[PATCH] extcon: intel-int3496: Fix oops on probe

[PATCH] extcon: intel-int3496: Fix oops on probe

[Hans de Goede]

Commit 41d600274fbf ("extcon: int3496: process id-pin first so that we
start with the right status") starts the work on the workqueue before
registration to make sure we've a valid cable state directly after
registration.

But that commit moves the queuing of the work to before we even alloc the
extcon, causing a NULL pointer deref in the worker.

This commit moves the queuing of the work to after we alloc the extcon,
fixing the NULL pointer deref.

Fixes: 41d600274fbf ("extcon: int3496: process id-pin first ...")
Cc: stable@vger.kernel.org
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
---
 drivers/extcon/extcon-intel-int3496.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/extcon/extcon-intel-int3496.c b/drivers/extcon/extcon-intel-int3496.c
index b23ee9d993a3..673bb26a9a2a 100644
--- a/drivers/extcon/extcon-intel-int3496.c
+++ b/drivers/extcon/extcon-intel-int3496.c
@@ -131,15 +131,15 @@ static int int3496_probe(struct platform_device *pdev)
 	if (IS_ERR(data->gpio_usb_mux))
 		dev_info(dev, "can't request USB MUX GPIO\n");
 
+	data->edev = devm_extcon_dev_allocate(dev, int3496_cable);
+	if (IS_ERR(data->edev))
+		return -ENOMEM;
+
 	/* process id-pin first so that we start with the right status */
 	queue_delayed_work(system_wq, &data->work, 0);
 	flush_delayed_work(&data->work);
 
 	/* register extcon device */
-	data->edev = devm_extcon_dev_allocate(dev, int3496_cable);
-	if (IS_ERR(data->edev))
-		return -ENOMEM;
-
 	ret = devm_extcon_dev_register(dev, data->edev);
 	if (ret < 0) {
 		dev_err(dev, "can't register extcon device: %d\n", ret);
-- 
2.14.3

Re: [PATCH] extcon: intel-int3496: Fix oops on probe

[Chanwoo Choi]

On 2018년 02월 12일 08:26, Hans de Goede wrote:
> Commit 41d600274fbf ("extcon: int3496: process id-pin first so that we
> start with the right status") starts the work on the workqueue before
> registration to make sure we've a valid cable state directly after
> registration.
> 
> But that commit moves the queuing of the work to before we even alloc the
> extcon, causing a NULL pointer deref in the worker.
> 
> This commit moves the queuing of the work to after we alloc the extcon,
> fixing the NULL pointer deref.
> 
> Fixes: 41d600274fbf ("extcon: int3496: process id-pin first ...")
> Cc: stable@vger.kernel.org
> Signed-off-by: Hans de Goede <hdegoede@redhat.com>
> ---
>  drivers/extcon/extcon-intel-int3496.c | 8 ++++----
>  1 file changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/extcon/extcon-intel-int3496.c b/drivers/extcon/extcon-intel-int3496.c
> index b23ee9d993a3..673bb26a9a2a 100644
> --- a/drivers/extcon/extcon-intel-int3496.c
> +++ b/drivers/extcon/extcon-intel-int3496.c
> @@ -131,15 +131,15 @@ static int int3496_probe(struct platform_device *pdev)
>  	if (IS_ERR(data->gpio_usb_mux))
>  		dev_info(dev, "can't request USB MUX GPIO\n");
>  
> +	data->edev = devm_extcon_dev_allocate(dev, int3496_cable);
> +	if (IS_ERR(data->edev))
> +		return -ENOMEM;
> +
>  	/* process id-pin first so that we start with the right status */
>  	queue_delayed_work(system_wq, &data->work, 0);
>  	flush_delayed_work(&data->work);
>  
>  	/* register extcon device */
> -	data->edev = devm_extcon_dev_allocate(dev, int3496_cable);
> -	if (IS_ERR(data->edev))
> -		return -ENOMEM;
> -
>  	ret = devm_extcon_dev_register(dev, data->edev);
>  	if (ret < 0) {
>  		dev_err(dev, "can't register extcon device: %d\n", ret);
> 

Instead, You better to send v3 after merging your patch[1] and this patch.
[1] extcon: int3496: process id-pin first so that we start with the right status

-- 
Best Regards,
Chanwoo Choi
Samsung Electronics

Re: [PATCH] extcon: intel-int3496: Fix oops on probe

[Hans de Goede]

Hi,

On 12-02-18 00:36, Chanwoo Choi wrote:
> On 2018년 02월 12일 08:26, Hans de Goede wrote:
>> Commit 41d600274fbf ("extcon: int3496: process id-pin first so that we
>> start with the right status") starts the work on the workqueue before
>> registration to make sure we've a valid cable state directly after
>> registration.
>>
>> But that commit moves the queuing of the work to before we even alloc the
>> extcon, causing a NULL pointer deref in the worker.
>>
>> This commit moves the queuing of the work to after we alloc the extcon,
>> fixing the NULL pointer deref.
>>
>> Fixes: 41d600274fbf ("extcon: int3496: process id-pin first ...")
>> Cc: stable@vger.kernel.org
>> Signed-off-by: Hans de Goede <hdegoede@redhat.com>
>> ---
>>   drivers/extcon/extcon-intel-int3496.c | 8 ++++----
>>   1 file changed, 4 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/extcon/extcon-intel-int3496.c b/drivers/extcon/extcon-intel-int3496.c
>> index b23ee9d993a3..673bb26a9a2a 100644
>> --- a/drivers/extcon/extcon-intel-int3496.c
>> +++ b/drivers/extcon/extcon-intel-int3496.c
>> @@ -131,15 +131,15 @@ static int int3496_probe(struct platform_device *pdev)
>>   	if (IS_ERR(data->gpio_usb_mux))
>>   		dev_info(dev, "can't request USB MUX GPIO\n");
>>   
>> +	data->edev = devm_extcon_dev_allocate(dev, int3496_cable);
>> +	if (IS_ERR(data->edev))
>> +		return -ENOMEM;
>> +
>>   	/* process id-pin first so that we start with the right status */
>>   	queue_delayed_work(system_wq, &data->work, 0);
>>   	flush_delayed_work(&data->work);
>>   
>>   	/* register extcon device */
>> -	data->edev = devm_extcon_dev_allocate(dev, int3496_cable);
>> -	if (IS_ERR(data->edev))
>> -		return -ENOMEM;
>> -
>>   	ret = devm_extcon_dev_register(dev, data->edev);
>>   	if (ret < 0) {
>>   		dev_err(dev, "can't register extcon device: %d\n", ret);
>>
> 
> Instead, You better to send v3 after merging your patch[1] and this patch.
> [1] extcon: int3496: process id-pin first so that we start with the right status

Ok, I've rebased my tree on top of 4.16-rc1 and prepared a v3. During the
rebase I noticed that a patch for the axp288-extcon driver which was supposed
to get dropped did end up in 4.16-rc1, so I will also (resend) a revert for that.

Also another extcon-axp288 driver patch seems to not have made it into 4.16-rc1,
so I will resend that too.

Regards,

Hans