On Wed, 2016-08-17 at 11:05 +0300, Jarkko Nikula wrote:
> On 08/16/2016 05:07 PM, De Marchi, Lucas wrote:
> >
> > On Tue, 2016-08-16 at 17:00 +0300, Jarkko Nikula wrote:
> > >
> > > >
> > > > + */
> > > > + reg = dw_readl(dev, DW_IC_CON);
> > > > + dw_writel(dev, reg ^ DW_IC_CON_10BITADDR_MASTER,
> > > > DW_IC_CON);
> > > > +
> > > > + if ((dw_readl(dev, DW_IC_CON) &
> > > > DW_IC_CON_10BITADDR_MASTER) ==
> > > > + (reg & DW_IC_CON_10BITADDR_MASTER)) {
> > > > + dev->dynamic_tar_update_enabled = true;
> > > > + dev_dbg(dev->dev, "Dynamic TAR update
> > > > enabled");
> > > > + }
> > >
> > > Is this possible to move to i2c_dw_probe()? I guess the enabled
> > > status
> > > doesn't change runtime?
> >
> > It was actually useful at this place during development of this
> > patch
> > because we could check any unexpected change in behavior when
> > resuming.
> > We did catch a bug because of this and fixed.
> > I'm not sure if now it makes more sense to move to probe method.
> > I'd
> > leave it where it is, but I'm open to move it there.
> >
> Can you do a quick re-test that case to see does it change runtime?
> If
> it does then this needs a comment why there is need to do this check
> each time when HW is reinitialized. Otherwise there is chance
> someone
> may move this code to probe time in the future.
I already tested and it doesn't change. I'll move it to i2c_dw_probe()
then.
thanks
Lucas De Marchi
