Re: [PATCH V2] media: i2c: Add ADV761X support

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



[...]
>>>>>> The driver enables multiple interrupts on the chip, however, the
>>>>>> adv7604_isr callback doesn't seem to handle them correctly.
>>>>>> According to the docs:
>>>>>> "If an interrupt event occurs, and then a second interrupt event occurs
>>>>>> before the system controller has cleared or masked the first interrupt
>>>>>> event, the ADV7611 does not generate a second interrupt signal."
>>>>>>
>>>>>> However, the interrupt_service_routine doesn't account for that.
>>>>>> For example, in case fmt_change interrupt happens while
>>>>>> fmt_change_digital interrupt is being processed by the adv7604_isr
>>>>>> routine. If fmt_change status is set just before we clear
>>>>>> fmt_change_digital, we never clear fmt_change. Thus, we end up with
>>>>>> fmt_change interrupt missed and therefore further interrupts disabled.
>>>>>> I've tried to call the adv7604_isr routine in a loop and return from
>>>>>> the worlqueue only when all interrupt status bits are cleared. This did
>>>>>> help a bit, but sometimes I started getting lots of I2C read/write
>>>>>> errors for some reason.
>>>>>
>>>>> I'm not sure if there is much that can be done about this. The code
>>>>> reads the interrupt status, then clears the interrupts right after.
>>>>> There is always a race condition there since this isn't atomic ('read
>>>>> and clear'). Unless Lars-Peter has a better idea?
>>>>>
>>>>> What can be improved, though, is to clear not just the interrupts that
>>>>> were read, but all the interrupts that are unmasked. You are right, you
>>>>> could loose an interrupt that way.
>>>>
>>>> Wouldn't level-trigerred interrupts fix the issue ?
>>
>> In this case we need to disable the IRQ line in the IRQ handler and
>> re-enable it in the workqueue. (we can't call the interrupt service routine
>> from the interrupt context.)
> 
> Can't we just flag the interrupt in a non-threaded IRQ handler, acknowledge 
> the interrupt and then schedule work on a workqueue for the bottom half ?

Acknowledging the interrupt will require a non IRQ context, since it has to
do I2C transfers.

- Lars

--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html




[Index of Archives]     [Linux Input]     [Video for Linux]     [Gstreamer Embedded]     [Mplayer Users]     [Linux USB Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [Yosemite Backpacking]
  Powered by Linux