Skip to contentSkip to conversation

clocksource tsc unstable

DISCUSSION Renato S. Yamane 6 participants 7 messages Jul 9, 2007 – Jul 10, 2007

lore
Share & tools
lore
#0 lore raw

On Kernel 2.6.21.6 I see this message in dmesg:
Clocksource tsc unstable (delta = 732182695 ns)

It's normal?

Regards,
Renato

#1 lore raw
Quoted 5 lines · Renato S. Yamane

On Kernel 2.6.21.6 I see this message in dmesg:
Clocksource tsc unstable (delta = 732182695 ns)

It's normal?

AMD CPU? SMP? Details, please.

- Arnaldo

#2 lore raw
Quoted 7 lines · Arnaldo Carvalho de Melo
Quoted 4 lines · Renato S. Yamane

On Kernel 2.6.21.6 I see this message in dmesg:
Clocksource tsc unstable (delta = 732182695 ns)

It's normal?

AMD CPU? SMP? Details, please.

Here (after resume from ram btw):

[   10.726665] Marking TSC unstable due to: possible TSC halt in C2.
[   47.203328] Clocksource tsc unstable (delta = -198170415 ns)

2.6.22 #77 Mon Jul 9 09:14:25 CEST 2007 i686

vendor_id : GenuineIntel
cpu family : 6
model : 13
model name : Intel(R) Pentium(R) M processor 2.00GHz

It's thinkpad z60m.

Quoted 1 line

- Arnaldo

Signature · 2 lines
Arkadiusz Miśkiewicz        PLD/Linux Team
arekm / maven.pl            http://ftp.pld-linux.org/
#3 lore raw
In reply to Arnaldo Carvalho de Melo

Arnaldo Carvalho de Melo escreveu:

Quoted 6 lines
Quoted 3 lines · Renato S. Yamane

On Kernel 2.6.21.6 I see this message in dmesg:
Clocksource tsc unstable (delta = 732182695 ns)
It's normal?

AMD CPU? SMP? Details, please.

Oh, sorry!
cat /proc/cpuinfo

processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 13
model name : Intel(R) Pentium(R) M processor 1.86GHz
stepping : 8
cpu MHz : 800.000
cache size : 2048 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat clflush dts acpi mmx fxsr sse sse2 ss tm pbe est tm2
bogomips : 1596.70
clflush size : 64

Regards,
Renato

#4 lore raw

As far as i know Pentium M processors have dynamically changed clock speed (ofc to save power). That's why kernel notice that TSC is unstable (it is indeed). On my Athlon 64 I have similar situation, because CPU frequency is dynamically changed.
I don't think that there is an easy way to fix it (it is not even a bug), time-stamp counter is not perfect...

Regards,
Pawel

#5 lore raw
Quoted 6 lines · Pawel Dziepak

As far as i know Pentium M processors have dynamically changed clock speed (ofc to save power). That's why kernel notice that TSC is unstable (it is indeed). On my Athlon 64 I have similar situation, because CPU frequency is dynamically changed.
I don't think that there is an easy way to fix it (it is not even a bug), time-stamp counter is not perfect...

Also see http://lkml.org/lkml/2007/5/6/29

The TSC is supposed to tick at the CPU rate (unless it's a constant_tsc, see /proc/cpuinfo flags:), so on frequency change, this ought to happen. The kernel will automatically switch to something else. There's acpi_pm and, IIRC, pit.

Jan

#6 lore raw
In reply to Arnaldo Carvalho de Melo

Now that you mention it - I am seeing something similar with kernel 2.6.22 on an Intel Pentium D 940 dual core processor (arbitrary selection of dmesg lines that appeared relevant):

<5>[    0.000000] Linux version 2.6.22-testing (ts@xenon) (gcc version 4.1.2 20061115 (prerelease) (SUSE Linux)) #1 SMP PREEMPT Mon Jul 9 10:57:22 CEST 2007
<6>[   56.367608] hpet0: at MMIO 0xfed00000, IRQs 2, 8, 0
<6>[   56.367867] hpet0: 3 64-bit timers, 14318180 Hz
<4>[   56.428897] Calibrating delay using timer specific routine.. 6399.40 BogoMIPS (lpj=3199704)
<4>[   56.438058] CPU0: Intel(R) Pentium(R) D CPU 3.20GHz stepping 04
<4>[   56.509321] CPU1: Intel(R) Pentium(R) D CPU 3.20GHz stepping 04
<6>[   56.510299] Total of 2 processors activated (12793.06 BogoMIPS).
<4>[   56.510527] ENABLING IO-APIC IRQs
<6>[   56.510792] ..TIMER: vector=0x31 apic1=0 pin1=2 apic2=-1 pin2=-1
<6>[   56.622869] checking TSC synchronization [CPU#0 -> CPU#1]:
<4>[   56.642971] Measured 16 cycles TSC warp between CPUs, turning off TSC clock.
<4>[    0.152000] Marking TSC unstable due to: check_tsc_sync_source failed.
<6>[    0.153000] Brought up 2 CPUs
<4>[    0.241000] migration_cost=2000

2.6.22-rc6-mm1 has it too, though without the backward jump in printk timestamps:

<5>[    0.000000] Linux version 2.6.22-rc6-mm1-testing (ts@xenon) (gcc version 4.1.2 20061115 (prerelease) (SUSE Linux)) #10 SMP PREEMPT Thu Jul 5 23:41:32 CEST 2007
<6>[    0.160000] Total of 2 processors activated (12791.39 BogoMIPS).
<4>[    0.160000] ENABLING IO-APIC IRQs
<6>[    0.161000] ..TIMER: vector=0x31 apic1=0 pin1=2 apic2=-1 pin2=-1
<6>[    0.171000] checking TSC synchronization [CPU#0 -> CPU#1]:
<4>[    0.171005] Measured 96 cycles TSC warp between CPUs, turning off TSC clock.
<4>[    0.171005] Marking TSC unstable due to: check_tsc_sync_source failed.
<6>[    0.172000] Brought up 2 CPUs

2.6.21.6 is fine here, though:

<5>[    0.000000] Linux version 2.6.21.6-noinitrd (ts@xenon) (gcc version 4.1.2 20061115 (prerelease) (SUSE Linux)) #1 SMP PREEMPT Sat Jul 7 16:40:13 CEST 2007
<6>[   59.776179] Total of 2 processors activated (12794.21 BogoMIPS).
<4>[   59.776403] ENABLING IO-APIC IRQs
<6>[   59.776657] ..TIMER: vector=0x31 apic1=0 pin1=2 apic2=-1 pin2=-1
<6>[   59.923347] checking TSC synchronization [CPU#0 -> CPU#1]: passed.
<6>[   59.943479] Brought up 2 CPUs
<4>[   60.198497] migration_cost=496
<6>[   60.303586] Time: tsc clocksource has been installed.
<6>[   60.303678] Switched to high resolution mode on CPU 0
<6>[   60.303974] Switched to high resolution mode on CPU 1

HTH

Signature · 4 lines
Tilman Schmidt                          E-Mail: [email protected]
Bonn, Germany
Diese Nachricht besteht zu 100% aus wiederverwerteten Bits.
Ungeöffnet mindestens haltbar bis: (siehe Rückseite)