ACPI and kernel 2.6.9-1.6.FC2

Jonathan Berry berryja at gmail.com
Sat Dec 4 19:45:21 UTC 2004


On Sat, 04 Dec 2004 10:07:24 -0500, Jim Cornette
<fc-cornette at insight.rr.com> wrote:
> Jonathan Berry wrote:
> > An option for that would be another work around, nicer than disabling
> > acpi all together.
> 
> acpi works well for fan, battery and cpu control. Disabling it
> altogether is a poorer option than using 2.6.5 series kernels.

And just living with the fact that you have to manually power off your
computer is better than both.  If you just cannot stand that, use a
2.6.8 or 2.6.7 kernel (2.6.8 has some issues with CD burning).  Just
so that it is clear, the computer is completely shut down, it just
fails to turn the power off using software.  It should be fine to just
turn the compter off manually.  But I agree, simply turning off acpi
is a bad option, especially for laptops.  For most desktops, it
probably doesn't matter as much.

>  > The real issue is something being broken, though,
> > or at least it seems that way.  It works for kernels through the 2.6.8
> > kernel (you should really use 2.6.8, not 2.6.5), and then breaks in
> > the 2.6.9 series on some machines.  Whatever changed in the
> > acpi_power_off function (I think that is the one that announces it was
> > called when it should power off) should be looked at.
> 
> It would be rather strange that the announcing the process would break
> the functionality of the program. I'd rather have the power-off feature
> shutdown the computer, rather than announce that it was called. Maybe
> the halted process is caused by shutting down the process before the
> display?

I wasn't saying that the announcing brakes the program, there is just
a message printed to the screen upon shutdown saying that the above
mentioned function was called.  Presumably, this always happens, but
because the function is supposed to turn off the computer you never
see it.  Now it doesn't turn off the power for some reason, so you see
it.

> I know the kernel developers are working hard to correct this problem.
> 
> Jim

Hopefully they will figure it out.
Jonathan




More information about the fedora-list mailing list