CONFIG_X86_PM_TIMER broken in 2.6.15.1 (since before 2.6.14?)

[Kalin KOZHUHAROV <kalin@thinrope.net>]

Hi all.

Now I have a repeatable confirmation that selecting CONFIG_X86_PM_TIMER=y
does "break" printk on a few systems.

With CONFIG_X86_PM_TIMER=y pmtmr is preferred to tsc and as a result, printk
time is not reset on CPU init.

The difference (hand diff) between the dmesg in the two configs is:

normal( # CONFIG_X86_PM_TIMER is not set ):
...

...
[17179569.184000] PID hash table entries: 4096 (order: 12, 65536 bytes)
          [    0.000000] Detected 3011.142 MHz processor.
                    [   25.672059] Using tsc for high-res timesource
                              [   25.673931] Console: colour VGA+ 80x25



broken ( CONFIG_X86_PM_TIMER=y ):
...
[17179569.184000] ACPI: PM-Timer IO Port: 0x808
...
[17179569.184000] PID hash table entries: 4096 (order: 12, 65536 bytes)

[17179569.184000] Detected 3011.098 MHz processor.

[17179569.184000] Using pmtmr for high-res timesource

[17179569.184000] Console: colour VGA+ 80x25


Apart from that everything is the same (apart from the prink time).

Any clues as to why might this be broken?

I tired to look under arch/i386/kernel/timers but it was too much for me,
got buried under too many jiffies :-)


Kalin.

-- 
|[ ~~~~~~~~~~~~~~~~~~~~~~ ]|
+-> http://ThinRope.net/ <-+
|[ ______________________ ]|