[lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c on

[lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c on

[Mike Campin]

Hi,

Only 2 of the 5 fans are showing up on my ASRock P67 Extreme4 motherboard.
It due to the NCT6776 global configuration register 0x1c and 0x24 values.
The fans work if I modify these registers prior to loading the w83627ehf
driver.

My question is where are these registers initialized? Should I ask ASRock
to fix the BIOS?

  isadump -y -k 0x87,0x87 0x2e 0x2f

          0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
     00: ff ff 00 ff ff ff ff 02 ff ff ff ff ff ff ff ff 
     10: ff ff ff ff ff ff ff ff ff ff f8 0e 00 00 ff ff 
     20: c3 33 ff 00 5c 00 00 98 00 ff 20 00 80 00 00 01 
     30: 00 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff 

  isaset -y -f 0x2e 0x87
  isaset -y -f 0x2e 0x87
  isaset -y 0x2e 0x2f 0x1c 0x3
  isaset -y 0x2e 0x2f 0x24 0x1c

Thanks, Mike

-- 
Mike Campin
lm_sensors@ootsa.homelinux.net

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

Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c

[Ian Dobson]

Hi,

--------------------------------------------------
From: "Mike Campin" <lm_sensors@ootsa.homelinux.net>
Sent: Sunday, April 10, 2011 6:04 PM
To: <lm-sensors@lm-sensors.org>
Subject: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c onASRock 
Extreme4

> Hi,
>
> Only 2 of the 5 fans are showing up on my ASRock P67 Extreme4 motherboard.
> It due to the NCT6776 global configuration register 0x1c and 0x24 values.
> The fans work if I modify these registers prior to loading the w83627ehf
> driver.
>
> My question is where are these registers initialized? Should I ask ASRock
> to fix the BIOS?
>
>  isadump -y -k 0x87,0x87 0x2e 0x2f
>
>          0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
>     00: ff ff 00 ff ff ff ff 02 ff ff ff ff ff ff ff ff
>     10: ff ff ff ff ff ff ff ff ff ff f8 0e 00 00 ff ff
>     20: c3 33 ff 00 5c 00 00 98 00 ff 20 00 80 00 00 01
>     30: 00 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
>
>  isaset -y -f 0x2e 0x87
>  isaset -y -f 0x2e 0x87
>  isaset -y 0x2e 0x2f 0x1c 0x3
>  isaset -y 0x2e 0x2f 0x24 0x1c
>
> Thanks, Mike
>
> -- 
> Mike Campin
> lm_sensors@ootsa.homelinux.net
>
> _______________________________________________
> lm-sensors mailing list
> lm-sensors@lm-sensors.org
> http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

ASRock designed the board, so they should setup the hardware monitoring chip 
correctly. Allowing users to change the chip configuration is a really bad 
idea, as many of the pins can have different functionality/electrical 
spcifications depending on the config. Setting up a pin incorrectly could 
blow the chip (Configuring a pin so that it's a source that's connected to 
an unprotected/limited sink for example).

So ASRock should fix the BIOS.

Regards
Ian Dobson
 


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

Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c

[Jean Delvare]

On Tue, 12 Apr 2011 19:02:19 +0200, Ian Dobson wrote:
> Hi,
> 
> --------------------------------------------------
> From: "Mike Campin" <lm_sensors@ootsa.homelinux.net>
> Sent: Sunday, April 10, 2011 6:04 PM
> To: <lm-sensors@lm-sensors.org>
> Subject: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c onASRock 
> Extreme4
> 
> > Hi,
> >
> > Only 2 of the 5 fans are showing up on my ASRock P67 Extreme4 motherboard.
> > It due to the NCT6776 global configuration register 0x1c and 0x24 values.
> > The fans work if I modify these registers prior to loading the w83627ehf
> > driver.
> >
> > My question is where are these registers initialized? Should I ask ASRock
> > to fix the BIOS?
> >
> >  isadump -y -k 0x87,0x87 0x2e 0x2f
> >
> >          0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
> >     00: ff ff 00 ff ff ff ff 02 ff ff ff ff ff ff ff ff
> >     10: ff ff ff ff ff ff ff ff ff ff f8 0e 00 00 ff ff
> >     20: c3 33 ff 00 5c 00 00 98 00 ff 20 00 80 00 00 01
> >     30: 00 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> >
> >  isaset -y -f 0x2e 0x87
> >  isaset -y -f 0x2e 0x87
> >  isaset -y 0x2e 0x2f 0x1c 0x3
> >  isaset -y 0x2e 0x2f 0x24 0x1c
> >
> > Thanks, Mike
> >
> > -- 
> > Mike Campin
> > lm_sensors@ootsa.homelinux.net
> >
> > _______________________________________________
> > lm-sensors mailing list
> > lm-sensors@lm-sensors.org
> > http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
> 
> ASRock designed the board, so they should setup the hardware monitoring chip 
> correctly. Allowing users to change the chip configuration is a really bad 
> idea, as many of the pins can have different functionality/electrical 
> spcifications depending on the config. Setting up a pin incorrectly could 
> blow the chip (Configuring a pin so that it's a source that's connected to 
> an unprotected/limited sink for example).
> 
> So ASRock should fix the BIOS.

