[PATCH linux dev-4.10] drivers: i2c: fsi: Add proper abort method

[Eddie James]

On 10/19/2017 11:45 PM, Andrew Jeffery wrote:
> On Fri, 2017-10-13 at 10:20 -0500, Eddie James wrote:
>> On 10/13/2017 01:18 AM, Andrew Jeffery wrote:
>>> On Thu, 2017-10-12 at 15:42 -0500, Eddie James wrote:
>>>>> From: "Edward A. James" <eajames at us.ibm.com>
>>>>    
>>>> Driver wasn't cleaning up on timeout or in an error situation properly.
>>>> Need to do a full reset if we fail in order to re-stablish a good state
>>>> of the engine.
>>> Was there a particular use-case that triggered this behaviour or was this just
>>> found in general use? I feel a bit more context would be helpful here. It's a
>>> fairly sizeable patch and I'm surprised recovery was overlooked initially.
>> You'd have to hit either a timeout or some non-NACK error condition in
>> order to trigger the problem. Somehow NACK doesn't cause the I2C engine
>> to get into a bad state. And timeouts it seems to be somewhat
>> intermittent. Probably depends what state the bus is in when the driver
>> actually bails out.
> I asked off-list about finding someone to provide at least a Tested-by
> (and I'm hoping for a Reviewed-by) tag, did we get anyone?

Sivas ran it through the usual suite of power-on and reboot tests, but 
I'm not sure how extensively the i2c path is exercised there. I can ask 
him to add Tested-by when he returns from vacation.

Eddie

>
> Cheers,
>
> Andrew