Comment 351 for bug 272247

@67GTA, the compile log seems not to show anything touching the Processor declarations. Its a bit weird. I wonder whether you see something noticeably different in powertop (on the C-state usage) between using that modified DSDT and unmodified with nohpet (but thats probably more out of curiosity).

@DEFHol, wribeiro, the problem with that sort of thing is that it is hard to get hold of and likely depends on having certain hardware. The "idle=poll" is not really desirable because it prevents the cpu from any idle method which is not doing good on battery life. If neither "nohpet" nor "acpi_skip_timer_override" work for you, you might give "idle=halt" a try, but the other two options would probably be better.

In general a fix seems to require a dmesg (with "debug apic=debug") and "sudo lspci -vvvnnxxxx" and a good documentation on the chipset. It seems it is not the first time HP get the IRQ0 override wrong.