Seconded. But before flaming Asrock, two things worth checking:
* Availability of a BIOS update fixing the issue.
* Options in the BIOS to enable/disable the monitoring of specific
  fans. I can imagine that the BIOS skips the configuration steps for
  fans for which monitoring was disabled (although that would probably
  mean more code than just doing it unconditionally...)

-- 
Jean Delvare

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

[lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c onASRock

[Harald Judt]

Hi,

> On Tue, 12 Apr 2011 19:02:19 +0200, Ian Dobson wrote:
>> Hi,
>>
>> --------------------------------------------------
>> From: "Mike Campin" <lm_sensors at ootsa.homelinux.net>
>> Sent: Sunday, April 10, 2011 6:04 PM
>> To: <lm-sensors at lm-sensors.org>
>> Subject: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c onASRock
>> Extreme4
>>
>> > Hi,
>> >
>> > Only 2 of the 5 fans are showing up on my ASRock P67 Extreme4 motherboard.
>> > It due to the NCT6776 global configuration register 0x1c and 0x24 values.
>> > The fans work if I modify these registers prior to loading the w83627ehf
>> > driver.
>> >
>> > My question is where are these registers initialized? Should I ask ASRock
>> > to fix the BIOS?
>> >
>> >  isadump -y -k 0x87,0x87 0x2e 0x2f
>> >
>> >          0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
>> >     00: ff ff 00 ff ff ff ff 02 ff ff ff ff ff ff ff ff
>> >     10: ff ff ff ff ff ff ff ff ff ff f8 0e 00 00 ff ff
>> >     20: c3 33 ff 00 5c 00 00 98 00 ff 20 00 80 00 00 01
>> >     30: 00 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
>> >
>> >  isaset -y -f 0x2e 0x87
>> >  isaset -y -f 0x2e 0x87
>> >  isaset -y 0x2e 0x2f 0x1c 0x3
>> >  isaset -y 0x2e 0x2f 0x24 0x1c
>> >
>> > Thanks, Mike
>> >
>> > --
>> > Mike Campin
>> > lm_sensors at ootsa.homelinux.net
>> >
>> > _______________________________________________
>> > lm-sensors mailing list
>> > lm-sensors at lm-sensors.org
>> > http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
>>
>> ASRock designed the board, so they should setup the hardware monitoring chip
>> correctly. Allowing users to change the chip configuration is a really bad
>> idea, as many of the pins can have different functionality/electrical
>> spcifications depending on the config. Setting up a pin incorrectly could
>> blow the chip (Configuring a pin so that it's a source that's connected to
>> an unprotected/limited sink for example).
>>
>> So ASRock should fix the BIOS.
>
> Seconded. But before flaming Asrock, two things worth checking:
> * Availability of a BIOS update fixing the issue.
> * Options in the BIOS to enable/disable the monitoring of specific
>   fans. I can imagine that the BIOS skips the configuration steps for
>   fans for which monitoring was disabled (although that would probably
>   mean more code than just doing it unconditionally...)
>
> --
> Jean Delvare

I own an Asrock P67 Extreme4 too and have the same problem. It can be 
solved by following Mike Campin's instructions, which I successfully 
incorporated into /etc/init.d/lm_sensors. There is no BIOS available 
that fixes this issue, but I noticed that SpeedFan for Windows 
(www.almico.com/speedfan.php) seems to do something similar, otherwise 
it would show only 2 fans too. So maybe we could get that into 
lm_sensors, even if the BIOS is at fault here?

Regards,
Harald

-- 
`Experience is the best teacher.'

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

Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c

[Guenter Roeck]

