[lm-sensors] w83627ehf: Missing fan input and minor problems w/ PWM

[lm-sensors] w83627ehf: Missing fan input and minor problems w/ PWM

[Matthias Hentges]

Hi all,

since a few weeks ago I am the proud owner of an Asus P5W DH Deluxe
mainboard. It features the w83627ehf chip for hardware monitoring. While
all in all the monitoring works allright, there are a few minor
bugs in this driver. Before I continue let me say that I have fans
connected to every connector on this board.

- The board has 5 fan connectors (1x CPU, 2xCHA, 2xPWR) and 5 RPM
readings can be accessed in the BIOS, the driver however, only shows 4
fans. It would be nice to get a reading of the missing fan =)

- PWM control is a little bit confused:
	- pwm1 controls _both_ CHA fans, not only one of them (the 2nd CHA fan
is the one missing from 
	the driver)
	- pwm2 works just fine and controls the CPU fan
	- pwm3 doesn't change any fans at all
	- pwm4 controls _both_ PWR fans, not only one of them

It would be nice to get a pwm5 for the 5th fan and have every pwm
control a single fan =)

FWIW: Im running 2.6.18-rc4-mm3 right now.
If you need more information or a willing test-subject, let me know.

-- 
Matthias 'CoreDump' Hentges 

Webmaster of hentges.net and OpenZaurus developer.
You can reach me in #openzaurus on Freenode.

My OS: Debian SID. Geek by Nature, Linux by Choice
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Dies ist ein digital signierter Nachrichtenteil
Url : http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20060905/26cacaa5/attachment.bin 

[lm-sensors] w83627ehf: Missing fan input and minor problems w/

[David Hubbard]

Hi Matthias,

Nice to get some feedback on w83627ehf. The datasheet documents only 4
RPM sensors, but 5 pwm outputs. That's if your motherboard uses the
standard w83627ehf without any additional electronics.

Can you disconnect one fan at a time while in the BIOS readings and
verify that each is independent of all other fans? I wonder if the
BIOS is actually getting 5 independent RPM readings.

Also, is there some way (a different OS, or in the BIOS) that you can
actually control each of the fans independently? I'm looking for more
info on your motherboard, but there is the possibility that the two
CHA fans and the two PWR fans are actually hard-wired together. The
Asus page (http://www.asus.com/products4.aspx?modelmenu=1&model\x1198&l1=3&l2\x11&l3$8)
says you have the Intel 975X and Intel ICH7R chipsets, but that's all
the time I have right now to google...  If you discover on google that
there are similar questions on forums, etc., that might provide a
clue. Otherwise, look at the board and see if there are any additional
chips - I have an Asus motherboard and the w83627ehf chip is actually
not hard to find. On mine, it's in the corner next to the ram slots.

Do the fans have three or four pins? (Just curious)

Can you provide a listing of the sysfs files exported by the driver on
your system? The w83627ehf pins for fans 4 and 5 can be programmed as
general purpose I/O, and the driver detects whether you have 3, 4, or
5 fan pins configured. (The setting is read, but not changed.) So...do
you have pwm4 and pwm5 in your sysfs?

HTH,
David

On 9/5/06, Matthias Hentges <oe at hentges.net> wrote:
> Hi all,
>
> since a few weeks ago I am the proud owner of an Asus P5W DH Deluxe
> mainboard. It features the w83627ehf chip for hardware monitoring. While
> all in all the monitoring works allright, there are a few minor
> bugs in this driver. Before I continue let me say that I have fans
> connected to every connector on this board.
>
> - The board has 5 fan connectors (1x CPU, 2xCHA, 2xPWR) and 5 RPM
> readings can be accessed in the BIOS, the driver however, only shows 4
> fans. It would be nice to get a reading of the missing fan =)
>
> - PWM control is a little bit confused:
>         - pwm1 controls _both_ CHA fans, not only one of them (the 2nd CHA fan
> is the one missing from
>         the driver)
>         - pwm2 works just fine and controls the CPU fan
>         - pwm3 doesn't change any fans at all
>         - pwm4 controls _both_ PWR fans, not only one of them
>
> It would be nice to get a pwm5 for the 5th fan and have every pwm
> control a single fan =)
>
> FWIW: Im running 2.6.18-rc4-mm3 right now.
> If you need more information or a willing test-subject, let me know.
>
> --
> Matthias 'CoreDump' Hentges

[lm-sensors] w83627ehf: Missing fan input and minor problems w/

[Matthias Hentges]

Hello David,

thanks for your reply.

