Jean Delvare wrote:
> Hi Hans,
>
> On Thu, 19 Jul 2007 15:57:20 +0200, Hans de Goede wrote:
>> Here is a small fscher bugfix for 2.6.23 merging, lifted from my other fscher
>> work, as requested by Jean. The current driver has a control sysfs attribute,
>> which shows the contents of the control register, but the underlying
>> global_control value in the data structure currently never gets filled with
>> the actual contents of this register.
>>
>> Signed-off-by: Hans de Goede <j.w.r.degoede at hhs.nl>
>
> Acked-by: Jean Delvare <khali at linux-fr.org>
>
> Note that the Linux 2.4 driver in the lm-sensors SVN repository has the
> same bug, so it should be fixed there too.
>
> While we're here...
>
> static ssize_t set_control(struct i2c_client *client, struct fscher_data *data,
> const char *buf, size_t count, int nr, int reg)
> {
> /* bits 1..7 reserved => mask with 0x01 */
> unsigned long v = simple_strtoul(buf, NULL, 10) & 0x01;
>
> mutex_lock(&data->update_lock);
> data->global_control &= ~v;
> fscher_write_value(client, reg, v);
> mutex_unlock(&data->update_lock);
> return count;
> }
>
> Is the code above correct? It behaves as if bit 0 was "write 1 to clear"
> while the datasheet says this bit is regular read-write, and the Linux
> 2.4 driver treats it as such.
>
You are right, good catch. I'll send a patch fixing this right after this mail.
Regards,
Hans
