On Mon, Mar 8, 2010 at 10:31 PM, Phillip Pi <ant@xxxxxxxxxx> wrote: >> On Mon, Mar 8, 2010 at 10:10 PM, Phillip Pi <ant@xxxxxxxxxx> wrote: >> >> > >> This morning, I rebooted my old Debian box to start using its new Kernel >> >> > >> 2.6.32 from 2.6.30, but I noticed my lm_sensors didn't show everything >> >> > >> (voltages, fan RPMs, etc.) anymore compared to 2.6.30 and earlier. I am >> >> > >> using a MSI K8N NEO4-F (MS-7125) motherboard. >> >> > > [...] >> >> > >> I looked at the FAQ and saw >> >> > >> http://www.lm-sensors.org/wiki/FAQ/Chapter3#Mysensorshavestoppedworkinginkernel2.6.31 >> >> > >> which look like my issue and matches dmesg: http://pastie.org/856424 ... >> >> > >> >> >> > >> I am scared to use the lax method since it is not recommended. Is there >> >> > >> another way to show the datas safely? Or do I have to use this trick or >> >> > >> live without them? >> >> > > >> >> > > Asus provides an ACPI interface for accessing the sensors in a safe >> >> > > way; I don't know what it's used on MSI boards, but I can take a look: >> >> > > please send me a dump of the DSDT table >> >> > >> >> > Nothing interesting in the DSDT; the hwmon chip is used only for >> >> > reading a temperature (TZ). >> >> > I think you can use "acpi_enforce_resources=lax", there not other ways >> >> > to access the sensor data. >> > >> > Thanks Luca. I assume it is safe to use for my motherboard to see other >> > datas like voltages? >> >> Not 100%, there's still a race window. But it's pretty small. > > What do you mean by race window? I am not familiar with this > term/phrase. There a small temporal window when the two drivers (thermal and native driver) might access the chip at the same time leading to incorrect readings (or worse). >> > Are there be any future plans to show voltages and other datas without >> > the old method ("acpi_enforce_resources=lax")? >> >> We can't do much... the IO ports used by the monitoring chip are >> claimed by ACPI (supplied by the board vendor), so technically it's >> wrong for another driver to poke it. Asus provides an interface for >> reading the sensors in safe way, MSI does not... > > So no one was able to hack MSI hardwares even after a few years? And MSI > never released their datas? Is that why? No, ACPI code is basically embedded in the BIOS, so there's no way to expand the interface - short of providing a custom BIOS for each board. The problem is that two pieces of software are trying to control the same hw resource without any coordination. Luca _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