On Tue, September 5, 2006 16:55, David Hubbard wrote:
> Hi Matthias,
>
> Nice to get some feedback on w83627ehf. The datasheet documents only 4
> RPM sensors, but 5 pwm outputs. That's if your motherboard uses the
> standard w83627ehf without any additional electronics.
>
> Can you disconnect one fan at a time while in the BIOS readings and
> verify that each is independent of all other fans? I wonder if the
> BIOS is actually getting 5 independent RPM readings.

I will do so when I get home. However, I am pretty sure that these 5
readings are independent of each other.

> Also, is there some way (a different OS, or in the BIOS) that you can
> actually control each of the fans independently?

Windows users report that they can indeed controll all 5 fans  with a tool
called "SpeedFan" or something like that. I will investigate =)

> I'm looking for more
> info on your motherboard, but there is the possibility that the two
> CHA fans and the two PWR fans are actually hard-wired together. The
> Asus page
> (http://www.asus.com/products4.aspx?modelmenu=1&model\x1198&l1=3&l2\x11&l3$8)
> says you have the Intel 975X and Intel ICH7R chipsets, but that's all
> the time I have right now to google...  If you discover on google that
> there are similar questions on forums, etc., that might provide a
> clue. Otherwise, look at the board and see if there are any additional
> chips - I have an Asus motherboard and the w83627ehf chip is actually
> not hard to find. On mine, it's in the corner next to the ram slots.
>

I will have a look.

> Do the fans have three or four pins? (Just curious)

The CPU fan has 4 pins, the other 4 only have 3.

> Can you provide a listing of the sysfs files exported by the driver on
> your system? The w83627ehf pins for fans 4 and 5 can be programmed as
> general purpose I/O, and the driver detects whether you have 3, 4, or
> 5 fan pins configured. (The setting is read, but not changed.) So...do
> you have pwm4 and pwm5 in your sysfs?

I will send you the complete listing when I get home but I can already say
that there is

pwm1 to pwm4, no pwm5

Thanks for your time

-- 
Matthias Hentges

[lm-sensors] w83627ehf: Missing fan input and minor problems w/

[Jean Delvare]

David, Matthias,

> Nice to get some feedback on w83627ehf. The datasheet documents only 4
> RPM sensors, but 5 pwm outputs. That's if your motherboard uses the
> standard w83627ehf without any additional electronics.

No, that's the other way around. 5 RPM sensors, 4 PWM output.

> Can you disconnect one fan at a time while in the BIOS readings and
> verify that each is independent of all other fans? I wonder if the
> BIOS is actually getting 5 independent RPM readings.

The number of readings in the BIOS matches the number of fan monitoring
inputs of the chip - looks clear enough to me.

> Also, is there some way (a different OS, or in the BIOS) that you can
> actually control each of the fans independently? I'm looking for more
> info on your motherboard, but there is the possibility that the two
> CHA fans and the two PWR fans are actually hard-wired together.

This is my conclusion as well. The PWM outputs wiring is decided by the
motherboard maker, and they may use the same PWM output for several
fans. Given that there are only 4 PWM outputs they actually had little
choice for at least two fans. It makes sense to me to control related
fans with a single command too.

> Can you provide a listing of the sysfs files exported by the driver on
> your system? The w83627ehf pins for fans 4 and 5 can be programmed as
> general purpose I/O, and the driver detects whether you have 3, 4, or
> 5 fan pins configured. (The setting is read, but not changed.) So...do
> you have pwm4 and pwm5 in your sysfs?

