[lm-sensors] w83627ehf informations

[lm-sensors] w83627ehf informations

[slompf]

Hallo,

I'm trying to reduce the speed of a cpu fan on a k8n-vm. The chip is for
controlling the temp is a winbond w83627ehf. My kernel is 2.6.18.2.
lm-sensors seems to work correctly, because on the 'sensors' command I'm
getting the following output:

w83627ehf-isa-0290
Adapter: ISA adapter
Case Fan:  888 RPM  (min = 1205 RPM, div = 8)
CPU Fan:  2109 RPM  (min = 1704 RPM, div = 8)
fan3:        0 RPM  (min =   45 RPM, div = 128)
fan4:        0 RPM  (min = 10546 RPM, div = 128)
Sys Temp:    +28?C  (high =   +45?C, hyst =   +40?C)  
CPU Temp:  +23.0?C  (high = +45.0?C, hyst = +40.0?C)  
temp3:     +25.0?C  (high = +80.0?C, hyst = +75.0?C)  

Now the first question: I thought support for the voltage inputs was
added at one of the last kernels. Is that correct? 

Now the PROBLEM: After calling pwmconfig (which seems to be the tool for
configuring fancontrol) I'm getting this output:

/usr/sbin/pwmconfig: There are no pwm-capable sensor modules installed

After checking this script, it became clear that it needs something
called pwm.. or fan.._pwm in the /sys/bus/i2c/devices/9191-0290
directory. ls does only show that: 

bus         fan2_div    fan4_alarm    in0_max    in2_input  in4_alarm
in5_min    in7_max    in9_input    temp1_input     temp3_alarm
driver      fan2_input  fan4_div      in0_min    in2_max    in4_input
in6_alarm  in7_min    in9_max      temp1_max       temp3_input
fan1_alarm  fan2_min    fan4_input    in1_alarm  in2_min    in4_max
in6_input  in8_alarm  in9_min      temp1_max_hyst  temp3_max
fan1_div    fan3_alarm  fan4_min      in1_input  in3_alarm  in4_min
in6_max    in8_input  name         temp2_alarm     temp3_max_hyst
fan1_input  fan3_div    hwmon:hwmon0  in1_max    in3_input  in5_alarm
in6_min    in8_max    power        temp2_input     uevent
fan1_min    fan3_input  in0_alarm     in1_min    in3_max    in5_input
in7_alarm  in8_min    subsystem    temp2_max
fan2_alarm  fan3_min    in0_input     in2_alarm  in3_min    in5_max
in7_input  in9_alarm  temp1_alarm  temp2_max_hyst

Next question: where are my pwm entries? The in_... where printed first,
after I compiled the new kernel. (2.6.17->2.6.18) Is the pwm function
not supported for this chipset in the above mentioned kernel?

Maybe you guys can give some general information about the status of the
chipset. Whats supported by which kernel and what is planned and what is
not. That would be quiet helpful. Maybe there is a document more
detailed than that: http://www.lm-sensors.org/wiki/Devices?

Kind Regards
Torsten

[lm-sensors] w83627ehf informations

[David Hubbard]

Hello Torsten,

On 11/11/06, slompf <2crazy4you at gmx.de> wrote:
> Hallo,
>
> I'm trying to reduce the speed of a cpu fan on a k8n-vm. The chip is for
> controlling the temp is a winbond w83627ehf. My kernel is 2.6.18.2.
> lm-sensors seems to work correctly, because on the 'sensors' command I'm
> getting the following output:
>
> w83627ehf-isa-0290
> Adapter: ISA adapter
> Case Fan:  888 RPM  (min = 1205 RPM, div = 8)
> CPU Fan:  2109 RPM  (min = 1704 RPM, div = 8)
> fan3:        0 RPM  (min =   45 RPM, div = 128)
> fan4:        0 RPM  (min = 10546 RPM, div = 128)
> Sys Temp:    +28?C  (high =   +45?C, hyst =   +40?C)
> CPU Temp:  +23.0?C  (high = +45.0?C, hyst = +40.0?C)
> temp3:     +25.0?C  (high = +80.0?C, hyst = +75.0?C)
>
> Now the first question: I thought support for the voltage inputs was
> added at one of the last kernels. Is that correct?

Yes, that is correct.

