[PATCH v5 2/3] I2C: mediatek: Add driver for MediaTek I2C controller

[eddie.huang@mediatek.com (Eddie Huang)]

Hi Sascha,

On Mon, 2015-03-30 at 19:23 +0200, Sascha Hauer wrote:
> On Mon, Mar 30, 2015 at 04:14:12PM +0800, Eddie Huang wrote:
> > Hi Sascha,
> > 
> > > 
> > > [...]
> > > 
> > > > +	if (i2c->speed_hz > 400000)
> > > > +		control_reg |= I2C_CONTROL_RS;
> > > > +	if (i2c->op == I2C_MASTER_WRRD)
> > > > +		control_reg |= I2C_CONTROL_DIR_CHANGE | I2C_CONTROL_RS;
> > > > +	mtk_i2c_writew(control_reg, i2c, OFFSET_CONTROL);
> > > > +
> > > > +	/* set start condition */
> > > > +	if (i2c->speed_hz <= 100000)
> > > > +		mtk_i2c_writew(I2C_ST_START_CON, i2c, OFFSET_EXT_CONF);
> > > > +	else
> > > > +		mtk_i2c_writew(I2C_FS_START_CON, i2c, OFFSET_EXT_CONF);
> > > > +
> > > > +	if (~control_reg & I2C_CONTROL_RS)
> > > > +		mtk_i2c_writew(I2C_DELAY_LEN, i2c, OFFSET_DELAY_LEN);
> > > 
> > > speed <= 400000 here to make this more obvious?
> > There are two cases, not only speed<=400000, but I2C_MASTER_WRRD. I tend
> > to keep it.
> 
> Still it looks strange. You only ever write this default value to the
> register. Putting this register write under an if() seems bogus since
> the same value will be in the register the next time this code is
> executed. It looks like you should move this register write to some
> initialization function.
OK, move to mtk_i2c_init_hw function

> 
> > > > +
> > > > +	/* Enable interrupt */
> > > > +	mtk_i2c_writew(I2C_HS_NACKERR | I2C_ACKERR | I2C_TRANSAC_COMP,
> > > > +		i2c, OFFSET_INTR_MASK);
> > > 
> > > Why do you enable/disable interrupts for each transfer? Enabling them
> > > once and just acknowledge them in the interrupt handler should be
> > > enough.
> > This can avoid unwanted I2C interrupt. For example, I2C transfer error,
> > and cause timeout, I2C driver report error to caller. Then I2C error
> > interrupt happen.
> 
> So isn't the same unwanted interrupt then just delayed until you enable
> the interrupts again? Is this something that really happens or just
> something you think that might happen?
> 
Clear interrupt status before enable interrupt, so won't get unwanted
interrupt again. I just think this might happen, and it's not harmful to
enable/disable interrupt in transfer function and can get extra benefit
to avoid unnecessary interrupt.  Tegra I2C driver i2c-tegra.c also do
the same thing.

Eddie