Hi Guenter, Durgadoss, Sorry for the very late comment. On Wed, 13 Jul 2011 13:48:11 -0700, Guenter Roeck wrote: > On Tue, 2011-07-12 at 07:07 -0400, Durgadoss R wrote: > > This patch adds the core and pkg support to coretemp. > > These thresholds can be configured via the sysfs interfaces tempX_max > > and tempX_max_hyst. An interrupt is generated when CPU temperature reaches > > or crosses above tempX_max OR drops below tempX_max_hyst. > > > > This patch is based on the documentation in IA Manual vol 3A, that can be > > downloaded from here: > > http://download.intel.com/design/processor/manuals/253668.pdf > > > > Signed-off-by: Durgadoss R <durgadoss.r@xxxxxxxxx> > > Looks like it is working now, so I applied the patch to -next. Working, really? That's not how it looks like from here. On my Xeon E5520, the output looked like this before the patch: coretemp-isa-0000 Core 0: +61.0°C (high = +87.0°C, crit = +97.0°C) Core 1: +56.0°C (high = +87.0°C, crit = +97.0°C) Core 2: +57.0°C (high = +87.0°C, crit = +97.0°C) Core 3: +56.0°C (high = +87.0°C, crit = +97.0°C) And now it looks like this : coretemp-isa-0000 Core 0: +61.0°C (high = +97.0°C, hyst = +0.0°C) (crit = +97.0°C) Core 1: +57.0°C (high = +97.0°C, hyst = +0.0°C) (crit = +97.0°C) Core 2: +60.0°C (high = +97.0°C, hyst = +0.0°C) (crit = +97.0°C) Core 3: +56.0°C (high = +97.0°C, hyst = +0.0°C) (crit = +97.0°C) High == crit doesn't make much sense, and neither does hyst == 0. Looking at the code, it seems that the hyst value (tmin) is simply never read from the CPU. I wrote a fix, I'll send it shortly. After properly initializing tmin, I get the following: coretemp-isa-0000 Core 0: +61.0°C (high = +97.0°C, hyst = +97.0°C) (crit = +97.0°C) Core 1: +57.0°C (high = +97.0°C, hyst = +97.0°C) (crit = +97.0°C) Core 2: +59.0°C (high = +97.0°C, hyst = +97.0°C) (crit = +97.0°C) Core 3: +56.0°C (high = +97.0°C, hyst = +97.0°C) (crit = +97.0°C) I am still suspicious... Does this mean that the thermal interrupt mechanism is disabled by default? What happened to the old high limit value (87°C)? I don't know what meaning it had physically but at least the value was reasonable. How comes that the driver doesn't use THERM_INT_THRESHOLD0_ENABLE and THERM_INT_THRESHOLD1_ENABLE? If the thresholds can be disabled, then the driver should certainly handle these cases. On my Core Duo T2600n the output looks like: coretemp-isa-0000 Core 0: +54.0°C (high = +47.0°C, hyst = +64.0°C) ALARM (crit = +100.0°C) Core 1: +55.0°C (high = +47.0°C, hyst = +64.0°C) ALARM (crit = +100.0°C) High at 47°C by default seems unreasonable, and hyst > high even more so. But at least the ALARM flags are consistent with these limits. Lastly, on Core2 Duo E6400, I get this : coretemp-isa-0000 Core 0: +71.0°C (high = +100.0°C, hyst = +100.0°C) (crit = +100.0°C) Core 1: +69.0°C (high = +100.0°C, hyst = +100.0°C) (crit = +100.0°C) So problem is the same as on my Xeon E5520: it looks like the thresholds are disabled? -- Jean Delvare _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
