On Fri, Oct 12, 2012 at 11:12:04AM +0200, Jean Delvare wrote:
> On Tue, 9 Oct 2012 14:09:00 -0700, Guenter Roeck wrote:
> > Make the code easier to extend and easier to adjust by using a model table
> > listing CPU models, stepping/mask, and associated TjMax.
> >
> > Signed-off-by: Guenter Roeck <linux@xxxxxxxxxxxx>
> > ---
> > drivers/hwmon/coretemp.c | 36 ++++++++++++++++++++++--------------
> > 1 file changed, 22 insertions(+), 14 deletions(-)
>
> I like the idea. Just a couple things about the implementation:
>
> > diff --git a/drivers/hwmon/coretemp.c b/drivers/hwmon/coretemp.c
> > index 1937cd4..7107096 100644
> > --- a/drivers/hwmon/coretemp.c
> > +++ b/drivers/hwmon/coretemp.c
> > @@ -208,6 +208,23 @@ static const struct tjmax __cpuinitconst tjmax_table[] = {
> > { "CPU 330", 125000 },
> > };
> >
> > +struct tjmax_model {
> > + u8 model;
> > + u8 mask;
> > + int tjmax;
> > +};
> > +
> > +static const struct tjmax_model __cpuinitconst tjmax_model_table[] = {
> > + { 0x1c, 10, 100000 }, /* D4xx, N4xx, D5xx, N5xx */
> > + { 0x1c, 0, 90000 }, /* Z5xx, N2xx, 230, 330, others */
>
> I don't think the 230 and 330 should be listed there, as they are
> overridden by an earlier match in tjmax_table[].
>
I rephrased it:
/* Z5xx, N2xx, possibly others
* Note: Also matches 230 and 330,
* which are covered by tjmax_table
*/
> > + { 0x26, 0, 90000 }, /* Atom Tunnel Creek (Exx), Lincroft (Z6xx)
> > + * Note: TjMax for E6xxT is 110C, but CPU type
> > + * is undetectable by software
> > + */
> > + { 0x27, 0, 90000 }, /* Atom Medfield (Z2460) */
> > + { 0x36, 0, 100000 }, /* Atom Cedar Trail/Cedarview (N2xxx, D2xxx) */
> > +};
>
> Using 0 as a special value may be problematic, as I think it is a
> possible stepping value:
>
> http://openbenchmarking.org/system/1109120-IV-TSCP1539338/tscp1/cpuinfo
>
> (I agree model name looks odd, not sure if it is a public model or some
> engineering sample.)
>
> In hwmon-vid we have:
>
> #define ANY 0xFF
>
Good idea. I changed the code to use ANY.
> and we use that when the stepping doesn't matter. Note that we also
> support stepping ranges there, maybe we'll need it for coretemp too
> some day, but I guess we can add it then.
>
Agreed.
> > +
> > static int __cpuinit adjust_tjmax(struct cpuinfo_x86 *c, u32 id,
> > struct device *dev)
> > {
> > @@ -226,20 +243,11 @@ static int __cpuinit adjust_tjmax(struct cpuinfo_x86 *c, u32 id,
> > return tjmax_table[i].tjmax;
> > }
> >
> > - /* Atom CPUs */
> > -
> > - if (c->x86_model == 0x1c) {
> > - /*
> > - * TjMax for stepping 10 CPUs (N4xx, N5xx, D4xx, D5xx)
> > - * is 100 degrees C, for all others it is 90 degrees C.
> > - */
> > - if (c->x86_mask == 10)
> > - return 100000;
> > - return 90000;
> > - } else if (c->x86_model == 0x26 || c->x86_model == 0x27) {
> > - return 90000;
> > - } else if (c->x86_model == 0x36) {
> > - return 100000;
> > + for (i = 0; i < ARRAY_SIZE(tjmax_model_table); i++) {
> > + const struct tjmax_model *tm = &tjmax_model_table[i];
> > + if (c->x86_model == tm->model &&
> > + (!tm->mask || c->x86_mask == tm->mask))
> > + return tm->tjmax;
> > }
> >
> > /* Early chips have no MSR for TjMax */
>
> Other than these implementation details:
>
> Acked-by: Jean Delvare <khali@xxxxxxxxxxxx>
>
Thanks!
Guenter
_______________________________________________
lm-sensors mailing list
lm-sensors@xxxxxxxxxxxxxx
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