Matthias, can you please provide a dump of your chip (isadump 0x295
0x296) at init time, before loading the w83627ehf driver?

I'd try forcing the driver to use all 5 fan inputs. Just run "isaset
0x295 0x296 0x47 0x05 0x05" before loading the w83627ehf driver, and
see what happens. It's supposed to be the default, but...

Maybe our heuristics to detect the use of the pins are not correct,
although they are simple enough and look OK to me. Or maybe the Asus
BIOS isn't configuring the chip properly (that'd be odd as they need
proper setup for displaying the values themselves.)

-- 
Jean Delvare

[lm-sensors] w83627ehf: Missing fan input and minor problems w/

[David Hubbard]

Hi Matthias,

Jean is right. I mixed up the PWM and RPM numbers there.

These things would be useful:

> Matthias, can you please provide a dump of your chip (isadump 0x295
> 0x296) at init time, before loading the w83627ehf driver?
>
> I'd try forcing the driver to use all 5 fan inputs. Just run "isaset
> 0x295 0x296 0x47 0x05 0x05" before loading the w83627ehf driver, and
> see what happens. It's supposed to be the default, but...
>
> Maybe our heuristics to detect the use of the pins are not correct,
> although they are simple enough and look OK to me. Or maybe the Asus
> BIOS isn't configuring the chip properly (that'd be odd as they need
> proper setup for displaying the values themselves.)

Matthias wrote:
> Windows users report that they can indeed controll all 5 fans  with a
> tool called "SpeedFan" or something like that. I will investigate =)

SpeedFan is indeed a useful tool. The author develops hardware
monitoring drivers for quite a lot of chips out there. I have used
SpeedFan on my machine, and the results match what I get from the
w83627ehf driver. But your motherboard is doing something new, in
order to drive 5 fans.

David

[lm-sensors] w83627ehf: Missing fan input and minor problems w/

[Matthias Hentges]

Am Dienstag, den 05.09.2006, 18:20 +0200 schrieb Jean Delvare:
> David, Matthias,
> 
> > Nice to get some feedback on w83627ehf. The datasheet documents only 4
> > RPM sensors, but 5 pwm outputs. That's if your motherboard uses the
> > standard w83627ehf without any additional electronics.
> 
> No, that's the other way around. 5 RPM sensors, 4 PWM output.
> 
> > Can you disconnect one fan at a time while in the BIOS readings and
> > verify that each is independent of all other fans? I wonder if the
> > BIOS is actually getting 5 independent RPM readings.
> 
> The number of readings in the BIOS matches the number of fan monitoring
> inputs of the chip - looks clear enough to me.

Indeed, here's what the BIOS looks like:
http://hentges.net/tmp/photos/p5w%20dh%20deluxe/hwmon/

> > Also, is there some way (a different OS, or in the BIOS) that you can
> > actually control each of the fans independently? I'm looking for more
> > info on your motherboard, but there is the possibility that the two
> > CHA fans and the two PWR fans are actually hard-wired together.
> 
> This is my conclusion as well. The PWM outputs wiring is decided by the
> motherboard maker, and they may use the same PWM output for several
> fans. Given that there are only 4 PWM outputs they actually had little
> choice for at least two fans. It makes sense to me to control related
> fans with a single command too.

Looks like you are right, SpeedFan on the "other" OS behaves exactly
like lm-sensors.
both CHA and both PWR fans can only be controlled together :\

> > Can you provide a listing of the sysfs files exported by the driver on
> > your system?

The listing is attached.

>  The w83627ehf pins for fans 4 and 5 can be programmed as
> > general purpose I/O, and the driver detects whether you have 3, 4, or
> > 5 fan pins configured. (The setting is read, but not changed.) So...do
> > you have pwm4 and pwm5 in your sysfs?
> 
> Matthias, can you please provide a dump of your chip (isadump 0x295
> 0x296) at init time, before loading the w83627ehf driver?

The dump is attached. It was taken before loading any modules
(init=/bin/sh).

