On Mon, Feb 07, 2011 at 08:56:29AM -0500, Andrew Lutomirski wrote: > On Mon, Feb 7, 2011 at 2:33 AM, Jean Delvare <khali@xxxxxxxxxxxx> wrote: > > On Sun, 6 Feb 2011 20:50:05 -0800, Guenter Roeck wrote: > >> On Sun, Feb 06, 2011 at 10:50:02PM -0500, Andy Lutomirski wrote: > >> > On 02/05/2011 12:58 PM, Guenter Roeck wrote: > >> > > Hi, > >> > > > >> > > a standalone driver for W83667HG-I (NCT6771F, NCT6775F) and NCT6776F > >> > > is available for testing at http://roeck-us.net/linux/drivers/w83627ehf/. > >> > > > >> > > Please test and let me know if there are any problems. > >> > > >> > It appears to mostly work on an Intel DH67GD (nct6775). There are two > >> > issues: > >> > > >> > 1. The first time I ran 'sensors', my CPU fan showed 0 RPM instead of > >> > the correct value (~134). Running it again shows the correct value. > >> > > >> If you check the log, you might see a notice that the fan divisors were modified. > >> Can you check ? > > Yes, I got a message. dmesg shows: > > [51676.259546] w83627ehf: Found NCT6775 chip at 0x290 > [51678.246682] w83627ehf w83627ehf.656: Increasing fan2 clock divider > from 32 to 64 > > The second line happens on the first call to 'sensors'. Rebooting > triggers it again, but rmmod/insmod does not. > That is as expected. The driver loads the sensor value, detects that it is out of range, and automatically adjusts the divisor register. Usually, in other drivers, it is possible to write into the divisor register. This can then be used to initialize the divisor to a reasonable value at startup. For some reason that is not implemented for this driver; I don't know why. > Should the driver just fix up the divisors on load instead of on first read? > No, the divisor should be writable so you can initialize it at startup. I'll see if I can fix that. Thanks, Guenter _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
