On 11/19/19 12:43 PM, Hans Verkuil wrote:
> On 11/19/19 12:38 PM, Lucas Stach wrote:
>> Hi Hans,
>>
>> On Di, 2019-11-19 at 11:51 +0100, Hans Verkuil wrote:
>>> The irq_find_mapping() function can return 0 when called in the
>>> rmi_process_interrupt_requests() function.
>>>
>>> This causes a kernel crash. Check for a 0 value and skip calling
>>> handle_nested_irq() in that case.
>>>
>>> This was tested with the F54 function enabled on a Lenovo X1 Carbon.
>>>
>>> Signed-off-by: Hans Verkuil <hverkuil-cisco@xxxxxxxxx>
>>> Fixes: 24d28e4f1271 ("Input: synaptics-rmi4 - convert irq distribution to irq_domain")
>>
>> This is already fixed upstream by 549766ac2ac1
>> "Input: synaptics-rmi4 - clear IRQ enables for F54"
>
> Good news. I'm not subscribed to the linux-input ML, so I never saw that.
>
> Ah, I now see that I'm missing a whole bunch of patches that were added
> after v5.4-rc1. I'll test this again next week (I don't have access to my
> Lenovo at the moment).
Tested with v5.4-rc8 and I can confirm that this patch is not needed anymore
and can be dropped.
Regards,
Hans
>
> Regards,
>
> Hans
>
>>
>> Regards,
>> Lucas
>>
>>> ---
>>> drivers/input/rmi4/rmi_driver.c | 8 ++++++--
>>> 1 file changed, 6 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/drivers/input/rmi4/rmi_driver.c b/drivers/input/rmi4/rmi_driver.c
>>> index 772493b1f665..6085ec424a84 100644
>>> --- a/drivers/input/rmi4/rmi_driver.c
>>> +++ b/drivers/input/rmi4/rmi_driver.c
>>> @@ -154,8 +154,12 @@ static int rmi_process_interrupt_requests(struct rmi_device *rmi_dev)
>>> */
>>> mutex_unlock(&data->irq_mutex);
>>>
>>> - for_each_set_bit(i, data->irq_status, data->irq_count)
>>> - handle_nested_irq(irq_find_mapping(data->irqdomain, i));
>>> + for_each_set_bit(i, data->irq_status, data->irq_count) {
>>> + unsigned int irq = irq_find_mapping(data->irqdomain, i);
>>> +
>>> + if (irq)
>>> + handle_nested_irq(irq);
>>> + }
>>>
>>> if (data->input)
>>> input_sync(data->input);
>>
>