> I'd try forcing the driver to use all 5 fan inputs. Just run "isaset
> 0x295 0x296 0x47 0x05 0x05" before loading the w83627ehf driver, and
> see what happens. It's supposed to be the default, but...

I tried it, no success :\ I still only get 4 fans.

> Maybe our heuristics to detect the use of the pins are not correct,
> although they are simple enough and look OK to me. Or maybe the Asus
> BIOS isn't configuring the chip properly (that'd be odd as they need
> proper setup for displaying the values themselves.)
> 
-- 
Matthias 'CoreDump' Hentges 

Webmaster of hentges.net and OpenZaurus developer.
You can reach me in #openzaurus on Freenode.

My OS: Debian SID. Geek by Nature, Linux by Choice
-------------- next part --------------
     0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
00: 04 ff 04 ff 00 2b 3f 33 c0 9b c0 9b 3c 3c 01 01 
10: 01 5e 70 00 00 01 01 3c 43 17 00 00 57 00 00 d3 
20: 8f ef d3 d3 d5 cb cc 27 ad 89 a9 da 00 f7 fe ff 
30: 7f 57 ed 1f fd f9 fe bf df 3c 37 0f 78 0f 0a 55 
40: 03 fa af de ff ff 00 f4 2d ff 00 c4 98 95 06 a3 
50: e4 1f 59 96 3a 39 3d 3d ff ff ff ff ff ff ff ff 
60: 04 ff 03 21 f0 f0 3c ff 01 ff 01 ff ff ff ff ff 
70: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff 
80: 04 ff 04 ff 00 2b 3f 33 c0 9b c0 9b 3c 3c 01 01 
90: 01 5e 70 00 00 01 01 3c 43 17 00 00 57 00 00 d3 
a0: 8f ef d3 d3 d5 cb cc 27 ad 89 a9 da 00 f7 fe ff 
b0: 7f 57 ed 1f fd f9 fe bf df 3c 37 0f 78 0f 0a 55 
c0: 03 00 00 de ff ff 00 f4 2d ff 00 c4 98 95 06 a3 
d0: e4 1f 59 96 3a 39 3d 3d ff ff ff ff ff ff ff ff 
e0: 04 ff 03 21 f0 f0 3c ff 01 ff 01 ff ff ff ff ff 
f0: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff 
-------------- next part --------------
bus
driver
fan1_alarm
fan1_div
fan1_input
fan1_min
fan2_alarm
fan2_div
fan2_input
fan2_min
fan3_alarm
fan3_div
fan3_input
fan3_min
fan4_alarm
fan4_div
fan4_input
fan4_min
hwmon:hwmon0
in0_alarm
in0_input
in0_max
in0_min
in1_alarm
in1_input
in1_max
in1_min
in2_alarm
in2_input
in2_max
in2_min
in3_alarm
in3_input
in3_max
in3_min
in4_alarm
in4_input
in4_max
in4_min
in5_alarm
in5_input
in5_max
in5_min
in6_alarm
in6_input
in6_max
in6_min
in7_alarm
in7_input
in7_max
in7_min
in8_alarm
in8_input
in8_max
in8_min
in9_alarm
in9_input
in9_max
in9_min
name
power
pwm1
pwm1_enable
pwm1_min_output
pwm1_mode
pwm1_stop_time
pwm1_target
pwm1_tolerance
pwm2
pwm2_enable
pwm2_min_output
pwm2_mode
pwm2_stop_time
pwm2_target
pwm2_tolerance
pwm3
pwm3_enable
pwm3_min_output
pwm3_mode
pwm3_stop_time
pwm3_target
pwm3_tolerance
pwm4
pwm4_enable
pwm4_min_output
pwm4_mode
pwm4_stop_time
pwm4_target
pwm4_tolerance
subsystem
temp1_alarm
temp1_input
temp1_max
temp1_max_hyst
temp2_alarm
temp2_input
temp2_max
temp2_max_hyst
temp3_alarm
temp3_input
temp3_max
temp3_max_hyst
uevent
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Dies ist ein digital signierter Nachrichtenteil
Url : http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20060906/33bd7775/attachment.bin 

[lm-sensors] w83627ehf: Missing fan input and minor problems w/

[Jean Delvare]

Guten Tag Matthias,

> Am Dienstag, den 05.09.2006, 18:20 +0200 schrieb Jean Delvare:
> > > Can you provide a listing of the sysfs files exported by the driver on
> > > your system?
> 
> The listing is attached.