Hi, 
On Wed, Dec 21, 2011 at 09:50:31AM -0500, Harald Judt wrote:
> Hi,
> 
> > On Tue, 12 Apr 2011 19:02:19 +0200, Ian Dobson wrote:
> >> Hi,
> >>
> >> --------------------------------------------------
> >> From: "Mike Campin" <lm_sensors at ootsa.homelinux.net>
> >> Sent: Sunday, April 10, 2011 6:04 PM
> >> To: <lm-sensors at lm-sensors.org>
> >> Subject: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c onASRock
> >> Extreme4
> >>
> >> > Hi,
> >> >
> >> > Only 2 of the 5 fans are showing up on my ASRock P67 Extreme4 motherboard.
> >> > It due to the NCT6776 global configuration register 0x1c and 0x24 values.
> >> > The fans work if I modify these registers prior to loading the w83627ehf
> >> > driver.
> >> >
> >> > My question is where are these registers initialized? Should I ask ASRock
> >> > to fix the BIOS?
> >> >
> >> >  isadump -y -k 0x87,0x87 0x2e 0x2f
> >> >
> >> >          0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
> >> >     00: ff ff 00 ff ff ff ff 02 ff ff ff ff ff ff ff ff
> >> >     10: ff ff ff ff ff ff ff ff ff ff f8 0e 00 00 ff ff
> >> >     20: c3 33 ff 00 5c 00 00 98 00 ff 20 00 80 00 00 01
> >> >     30: 00 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> >> >
> >> >  isaset -y -f 0x2e 0x87
> >> >  isaset -y -f 0x2e 0x87
> >> >  isaset -y 0x2e 0x2f 0x1c 0x3
> >> >  isaset -y 0x2e 0x2f 0x24 0x1c
> >> >
> >> > Thanks, Mike
> >> >
> >> > --
> >> > Mike Campin
> >> > lm_sensors at ootsa.homelinux.net
> >> >
> >> > _______________________________________________
> >> > lm-sensors mailing list
> >> > lm-sensors at lm-sensors.org
> >> > http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
> >>
> >> ASRock designed the board, so they should setup the hardware monitoring chip
> >> correctly. Allowing users to change the chip configuration is a really bad
> >> idea, as many of the pins can have different functionality/electrical
> >> spcifications depending on the config. Setting up a pin incorrectly could
> >> blow the chip (Configuring a pin so that it's a source that's connected to
> >> an unprotected/limited sink for example).
> >>
> >> So ASRock should fix the BIOS.
> >
> > Seconded. But before flaming Asrock, two things worth checking:
> > * Availability of a BIOS update fixing the issue.
> > * Options in the BIOS to enable/disable the monitoring of specific
> >   fans. I can imagine that the BIOS skips the configuration steps for
> >   fans for which monitoring was disabled (although that would probably
> >   mean more code than just doing it unconditionally...)
> >
> > --
> > Jean Delvare
> 
> I own an Asrock P67 Extreme4 too and have the same problem. It can be 
> solved by following Mike Campin's instructions, which I successfully 
> incorporated into /etc/init.d/lm_sensors. There is no BIOS available 
> that fixes this issue, but I noticed that SpeedFan for Windows 
> (www.almico.com/speedfan.php) seems to do something similar, otherwise 
> it would show only 2 fans too. So maybe we could get that into 
> lm_sensors, even if the BIOS is at fault here?
> 
See above ... worst case this can result in defective hardware if misconfigured.

We could possibly introduce some kind of module parameter, but I don't feel
too comfortable with that - wrong values may still blow up the hardware.
Any opinions, anyone ?

Guenter

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

Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c

[Ian Dobson]

Hi,
--------------------------------------------------
From: "Guenter Roeck" <guenter.roeck@ericsson.com>
Sent: Wednesday, December 21, 2011 6:01 PM
To: "Harald Judt" <h.judt@gmx.at>
Cc: <lm-sensors@lm-sensors.org>
Subject: Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c 
