Console options for legacy-free mini-itx server?
Andriy Gapon
avg at freebsd.org
Fri Nov 19 13:30:31 UTC 2010
on 19/11/2010 05:14 Ian Smith said the following:
> On Thu, 18 Nov 2010, John Baldwin wrote:
> > On Thursday, November 18, 2010 4:07:27 pm Andrew Reilly wrote:
> > > Hi John,
> > >
> > > On 18/11/2010, at 23:59 , John Baldwin wrote:
> > >
> > > > You used devinfo -v, so it shows the devices that exist even if they failed
> > > > to attach. Try just using 'devinfo' and seeing if est1 still shows up.
> > >
> > > OK, you're right: running without -v shows only est0:
> > >
> > > nexus0
> > > apic0
> > > ram0
> > > acpi0
> > > cpu0
> > > acpi_perf0
> ^^^^^^^^^^
> > > est0
> > > p4tcc0
> > > cpufreq0
> > > cpu1
> > > p4tcc1
> > > cpufreq1
> > > acpi_button0
> > > pcib0
> > > <...>
> > >
> > > Is est. involved in the reboot question?
> >
> > Probably not.
>
> I'm not so sure, having just noticed something that I missed last time:
>
> device_attach: acpi_perf1 attach returned 6
> est0: <Enhanced SpeedStep Frequency Control> on cpu0
> p4tcc0: <CPU Frequency Thermal Control> on cpu0
> device_attach: acpi_perf1 attach returned 6
> est1: <Enhanced SpeedStep Frequency Control> on cpu1
> est: CPU supports Enhanced Speedstep, but is not recognized.
> est: cpu_vendor GenuineIntel, msr 16
> device_attach: est1 attach returned 6
> p4tcc1: <CPU Frequency Thermal Control> on cpu1
>
> I'd mis-read or assumed that first line as being for acpi_perf0, rather
> than an earlier? attempt to attach acpi_perf1 at that time.
>
>>From cpufreq(4):
>
> The following device drivers offer absolute frequency control via the
> cpufreq interface. Usually, only one of these can be active at a time.
>
> acpi_perf ACPI CPU performance states
> est Intel Enhanced SpeedStep
> ichss Intel SpeedStep for ICH
> powernow AMD PowerNow! for K7 and K8
> smist Intel SMI-based SpeedStep for PIIX4
>
> Apart from never having seen est0 attach but est1 not, my understanding
Very likely that it could be a BIOS bug.
My guess is that DSDT defines _PSS method in one processor object, but not in the
other,
> from both the manual and from having read some of the code over the time
> is that acpi_perf0 should only attach if eg est0 didn't - is that wrong?
To some degree :-)
On most modern systems acpi_perf attaches in "information only" mode where it
provides P-state information to a driver like est or hwpstate that performs actual
P-state switching.
In some cases acpi_perf may work in fill mode.
In other other cases a hardware-specific driver like est can work without acpi_perf.
> Note also that there's no mention of acpi_perf0 attaching in the dmesg,
> where I'm used to seeing that listed (eg as with my P3-Mobile CPU)
--
Andriy Gapon
More information about the freebsd-stable
mailing list