Error on Boot: [Firmware Bug]:Powernow-k8: Your BIOS does not provide ACPI_PSS objects in a way that Linux understands.

[Shaved Wookie]

Binary package hint: acpi-support

I update daily and thismorning when I booted my machine booted fine, however after updating today I get the error:
"[Firmware Bug]:Powernow-k8: Your BIOS does not provide ACPI_PSS objects in a way that Linux understands.Please report this to the Linux ACPI maintainers and complain to your BIOS vendor."

After that there was a kernel panic:
"Kernel panic - not syncing: VFS: Unable to mount root fs on Unknown-block(0,0)"

I booted into the older .7 version of the kernel which worked fine, did a dpkg --configure -a which configured a bunch of stuff and reset. Now my .8 version of the kernel boots, but for a split second still flashes up an error message, which I think is the ACPI_PSS one above, although it goes by too fast to say for sure. Strange thing is that this error (either the one above or a new one) never appeared before today's updates.

I'm on Kubuntu Jaunty - AMD64 on an ASUS M3A78-T motherboard with acpi-support 0.119.

[Steve Langasek (vorlon)]

Thank you for taking the time to report this bug and helping to improve Ubuntu.
acpi-support is unrelated to this problem; reassigning to the kernel.

The error message is related to a change in how the CPU frequency drivers are built into the kernel, and is probably harmless, but in that case there's probably no reason to have the error message either.

[Kyle Jones (mutiny32)]

Have you tried the ASUS M3A78-T BIOS 0802 BIOS dated January 14, 2009 on ASUS' support site? it has this specific fix:

02. Fix the problem that the system may show boot fail message if ACC enabled.

ACC stands for Advanced Clock Calibration and would fall under the control of PowerNow-K8.

It may be a mixture of both kernel and BIOS issues, but I'd try the BIOS update first.

[Kyle Jones (mutiny32)]

Looks like this has been fixed in Kernel 2.6.29-RC5:

commit 732553e567c2700ba5b9bccc6ec885c75779a94b
Author: Mark Langsdorf <email address hidden>
Date: Tue Feb 3 17:46:43 2009 +0100

    [CPUFREQ] powernow-k8: Get transition latency from ACPI _PSS table

    At this time, the PowerNow! driver for K8 uses an experimentally
    derived formula to calculate transition latency. The value it
    provides is orders of magnitude too large on modern systems.
    This patch replaces the formula with ACPI _PSS latency values
    for more accuracy and better performance.

    I've tested it on two 2nd generation Opteron systems, a 3rd
    generation Operton system, and a Turion X2 without seeing any
    stability problems.

    Signed-off-by: Mark Langsdorf <email address hidden>
    Signed-off-by: Thomas Renninger <email address hidden>
    Signed-off-by: Dave Jones <email address hidden>

[Shaved Wookie (shavedwookie)]

Thanks Kyle, given the choice between:

a) Being patient and waiting for 2.6.29-RC5 to come down the line and
b) Potentially hosing an otherwise well working system with a BIOS update filled with dire warnings about its dangers

I'll take option A. Thanks for looking into this though!

[Andy Whitcroft (apw)]

@Kyle -- its not clear that the commit you reference there will have any effect on the message that @Shaved is seeing. That commit is an improvement in the way that the driver calculates a latency calculation. The check that triggers the error above seems to be unchanged before and after this change.

You could confirm that by trying the latest mainline kernel which should include the fix specified. This kernel can be found here:

    http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.29-rc6/

The 'Your BIOS does not provide ACPI _PSS objects ..." seems to be triggered when we cannot lookup the _PSS stuff and we have more than one CPU enabled. That seems to trip when powernow_k8_cpu_init_acpi fails, we might get more information from that if you enable cpufreq debugging. I believe booting with the following added to the kernel command line should turn that on:

    cpufreq.debug=7

Expect additional information between these two lines in your dmesg output:

    Feb 16 22:36:42 the-beast kernel: [ 12.412954] powernow-k8: Found 1 AMD Phenom(tm) II X4 940 Processor processors (4 cpu cores) (version 2.20.00)
    Feb 16 22:36:42 the-beast kernel: [ 12.412968] [Firmware Bug]: powernow-k8: Your BIOS does not provide ACPI _PSS objects in a way that Linux understands. Please report this to the Linux ACPI maintainers and complain to your BIOS vendor.