onASRock Extreme4

> Hi,
> On Wed, Dec 21, 2011 at 09:50:31AM -0500, Harald Judt wrote:
>> Hi,
>>
>> > On Tue, 12 Apr 2011 19:02:19 +0200, Ian Dobson wrote:
>> >> Hi,
>> >>
>> >> --------------------------------------------------
>> >> From: "Mike Campin" <lm_sensors at ootsa.homelinux.net>
>> >> Sent: Sunday, April 10, 2011 6:04 PM
>> >> To: <lm-sensors at lm-sensors.org>
>> >> Subject: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c 
>> >> onASRock
>> >> Extreme4
>> >>
>> >> > Hi,
>> >> >
>> >> > Only 2 of the 5 fans are showing up on my ASRock P67 Extreme4 
>> >> > motherboard.
>> >> > It due to the NCT6776 global configuration register 0x1c and 0x24 
>> >> > values.
>> >> > The fans work if I modify these registers prior to loading the 
>> >> > w83627ehf
>> >> > driver.
>> >> >
>> >> > My question is where are these registers initialized? Should I ask 
>> >> > ASRock
>> >> > to fix the BIOS?
>> >> >
>> >> >  isadump -y -k 0x87,0x87 0x2e 0x2f
>> >> >
>> >> >          0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
>> >> >     00: ff ff 00 ff ff ff ff 02 ff ff ff ff ff ff ff ff
>> >> >     10: ff ff ff ff ff ff ff ff ff ff f8 0e 00 00 ff ff
>> >> >     20: c3 33 ff 00 5c 00 00 98 00 ff 20 00 80 00 00 01
>> >> >     30: 00 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
>> >> >
>> >> >  isaset -y -f 0x2e 0x87
>> >> >  isaset -y -f 0x2e 0x87
>> >> >  isaset -y 0x2e 0x2f 0x1c 0x3
>> >> >  isaset -y 0x2e 0x2f 0x24 0x1c
>> >> >
>> >> > Thanks, Mike
>> >> >
>> >> > --
>> >> > Mike Campin
>> >> > lm_sensors at ootsa.homelinux.net
>> >> >
>> >> > _______________________________________________
>> >> > lm-sensors mailing list
>> >> > lm-sensors at lm-sensors.org
>> >> > http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
>> >>
>> >> ASRock designed the board, so they should setup the hardware 
>> >> monitoring chip
>> >> correctly. Allowing users to change the chip configuration is a really 
>> >> bad
>> >> idea, as many of the pins can have different functionality/electrical
>> >> spcifications depending on the config. Setting up a pin incorrectly 
>> >> could
>> >> blow the chip (Configuring a pin so that it's a source that's 
>> >> connected to
>> >> an unprotected/limited sink for example).
>> >>
>> >> So ASRock should fix the BIOS.
>> >
>> > Seconded. But before flaming Asrock, two things worth checking:
>> > * Availability of a BIOS update fixing the issue.
>> > * Options in the BIOS to enable/disable the monitoring of specific
>> >   fans. I can imagine that the BIOS skips the configuration steps for
>> >   fans for which monitoring was disabled (although that would probably
>> >   mean more code than just doing it unconditionally...)
>> >
>> > --
>> > Jean Delvare
>>
>> I own an Asrock P67 Extreme4 too and have the same problem. It can be
>> solved by following Mike Campin's instructions, which I successfully
>> incorporated into /etc/init.d/lm_sensors. There is no BIOS available
>> that fixes this issue, but I noticed that SpeedFan for Windows
>> (www.almico.com/speedfan.php) seems to do something similar, otherwise
>> it would show only 2 fans too. So maybe we could get that into
>> lm_sensors, even if the BIOS is at fault here?
>>
> See above ... worst case this can result in defective hardware if 
> misconfigured.
>
> We could possibly introduce some kind of module parameter, but I don't 
> feel
> too comfortable with that - wrong values may still blow up the hardware.
> Any opinions, anyone ?
>
> Guenter
>
> _______________________________________________
> lm-sensors mailing list
> lm-sensors@lm-sensors.org
> http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