OK, it actually lacks the fan5 files, so that's a kernel issue and not
a user-space issue.

> >  The w83627ehf pins for fans 4 and 5 can be programmed as
> > > general purpose I/O, and the driver detects whether you have 3, 4, or
> > > 5 fan pins configured. (The setting is read, but not changed.) So...do
> > > you have pwm4 and pwm5 in your sysfs?
> > 
> > Matthias, can you please provide a dump of your chip (isadump 0x295
> > 0x296) at init time, before loading the w83627ehf driver?
> 
> The dump is attached. It was taken before loading any modules
> (init=/bin/sh).

It indeed shows fan5 as disabled.

> > I'd try forcing the driver to use all 5 fan inputs. Just run "isaset
> > 0x295 0x296 0x47 0x05 0x05" before loading the w83627ehf driver, and
> > see what happens. It's supposed to be the default, but...
> 
> I tried it, no success :\ I still only get 4 fans.

Odd. Did "isaset" complain about failed readback?

The easiest way to make sure is:

isadump -y 0x295 0x296
isaset -y 0x295 0x296 0x47 0xf5
isadump -y 0x295 0x296

The value of register 0x47 should change in the second dump.

-- 
Jean Delvare

[lm-sensors] w83627ehf: Missing fan input and minor problems w/

[Matthias Hentges]

Am Mittwoch, den 06.09.2006, 16:01 +0200 schrieb Jean Delvare:
> Guten Tag Matthias,
> 
> > Am Dienstag, den 05.09.2006, 18:20 +0200 schrieb Jean Delvare:
> > > > Can you provide a listing of the sysfs files exported by the driver on
> > > > your system?
> > 
> > The listing is attached.
> 
> OK, it actually lacks the fan5 files, so that's a kernel issue and not
> a user-space issue.
> 
> > >  The w83627ehf pins for fans 4 and 5 can be programmed as
> > > > general purpose I/O, and the driver detects whether you have 3, 4, or
> > > > 5 fan pins configured. (The setting is read, but not changed.) So...do
> > > > you have pwm4 and pwm5 in your sysfs?
> > > 
> > > Matthias, can you please provide a dump of your chip (isadump 0x295
> > > 0x296) at init time, before loading the w83627ehf driver?
> > 
> > The dump is attached. It was taken before loading any modules
> > (init=/bin/sh).
> 
> It indeed shows fan5 as disabled.
> 
> > > I'd try forcing the driver to use all 5 fan inputs. Just run "isaset
> > > 0x295 0x296 0x47 0x05 0x05" before loading the w83627ehf driver, and
> > > see what happens. It's supposed to be the default, but...
> > 
> > I tried it, no success :\ I still only get 4 fans.
> 
> Odd. Did "isaset" complain about failed readback?

Err now that you mention it heh.

> The easiest way to make sure is:
> 
> isadump -y 0x295 0x296
> isaset -y 0x295 0x296 0x47 0xf5
> isadump -y 0x295 0x296
> 
> The value of register 0x47 should change in the second dump.

I'm indeed getting an error from isaset: 
"Data mismatch, wrote 0xf5, read 0xf4 back."

I just tried again to be sure.
-- 
Matthias 'CoreDump' Hentges 

Webmaster of hentges.net and OpenZaurus developer.
You can reach me in #openzaurus on Freenode.