You would also be recommended to report this at least to the acpi list.

[Pete Graner (pgraner)]

@Andy - I came across another patch you might want to look at:

http://patchwork.kernel.org/patch/5255/

[Tom Rini (trini)]

@Kyle: I have this same mobo but it shipped with AMIBIOS version 0802. I saw the _PSS error until I enabled the "AMD Cool'n'Quiet Function" in the BIOS. While the printed manual isn't helpful, the built-in BIOS help specifically says that this enables generation of ACPI _PSS objects (and a few others).

[Peter Schaefer (derschaefer)]

I'm seeing the same on Jaunty Kernel version 2.6.28-11-generic since i swapped my Asus M2A-VM Mobo with an Asrock A770Crossfire Mobo (newest BIOS P1.20, C'n'Q enabled in the BIOS).

Attaching the output of dmidecode and my dmesg.

[Peter Schaefer (derschaefer)]

And seems we have the same on another Asrock Mobo: https://bugs.launchpad.net/ubuntu/+bug/347002

[Shaved Wookie (shavedwookie)]

Except my fan isn't going off all the time like described in that report. In fact, it hardly ever goes beyond minimal.

[Peter Schaefer (derschaefer)]

Well, i've entered the following bug in the kernel bug tracker:
http://bugzilla.kernel.org/show_bug.cgi?id=12948

It was confirmed that the Asrock BIOS doesn't contained the required _PSS tables - which is an error on their part - nothing Linux can do about. The _PSS tables are needed for powernow-k8 in case of a multi-core CPU.

So, i emailed Asrock about the problem. Guess what they'll tell you:

"regarding Linux: have to say sorry, ASRock does not test or support the use of Linux OS. We cannot guarantee that ASRock motherboards will work with Linux without problems."

So, better refrain from Asrock for future buys. I surely won't recommend them to anyone in future.

[Peter Schaefer (derschaefer)]

One last feedback from me: At least on my motherboard (Asrock A770Crossfire) the reason for the missing _PSS tables is solved:

The BIOS disables C'n'Q *silently* (even when the setting is set to "Enabled") if memory modules from a certain vendor are used. In my case it were those:

A-Data AD2800001GOU2 2x1024MB Kit A-DATA DDR2 800MHz CL5

If i use DIMMs from another manufacturer C'n'Q works.

However, i still think that's a BIOS bug, since all DIMMs i use are JEDEC-compliant, CL5, 1,8V so no "fancy" overclocking/overvoltage modules.

And for the decision to silently ignore a user setting without printing a message at POST the BIOS programmer needs a visit with a cluebat.

[bj mccormick (bjmccormick)]

I got this error trying to boot from live usb on jaunty beta...

I have an Asus M4A79 mobo with a phenom2 940. Turning on Cool n Quiet got rid of the error, but the live session would not boot.

8.10 kernel works fine with CnQ disabled or enabled and all 4 cores blazing so....

[bj mccormick (bjmccormick)]

Can sort of disregard last comment. I updated my bios and everything seems to work right now. I originally thought that updating the bios was a rather large action to take for something that worked before, but mobo manufacturers have made it easier these days to update the bios. Can do it off of a usb stick no problem. Pretty cool.

[Andy Whitcroft (apw)]

It is sounding like this is likely a BIOS issue, there are a number of contributers here reporting problems solved when updating theirs. I will therefore close this one off Invalid. If any of you have the latest BIOS and are still seeing issues please reopen this bug by moving it to New. Thanks!