Bug 156369 - powernow-k8 (AMD Athlon 64 X2): error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Summary: powernow-k8 (AMD Athlon 64 X2): error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Status: RESOLVED DUPLICATE of bug 148014
Alias: None
Product: SUSE LINUX 10.0
Classification: openSUSE
Component: Basesystem (show other bugs)
Version: Final
Hardware: x86-64 Other
: P5 - None : Normal
Target Milestone: ---
Assignee: Thomas Renninger
QA Contact: E-mail List
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2006-03-09 08:00 UTC by Ulrich Windl
Modified: 2006-03-27 10:42 UTC (History)
1 user (show)

See Also:
Found By: Other
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---


Attachments
/var/log/boot.msg (30.31 KB, text/plain)
2006-03-09 14:57 UTC, Ulrich Windl
Details
/proc/cpuinfo (1.07 KB, text/plain)
2006-03-24 07:10 UTC, Ulrich Windl
Details
hwinfo --cpu (1005 bytes, text/plain)
2006-03-24 07:11 UTC, Ulrich Windl
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Ulrich Windl 2006-03-09 08:00:44 UTC
For my AMD Athlon 64 X2 3800+ syslog has frequent "out of sync" messages from powernow-k8. Messages look like this:
kernel: powernow-k8: Found 2 AMD Athlon 64 / Opteron processors (version 1.50.3)
kernel: powernow-k8:    0 : fid 0xc (2000 MHz), vid 0x8 (1350 mV)
kernel: powernow-k8:    1 : fid 0x2 (1000 MHz), vid 0x12 (1100 mV)
kernel: cpu_init done, current fid 0xc, vid 0x8
kernel: powernow-k8:    0 : fid 0xc (2000 MHz), vid 0x8 (1350 mV)
kernel: powernow-k8:    1 : fid 0x2 (1000 MHz), vid 0x12 (1100 mV)
kernel: cpu_init done, current fid 0xc, vid 0x8
kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
...
Comment 1 Michael Gross 2006-03-09 13:30:51 UTC
Please attach `hwinfo --cpu', /var/log/boot.msg and the last 200 lines of your syslog. Does this have any effects? What does /proc/cpuinfo show?
Comment 2 Ulrich Windl 2006-03-09 14:57:35 UTC
Created attachment 71991 [details]
/var/log/boot.msg
Comment 3 Ulrich Windl 2006-03-09 15:02:29 UTC
(The CPU info has to follow later, I have no access right now). The last 200 lines of syslog did not contain anything about "powernow", so I preferred a grep over tail:
> grep powernow /var/log/messages
Mar  8 20:28:30 aax264 rcpowersaved: enter 'powernow_k8' into CPUFREQD_MODULE in /etc/sysconfig/powersave/cpufreq.
Mar  8 20:28:33 aax264 kernel: powernow-k8: Found 2 AMD Athlon 64 / Opteron processors (version 1.50.3)
Mar  8 20:28:33 aax264 kernel: powernow-k8:    0 : fid 0xc (2000 MHz), vid 0x8 (1350 mV)
Mar  8 20:28:33 aax264 kernel: powernow-k8:    1 : fid 0x2 (1000 MHz), vid 0x12 (1100 mV)
Mar  8 20:28:33 aax264 kernel: powernow-k8:    0 : fid 0xc (2000 MHz), vid 0x8 (1350 mV)
Mar  8 20:28:33 aax264 kernel: powernow-k8:    1 : fid 0x2 (1000 MHz), vid 0x12 (1100 mV)
Mar  8 20:28:33 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Mar  8 20:28:34 aax264 kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Mar  8 20:28:34 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Mar  8 20:37:40 aax264 kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Mar  8 20:37:42 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Mar  8 20:37:44 aax264 kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Mar  8 20:37:47 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Mar  8 20:40:24 aax264 kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Mar  8 20:40:24 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Mar  8 20:40:29 aax264 kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Mar  8 20:40:32 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Mar  8 20:43:12 aax264 kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Mar  8 20:43:13 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Mar  8 20:44:52 aax264 kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Mar  8 20:44:55 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Mar  8 21:06:23 aax264 kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Mar  8 21:06:24 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Mar  8 21:06:34 aax264 kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Mar  8 21:06:36 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Mar  8 21:06:59 aax264 kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Mar  8 21:07:00 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Mar  8 21:07:02 aax264 kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Mar  8 21:07:02 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Mar  8 21:12:11 aax264 kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Mar  8 21:12:11 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Mar  8 21:39:39 aax264 kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Mar  8 21:39:39 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Mar  8 21:48:33 aax264 kernel: powernow-k8: error - out of sync, fix 0x2 0xc, vid 0x12 0x8
Mar  8 21:48:33 aax264 kernel: powernow-k8: error - out of sync, fix 0xc 0x2, vid 0x8 0x12
Comment 4 Michael Gross 2006-03-14 11:50:54 UTC
Try booting with ACPI=off or even with the safe-settings and check if this still happens.
Comment 5 Michael Gross 2006-03-20 13:05:07 UTC
Let's ask the kernel-maintainers for some input here. This issue might be known.
Comment 7 Greg Kroah-Hartman 2006-03-23 06:45:29 UTC
No response for quite some time, am closing for now.
Comment 8 Ulrich Windl 2006-03-23 07:06:02 UTC
(In reply to comment #7)
> No response for quite some time, am closing for now.
> 

