Hi David,
On Thu, 13 Dec 2012 00:58:35 -0700, David Hubbard wrote:
> On Wed, Dec 12, 2012 at 7:45 AM, Jean Delvare <khali@xxxxxxxxxxxx> wrote:
> > We do not support any VID decoding since family 10h CPUs for AMD:
> >
> > {X86_VENDOR_AMD, 0x10, 0x0, ANY, ANY, 25}, /* NPT family 10h
> > */
> >
> > Starting with family 11h CPUs, AMD has given up on parallel VID in
> > favor of serial VID which uses only 2 pins. I know that the IT8720F has
> > support for this, but I did not find any trace in other ITE datasheets.
> >
> > The serial VID uses 7-bit codes which are incompatible with the
> > previous 6-bit parallel VID codes used by family 10h CPUs. BTW the
> > family 10h CPUs already supported serial VID, and I have no idea how to
> > support that properly as hwmon-vid has no idea whether the code came
> > from the serial or parallel VID interface.
> >
> > The following should at least get rid of the warning in the kernel logs:
> >
> > * * * * *
> >
> > Since family 11h processors, AMD is exclusively using 7-bit VID codes
> > transmitted using a serial protocol over two pins (clock and data.)
> >
> > Signed-off-by: Jean Delvare <khali@xxxxxxxxxxxx>
> > ---
> > drivers/hwmon/hwmon-vid.c | 10 ++++++++++
> > 1 file changed, 10 insertions(+)
> >
> > --- linux-3.7-rc8.orig/drivers/hwmon/hwmon-vid.c 2012-10-20
> > 21:11:32.000000000 +0200
> > +++ linux-3.7-rc8/drivers/hwmon/hwmon-vid.c 2012-12-12
> > 15:36:36.309882929 +0100
> > @@ -115,6 +115,12 @@ int vid_from_reg(int val, u8 vrm)
> > return (val < 32) ? 1550 - 25 * val
> > : 775 - (25 * (val - 31)) / 2;
> >
> > + case 26: /* AMD family 10h to 15h, serial VID */
> > + val &= 0x7f;
> > + if (val >= 0x7c)
> > + return 0;
> > + return DIV_ROUND_CLOSEST(15500 - 125 * val, 10);
> > +
> > case 91: /* VRM 9.1 */
> > case 90: /* VRM 9.0 */
> > val &= 0x1f;
> > @@ -195,6 +201,10 @@ static struct vrm_model vrm_models[] = {
> > {X86_VENDOR_AMD, 0xF, 0x40, 0x7F, ANY, 24}, /* NPT family 0Fh
> > */
> > {X86_VENDOR_AMD, 0xF, 0x80, ANY, ANY, 25}, /* future fam. 0Fh
> > */
> > {X86_VENDOR_AMD, 0x10, 0x0, ANY, ANY, 25}, /* NPT family 10h
> > */
> > + {X86_VENDOR_AMD, 0x11, 0x0, ANY, ANY, 26}, /* family 11h */
> > + {X86_VENDOR_AMD, 0x12, 0x0, ANY, ANY, 26}, /* family 12h */
> > + {X86_VENDOR_AMD, 0x14, 0x0, ANY, ANY, 26}, /* family 14h */
> > + {X86_VENDOR_AMD, 0x15, 0x0, ANY, ANY, 26}, /* family 15h */
> >
> > {X86_VENDOR_INTEL, 0x6, 0x0, 0x6, ANY, 82}, /* Pentium Pro,
> > * Pentium II,
> > Xeon,
> >
> > Also available as a standalone driver at:
> > http://khali.linux-fr.org/devel/misc/hwmon-vid/
> >
> >
> After applying the patch you sent cpu0_vid reads as 1550. Thanks much, that
> was very helpful.
1.550 V is the maximum value supported by the VID encoding on these
processors. This corresponds to value 00h. This means that either the
chip doesn't support VID, or not in this register, or it is supported
but not wired on the board in question.
Either way, thanks for testing, I'll push this upstream for other AMD
family 11h+ processor users.
--
Jean Delvare
_______________________________________________
lm-sensors mailing list
lm-sensors@xxxxxxxxxxxxxx
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