I wouldn't recommend a module parameter, leave the driver as it is.
If there is a hack (using isaset) just use that. It's too easy for a user to 
set a module parameter hoping that it'll fix some random problem, and if the 
module kills the hardware, then the module is too blame.

Asrock should fix the BIOS, that's the correct option.

Regards
Ian Dobson
 


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

Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c

[Ira W. Snyder]

On Wed, Dec 21, 2011 at 06:55:31PM +0100, Ian Dobson wrote:
> Hi,
> --------------------------------------------------
> From: "Guenter Roeck" <guenter.roeck@ericsson.com>
> Sent: Wednesday, December 21, 2011 6:01 PM
> To: "Harald Judt" <h.judt@gmx.at>
> Cc: <lm-sensors@lm-sensors.org>
> Subject: Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c 
> onASRock Extreme4
> 
> > Hi,
> > On Wed, Dec 21, 2011 at 09:50:31AM -0500, Harald Judt wrote:
> >> Hi,
> >>
> >> > On Tue, 12 Apr 2011 19:02:19 +0200, Ian Dobson wrote:
> >> >> Hi,
> >> >>
> >> >> --------------------------------------------------
> >> >> From: "Mike Campin" <lm_sensors at ootsa.homelinux.net>
> >> >> Sent: Sunday, April 10, 2011 6:04 PM
> >> >> To: <lm-sensors at lm-sensors.org>
> >> >> Subject: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c 
> >> >> onASRock
> >> >> Extreme4
> >> >>
> >> >> > Hi,
> >> >> >
> >> >> > Only 2 of the 5 fans are showing up on my ASRock P67 Extreme4 
> >> >> > motherboard.
> >> >> > It due to the NCT6776 global configuration register 0x1c and 0x24 
> >> >> > values.
> >> >> > The fans work if I modify these registers prior to loading the 
> >> >> > w83627ehf
> >> >> > driver.
> >> >> >
> >> >> > My question is where are these registers initialized? Should I ask 
> >> >> > ASRock
> >> >> > to fix the BIOS?
> >> >> >
> >> >> >  isadump -y -k 0x87,0x87 0x2e 0x2f
> >> >> >
> >> >> >          0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
> >> >> >     00: ff ff 00 ff ff ff ff 02 ff ff ff ff ff ff ff ff
> >> >> >     10: ff ff ff ff ff ff ff ff ff ff f8 0e 00 00 ff ff
> >> >> >     20: c3 33 ff 00 5c 00 00 98 00 ff 20 00 80 00 00 01
> >> >> >     30: 00 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> >> >> >
> >> >> >  isaset -y -f 0x2e 0x87
> >> >> >  isaset -y -f 0x2e 0x87
> >> >> >  isaset -y 0x2e 0x2f 0x1c 0x3
> >> >> >  isaset -y 0x2e 0x2f 0x24 0x1c
> >> >> >
> >> >> > Thanks, Mike
> >> >> >
> >> >> > --
> >> >> > Mike Campin
> >> >> > lm_sensors at ootsa.homelinux.net
> >> >> >
> >> >> > _______________________________________________
> >> >> > lm-sensors mailing list
> >> >> > lm-sensors at lm-sensors.org
> >> >> > http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
> >> >>
> >> >> ASRock designed the board, so they should setup the hardware 
> >> >> monitoring chip
> >> >> correctly. Allowing users to change the chip configuration is a really 
> >> >> bad
> >> >> idea, as many of the pins can have different functionality/electrical
> >> >> spcifications depending on the config. Setting up a pin incorrectly 
> >> >> could
> >> >> blow the chip (Configuring a pin so that it's a source that's 
> >> >> connected to
> >> >> an unprotected/limited sink for example).
> >> >>
> >> >> So ASRock should fix the BIOS.
> >> >
> >> > Seconded. But before flaming Asrock, two things worth checking:
> >> > * Availability of a BIOS update fixing the issue.
> >> > * Options in the BIOS to enable/disable the monitoring of specific
> >> >   fans. I can imagine that the BIOS skips the configuration steps for
> >> >   fans for which monitoring was disabled (although that would probably
> >> >   mean more code than just doing it unconditionally...)
> >> >
> >> > --
> >> > Jean Delvare
> >>
> >> I own an Asrock P67 Extreme4 too and have the same problem. It can be
> >> solved by following Mike Campin's instructions, which I successfully
> >> incorporated into /etc/init.d/lm_sensors. There is no BIOS available
> >> that fixes this issue, but I noticed that SpeedFan for Windows
> >> (www.almico.com/speedfan.php) seems to do something similar, otherwise
> >> it would show only 2 fans too. So maybe we could get that into
> >> lm_sensors, even if the BIOS is at fault here?
> >>
> > See above ... worst case this can result in defective hardware if 
> > misconfigured.
> >
> > We could possibly introduce some kind of module parameter, but I don't 
> > feel
> > too comfortable with that - wrong values may still blow up the hardware.
> > Any opinions, anyone ?
> >
> > Guenter
> >
> > _______________________________________________
> > lm-sensors mailing list
> > lm-sensors@lm-sensors.org
> > http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
> 
> I wouldn't recommend a module parameter, leave the driver as it is.
> If there is a hack (using isaset) just use that. It's too easy for a user to 
> set a module parameter hoping that it'll fix some random problem, and if the 
> module kills the hardware, then the module is too blame.
> 
> Asrock should fix the BIOS, that's the correct option.
> 

