10 @@ static int i2c_device_probe(struct device *dev)
> err_clear_wakeup_irq:
> dev_pm_clear_wake_irq(&client->dev);
> device_init_wakeup(&client->dev, false);
> +put_sync_adapter:
> + if (client->flags & I2C_CLIENT_HOST_NOTIFY)
> + pm_runtime_put_sync(&client->adapter->dev);
> +
> return status;
> }
>
> --
> 2.17.1
>


Attachments:
(No filename) (2.26 kB)
signature.asc (849.00 B)
Download all attachments

2020-05-04 06:04:40

by Jarkko Nikula

[permalink] [raw]
Subject: Re: [PATCH] i2c: fix missing pm_runtime_put_sync in i2c_device_probe

On 4/30/20 7:35 PM, Wolfram Sang wrote:
> On Thu, Apr 30, 2020 at 05:43:21PM +0200, Alain Volmat wrote:
>> In case of the I2C client exposes the flag I2C_CLIENT_HOST_NOTIFY,
>> pm_runtime_get_sync is called in order to always keep active the
>> adapter. However later on, pm_runtime_put_sync is never called
>> within the function in case of an error. This commit add this
>> error handling.
>>
>> Fixes: 72bfcee11cf8 ("i2c: Prevent runtime suspend of adapter when Host Notify is required")
>
Right, I was blind to see this.

Reviewed-by: Jarkko Nikula <[email protected]>

2020-05-05 14:32:32

by Wolfram Sang

[permalink] [raw]
Subject: Re: [PATCH] i2c: fix missing pm_runtime_put_sync in i2c_device_probe

On Thu, Apr 30, 2020 at 05:43:21PM +0200, Alain Volmat wrote:
> In case of the I2C client exposes the flag I2C_CLIENT_HOST_NOTIFY,
> pm_runtime_get_sync is called in order to always keep active the
> adapter. However later on, pm_runtime_put_sync is never called
> within the function in case of an error. This commit add this
> error handling.
>
> Fixes: 72bfcee11cf8 ("i2c: Prevent runtime suspend of adapter when Host Notify is required")
> Signed-off-by: Alain Volmat <[email protected]>

Applied to for-current, thanks!


Attachments:
(No filename) (541.00 B)
signature.asc (849.00 B)
Download all attachments