Re: [PATCH] i2c: designware: deduce speed mode from device tree setting

[atull]

On Wed, 20 Aug 2014, Romain Baeriswyl wrote:

> Hi Alan,
> 
> We got board issue using I2C and they were solved by changing the i2c timing.
> 
> It is possible to change the tLOW and tHIGH period with the following parameters:
>   i2c-sda-hold-time-ns
>   i2c-sda-falling-time-ns
>   i2c-scl-falling-time-ns
> 
> Romain

Hi Romain,

This thread is getting kind of messed up.  Yep, that's how I am slowing 
down the bus, using i2c-sda-falling-time-ns and i2c-scl-falling-time-ns.
It is fine with me if you want to test for == 100KHz, just not my 
(probably hair-splitting) preference. :)

Alan

> 
> ----- Original Message -----
> From: "atull" <atull@xxxxxxxxxxxxxxxxxxxxx>
> To: "Romain Baeriswyl" <Romain.Baeriswyl@xxxxxxxxxx>
> Cc: "Mark Rutland" <mark.rutland@xxxxxxx>, wsa@xxxxxxxxxxxxx, baruch@xxxxxxxxxx, "mika westerberg" <mika.westerberg@xxxxxxxxxxxxxxx>, "grant likely" <grant.likely@xxxxxxxxxx>, robh+dt@xxxxxxxxxx, skuribay@xxxxxxxxx, "rafael j wysocki" <rafael.j.wysocki@xxxxxxxxx>, alan@xxxxxxxxxxxxxxx, linux-i2c@xxxxxxxxxxxxxxx, linux-kernel@xxxxxxxxxxxxxxx, devicetree@xxxxxxxxxxxxxxx, "delicious quinoa" <delicious.quinoa@xxxxxxxxx>, dinguyen@xxxxxxxxxxxxxxxxxxxxx, yvanderv@xxxxxxxxxxxxxxxxxxxxx
> Sent: Wednesday, August 20, 2014 4:24:45 PM
> Subject: Re: [PATCH] i2c: designware: deduce speed mode from device tree setting
> 
> 
> 
> On Wed, 20 Aug 2014, Romain Baeriswyl wrote:
> 
> > Hi,
> > 
> > With the patch "i2c designware add support of I2C standard mode" I already proposed:
> > - I2C standard mode is selected with 100kHz clock frequency.
> > - I2C fast mode is selected with 400kHy clock frequency.
> > - EINVAL error is returned if clock frequency is not 100000 and not 400000.
> > 
> > but this patch seems not available yet.
> > What about the other patch "i2c designware make SCL and SDA falling time configurable" ?
> > 
> > In i2c-designware_platdrv.c the flag DW_IC_CON_SPEED_xxx is well set
> > depending on the mode:
> > 
> >   if (clk_freq == 100000)
> 
> Romain,
> 
> I'm really happy if your patches get accepted.  Can this be <= 100000?
> It is really common to run I2C at a lower speed if you have some
> board issues with the i2c bus.
> 
> Alan
> 
> >      dev->master_cfg =  DW_IC_CON_MASTER | DW_IC_CON_SLAVE_DISABLE |
> >        DW_IC_CON_RESTART_EN | DW_IC_CON_SPEED_STD;
> >    else
> >      dev->master_cfg =  DW_IC_CON_MASTER | DW_IC_CON_SLAVE_DISABLE |
> >        DW_IC_CON_RESTART_EN | DW_IC_CON_SPEED_FAST;
> > 
> > 
> > So for me everything should be fine if there patches are applied.
> > 
> > Regards,
> > 
> > Romain
> > 
> > ----- Original Message -----
> > From: "Mark Rutland" <mark.rutland@xxxxxxx>
> > To: atull@xxxxxxxxxxxxxxxxxxxxx
> > Cc: wsa@xxxxxxxxxxxxx, baruch@xxxxxxxxxx, "mika westerberg" <mika.westerberg@xxxxxxxxxxxxxxx>, "grant likely" <grant.likely@xxxxxxxxxx>, robh+dt@xxxxxxxxxx, skuribay@xxxxxxxxx, "Romain Baeriswyl" <Romain.Baeriswyl@xxxxxxxxxx>, "rafael j wysocki" <rafael.j.wysocki@xxxxxxxxx>, alan@xxxxxxxxxxxxxxx, linux-i2c@xxxxxxxxxxxxxxx, linux-kernel@xxxxxxxxxxxxxxx, devicetree@xxxxxxxxxxxxxxx, "delicious quinoa" <delicious.quinoa@xxxxxxxxx>, dinguyen@xxxxxxxxxxxxxxxxxxxxx, yvanderv@xxxxxxxxxxxxxxxxxxxxx