My OS: Debian SID. Geek by Nature, Linux by Choice
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Dies ist ein digital signierter Nachrichtenteil
Url : http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20060907/a47f1a43/attachment.bin 

[lm-sensors] w83627ehf: Missing fan input and minor problems w/

[Jean Delvare]

Hallo Matthias,

> > > I tried it, no success :\ I still only get 4 fans.
> > 
> > Odd. Did "isaset" complain about failed readback?
> 
> Err now that you mention it heh.
> 
> > The easiest way to make sure is:
> > 
> > isadump -y 0x295 0x296
> > isaset -y 0x295 0x296 0x47 0xf5
> > isadump -y 0x295 0x296
> > 
> > The value of register 0x47 should change in the second dump.
> 
> I'm indeed getting an error from isaset: 
> "Data mismatch, wrote 0xf5, read 0xf4 back."
> 
> I just tried again to be sure.

Odd, this register is supposed to be writable. Can you please try other
values (e.g. 0xa4) to see if other bits in the register can be written?

You may also try writing to register 0x48, we don't care about the
value there and it'll be an interesting test.

Pin 58 is shared between two functions, SI and AUXFANIN1. Your problem
might have something to do with this, but I can't find information
about function selection for this pin in the datasheet.

David, any idea?

-- 
Jean Delvare

[lm-sensors] w83627ehf: Missing fan input and minor problems w/

[Rudolf Marek]

> Hallo Matthias,

Please can you send me a DSDT table from your computer? (just in private because
binary blobs are discarded by the list)

So please do following:

cat > /proc/acpi/dsdt > /tmp/dsdt.bin.

I will check the ACPI bytecode to see how the fans are handled.

Regards
Rudolf

[lm-sensors] w83627ehf: Missing fan input and minor problems w/

[Matthias Hentges]

Hello Jean,

Am Sonntag, den 10.09.2006, 19:48 +0200 schrieb Jean Delvare:
> Hallo Matthias,

> > I'm indeed getting an error from isaset: 
> > "Data mismatch, wrote 0xf5, read 0xf4 back."
> > 
> > I just tried again to be sure.
> 
> Odd, this register is supposed to be writable. Can you please try other
> values (e.g. 0xa4) to see if other bits in the register can be written?

Interesting, I can set the value 0xa4 (and then back to 0xf4) just fine.
BTW: I did this while the driver was loaded.

> You may also try writing to register 0x48, we don't care about the
> value there and it'll be an interesting test.

+root at mhcln01:/tmp >isaset -y 0x295 0x296 0x48 0xa5 
Data mismatch, wrote 0xa5, read 0x25 back.
+root at mhcln01:/tmp >isaset -y 0x295 0x296 0x48 0xa6
Data mismatch, wrote 0xa6, read 0x26 back.
+root at mhcln01:/tmp >isaset -y 0x295 0x296 0x48 0xb6
Data mismatch, wrote 0xb6, read 0x36 back.
+root at mhcln01:/tmp >isaset -y 0x295 0x296 0x48 0xc6
Data mismatch, wrote 0xc6, read 0x46 back.
+root at mhcln01:/tmp >

Dunno if that was to be expected heh.

> Pin 58 is shared between two functions, SI and AUXFANIN1. Your problem
> might have something to do with this, but I can't find information
> about function selection for this pin in the datasheet.
> 
> David, any idea?
> 
-- 
Matthias 'CoreDump' Hentges 

Webmaster of hentges.net and OpenZaurus developer.
You can reach me in #openzaurus on Freenode.

My OS: Debian SID. Geek by Nature, Linux by Choice
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Dies ist ein digital signierter Nachrichtenteil
Url : http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20060910/d6e9aca6/attachment.bin 

[lm-sensors] w83627ehf: Missing fan input and minor problems w/

[Rudolf Marek]

Hello all

I checked the ACPI DSDT and they set the bit0 of 0x47... in every change of the
the 47reg.

               And (FDR1, 0x3F, Local3)

                Or (Local2, Local3, Local3)

                Or (Local3, 0x01, Local3)

                Store (Local3, FDR1)

(FDR1 is 0x47)

I did not find out how they force the bit to be R/W. I remember I asked already
at Winbond how to know when fan5 is enabled, and they told me that the serial
flash interface should be disabled (ENROM cr24) And then FAN5 (AUXFANIN1) works.
This way it is implemented in newer EHF patches.