I wonder if you could set these register values only on this specific
motherboard by checking the DMI table to determine the motherboard
vendor and model number.

At least then you can work around a broken BIOS until the vendor
provides a fix.

Ira

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

Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c

[Guenter Roeck]

On Wed, Dec 21, 2011 at 03:55:55PM -0500, Ira W. Snyder wrote:
> On Wed, Dec 21, 2011 at 06:55:31PM +0100, Ian Dobson wrote:
> > Hi,
> > --------------------------------------------------
> > From: "Guenter Roeck" <guenter.roeck@ericsson.com>
> > Sent: Wednesday, December 21, 2011 6:01 PM
> > To: "Harald Judt" <h.judt@gmx.at>
> > Cc: <lm-sensors@lm-sensors.org>
> > Subject: Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c 
> > onASRock Extreme4
> > 
> > > Hi,
> > > On Wed, Dec 21, 2011 at 09:50:31AM -0500, Harald Judt wrote:
> > >> Hi,
> > >>
> > >> > On Tue, 12 Apr 2011 19:02:19 +0200, Ian Dobson wrote:
> > >> >> Hi,
> > >> >>
> > >> >> --------------------------------------------------
> > >> >> From: "Mike Campin" <lm_sensors at ootsa.homelinux.net>
> > >> >> Sent: Sunday, April 10, 2011 6:04 PM
> > >> >> To: <lm-sensors at lm-sensors.org>
> > >> >> Subject: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c 
> > >> >> onASRock
> > >> >> Extreme4
> > >> >>
> > >> >> > Hi,
> > >> >> >
> > >> >> > Only 2 of the 5 fans are showing up on my ASRock P67 Extreme4 
> > >> >> > motherboard.
> > >> >> > It due to the NCT6776 global configuration register 0x1c and 0x24 
> > >> >> > values.
> > >> >> > The fans work if I modify these registers prior to loading the 
> > >> >> > w83627ehf
> > >> >> > driver.
> > >> >> >
> > >> >> > My question is where are these registers initialized? Should I ask 
> > >> >> > ASRock
> > >> >> > to fix the BIOS?
> > >> >> >
> > >> >> >  isadump -y -k 0x87,0x87 0x2e 0x2f
> > >> >> >
> > >> >> >          0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
> > >> >> >     00: ff ff 00 ff ff ff ff 02 ff ff ff ff ff ff ff ff
> > >> >> >     10: ff ff ff ff ff ff ff ff ff ff f8 0e 00 00 ff ff
> > >> >> >     20: c3 33 ff 00 5c 00 00 98 00 ff 20 00 80 00 00 01
> > >> >> >     30: 00 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > >> >> >
> > >> >> >  isaset -y -f 0x2e 0x87
> > >> >> >  isaset -y -f 0x2e 0x87
> > >> >> >  isaset -y 0x2e 0x2f 0x1c 0x3
> > >> >> >  isaset -y 0x2e 0x2f 0x24 0x1c
> > >> >> >
> > >> >> > Thanks, Mike
> > >> >> >
> > >> >> > --
> > >> >> > Mike Campin
> > >> >> > lm_sensors at ootsa.homelinux.net
> > >> >> >
> > >> >> > _______________________________________________
> > >> >> > lm-sensors mailing list
> > >> >> > lm-sensors at lm-sensors.org
> > >> >> > http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
> > >> >>
> > >> >> ASRock designed the board, so they should setup the hardware 
> > >> >> monitoring chip
> > >> >> correctly. Allowing users to change the chip configuration is a really 
> > >> >> bad
> > >> >> idea, as many of the pins can have different functionality/electrical
> > >> >> spcifications depending on the config. Setting up a pin incorrectly 
> > >> >> could
> > >> >> blow the chip (Configuring a pin so that it's a source that's 
> > >> >> connected to
> > >> >> an unprotected/limited sink for example).
> > >> >>
> > >> >> So ASRock should fix the BIOS.
> > >> >
> > >> > Seconded. But before flaming Asrock, two things worth checking:
> > >> > * Availability of a BIOS update fixing the issue.
> > >> > * Options in the BIOS to enable/disable the monitoring of specific
> > >> >   fans. I can imagine that the BIOS skips the configuration steps for
> > >> >   fans for which monitoring was disabled (although that would probably
> > >> >   mean more code than just doing it unconditionally...)
> > >> >
> > >> > --
> > >> > Jean Delvare
> > >>
> > >> I own an Asrock P67 Extreme4 too and have the same problem. It can be
> > >> solved by following Mike Campin's instructions, which I successfully
> > >> incorporated into /etc/init.d/lm_sensors. There is no BIOS available
> > >> that fixes this issue, but I noticed that SpeedFan for Windows
> > >> (www.almico.com/speedfan.php) seems to do something similar, otherwise
> > >> it would show only 2 fans too. So maybe we could get that into
> > >> lm_sensors, even if the BIOS is at fault here?
> > >>
> > > See above ... worst case this can result in defective hardware if 
> > > misconfigured.
> > >
> > > We could possibly introduce some kind of module parameter, but I don't 
> > > feel
> > > too comfortable with that - wrong values may still blow up the hardware.
> > > Any opinions, anyone ?
> > >
> > > Guenter
> > >
> > > _______________________________________________
> > > lm-sensors mailing list
> > > lm-sensors@lm-sensors.org
> > > http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
> > 
> > I wouldn't recommend a module parameter, leave the driver as it is.
> > If there is a hack (using isaset) just use that. It's too easy for a user to 
> > set a module parameter hoping that it'll fix some random problem, and if the 
> > module kills the hardware, then the module is too blame.
> > 
> > Asrock should fix the BIOS, that's the correct option.
> > 
> 
> I wonder if you could set these register values only on this specific
> motherboard by checking the DMI table to determine the motherboard
> vendor and model number.
> 
> At least then you can work around a broken BIOS until the vendor
> provides a fix.
> 
Possibly, and it might be what SpeedFan is doing.
That should be done in drivers/platform/x86, though.

Guenter

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

Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c

[Harald Judt]

