[PATCH] possible bug in i2c-algo-bit's inb function

Andreas Oberritter obi at saftware.de
Sun Feb 27 22:55:47 CET 2005


On Sun, 2005-02-27 at 19:52 +0100, Jean Delvare wrote:
> > Here's my I2C code:
> > (...)
> 
> Looks sane.
> 
> > The bus master is a so called Pluto2 by SCM (part of a DVB-T card sold
> > by Satelco). If this is a hardware bug, is it possible to add a flag
> > to struct i2c_algo_bit_data to workaround this bug?
> 
> I would try to find out whether the culprit is setscl or getsda
> (assuming I am correct and either of these actually is the problem).
> Once you know which it is, you could modifiy that function to restore
> the SDA line, that should do the trick.
> 
> Hope that helps,

Thank you, using the workaround below makes it possible to use the
bitbanging code without modifications. It tries to detect i2c_inb() and
resets SDA to high after SCL has been set to high and then to low.

static void pluto_setscl(void *data, int state)
{
        struct pluto *pluto = data;

        if (state)
                pluto_rw(pluto, REG_SLCS, SLCS_SCL, SLCS_SCL);
        else
                pluto_rw(pluto, REG_SLCS, SLCS_SCL, 0);

        if ((state) && (pluto->i2cbug == 0)) {
                pluto->i2cbug = 1;
        } else {
                if ((!state) && (pluto->i2cbug == 1))
                        pluto_setsda(pluto, 1);
                pluto->i2cbug = 0;
        }
}

Regards,
Andreas



More information about the lm-sensors mailing list