> > Sent: Wednesday, August 20, 2014 11:22:57 AM
> > Subject: Re: [PATCH] i2c: designware: deduce speed mode from device tree setting
> > 
> > On Tue, Aug 19, 2014 at 09:18:49PM +0100, atull@xxxxxxxxxxxxxxxxxxxxx wrote:
> > > From: Alan Tull <atull@xxxxxxxxxxxxxxxxxxxxx>
> > > 
> > > Use the documented, but unimplemented "clock-frequency"
> > > Device Tree setting as a guide on whether to set the speed
> > > mode bits in DW_IC_CON to standard or fast i2c mode.
> > > 
> > > Previously, the driver was hardwired to fast mode.  Default
> > > to fast mode if the "clock-frequency" property is not present
> > > for backwards compatiblity.
> > > 
> > > Signed-off-by: Alan Tull <atull@xxxxxxxxxxxxxxxxxxxxx>
> > > ---
> > >  drivers/i2c/busses/i2c-designware-platdrv.c |   10 ++++++++--
> > >  1 file changed, 8 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/drivers/i2c/busses/i2c-designware-platdrv.c b/drivers/i2c/busses/i2c-designware-platdrv.c
> > > index bc87733..18cd3d9 100644
> > > --- a/drivers/i2c/busses/i2c-designware-platdrv.c
> > > +++ b/drivers/i2c/busses/i2c-designware-platdrv.c
> > > @@ -122,7 +122,8 @@ static int dw_i2c_probe(struct platform_device *pdev)
> > >  	struct dw_i2c_dev *dev;
> > >  	struct i2c_adapter *adap;
> > >  	struct resource *mem;
> > > -	int irq, r;
> > > +	int irq, r, ret, speed = DW_IC_CON_SPEED_FAST;
> > > +	u32 bus_rate;
> > >  
> > >  	irq = platform_get_irq(pdev, 0);
> > >  	if (irq < 0) {
> > > @@ -167,6 +168,11 @@ static int dw_i2c_probe(struct platform_device *pdev)
> > >  		of_property_read_u32(pdev->dev.of_node,
> > >  				     "i2c-scl-falling-time-ns",
> > >  				     &dev->scl_falling_time);
> > > +
> > > +		ret = of_property_read_u32(pdev->dev.of_node,
> > > +					   "clock-frequency", &bus_rate);
> > > +		if (!ret && (bus_rate <= 100000))
> > > +			speed = DW_IC_CON_SPEED_STD;
> > 
> > This looks a bit odd.
> > 
> > If the device only supports two particular speeds why do we accept any
> > other speed in the clock-frequency property? Surely we should at least
> > warn that something was off?
> > 
> > Thanks,
> > Mark
> > 
> > >  	}
> > >  
> > >  	dev->functionality =
> > > @@ -177,7 +183,7 @@ static int dw_i2c_probe(struct platform_device *pdev)
> > >  		I2C_FUNC_SMBUS_WORD_DATA |
> > >  		I2C_FUNC_SMBUS_I2C_BLOCK;
> > >  	dev->master_cfg =  DW_IC_CON_MASTER | DW_IC_CON_SLAVE_DISABLE |
> > > -		DW_IC_CON_RESTART_EN | DW_IC_CON_SPEED_FAST;
> > > +		DW_IC_CON_RESTART_EN | speed;
> > >  
> > >  	/* Try first if we can configure the device from ACPI */
> > >  	r = dw_i2c_acpi_configure(pdev);
> > > -- 
> > > 1.7.9.5
> > > 
> > > --
> > > To unsubscribe from this list: send the line "unsubscribe devicetree" in
> > > the body of a message to majordomo@xxxxxxxxxxxxxxx
> > > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> > > 
> > 
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/