Via KT600 as I2C controller

Via KT600 as I2C controller

[Lou]

I have an A7V600 motherboard, which has a Via 8237 on it.
It is not supported by current 2.6 kernels (test8, for example).
I have added support by adding the following lines to vt596_ids[] in
/usr/src/linux-2.6.0-test8/drivers/i2c/busses/i2c-viapro.c.

    {
        .vendor     = PCI_VENDOR_ID_VIA,
        .device     = PCI_DEVICE_ID_VIA_8237,
        .subvendor  = PCI_ANY_ID,
        .subdevice  = PCI_ANY_ID,
        .driver_data    = SMBBA3
    },

I have not observed any brokenness, having used it for several weeks.
I could, of course, continue making the change to future kernel
versions, but this would surely benefit others if in the regular dist.

Via KT600 as I2C controller

[Jean Delvare]

> I have an A7V600 motherboard, which has a Via 8237 on it.
> It is not supported by current 2.6 kernels (test8, for example).
> I have added support by adding the following lines to vt596_ids[] in
> /usr/src/linux-2.6.0-test8/drivers/i2c/busses/i2c-viapro.c.
> 
>     {
>         .vendor     = PCI_VENDOR_ID_VIA,
>         .device     = PCI_DEVICE_ID_VIA_8237,
>         .subvendor  = PCI_ANY_ID,
>         .subdevice  = PCI_ANY_ID,
>         .driver_data    = SMBBA3
>     },
> 
> I have not observed any brokenness, having used it for several weeks.
> I could, of course, continue making the change to future kernel
> versions, but this would surely benefit others if in the regular dist.

I've added support in our CVS repository. I will also send a patch to
Greg KH for Linux 2.6. Don't expect it to be added before 2.6.0 final
though, since only bug fixes will be accepted for now.

-- 
Jean Delvare
http://www.ensicaen.ismra.fr/~delvare/

Via KT600 as I2C controller

[Mark Studebaker]

thanks. would you please download the latest lm_sensors CVS
and compile the userspace tools with 2.6 support
(make user and make user_install)
and use i2cdetect and i2cdump to test the driver.
use all modes of i2cdump that work (b,w,c,s)
On a DIMM eeprom, for example, all modes should return the
same data.


Jean Delvare wrote:
> 
> > I have an A7V600 motherboard, which has a Via 8237 on it.
> > It is not supported by current 2.6 kernels (test8, for example).
> > I have added support by adding the following lines to vt596_ids[] in
> > /usr/src/linux-2.6.0-test8/drivers/i2c/busses/i2c-viapro.c.
> >
> >     {
> >         .vendor     = PCI_VENDOR_ID_VIA,
> >         .device     = PCI_DEVICE_ID_VIA_8237,
> >         .subvendor  = PCI_ANY_ID,
> >         .subdevice  = PCI_ANY_ID,
> >         .driver_data    = SMBBA3
> >     },
> >
> > I have not observed any brokenness, having used it for several weeks.
> > I could, of course, continue making the change to future kernel
> > versions, but this would surely benefit others if in the regular dist.
> 
> I've added support in our CVS repository. I will also send a patch to
> Greg KH for Linux 2.6. Don't expect it to be added before 2.6.0 final
> though, since only bug fixes will be accepted for now.
> 
> --
> Jean Delvare
> http://www.ensicaen.ismra.fr/~delvare/

Via KT600 as I2C controller

[Jean Delvare]

> thanks. would you please download the latest lm_sensors CVS
> and compile the userspace tools with 2.6 support
> (make user and make user_install)
> and use i2cdetect and i2cdump to test the driver.
> use all modes of i2cdump that work (b,w,c,s)
> On a DIMM eeprom, for example, all modes should return the
> same data.

I don't have DIMM eeproms on my i2c-viapro bus, so I tried on an
as99127f. b produces a good dump, c doesn't. The byte at 0x00 (value
0xbd) is repeated everywhere. I guess this is expected because the chip
doesn't have autoincrement? s gave me a kernel panic (stack trashed with
0xbd). w is OK (each byte is repeated twice to form a word).

-- 
Jean Delvare
http://www.ensicaen.ismra.fr/~delvare/

Via KT600 as I2C controller

[Mark Studebaker]

right, c mode won't work except on eeproms.
s mode won't generally work on most chips, I was wrong in saying it
would.
But it shouldn't panic.
Block transactions aren't well-tested in general.
Glanced at i2c-viapro.c, nothing jumped out at me.

Jean Delvare wrote:
> 
> > thanks. would you please download the latest lm_sensors CVS
> > and compile the userspace tools with 2.6 support
> > (make user and make user_install)
> > and use i2cdetect and i2cdump to test the driver.
> > use all modes of i2cdump that work (b,w,c,s)
> > On a DIMM eeprom, for example, all modes should return the
> > same data.
> 
> I don't have DIMM eeproms on my i2c-viapro bus, so I tried on an
> as99127f. b produces a good dump, c doesn't. The byte at 0x00 (value
> 0xbd) is repeated everywhere. I guess this is expected because the chip
> doesn't have autoincrement? s gave me a kernel panic (stack trashed with
> 0xbd). w is OK (each byte is repeated twice to form a word).
> 
> --
> Jean Delvare
> http://www.ensicaen.ismra.fr/~delvare/