On Mon, Apr 11, 2016 at 05:28:38PM +0200, Jan Glauber wrote:
> Switch to the i2c bus recovery framework using generic SCL recovery.
> If this fails try to reset the hardware. The recovery is triggered
> during START on timeout of the interrupt or failure to reach
> the START / repeated-START condition.
> 
> The START function is moved to xfer and while at it:
> - removed xfer debug message (i2c core already provides debugging)
> - removed length is zero check
> 
> Signed-off-by: Jan Glauber <jglauber@cavium.com>
> ---
>  drivers/i2c/busses/i2c-octeon.c | 178 +++++++++++++++++++++++++---------------
>  1 file changed, 111 insertions(+), 67 deletions(-)

Interesting, it got larger...

> 
> +/**
> + * octeon_i2c_write_int - read the TWSI_INT register

read_int

> +	int ret, retries = 2;

I don't think 'retries' makes sense here. On failure, you return
-EAGAIN, so the core will retry 'adapter->retries' times anyhow.

> -	if (length < 1)
> -		return -EINVAL;

So, the adapter support 0-length messages now? Or why was it there? I
have the feeling this is a seperate patch.

> +static void octeon_i2c_prepare_recovery(struct i2c_adapter *adap)
> +{
> +	struct octeon_i2c *i2c = i2c_get_adapdata(adap);
> +
> +	/*
> +	 * The stop resets the state machine, does not _transmit_ STOP unless
> +	 * engine was active.
> +	 */
> +	octeon_i2c_stop(i2c);
> +
> +	octeon_i2c_write_int(i2c, 0);

Maybe a comment why the delay?

> +	udelay(5);
> +}