Possible bug in i2c-nforce2

Possible bug in i2c-nforce2

[Jean Delvare]

Hi Hans-Frieder,

Recent failure reports with the i2c-nforce2+eeprom combination led me to
the conclusion that your i2c-norce2 module might not support
i2c_smbus_write_byte() correctly. It is used in the eeprom driver and
doesn't seem to set the address register as it should.

Could you please take a look?

Thanks.

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

Possible bug in i2c-nforce2

[Jean Delvare]

Hi again Hans-Frieder,

> Recent failure reports with the i2c-nforce2+eeprom combination led me
> to the conclusion that your i2c-norce2 module might not support
> i2c_smbus_write_byte() correctly. It is used in the eeprom driver and
> doesn't seem to set the address register as it should.

Mark D. Studebaker probably found and fixed that bug. The following change
was made to your driver:

Index: i2c-nforce2.c
=================================RCS file: /home/cvs/lm_sensors2/kernel/busses/i2c-nforce2.c,v
retrieving revision 1.6
retrieving revision 1.7
diff -u -r1.6 -r1.7
--- i2c-nforce2.c	3 Aug 2003 05:32:01 -0000	1.6
+++ i2c-nforce2.c	27 Nov 2003 02:37:43 -0000	1.7
@@ -143,7 +143,7 @@
 
 		case I2C_SMBUS_BYTE:
 			if (read_write = I2C_SMBUS_WRITE)
-				outb_p(data->byte, NVIDIA_SMB_DATA);
+				outb_p(command, NVIDIA_SMB_CMD);
 			protocol |= NVIDIA_SMB_PRTCL_BYTE;
 			break;
 

Comments on this change are welcome. If you confirm it's OK, I'll send a
patch to Greg KH to update the driver present in Linux 2.6.

Thanks.

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

Possible bug in i2c-nforce2

[Hans-Frieder Vogt]

Jean, Mark,

thanks, you spotted this bug much quicker than me. 
I have also tested the patch and found no unexpected behaviour, e.g. both
i2cdump 0 0x50
and
i2cdump 0 0x50 c
work and give the same output, which indicates for me that the patch must have 
solved this bug.

Thanks.
Hans-Frieder

> Hi again Hans-Frieder,
>
> > Recent failure reports with the i2c-nforce2+eeprom combination led me
> > to the conclusion that your i2c-norce2 module might not support
> > i2c_smbus_write_byte() correctly. It is used in the eeprom driver and
> > doesn't seem to set the address register as it should.
>
> Mark D. Studebaker probably found and fixed that bug. The following change
> was made to your driver:
>
> Index: i2c-nforce2.c
> =================================> RCS file: /home/cvs/lm_sensors2/kernel/busses/i2c-nforce2.c,v
> retrieving revision 1.6
> retrieving revision 1.7
> diff -u -r1.6 -r1.7
> --- i2c-nforce2.c	3 Aug 2003 05:32:01 -0000	1.6
> +++ i2c-nforce2.c	27 Nov 2003 02:37:43 -0000	1.7
> @@ -143,7 +143,7 @@
>
>  		case I2C_SMBUS_BYTE:
>  			if (read_write = I2C_SMBUS_WRITE)
> -				outb_p(data->byte, NVIDIA_SMB_DATA);
> +				outb_p(command, NVIDIA_SMB_CMD);
>  			protocol |= NVIDIA_SMB_PRTCL_BYTE;
>  			break;
>
>
> Comments on this change are welcome. If you confirm it's OK, I'll send a
> patch to Greg KH to update the driver present in Linux 2.6.
>
> Thanks.

-- 
--
Hans-Frieder Vogt                 e-mail: hfvogt@arcor.de