I really think that boot.msg has essentially the same info as /proc/cpuinfo would have, and I'm unsure "hwinfo --cpu" would add more. Despite of that I doubt, that this is really preventing anyone to try to analyze what the problem might be. So what are you really waiting for?
Comment 9 Olaf Kirch 2006-03-23 10:28:48 UTC
Ulrich, there have been several questions that you didn't answer, e.g.

 -	Does this have any negative effects on the system?
 -	Do safe settings and/or acpi=off change the picture?

Frankly, from our point of view this seems to be just an annoying log
message with no (major) ill side effects, so I would close this as WONTFIX.

I am reassigning this to Thomas Renninger for tracking. Please close unless
this turns out to be a serious issue.
Comment 10 Ulrich Windl 2006-03-24 07:10:37 UTC
Created attachment 74848 [details]
/proc/cpuinfo
Comment 11 Ulrich Windl 2006-03-24 07:11:12 UTC
Created attachment 74849 [details]
hwinfo --cpu
Comment 12 Ulrich Windl 2006-03-24 07:16:29 UTC
Trying to boot with "acpi=off" reliably hands the machine after "nv_sata" has detected drive "ata1". So this is not a valid option or work-around. Regarding negative effects (other than filling the syslog): I haven't found any.

Why I did send the report? I've read an article (I think in German c't magazine) about problems with dual core CPUs and multimedia applications in MS Windows. From what I remember they said that the problem is related to power management and frequency throttling, resuting in the cores using sync. I also think they said the proper solution requires a change in hardware (or the power management interface?). Another reason why I sent the report is that only few people have such a modern CPU at home.
So possibly the proper test is using hight precision timing routines.
Comment 13 Thomas Renninger 2006-03-24 10:16:53 UTC
acpi=off is no option. CPUfreq won't be supported there at all.

The system regularly mixes up the speed it currently runs on and the speed it think it runs on.

Ulrich: Can you try to boot a debug kernel or compile a kernel yourself with
CONFIG_CPU_FREQ_DEBUG=y
Then: echo 7 >/sys/modules/cpufreq/parameters/debug
Be sure you run into the same problem again and send relevant /var/log/messages output.

Tell me if you have problems compiling a kernel yourself.
Short How-To:
- install kernel-source.rpm
- cd /usr/src/linux
- cp arch/x86_64/defconfig.smp .config
- vi .config (search for CONFIG_CPU_FREQ_DEBUG and set it to yes)
- make -j4
- make install modules_install

Check /boot/vmlinuz (should point to the new kernel) and your /boot/grub/menu.lst)
Comment 14 Jacob Shin 2006-03-24 15:02:49 UTC
This is result of:

Bug 148014 - i386 kernel does not initialize cpu_core_id and phys_core_id properly.

Please apply the patch in Bug 148014 to fix..

-Jacob Shin
AMD, Inc.
Comment 15 Ulrich Windl 2006-03-27 07:03:51 UTC
(In reply to comment #14)
> This is result of bug 148014 - i386 kernel does not initialize cpu_core_id and
> phys_core_id properly.

(In reply to comment #13)
> Ulrich: Can you try to boot a debug kernel or compile a kernel yourself with
> CONFIG_CPU_FREQ_DEBUG=y
> Then: echo 7 >/sys/modules/cpufreq/parameters/debug

Do you still need this information from me when considering comment #14?
Comment 16 Thomas Renninger 2006-03-27 10:38:49 UTC
comment #15 -> setting as duplicate.

Ulrich, no we probably don't need this anymore (if Andi's patch works).


*** This bug has been marked as a duplicate of 148014 ***
Comment 17 Thomas Renninger 2006-03-27 10:42:41 UTC
Ulrich: Please stay tuned. I expect Andi will backport his patch to 10.0, I am not sure, though. He probably has some higher priority tasks concerning SLES10/NLD10 at the moment and this may take could take some time.