Well in my case is the bit0/47 is RO too (I own the EHF chip) - so writing 0xf5
fails same as in Matthias case... even if the serial flash is disabled,
monitoring stopped...

I'm out of ideas... but ;)

Yuan, please can you ask someone:

How is it possible that the bit0/of reg 47 in hardware monitor is read only (1
cannot be written) even when serial flash interface is disabled and CR24 bit 1
in sio (ENROM strap says 0)?

In other words:
What should I do to enable the AUXFANIN1 or why is the bit0 at 47 readonly.

I have ID 88 63  so it seems it is version C


Thank you very much.
Regards
Rudolf

[lm-sensors] w83627ehf: Missing fan input and minor problems w/

[David Hubbard]

Hi Jean,

> Odd, this register is supposed to be writable. Can you please try other
> values (e.g. 0xa4) to see if other bits in the register can be written?
>
> You may also try writing to register 0x48, we don't care about the
> value there and it'll be an interesting test.
>
> Pin 58 is shared between two functions, SI and AUXFANIN1. Your problem
> might have something to do with this, but I can't find information
> about function selection for this pin in the datasheet.
>
> David, any idea?

There's been a lot going on around here, so sorry for the delay. I
looked at the datasheet to see how pin 58 is configured. SI (serial
flash input) and AUXFANIN1 are affected by two registers:

1. register 47h: force AUXFANIN1 to act as a fan output
2. CR[0x24] bit 1. ENROM. The datasheet says a hardware reset is necessary.

My guess is that the serial flash interface is set by default. I think
there's a way to reset the w83627ehf chip by writing to certain ports.
What do you think, Jean?

David

P.S. Working on an updated w83627dhg patch!

[lm-sensors] w83627ehf: Missing fan input and minor problems

[Rudolf Marek]

Hi all,

Good news everyone, the mystery of fan5 is now uncloaked. It is short story. The
bit0 in reg 47 is write only and reads always 0.

> 1. register 47h: force AUXFANIN1 to act as a fan output
> 2. CR[0x24] bit 1. ENROM. The datasheet says a hardware reset is necessary.

Yes this is true, we have the code which detects this already in the
wmon-w83627ehf-add-pwm-support.patch

I have hacked together a patch that should solve the problem. It checks for the
ENROM as it was, later it checks bit 1 instead of bit 0 which is zero anyway.
Then all writes to reg 47 are modified so the bit0 is 1 if the ENROM is disabled.

The patch contains some schematics that is my idea how the pins are connected.
Problem is that we really never know if the pin is fan output or input because
we cannot read the bit :/

So bit1 means that output should be 1, this is not the default, so then I assume
this is not fan5 input and fan5 is disabled in driver.

If the pin is configured as output and the level in bit1 is set to 0, then bad
luck we will still think that it is input and the chip will be reprogrammed.

The schematics shows what will happen. The output of 0 will be switched to 1.
But there is no way how we can detect this. If there is really fan connected,
the side effect will be that the fan starts to spin ;) (if the polarity is set 1
on 0 off - it is mostly) The schematics shows the typical programmable I/O  pin.


All in all I think the usage as fan output for fan on/off is very very unlikely
and I have done what I could to detect if it is this way.

> P.S. Working on an updated w83627dhg patch!

Good.

Matthias, please can you check if the patch is working for you? There might be
some typos in code, please fix them if you can.

The patch should be applied on the top of:
http://khali.linux-fr.org/devel/i2c/linux-2.6/hwmon-w83627ehf-add-pwm-support.patch
http://khali.linux-fr.org/devel/i2c/linux-2.6/hwmon-unchecked-return-status-fixes-w83627ehf.patch

Thanks,
Rudolf

-------------- next part --------------
A non-text attachment was scrubbed...
Name: fan5fix.patch
Type: text/x-patch
Size: 2769 bytes
Desc: not available
Url : http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20060915/c0775b63/attachment.bin