> Now the PROBLEM: After calling pwmconfig (which seems to be the tool for
> configuring fancontrol) I'm getting this output:
>
> /usr/sbin/pwmconfig: There are no pwm-capable sensor modules installed
>
> After checking this script, it became clear that it needs something
> called pwm.. or fan.._pwm in the /sys/bus/i2c/devices/9191-0290
> directory. ls does only show that:
>
> bus         fan2_div    fan4_alarm    in0_max    in2_input  in4_alarm
> in5_min    in7_max    in9_input    temp1_input     temp3_alarm
> driver      fan2_input  fan4_div      in0_min    in2_max    in4_input
> in6_alarm  in7_min    in9_max      temp1_max       temp3_input
> fan1_alarm  fan2_min    fan4_input    in1_alarm  in2_min    in4_max
> in6_input  in8_alarm  in9_min      temp1_max_hyst  temp3_max
> fan1_div    fan3_alarm  fan4_min      in1_input  in3_alarm  in4_min
> in6_max    in8_input  name         temp2_alarm     temp3_max_hyst
> fan1_input  fan3_div    hwmon:hwmon0  in1_max    in3_input  in5_alarm
> in6_min    in8_max    power        temp2_input     uevent
> fan1_min    fan3_input  in0_alarm     in1_min    in3_max    in5_input
> in7_alarm  in8_min    subsystem    temp2_max
> fan2_alarm  fan3_min    in0_input     in2_alarm  in3_min    in5_max
> in7_input  in9_alarm  temp1_alarm  temp2_max_hyst
>
> Next question: where are my pwm entries? The in_... where printed first,
> after I compiled the new kernel. (2.6.17->2.6.18) Is the pwm function
> not supported for this chipset in the above mentioned kernel?

You need a newer kernel. The fan control support is available in the
-mm branch (which is for testing, before being integrated into the
mainline kernel). I looked at
http://www.kernel.org/pub/linux/kernel/v2.6/linux-2.6.18.tar.bz2 and
it has an older version of drivers/hwmon/w83627ehf.c which does not
include pwm/fan control support. It would be great if you could
upgrade to a testing kernel and report your success / failure with the
new driver. Does this help?

> Maybe you guys can give some general information about the status of the
> chipset. Whats supported by which kernel and what is planned and what is
> not. That would be quiet helpful. Maybe there is a document more
> detailed than that: http://www.lm-sensors.org/wiki/Devices?

From the wiki: "Fan control support is available in -mm."

I'm probably the person you should contact to get work done on the
w83627ehf driver. I've had a backlog of schoolwork recently, but I am
looking at getting the w83627dhg support into the w83627ehf driver in
the next two weeks. When I submit that patch, it will go into the
latest -mm kernel.

Thank you for writing,
David

[lm-sensors] w83627ehf informations

[torsten goroll]

Ok, I will try Kernel 2.6.18.2 with the mm2 patch. This seems to be the correct one. Alternativly I could try .19-rc4, what would might be the best choice? Is the driver in the release version more recent?

Torsten
-- 
"Ein Herz f?r Kinder" - Ihre Spende hilft! Aktion: www.deutschlandsegelt.de
Unser Dankesch?n: Ihr Name auf dem Segel der 1. deutschen America's Cup-Yacht!

[lm-sensors] w83627ehf informations

[David Hubbard]

Hi Torsten,

I don't know which kernel the 2.6.18-mm2 patch applies against. I
wonder if someone on the lm-sensors list knows more about this?

David

On 11/13/06, torsten goroll <2crazy4you at gmx.de> wrote:
> Ok, I will try Kernel 2.6.18.2 with the mm2 patch. This seems to be the correct one. Alternativly I could try .19-rc4, what would might be the best choice? Is the driver in the release version more recent?
>
> Torsten

[lm-sensors] w83627ehf informations

[Jean Delvare]

Hi David, Torsten,

On Mon, 13 Nov 2006 10:46:55 -0700, David Hubbard wrote:
> Hi Torsten,
> 
> I don't know which kernel the 2.6.18-mm2 patch applies against. I
> wonder if someone on the lm-sensors list knows more about this?

The 2.6.18-mm2 patch applies on top of 2.6.18, but I wouldn't recommend
using an old -mm kernel.

> On 11/13/06, torsten goroll <2crazy4you at gmx.de> wrote:
> > Ok, I will try Kernel 2.6.18.2 with the mm2 patch.
> > This seems to be the correct one. Alternativly I could try .19-rc4,
> > what would might be the best choice? Is the driver in the release
> > version more recent?

The driver in 2.6.19-rcX has one more fix from Rudolf Marek for fan5
detection, that wasn't present in 2.6.18-mmX.

I suggest that you try 2.6.19-rc5 (or even 2.6.19-rc5-git2 or whatever
is the latest Linus' tree when you fetch it.) It's probably the most
stable kernel tree you can hope for which will include the support you
need.