Am 21.12.2011 21:55, schrieb Ira W. Snyder:
> On Wed, Dec 21, 2011 at 06:55:31PM +0100, Ian Dobson wrote:
>> Hi,
>> --------------------------------------------------
>> From: "Guenter Roeck"<guenter.roeck@ericsson.com>
>> Sent: Wednesday, December 21, 2011 6:01 PM
>> To: "Harald Judt"<h.judt@gmx.at>
>> Cc:<lm-sensors@lm-sensors.org>
>> Subject: Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c
>> onASRock Extreme4
>>
>>> Hi,
>>> On Wed, Dec 21, 2011 at 09:50:31AM -0500, Harald Judt wrote:
>>>> Hi,
>>>>
>>>>> On Tue, 12 Apr 2011 19:02:19 +0200, Ian Dobson wrote:
>>>>>> Hi,
>>>>>>
>>>>>> --------------------------------------------------
>>>>>> From: "Mike Campin"<lm_sensors at ootsa.homelinux.net>
>>>>>> Sent: Sunday, April 10, 2011 6:04 PM
>>>>>> To:<lm-sensors at lm-sensors.org>
>>>>>> Subject: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c
>>>>>> onASRock
>>>>>> Extreme4
>>>>>>
>>>>>>> Hi,
>>>>>>>
>>>>>>> Only 2 of the 5 fans are showing up on my ASRock P67 Extreme4
>>>>>>> motherboard.
>>>>>>> It due to the NCT6776 global configuration register 0x1c and 0x24
>>>>>>> values.
>>>>>>> The fans work if I modify these registers prior to loading the
>>>>>>> w83627ehf
>>>>>>> driver.
>>>>>>>
>>>>>>> My question is where are these registers initialized? Should I ask
>>>>>>> ASRock
>>>>>>> to fix the BIOS?
>>>>>>>
>>>>>>>   isadump -y -k 0x87,0x87 0x2e 0x2f
>>>>>>>
>>>>>>>           0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
>>>>>>>      00: ff ff 00 ff ff ff ff 02 ff ff ff ff ff ff ff ff
>>>>>>>      10: ff ff ff ff ff ff ff ff ff ff f8 0e 00 00 ff ff
>>>>>>>      20: c3 33 ff 00 5c 00 00 98 00 ff 20 00 80 00 00 01
>>>>>>>      30: 00 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
>>>>>>>
>>>>>>>   isaset -y -f 0x2e 0x87
>>>>>>>   isaset -y -f 0x2e 0x87
>>>>>>>   isaset -y 0x2e 0x2f 0x1c 0x3
>>>>>>>   isaset -y 0x2e 0x2f 0x24 0x1c
>>>>>>>
>>>>>>> Thanks, Mike
>>>>>>>
>>>>>>> --
>>>>>>> Mike Campin
>>>>>>> lm_sensors at ootsa.homelinux.net
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> lm-sensors mailing list
>>>>>>> lm-sensors at lm-sensors.org
>>>>>>> http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
>>>>>>
>>>>>> ASRock designed the board, so they should setup the hardware
>>>>>> monitoring chip
>>>>>> correctly. Allowing users to change the chip configuration is a really
>>>>>> bad
>>>>>> idea, as many of the pins can have different functionality/electrical
>>>>>> spcifications depending on the config. Setting up a pin incorrectly
>>>>>> could
>>>>>> blow the chip (Configuring a pin so that it's a source that's
>>>>>> connected to
>>>>>> an unprotected/limited sink for example).
>>>>>>
>>>>>> So ASRock should fix the BIOS.
>>>>>
>>>>> Seconded. But before flaming Asrock, two things worth checking:
>>>>> * Availability of a BIOS update fixing the issue.
>>>>> * Options in the BIOS to enable/disable the monitoring of specific
>>>>>    fans. I can imagine that the BIOS skips the configuration steps for
>>>>>    fans for which monitoring was disabled (although that would probably
>>>>>    mean more code than just doing it unconditionally...)
>>>>>
>>>>> --
>>>>> Jean Delvare
>>>>
>>>> I own an Asrock P67 Extreme4 too and have the same problem. It can be
>>>> solved by following Mike Campin's instructions, which I successfully
>>>> incorporated into /etc/init.d/lm_sensors. There is no BIOS available
>>>> that fixes this issue, but I noticed that SpeedFan for Windows
>>>> (www.almico.com/speedfan.php) seems to do something similar, otherwise
>>>> it would show only 2 fans too. So maybe we could get that into
>>>> lm_sensors, even if the BIOS is at fault here?
>>>>
>>> See above ... worst case this can result in defective hardware if
>>> misconfigured.
>>>
>>> We could possibly introduce some kind of module parameter, but I don't
>>> feel
>>> too comfortable with that - wrong values may still blow up the hardware.
>>> Any opinions, anyone ?
>>>
>>> Guenter
>>>
>>> _______________________________________________
>>> lm-sensors mailing list
>>> lm-sensors@lm-sensors.org
>>> http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
>>
>> I wouldn't recommend a module parameter, leave the driver as it is.
>> If there is a hack (using isaset) just use that. It's too easy for a user to
>> set a module parameter hoping that it'll fix some random problem, and if the
>> module kills the hardware, then the module is too blame.

I understand and agree with that.

>> Asrock should fix the BIOS, that's the correct option.

Unfortunately, the correct option is not always the most likely to happen.

> I wonder if you could set these register values only on this specific
> motherboard by checking the DMI table to determine the motherboard
> vendor and model number.

This is what I currently do using dmidecode in /etc/init.d/lm_sensors, 
because I too fear to forget about this and accidently wreck a new 
mainboard I'm going to buy in the future:

# We need to do some initialization before loading the module,
# otherwise only 2 of the 5 fans will be shown
init_module_w83627ehf() {
     modprobe -q -r w83627ehf
     # make sure to do this only for the right mainboard
     if /usr/sbin/dmidecode | grep -i 'Base Board Information' -A2 -B1 | 
grep -q 'P67 Extreme4' \
         && /usr/sbin/dmidecode | grep -i 'BIOS Information' -A3 | grep 
-q 'Version: L1.61' && \
         ! lsmod | grep w83627ehf; then
         isaset -y -f 0x2e 0x87 2>/dev/null
         isaset -y -f 0x2e 0x87 2>/dev/null
         isaset -y 0x2e 0x2f 0x1c 0x3 2>/dev/null
         isaset -y 0x2e 0x2f 0x24 0x1c 2>/dev/null
     fi
}

Actually, checking the BIOS version might make no difference, but since 
I can change it very easily and know when I have to do so, I decided to 
include it.

I have asked the author of Speedfan how he did it, maybe he will 
respond. His approach seems to work and be safe. Somehow I don't think 
it will be very different from this solution though.

