[Intel-gfx] [PATCH 2/2] drm/i915: abstract i2c bit banging fallback in gmbus xfer

[Jani Nikula]

On Tue, 01 Dec 2015, Ville Syrjälä <ville.syrjala at linux.intel.com> wrote:
> On Tue, Dec 01, 2015 at 04:29:26PM +0200, Jani Nikula wrote:
>> Choose between i2c bit banging and gmbus in a new higher level function,
>> and let the i2c core retry the first time we fall back to bit banging.
>> 
>> Signed-off-by: Jani Nikula <jani.nikula at intel.com>
>> ---
>>  drivers/gpu/drm/i915/intel_i2c.c | 39 +++++++++++++++++++++++++--------------
>>  1 file changed, 25 insertions(+), 14 deletions(-)
>> 
>> diff --git a/drivers/gpu/drm/i915/intel_i2c.c b/drivers/gpu/drm/i915/intel_i2c.c
>> index ccb522c176bd..e26e22a72e3b 100644
>> --- a/drivers/gpu/drm/i915/intel_i2c.c
>> +++ b/drivers/gpu/drm/i915/intel_i2c.c
>> @@ -472,9 +472,7 @@ gmbus_xfer_index_read(struct drm_i915_private *dev_priv, struct i2c_msg *msgs)
>>  }
>>  
>>  static int
>> -gmbus_xfer(struct i2c_adapter *adapter,
>> -	   struct i2c_msg *msgs,
>> -	   int num)
>> +do_gmbus_xfer(struct i2c_adapter *adapter, struct i2c_msg *msgs, int num)
>>  {
>>  	struct intel_gmbus *bus = container_of(adapter,
>>  					       struct intel_gmbus,
>> @@ -483,14 +481,6 @@ gmbus_xfer(struct i2c_adapter *adapter,
>>  	int i = 0, inc, try = 0;
>>  	int ret = 0;
>>  
>> -	intel_display_power_get(dev_priv, POWER_DOMAIN_GMBUS);
>> -	mutex_lock(&dev_priv->gmbus_mutex);
>> -
>> -	if (bus->force_bit) {
>> -		ret = i2c_bit_algo.master_xfer(adapter, msgs, num);
>> -		goto out;
>> -	}
>> -
>>  retry:
>>  	I915_WRITE(GMBUS0, bus->reg0);
>>  
>> @@ -585,13 +575,34 @@ timeout:
>>  		 bus->adapter.name, bus->reg0 & 0xff);
>>  	I915_WRITE(GMBUS0, 0);
>>  
>> -	/* Hardware may not support GMBUS over these pins? Try GPIO bitbanging instead. */
>> +	/*
>> +	 * Hardware may not support GMBUS over these pins? Try GPIO bitbanging
>> +	 * instead. Use EAGAIN to have i2c core retry.
>> +	 */
>>  	bus->force_bit = 1;
>> -	ret = i2c_bit_algo.master_xfer(adapter, msgs, num);
>> +	ret = -EAGAIN;
>
> This does mean we're left to the mercy of the timeout handling in the
> i2c core. That is if our gmbus attempt takes too long, we may never
> get a chance to retry with bit banging. Not sure if that's a real
> concern or not.

IIUC the longest it can take to transfer 4 bytes without timing out is
just under 50 ms. The i2c core use a default timeout of 1000 ms. So if
we first succesfully, but slowly manage to transfer quite a bit (well,
at least 80 bytes, almost but not quite timing out), and then time out,
i2c core won't try again.

I do not think this is a real concern, but if it is, we can increase the
adapter->timeout in our gmbus setup code.

BR,
Jani.



>
> Otherwise lgtm.
>
>>  
>>  out:
>> -	mutex_unlock(&dev_priv->gmbus_mutex);
>> +	return ret;
>> +}
>> +
>> +static int
>> +gmbus_xfer(struct i2c_adapter *adapter, struct i2c_msg *msgs, int num)
>> +{
>> +	struct intel_gmbus *bus = container_of(adapter, struct intel_gmbus,
>> +					       adapter);
>> +	struct drm_i915_private *dev_priv = bus->dev_priv;
>> +	int ret;
>> +
>> +	intel_display_power_get(dev_priv, POWER_DOMAIN_GMBUS);
>> +	mutex_lock(&dev_priv->gmbus_mutex);
>> +
>> +	if (bus->force_bit)
>> +		ret = i2c_bit_algo.master_xfer(adapter, msgs, num);
>> +	else
>> +		ret = do_gmbus_xfer(adapter, msgs, num);
>>  
>> +	mutex_unlock(&dev_priv->gmbus_mutex);
>>  	intel_display_power_put(dev_priv, POWER_DOMAIN_GMBUS);
>>  
>>  	return ret;
>> -- 
>> 2.1.4
>> 
>> _______________________________________________
>> Intel-gfx mailing list
>> Intel-gfx at lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Jani Nikula, Intel Open Source Technology Center