Re: [PATCH v3] New-style I2C and SMBus EEPROM driver (with device_ids)

[David Brownell <david-b-yBeKhBN/0LDR7s880joybQ@public.gmane.org>]

On Sunday 08 June 2008, Jean Delvare wrote:
> You need to include <linux/bitops.h> for BIT().

That's handled by <linux/kernel.h> ... I rather think it's OK
to rely on a few basics like that.


> You're going to quite some extent to obfuscate simple things ;) All
> these defines are for the sole internal purpose of the at24 driver
> (custom eeprom types would use platform data instead) and should
> not be in the public header file... if they should defined at all.

The original notion was to get the driver out of the business of
holding a large table of device parameters including worst-case
pagesizes (e.g. Microchip pages being half or a quarter the size
of Atmel pages) and address consumption (e.g. Atmel 24c01 vs 24c01a,
or the SOT23 versions doing who-knows-undocumented-what).

So I think those #defines are somewhat a legacy of having had to
change direction mid-steam to cope with the new "i2c_device_id"
and its expectation that drivers *would* have such large tables
with worst-case parameters.

Just so you know.  :)

- Dave
 

_______________________________________________
i2c mailing list
i2c-GZX6beZjE8VD60Wz+7aTrA@public.gmane.org
http://lists.lm-sensors.org/mailman/listinfo/i2c