Re: [PATCH] greybus: lights: check return of get_channel_from_mode

[Alex Elder <elder@ieee.org>]

On 3/25/24 12:25 PM, Greg Kroah-Hartman wrote:
> On Thu, Mar 07, 2024 at 09:48:13AM +0000, Rui Miguel Silva wrote:
>> If channel for the given node is not found we return null from
>> get_channel_from_mode. Make sure we validate the return pointer
>> before using it in two of the missing places.
>>
>> This was originally reported in [0]:
>> Found by Linux Verification Center (linuxtesting.org) with SVACE.
>>
>> [0] https://lore.kernel.org/all/20240301190425.120605-1-m.lobanov@rosalinux.ru
>>
>> Fixes: 2870b52bae4c ("greybus: lights: add lights implementation")
>> Reported-by: Mikhail Lobanov <m.lobanov@rosalinux.ru>
>> Suggested-by: Mikhail Lobanov <m.lobanov@rosalinux.ru>
>> Suggested-by: Alex Elder <elder@ieee.org>
>> Signed-off-by: Rui Miguel Silva <rmfrfs@gmail.com>
>> ---
>>   drivers/staging/greybus/light.c | 6 +++++-
>>   1 file changed, 5 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/staging/greybus/light.c b/drivers/staging/greybus/light.c
>> index c6bd86a5335a..6f10b9e2c053 100644
>> --- a/drivers/staging/greybus/light.c
>> +++ b/drivers/staging/greybus/light.c
>> @@ -147,6 +147,9 @@ static int __gb_lights_flash_brightness_set(struct gb_channel *channel)
>>   		channel = get_channel_from_mode(channel->light,
>>   						GB_CHANNEL_MODE_TORCH);
>>   
>> +	if (!channel)
>> +		return -EINVAL;
>> +
>>   	/* For not flash we need to convert brightness to intensity */
>>   	intensity = channel->intensity_uA.min +
>>   			(channel->intensity_uA.step * channel->led->brightness);
>> @@ -549,7 +552,8 @@ static int gb_lights_light_v4l2_register(struct gb_light *light)
>>   	}
>>   
>>   	channel_flash = get_channel_from_mode(light, GB_CHANNEL_MODE_FLASH);
>> -	WARN_ON(!channel_flash);
>> +	if (WARN_ON(!channel_flash))
>> +		return -EINVAL;
> 
> We should NOT crash machines just because of this, the WARN_ON() should
> be removed and just properly handle the error please.

Greg, WARN_ON() doesn't normally crash the machine.  That said,
it's reasonable to remove the WARN_ON().

I think the purpose of the warning is that this is a case that
should "never happen," so if it does, it's making some noise.

The only caller is gb_lights_light_register(), and it checks
for an error and does "properly handle" the\ error (assuming
any error meaning "light has no flash" is correct).

Rui, please weigh in.

					-Alex


> 
> thanks,
> 
> greg k-h