Re: [PATCH v3 1/3] i3c: fix i2c and i3c scl rate by bus mode

[Boris Brezillon]

On Wed, 12 Jun 2019 11:16:34 +0000
Vitor Soares <Vitor.Soares@xxxxxxxxxxxx> wrote:

> From: Boris Brezillon <boris.brezillon@xxxxxxxxxxxxx>
> Date: Wed, Jun 12, 2019 at 07:15:33
> 
> > On Tue, 11 Jun 2019 16:06:43 +0200
> > Vitor Soares <Vitor.Soares@xxxxxxxxxxxx> wrote:
> >   
> > > Currently the I3C framework limits SCL frequency to FM speed when
> > > dealing with a mixed slow bus, even if all I2C devices are FM+ capable.
> > > 
> > > The core was also not accounting for I3C speed limitations when
> > > operating in mixed slow mode and was erroneously using FM+ speed as the
> > > max I2C speed when operating in mixed fast mode.
> > > 
> > > Fixes: 3a379bbcea0a ("i3c: Add core I3C infrastructure")
> > > Signed-off-by: Vitor Soares <vitor.soares@xxxxxxxxxxxx>
> > > Cc: Boris Brezillon <bbrezillon@xxxxxxxxxx>
> > > Cc: <stable@xxxxxxxxxxxxxxx>
> > > Cc: <linux-kernel@xxxxxxxxxxxxxxx>
> > > ---
> > > Changes in v3:
> > >   Change dev_warn() to dev_dbg()
> > > 
> > > Changes in v2:
> > >   Enhance commit message
> > >   Add dev_warn() in case user-defined i2c rate doesn't match LVR constraint
> > >   Add dev_warn() in case user-defined i3c rate lower than i2c rate
> > > 
> > >  drivers/i3c/master.c | 61 +++++++++++++++++++++++++++++++++++++++++-----------
> > >  1 file changed, 48 insertions(+), 13 deletions(-)
> > > 
> > > diff --git a/drivers/i3c/master.c b/drivers/i3c/master.c
> > > index 5f4bd52..f8e580e 100644
> > > --- a/drivers/i3c/master.c
> > > +++ b/drivers/i3c/master.c
> > > @@ -91,6 +91,12 @@ void i3c_bus_normaluse_unlock(struct i3c_bus *bus)
> > >  	up_read(&bus->lock);
> > >  }
> > >  
> > > +static struct i3c_master_controller *
> > > +i3c_bus_to_i3c_master(struct i3c_bus *i3cbus)
> > > +{
> > > +	return container_of(i3cbus, struct i3c_master_controller, bus);
> > > +}
> > > +
> > >  static struct i3c_master_controller *dev_to_i3cmaster(struct device *dev)
> > >  {
> > >  	return container_of(dev, struct i3c_master_controller, dev);
> > > @@ -565,20 +571,48 @@ static const struct device_type i3c_masterdev_type = {
> > >  	.groups	= i3c_masterdev_groups,
> > >  };
> > >  
> > > -int i3c_bus_set_mode(struct i3c_bus *i3cbus, enum i3c_bus_mode mode)
> > > +int i3c_bus_set_mode(struct i3c_bus *i3cbus, enum i3c_bus_mode mode,
> > > +		     unsigned long max_i2c_scl_rate)
> > >  {
> > > -	i3cbus->mode = mode;
> > >  
> > > -	if (!i3cbus->scl_rate.i3c)
> > > -		i3cbus->scl_rate.i3c = I3C_BUS_TYP_I3C_SCL_RATE;
> > > +	struct i3c_master_controller *master = i3c_bus_to_i3c_master(i3cbus);
> > >  
> > > -	if (!i3cbus->scl_rate.i2c) {
> > > -		if (i3cbus->mode == I3C_BUS_MODE_MIXED_SLOW)
> > > -			i3cbus->scl_rate.i2c = I3C_BUS_I2C_FM_SCL_RATE;
> > > -		else
> > > -			i3cbus->scl_rate.i2c = I3C_BUS_I2C_FM_PLUS_SCL_RATE;
> > > +	i3cbus->mode = mode;
> > > +
> > > +	switch (i3cbus->mode) {
> > > +	case I3C_BUS_MODE_PURE:
> > > +		if (!i3cbus->scl_rate.i3c)
> > > +			i3cbus->scl_rate.i3c = I3C_BUS_TYP_I3C_SCL_RATE;
> > > +		break;
> > > +	case I3C_BUS_MODE_MIXED_FAST:
> > > +		if (!i3cbus->scl_rate.i3c)
> > > +			i3cbus->scl_rate.i3c = I3C_BUS_TYP_I3C_SCL_RATE;
> > > +		if (!i3cbus->scl_rate.i2c)
> > > +			i3cbus->scl_rate.i2c = max_i2c_scl_rate;
> > > +		break;
> > > +	case I3C_BUS_MODE_MIXED_SLOW:
> > > +		if (!i3cbus->scl_rate.i2c)
> > > +			i3cbus->scl_rate.i2c = max_i2c_scl_rate;
> > > +		if (!i3cbus->scl_rate.i3c ||
> > > +		    i3cbus->scl_rate.i3c > i3cbus->scl_rate.i2c)
> > > +			i3cbus->scl_rate.i3c = i3cbus->scl_rate.i2c;
> > > +		break;
> > > +	default:
> > > +		return -EINVAL;
> > >  	}
> > >  
> > > +	if (i3cbus->scl_rate.i3c < i3cbus->scl_rate.i2c)
> > > +		dev_dbg(&master->dev,
> > > +			"i3c-scl-hz=%ld lower than i2c-scl-hz=%ld\n",
> > > +			i3cbus->scl_rate.i3c, i3cbus->scl_rate.i2c);
> > > +
> > > +	if (i3cbus->scl_rate.i2c != I3C_BUS_I2C_FM_SCL_RATE &&
> > > +	    i3cbus->scl_rate.i2c != I3C_BUS_I2C_FM_PLUS_SCL_RATE &&
> > > +	    i3cbus->mode != I3C_BUS_MODE_PURE)
> > > +		dev_dbg(&master->dev,
> > > +			"i2c-scl-hz=%ld not defined according MIPI I3C spec\n",
> > > +			i3cbus->scl_rate.i2c);
> > > +  
> > 
> > Again, that's not what I suggested, so I'll write it down:
> > 
> > 	dev_dbg(&master->dev, "i2c-scl = %ld Hz i3c-scl = %ld Hz\n",
> > 		i3cbus->scl_rate.i2c, i3cbus->scl_rate.i3c);
> >   
> 
> I'm not ok with that change. The reasons are:
>   i3cbus->scl_rate.i3c < i3cbus->scl_rate.i2c is an abnormal use case. As 
> discuss early it can be cause by a wrong DT definition or just for 
> testing purposes.