Note that you'll also need lm_sensors 2.10.1 for user-space support.

-- 
Jean Delvare

[lm-sensors] w83627ehf informations

[David Hubbard]

Hi Torsten,

The fix you found below may be interesting.

On 11/17/06, torsten goroll <2crazy4you at gmx.de> wrote:
> Hi David,
>
> it took me a long time, but now I habe a 2.6.19-rc5 kernel runnig. The sensors output looks OK:
>
> w83627ehf-isa-0290
> Adapter: ISA adapter
> VCore:     +1.38 V  (min =  +0.00 V, max =  +1.74 V)
> in1:      +12.25 V  (min =  +2.53 V, max =  +3.01 V) ALARM
> AVCC:      +3.25 V  (min =  +2.19 V, max =  +0.37 V) ALARM
> 3VCC:      +3.25 V  (min =  +0.34 V, max =  +1.34 V) ALARM
> in4:       +1.64 V  (min =  +0.38 V, max =  +0.28 V) ALARM
> in5:       +1.71 V  (min =  +0.04 V, max =  +1.59 V) ALARM
> in6:       +5.50 V  (min =  +2.92 V, max =  +1.51 V) ALARM
> VSB:       +3.25 V  (min =  +2.37 V, max =  +1.68 V) ALARM
> VBAT:      +0.51 V  (min =  +0.50 V, max =  +0.72 V)
> in9:       +1.58 V  (min =  +0.06 V, max =  +0.26 V) ALARM
> Case Fan:  883 RPM  (min =  677 RPM, div = 8)
> CPU Fan:  2295 RPM  (min = 3013 RPM, div = 4) ALARM
> Aux Fan:     0 RPM  (min =   45 RPM, div = 128) ALARM
> fan5:        0 RPM  (min = 23275 RPM, div = 2) ALARM
> Sys Temp:    +27?C  (high =  -115?C, hyst =    -8?C)   ALARM
> CPU Temp:  +21.5?C  (high = +80.0?C, hyst = +75.0?C)
> AUX Temp:  +24.0?C  (high = +80.0?C, hyst = +75.0?C)
>
> But when I trie to run pwm-config I'm getting those (error)messages:
> We will attempt to briefly stop each fan using the pwm controls.
> The program will attempt to restore each fan to full speed
> after testing. However, it is ** very important ** that you
> physically verify that the fans have been to full speed
> after the program has completed.
>
> Found the following PWM controls:
>    9191-0290/pwm1
> /usr/local/sbin/pwmconfig: line 102: echo: write error: Invalid argument
>    9191-0290/pwm2
> /usr/local/sbin/pwmconfig: line 102: echo: write error: Invalid argument
>    9191-0290/pwm3
> /usr/local/sbin/pwmconfig: line 102: echo: write error: Invalid argument
>
> Found the following fan sensors:
>    9191-0290/fan1_input     current speed: 888 RPM
>    9191-0290/fan2_input     current speed: 2311 RPM
>    9191-0290/fan3_input     current speed: 0 ... skipping!
>    9191-0290/fan5_input     current speed: 0 ... skipping!
>
> Warning!!! This program will stop your fans, one at a time,
> for approximately 5 seconds each!!!
> This may cause your processor temperature to rise!!!
> If you do not want to do this hit control-C now!!!
> Hit return to continue:
>
> Testing pwm control 9191-0290/pwm1 ...
>   9191-0290/fan1_input ... speed was 888 now 883
> /usr/local/sbin/pwmconfig: line 102: echo: write error: Invalid argument
>     no correlation
>   9191-0290/fan2_input ... speed was 2311 now 2295
> /usr/local/sbin/pwmconfig: line 102: echo: write error: Invalid argument
>     no correlation
>
> Checking the script I found the error. There must not be spaces in line 102.
> not: echo 0 > $ENABLE
> better:  echo 0>$ENABLE

This is important information. Who maintains the pwmconfig script?
(Jean, I CC'ed you in case you have more information.) Is this a
genuine bug?

>
> But that doesnt solve my problem. The "no correlation message" keeps. Even if I write a 0 into pwmx_enable the fan keeps rotating.
>
> Maybe the board(Asus K8N-VM) does not support fan control? Do you have another idea?

It does not surprise me that the rpm reading is noisy. Can you
manually experiment, now that you have a working kernel and the newest
w83627ehf driver?

