On Wed, Sep 17, 2014 at 05:27:28PM +0200, Jean Delvare wrote: > Le Wednesday 17 September 2014 à 06:24 -0700, Guenter Roeck a écrit : > > On 09/17/2014 03:23 AM, Jean Delvare wrote: > > > For a better long-term solution, I need to check if irq = 255 is any > > > special, maybe it means that interrupt is not available or something and > > > we shouldn't even try to request it. > > > > > Looks suspiciously like a bad (uninitialized ?) interrupt number to me. > > Apparently this was discussed previously: > > https://lkml.org/lkml/2013/2/18/115 > > but the discussion ultimately resulted in a USB-specific quirk, which > won't help us at all in this specific case. > Yes, but it helps us understand the problem. Apparently 255 means "no interrupt assigned", and the PCI subsystem ignores that for whatever reason but passes it on as real interrupt. The device is supposed to be controlled by ACPI, and presumably only accesses it in polling mode. I guess the BIOS developer(s) did not see a need to assign an interrupt to the device. Options we might have would be to check dev->pin as well (no idea what to expect, though), or accept the fact that interrupt assignment may fail. Guenter > Scott, can you please provide the output of: > # /sbin/lspci -vvv -s 00:1f.3 > # cat /proc/interrupts > > Maybe it will reveal something interesting... > > -- > Jean Delvare > SUSE L3 Support > _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
