--------------------------------------------------
From: "Guenter Roeck" <guenter.roeck@xxxxxxxxxxxx>
Sent: Thursday, February 10, 2011 7:59 PM
To: "Ian Dobson" <i.dobson@xxxxxxxxxxxxxx>
Cc: "Jean Delvare" <khali@xxxxxxxxxxxx>; "Andy Lutomirski" <luto@xxxxxxx>;
<lm-sensors@xxxxxxxxxxxxxx>
Subject: Re: [PATCH v2 5/5] hwmon: (w83627ehf) Add support for Nuvoton
NCT6775F and NCT6776F
On Thu, 2011-02-10 at 13:37 -0500, Ian Dobson wrote:
[ ... ]
Hi Guenter,
Are you sure the smartfan readout of pwm_* parameters actually works
correctly on nct6776f chips?
Not really.
root@alpha2:/sys/devices/platform/w83627ehf.656# cat pwm1_enable
5
root@alpha2:/sys/devices/platform/w83627ehf.656# cat pwm1
193
root@alpha2:/sys/devices/platform/w83627ehf.656# cat pwm1_mode
1
root@alpha2:/sys/devices/platform/w83627ehf.656# cat pwm1_target
0
root@alpha2:/sys/devices/platform/w83627ehf.656# cat pwm1_tolerance
0
root@alpha2:/sys/devices/platform/w83627ehf.656# cat pwm1_start_output
1
root@alpha2:/sys/devices/platform/w83627ehf.656# cat pwm1_step_output
255
root@alpha2:/sys/devices/platform/w83627ehf.656# cat pwm1_stop_output
1
It looks as if pwm1 is in smartfan mode but the remaining parameters look
totally wrong. Maybe there's parameters are only valid in thermal cruise
mode. I'll study the code at the weekend and maybe playing with my test
box.
Keep in mind that SmartFan IV isn't fully supported, ie it is not
possible to set any SmartFan IV specific registers. So question is what
we would want to do if the system is configured for SmartFan IV mode.
Right now the code just reports that SmartFan IV is configured; the
other registers don't really mean anything in that mode. An option might
be to make the other attributes read-only if SmartFan IV is configured.
And maybe I should add some text to the documentation explaining this in
more detail.
We could of course add full SmartFan IV support, but that goes a bit
beyond the time I have available, and I would not want to hold nct677X
chip support hostage for it.
Input is welcome ...
Thanks,
Guenter
Hi Guenter,
Maybe if the parameters (_target, _tolerance etc) are invalid in smartfan
mode, we should return invalid-ENOVAL? It would be interesting to be able to
actually control the setpoints, for example on my box running the fans is
silent mode the CPU temperature is abit high for my liking and and standard
mode the fans are too loud. Something in between would be good.
I have no problems with getting a driver out that monitors correctly, even
if control doesn't work. Most people would be happy with that, and when you
think that the new motherboards (1155) only came out 1 month ago, and the
lm-sensors project already has support for the superIO chips used on most
boards.
I'll start looking into the code at the weekend, and if I'm lucky I'll find
whats wrong with the smartfan stuff.
Regards and thanks again for the great work.
Ian Dobson
_______________________________________________
lm-sensors mailing list
lm-sensors@xxxxxxxxxxxxxx
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