> At least then you can work around a broken BIOS until the vendor
> provides a fix.
>
> Ira

Harald

-- 
`Experience is the best teacher.'

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

Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c

[Jean Delvare]

On Wed, 21 Dec 2011 23:43:47 +0100, Harald Judt wrote:
> Am 21.12.2011 21:55, schrieb Ira W. Snyder:
> > I wonder if you could set these register values only on this specific
> > motherboard by checking the DMI table to determine the motherboard
> > vendor and model number.
> 
> This is what I currently do using dmidecode in /etc/init.d/lm_sensors, 
> because I too fear to forget about this and accidently wreck a new 
> mainboard I'm going to buy in the future:
> 
> # We need to do some initialization before loading the module,
> # otherwise only 2 of the 5 fans will be shown
> init_module_w83627ehf() {
>      modprobe -q -r w83627ehf
>      # make sure to do this only for the right mainboard
>      if /usr/sbin/dmidecode | grep -i 'Base Board Information' -A2 -B1 | 
> grep -q 'P67 Extreme4' \
>          && /usr/sbin/dmidecode | grep -i 'BIOS Information' -A3 | grep 
> -q 'Version: L1.61' && \

dmidecode -s is your friend ;) Namely -s baseboard-product-name and -s
bios-version.

>          ! lsmod | grep w83627ehf; then
>          isaset -y -f 0x2e 0x87 2>/dev/null
>          isaset -y -f 0x2e 0x87 2>/dev/null
>          isaset -y 0x2e 0x2f 0x1c 0x3 2>/dev/null
>          isaset -y 0x2e 0x2f 0x24 0x1c 2>/dev/null
>      fi
> }

-- 
Jean Delvare

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

Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c

[Jean Delvare]

On Thu, 5 Jan 2012 15:39:01 +0100, Jean Delvare wrote:
> On Wed, 21 Dec 2011 23:43:47 +0100, Harald Judt wrote:
> > Am 21.12.2011 21:55, schrieb Ira W. Snyder:
> > > I wonder if you could set these register values only on this specific
> > > motherboard by checking the DMI table to determine the motherboard
> > > vendor and model number.
> > 
> > This is what I currently do using dmidecode in /etc/init.d/lm_sensors, 
> > because I too fear to forget about this and accidently wreck a new 
> > mainboard I'm going to buy in the future:
> > 
> > # We need to do some initialization before loading the module,
> > # otherwise only 2 of the 5 fans will be shown
> > init_module_w83627ehf() {
> >      modprobe -q -r w83627ehf
> >      # make sure to do this only for the right mainboard
> >      if /usr/sbin/dmidecode | grep -i 'Base Board Information' -A2 -B1 | 
> > grep -q 'P67 Extreme4' \
> >          && /usr/sbin/dmidecode | grep -i 'BIOS Information' -A3 | grep 
> > -q 'Version: L1.61' && \
> 
> dmidecode -s is your friend ;) Namely -s baseboard-product-name and -s
> bios-version.

Oh, and I almost forgot: the same info is typically available
from /sys/class/dmi/id/* for even cheaper.

-- 
Jean Delvare

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

Re: [lm-sensors] NCT6776 global registers 0x1c=0 and 0x24=0x5c

[Harald Judt]

Am 05.01.2012 15:45, schrieb Jean Delvare:
> On Thu, 5 Jan 2012 15:39:01 +0100, Jean Delvare wrote:
>> On Wed, 21 Dec 2011 23:43:47 +0100, Harald Judt wrote:
>>> Am 21.12.2011 21:55, schrieb Ira W. Snyder:
>>>> I wonder if you could set these register values only on this specific
>>>> motherboard by checking the DMI table to determine the motherboard
>>>> vendor and model number.
>>>
>>> This is what I currently do using dmidecode in /etc/init.d/lm_sensors,
>>> because I too fear to forget about this and accidently wreck a new
>>> mainboard I'm going to buy in the future:
>>>
>>> # We need to do some initialization before loading the module,
>>> # otherwise only 2 of the 5 fans will be shown
>>> init_module_w83627ehf() {
>>>       modprobe -q -r w83627ehf
>>>       # make sure to do this only for the right mainboard
>>>       if /usr/sbin/dmidecode | grep -i 'Base Board Information' -A2 -B1 |
>>> grep -q 'P67 Extreme4' \
>>>           &&  /usr/sbin/dmidecode | grep -i 'BIOS Information' -A3 | grep
>>> -q 'Version: L1.61'&&  \
>>
>> dmidecode -s is your friend ;) Namely -s baseboard-product-name and -s
>> bios-version.

To be honest, I didn't care enough to optimize a one-time call at 
startup, even if it looks terribly ugly.

> Oh, and I almost forgot: the same info is typically available
> from /sys/class/dmi/id/* for even cheaper.

Thanks, that's really helpful! It helps me to get rid of the dmidecode 
dependence and I've learned something new ;-)

Regards,
Harald

-- 
`Experience is the best teacher.'

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