Here is one way to manually test the pwm control:

cd /sys # I assume this is where your sysfs is mounted
# I am not at a linux machine, so the next line may be incorrect, but it's close
cd class/hwmon/hwmon0/device

ls # do you see the pwm device files? If you do, then you're in the
correct directory.

echo 1 > pwmX_enable # manual mode
echo 0 > pwmX

Does the fan stop?

echo 255 > pwmX

Does the fan run at full RPM?

Also set several intermediate values. Can you control the fan speed
successfully?

If you would be willing to perform these manual tests, that would
help. I'm at work and can't provide you with the exact directory where
the sysfs files are located, but you should be able to find them by
looking around in /sys/class/hwmon. The last directory will be a
symlink to another place in the sysfs where the files are.

David

[lm-sensors] w83627ehf informations

[David Hubbard]

Hi Torsten,

This is the way the driver is designed to work. 0 is an invalid value
for pwm1_enable, and the driver returns EINVAL. As a result, the
write() function that bash uses returns the error, and bash reports
it. (Did that explanation make sense?)

So echo 1 > pwm1_enable and then echo some values > pwm1

Do you get the results you expect?

The values which pwm1_enable will accept are:
1 = manual mode
2 = thermal cruise

In the future, we will support the following values as well, but right
now they return EINVAL:
3 = rpm cruise
4 = smartfan III (thermal cruise)

This information can also be found in
/usr/src/linux/Documentation/hwmon, which you may find useful.

David

On 11/18/06, torsten goroll <2crazy4you at gmx.de> wrote:
> Hi all,
>
> strange things are going on:
>
> "echo 1 > /sys/class/hwmon/hwmon0/device/pwm1_enable" does work
>
> "echo 2 > /sys/class/hwmon/hwmon0/device/pwm1_enable" does work
>
> "echo 0 > /sys/class/hwmon/hwmon0/device/pwm1_enable" does not work: "bash: echo: write error: Invalid argument"
>
> "echo 0>/sys/class/hwmon/hwmon0/device/pwm1_enable" does not work but doesn write an error message
>
> if I try to write the 0 with vi: ""pwm1_enable" E667: Fsync failed"
>
> I'm not that familiar with linux to guess whats going on here. I will try to google a bit today afternoon, but if anybody has an idea I would be thankful.
>
> Torsten
> --
> Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
> Ideal f?r Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
>

[lm-sensors] w83627ehf informations

[Jean Delvare]

On Fri, 17 Nov 2006 16:04:18 -0700, David Hubbard wrote:
> Hi Torsten,
> 
> The fix you found below may be interesting.
> 
> On 11/17/06, torsten goroll <2crazy4you at gmx.de> wrote:
> > Hi David,
> >
> > it took me a long time, but now I habe a 2.6.19-rc5 kernel runnig. The sensors output looks OK:
> >
> > w83627ehf-isa-0290
> > Adapter: ISA adapter
> > VCore:     +1.38 V  (min =  +0.00 V, max =  +1.74 V)
> > in1:      +12.25 V  (min =  +2.53 V, max =  +3.01 V) ALARM
> > AVCC:      +3.25 V  (min =  +2.19 V, max =  +0.37 V) ALARM
> > 3VCC:      +3.25 V  (min =  +0.34 V, max =  +1.34 V) ALARM
> > in4:       +1.64 V  (min =  +0.38 V, max =  +0.28 V) ALARM
> > in5:       +1.71 V  (min =  +0.04 V, max =  +1.59 V) ALARM
> > in6:       +5.50 V  (min =  +2.92 V, max =  +1.51 V) ALARM
> > VSB:       +3.25 V  (min =  +2.37 V, max =  +1.68 V) ALARM
> > VBAT:      +0.51 V  (min =  +0.50 V, max =  +0.72 V)
> > in9:       +1.58 V  (min =  +0.06 V, max =  +0.26 V) ALARM
> > Case Fan:  883 RPM  (min =  677 RPM, div = 8)
> > CPU Fan:  2295 RPM  (min = 3013 RPM, div = 4) ALARM
> > Aux Fan:     0 RPM  (min =   45 RPM, div = 128) ALARM
> > fan5:        0 RPM  (min = 23275 RPM, div = 2) ALARM
> > Sys Temp:    +27?C  (high =  -115?C, hyst =    -8?C)   ALARM
> > CPU Temp:  +21.5?C  (high = +80.0?C, hyst = +75.0?C)
> > AUX Temp:  +24.0?C  (high = +80.0?C, hyst = +75.0?C)
> >
> > But when I trie to run pwm-config I'm getting those (error)messages:
> > We will attempt to briefly stop each fan using the pwm controls.
> > The program will attempt to restore each fan to full speed
> > after testing. However, it is ** very important ** that you
> > physically verify that the fans have been to full speed
> > after the program has completed.
> >
> > Found the following PWM controls:
> >    9191-0290/pwm1
> > /usr/local/sbin/pwmconfig: line 102: echo: write error: Invalid argument
> >    9191-0290/pwm2
> > /usr/local/sbin/pwmconfig: line 102: echo: write error: Invalid argument
> >    9191-0290/pwm3
> > /usr/local/sbin/pwmconfig: line 102: echo: write error: Invalid argument
> >
> > Found the following fan sensors:
> >    9191-0290/fan1_input     current speed: 888 RPM
> >    9191-0290/fan2_input     current speed: 2311 RPM
> >    9191-0290/fan3_input     current speed: 0 ... skipping!
> >    9191-0290/fan5_input     current speed: 0 ... skipping!
> >
> > Warning!!! This program will stop your fans, one at a time,
> > for approximately 5 seconds each!!!
> > This may cause your processor temperature to rise!!!
> > If you do not want to do this hit control-C now!!!
> > Hit return to continue:
> >
> > Testing pwm control 9191-0290/pwm1 ...
> >   9191-0290/fan1_input ... speed was 888 now 883
> > /usr/local/sbin/pwmconfig: line 102: echo: write error: Invalid argument
> >     no correlation
> >   9191-0290/fan2_input ... speed was 2311 now 2295
> > /usr/local/sbin/pwmconfig: line 102: echo: write error: Invalid argument
> >     no correlation
> >
> > Checking the script I found the error. There must not be spaces in line 102.
> > not: echo 0 > $ENABLE
> > better:  echo 0>$ENABLE
> 
> This is important information. Who maintains the pwmconfig script?