Is it buggy, and if it is, what are the symptoms? And I'm not talking
about slow transfers here. Also, note that forcing the I2C/I3C rate
through the DT already means you want to tweak the bus speed (either
for debugging purposes or because slowing things down is needed to fix
a HW bug).

> 
>   i3cbus->scl_rate.i2c != I3C_BUS_I2C_FM_SCL_RATE && i3cbus->scl_rate.i2c 
> != I3C_BUS_I2C_FM_PLUS_SCL_RATE, the MIPI I3C Spec v1.0 clearly says that 
> all I2C devices on the bus shall have a LVR register and thus support FM 
> or FM+ modes.

Yet, you might want to apply a lower I2C freq, and this sounds like a
valid case that doesn't deserve a dev_warn().

> By  definition a FM bus works at 400kHz and a FM+ bus 1MHz.
> And for slaves, a FM device works up to 400kHz and a FM+ device works up 
> to 1MHz respectively.

*up to*, that's the important thing to keep in mind. There's no problem
driving the SCL signal at a lower freq.

> 
> Apart of that, if the I2C device support you can use a custom higher or 
> lower rate, yet not defined according MIPI I3C spec.

I'm not going to have this discussion again, sorry. I think I gave
enough arguments to explain why having an I2C SLC rate that's slower
than what I2C devices support is fine.

> 
> > dev_dbg() is not printed by default, so it's just fine to have a trace
> > that prints the I3C and I2C rate unconditionally.  
> 
> I'm ok to change the way that user is notified and I think that is here 
> the problem.
> Maybe the best is to change the first dev_dbg() to dev_warn() and the 
> second dev_info().

Same here. I'm fine having a dev_warn() when the rate is higher than
what's supported by devices present on the bus (because that case is
buggy), but not when it's lower and still in the valid range.