Bug 8634

Summary:	Unstable clocksource reported
Product:	Timers	Reporter:	Patrick Matthäi (patrick)
Component:	Other	Assignee:	Thomas Gleixner (tglx)
Status:	CLOSED PATCH_ALREADY_AVAILABLE
Severity:	normal	CC:	bunk, mingo, tglx
Priority:	P1
Hardware:	All
OS:	Linux
Kernel Version:	2.6.21.5	Subsystem:
Regression:	---	Bisected commit-id:
Attachments:	My Kernel config

Description Patrick Matthäi 2007-06-15 13:39:55 UTC

Most recent kernel where this bug did not occur: Not known
Distribution: Debian Sid
Hardware Environment: AMD Athlon64 San Diego 4000+
Software Environment: Tickless Timer Support (64bit CPU, but 32bit system)
Problem Description:
Every time if I activate the 'ondemand' powersave gorvener, dmesg reports following:

Time: acpi_pm clocksource has been installed.
Clocksource tsc unstable (delta = -245560788 ns)

This thing needs at booting up ~ 2-10 seconds, which is a lot of time, so I think this is a bug.

My kernel config with all my inbuilded powersave features etc. is in attached.

Steps to reproduce:
Just do a "/bin/echo 'ondemand' > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor".

Comment 1 Patrick Matthäi 2007-06-15 13:40:32 UTC

Created attachment 11762 [details]
My Kernel config

Comment 2 Patrick Matthäi 2007-06-16 00:08:54 UTC

Some additional infos:

the-me:~# cat /sys/devices/system/clocksource/clocksource0/available_clocksource
acpi_pm pit jiffies tsc

the-me:~# cat /proc/cpuinfo
processor       : 0
vendor_id       : AuthenticAMD
cpu family      : 15
model           : 55
model name      : AMD Athlon(tm) 64 Processor 4000+
stepping        : 2
cpu MHz         : 1000.000
cache size      : 1024 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 1
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 syscall nx mmxext fxsr_opt lm 3dnowext 3dnow pni lahf_lm ts fid vid ttp
bogomips        : 2011.84
clflush size    : 64

Comment 3 Thomas Gleixner 2007-06-17 23:54:20 UTC

Can you boot up with clocksource=acpi_pm on the kernel command line please and check, whether the long boot time goes away.

The "clock source unstable" message is _no_ BUG. The TSC is influenced by the CPU frequency and therefor considered unstable. That's a rule not a bug.

Comment 4 Patrick Matthäi 2007-06-18 00:46:27 UTC

I added it to my kernel parameters, but it still appears.
My grub entry:

title           Debian GNU/Linux, kernel 2.6.21.5
root            (hd0,0)
kernel          /boot/vmlinuz-2.6.21.5 root=/dev/hda1 ro clocksource=acpi_pm
initrd          /boot/initrd.img-2.6.21.5

Comment 5 Thomas Gleixner 2007-06-18 00:53:42 UTC

What does still appear ?

The TSC unstable message or the boot slowness ?

Comment 6 Patrick Matthäi 2007-06-18 01:03:18 UTC

Both.

Comment 7 Thomas Gleixner 2007-06-18 01:09:08 UTC

Can you please try:

http://www.tglx.de/projects/hrtimers/2.6.22-rc5/patch-2.6.22-rc5-hrt1.patch

on top of 2.6.22-rc5, whether the problem persist or not.

Comment 8 Thomas Gleixner 2007-06-18 01:17:53 UTC

Also please check, whether

nohz=off
or
highres=off
or
nohz=off highres=off

on the 2.6.21 kernel command line make any difference

Comment 9 Patrick Matthäi 2007-06-18 01:38:15 UTC

With these commandlines, it's working.
Bug I want to use a tickless kernel :(

Comment 10 Thomas Gleixner 2007-06-18 02:10:22 UTC

With all of them ?

even the highres=off one, where tickless is still active ?

Comment 11 Patrick Matthäi 2007-06-19 01:37:39 UTC

Seems like only with the tickless one.

Comment 12 Patrick Matthäi 2007-06-19 17:14:55 UTC

Seems to be fixed in 2.6.22-rc5.
Thanks a lot :)