Nobody in particular (unfortunately.)

> (Jean, I CC'ed you in case you have more information.) Is this a
> genuine bug?

No. Without the space between "0" and ">", you are doing I/O
redirection, which is something completely different. So it can't be a
valid fix.

It is expected that this echo command can fail, if the driver doesn't
support disabling fan speed control. However the failure should be
silent. I wonder why the error shows in Torsten's case. Torsten, what
version of bash are you using? Is the pwmconfig script somehow modified
by your distribution? Or did you modify it by yourself? For now, you
can just ignore the error messages.

-- 
Jean Delvare

[lm-sensors] w83627ehf informations

[Jean Delvare]

On Sun, 19 Nov 2006 16:51:33 +0100, torsten goroll wrote:
> thanks for your answer. The version of the bash is 3.1.
> I didn't modify it and compiled lm-sensors from the current
> sources.

Ok, it appears to be a new feature of bash 3.1: it displays an error
message when a write (echo) to a file returns an error. Previous
versions of bash didn't. So we need to hide the error in pwmconfig (and
fancontrol). And while we're at it, we could handle that error
properly, as it seems we weren't so far.

Patch is attached, I can't test it right now, so testers are wanted and
feedback is welcome. If the patch works OK, I'll apply it before we
release lm_sensors 2.10.2.

-- 
Jean Delvare
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: pwm-enable-error-check.patch
Url: http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20061226/784e1217/attachment.ksh 

[lm-sensors] w83627ehf informations

[slompf]

Hi,

the patch does work. Though, the script still finds no correlations on 
my board.

Torsten


> On Sun, 19 Nov 2006 16:51:33 +0100, torsten goroll wrote:
>   
>> thanks for your answer. The version of the bash is 3.1.
>> I didn't modify it and compiled lm-sensors from the current
>> sources.
>>     
>
> Ok, it appears to be a new feature of bash 3.1: it displays an error
> message when a write (echo) to a file returns an error. Previous
> versions of bash didn't. So we need to hide the error in pwmconfig (and
> fancontrol). And while we're at it, we could handle that error
> properly, as it seems we weren't so far.
>
> Patch is attached, I can't test it right now, so testers are wanted and
> feedback is welcome. If the patch works OK, I'll apply it before we
> release lm_sensors 2.10.2.
>   

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20070101/f54b7df0/attachment.html 

[lm-sensors] w83627ehf informations

[Jean Delvare]

Hi Torsten,

On Mon, 01 Jan 2007 12:03:50 +0100, slompf wrote:
> the patch does work. Though, the script still finds no correlations on 
> my board.

Well, maybe your board is actually not wired for fan speed control, or
you need to change the base PWM frequency - but the driver doesn't
support this yet. Do you have any reason to believe that your board
supports fan speed control (documentation, BIOS options, works under a
different OS)?

-- 
Jean Delvare