Re: [lm-sensors] Gigabyte GA-P35-DS3P - VID Zero value

From: Jean Delvare <khali@linux-fr.org>
To: lm-sensors@vger.kernel.org
Subject: Re: [lm-sensors] Gigabyte GA-P35-DS3P - VID Zero value
Date: Tue, 20 Oct 2009 12:02:00 +0000	[thread overview]
Message-ID: <20091020140200.3fb0448d@hyperion.delvare> (raw)
In-Reply-To: <20091019164516.701032uxlen5ai2k@mail.simonandkate.net>

Hi Adam,

On Tue, 20 Oct 2009 21:50:31 +1000, Adam Nielsen wrote:
> > The raw VID value is at 0xfc, value is 0x3f, which means all VID pins
> > are high. Which means they are not wired to the CPU, otherwise at least
> > one of them would be low.
> > 
> > Now, looking at register 0x27, it appears that all VID input pins have
> > been configured for their alternative function (GPIO.) This is not the
> > default value for this register, which means that the motherboard
> > vendor decided to not use these pins for VID input but for another
> > function.
> > 
> > BTW... the pins in question are in input mode. So who knows... maybe
> > they ARE used for VID monitoring, and GPIO was preferred over true VID
> > for obscure reasons (incompatible voltage levels?) You may want to run,
> > as root:
> > 
> > isadump -f 0x800 16
> > 
> > and see if the 3rd value would make sense as a VID value for your CPU.
> 
> Ah interesting.  It doesn't seem to for me though:
> 
>        0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
> 0800: ff 01 cf fa f7 07 ff ff ff ff ff ff ff ff ff ff
> 
> VRM v11 (in hwmon-vid) says values can't be > 0xb2, although I don't know how
> many bits are wired up to the VID.
> 
> So is 0x802 the GPIO input?  I wonder what it's wired up to...?

My instructions were for Simon. Unless you have the exact same
motherboard model, chances are high that these instructions do not
apply to you. I'd need the output of:

# isadump -k 0x87,0x01,0x55,0x55 0x2e 0x2f 7

as Simon already provided, to comment on your case. Just because the
symptoms look similar to Simon's, doesn't mean the problem is the same,
nor that the solution (if there exists any) would be the same.
Although, given the same motherboard manufacturer, there is a remote
chance.

> > One thing we could do is teach the it87 driver to not export the VID
> > value if any of VID pins 0-3 are configured for GPIO function. That
> > wouldn't be too difficult, but would you be able to test a kernel patch?
> 
> I can test an it87 patch if you need a guinea pig - especially if it'll
> compile against 2.6.30 so I don't have to reboot ;-)

Sure, I can provide a standalone driver for any kernel version, no
reboot needed unless I badly screw up ;)

But please provide a dump first.

-- 
Jean Delvare
http://khali.linux-fr.org/wishlist.html

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