cpuinfo_max_freq is (really) wrong

Bug #91553 reported by Davyd
8
Affects Status Importance Assigned to Milestone
linux-source-2.6.20 (Ubuntu)
Invalid
Undecided
Ben Collins

Bug Description

Binary package hint: linux-image-2.6-686

[davyd@frobisher cpufreq]$ pwd
/sys/devices/system/cpu/cpu0/cpufreq
[davyd@frobisher cpufreq]$ cat cpuinfo_max_freq
65535000

Should be the same as:

[davyd@frobisher cpufreq]$ cat scaling_max_freq
1200000

but:

[davyd@frobisher cpufreq]$ cat cpuinfo_min_freq
600000

is correct.

Revision history for this message
Davyd (davyd) wrote :

Oh, for reference:

[davyd@frobisher cpufreq]$ cat scaling_driver
acpi-cpufreq
[davyd@frobisher cpufreq]$ cat scaling_governor
conservative

Revision history for this message
Brian Murray (brian-murray) wrote :

Thanks for taking the time to report this bug and helping to make Ubuntu better. With which specific version of the kernel did you notice this issue? Thanks in advance.

Changed in linux-meta:
assignee: nobody → brian-murray
status: Unconfirmed → Needs Info
Revision history for this message
Davyd (davyd) wrote :

Certainly exists in 2.6.20-9-generic, have seen it in previous kernels too. Possibly in all Feisty 2.6.20 kernels.

Changed in linux-meta:
assignee: brian-murray → nobody
status: Needs Info → Unconfirmed
Revision history for this message
Ben Collins (ben-collins) wrote :

Please attach (do not paste into comment) the output of the following commands:

dmesg
cat /proc/cpuinfo

Thanks

Changed in linux-source-2.6.20:
assignee: nobody → ben-collins
status: Unconfirmed → Needs Info
Revision history for this message
Davyd (davyd) wrote :
Revision history for this message
Davyd (davyd) wrote :
Revision history for this message
Ben Collins (ben-collins) wrote :

Ok, this is definitely a problem with your BIOS reporting incorrect information. Check for a BIOS update. The reason you'll see this now in feisty with 2.6.20 is because of the upgraded acpi-cpufreq driver being able to get more info from the BIOS.

Also, I notice you have an oops in your dmesg. Please file a separate bug about that.

Changed in linux-source-2.6.20:
status: Needs Info → Rejected
Revision history for this message
Davyd (davyd) wrote :

Ok, I'll check to see if there is an update BIOS. For future reference, what hints that the BIOS is wrong?

Also, I hadn't spotted that oops, thanks.

Revision history for this message
Davyd (davyd) wrote :

I ran the latest BIOS update, but there was no improvement.

Revision history for this message
Tom Dison (fretinator) wrote :

FWIW, I have the same problem on an Athlon 2800+ laptop. My max frequency should be 2133 mhz (apprx) but instead it is 1466 (approx). Using cpufreq_set does not work. Echoing the value into scaling_max_freq does not work. I ran cpufreq-info and it reported the hardware was limiting it to 1467. For fun, I booted into Windows and checked with perfmon. It is 1467 there also!! This is either a hardware problem or somehow the bios got altered (certainly Linux would change the bios, would it??). Any suggestions would be welcome, but I thought this might help someone.

Revision history for this message
Tom Dison (fretinator) wrote :

Additional info to above:

If I add acpi=off to grub, I get the full 2.13 ghz. I then tried immediately rebooting into windows and re-ran perfmon, but windows was still limited to 1.467 ghz as per above. Perhaps ACPI is broken on this laptop? Also, I tried booting into the live 6.06 CD to see if it was a kernel issue (since it used to run at 2.13 ghz under dapper), but it was still limited to 1.467. This really makes me wonder if Feisty hasn't done something bad to the ACPI in the bios, but I can't imagine it WRITES to the bios. I hope not. Thanks!

Revision history for this message
Милош (misos) wrote :

I can confirm the same behavior on hardy.

# uname -a
Linux misos-laptop 2.6.24-20-generic #1 SMP Mon Jul 28 13:49:52 UTC 2008 i686 GNU/Linux
# cat cpuinfo_max_freq
1600000
# cat scaling_max_freq
800000

Revision history for this message
Милош (misos) wrote :
Revision history for this message
Милош (misos) wrote :
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.