Re: [lm-sensors] hwmon/w83627ehf

[Jean Delvare <khali@linux-fr.org>]

Hi Lambert, David,

On Tue, 15 May 2007 10:19:59 +0200, Lambert Carsten wrote:
> Actually I think it is working with the latest lm_sensors (2.10.3). I didn't 
> configure anything except uncommenting 'ignore in9' in /etc/sensors.conf as 
> advised by the emerge script (gentoo) for the W83627DHG chip. Here is the 
> output of sensors after installing and running the init script:
> 
> sensors
> w83627dhg-isa-0a00
> Adapter: ISA adapter
> VCore:     +1.26 V  (min =  +0.00 V, max =  +1.74 V)
> in1:      +12.30 V  (min =  +1.37 V, max =  +7.55 V) ALARM
> AVCC:      +3.26 V  (min =  +1.06 V, max =  +2.18 V) ALARM
> 3VCC:      +3.26 V  (min =  +1.71 V, max =  +3.74 V)
> in4:       +1.58 V  (min =  +1.11 V, max =  +1.04 V) ALARM
> in5:       +1.66 V  (min =  +1.52 V, max =  +0.09 V) ALARM
> in6:       +4.94 V  (min =  +5.43 V, max =  +1.31 V) ALARM
> VSB:       +3.26 V  (min =  +2.58 V, max =  +2.21 V) ALARM
> VBAT:      +3.17 V  (min =  +2.22 V, max =  +3.90 V)
> Case Fan:    0 RPM  (min =    0 RPM, div = 64) ALARM
> CPU Fan:   683 RPM  (min =  664 RPM, div = 8)
> Aux Fan:     0 RPM  (min = 1757 RPM, div = 64) ALARM
> fan4:        0 RPM  (min = 56250 RPM, div = 8) ALARM
> fan5:        0 RPM  (min = 9375 RPM, div = 8) ALARM
> Sys Temp:    +45 C  (high =  +115 C, hyst =   +27 C)
> CPU Temp:  +24.5 C  (high = +80.0 C, hyst = +75.0 C)
> AUX Temp: +121.5 C  (high = +80.0 C, hyst = +75.0 C)   ALARM
> 
> The CPU Temp reading seems low but I put the machine to work and the temp went 
> up as expected and seems right. The CPU Fan reading I think is corrct too. 
> The bios is configured to speed up the fan when the CPU reaches 50 C.

Could be that the Sys Temp and CPU Temp labels need to be swapped.
Check the order and values in your BIOS (if it prints them.)

> Here follows the isadumps you requested in case they are still usefull. It's 
> all 'middle clicked' this time so no typo.
> (...)
> isadump -y -k 0x87,0x87 0x4e 0x4f 0xb 0x7
>      0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
> 00: ff ff ff ff ff ff ff 0b ff ff ff ff ff ff ff ff
> 10: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> 20: a0 23 ff 00 44 00 40 ff 50 00 00 00 03 21 00 ff
> 30: 01 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> 40: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> 50: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> 60: 0a 00 ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> 70: 00 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> 80: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> 90: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> a0: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> b0: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> c0: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> d0: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> e0: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> f0: 81 ff 80 ff 00 01 00 00 ff ff ff ff ff ff ff ff

This confirms that the configuration space of the W83627DHG is mapped
to 0x4e/0x4f. This explains the log message that you reported: the
driver printed it after it failed to find a device at 0x2e/0x2f. Then
it tried 0x4e/0x4f and succeeded, so the message was only a warning,
not an error.

The w83627ehf driver should be changed to _not_ print a message when the
device ID reads 0xffff (or 0x0000, for that matter) as it means there
is _no_ chip at this address.

On top of that, the driver should _not_ print a message by default when
it finds an unsupported ID. It is perfectly valid to have a W83627DHG
at 0x4e/0x4f and another LPC chip at 0x2e/0x2f. I think we already
heard of boards with two LPC chips, and we don't want to fill the log
with irrelevant messages in that case. So I believe that the driver
should only print this message when compiled with debugging enabled, as
the w83627hf driver is doing.

David, can you please submit a patch doing that?

Thanks,
-- 
Jean Delvare

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors