linux-2.6.22.1
x86_64
[no dynticks for me, but I can still check powertop for when I do get them]
------------------------------------------------------------------------
P-states (frequencies)
2.14 Ghz 0.0%
1.60 Ghz 100.0%
< Detailed C-state information is only available on Mobile CPUs (laptops) >
Wakeups-from-idle per second : 132.4 interval: 10.0s
Top causes for wakeups:
93.2% (235.2) <interrupt> : extra timer interrupt
5.5% ( 14.0) hibernate.sh : queue_delayed_work_on
(delayed_work_timer_fn)
0.6% ( 1.5) <kernel core> : queue_delayed_work_on
(delayed_work_timer_fn)
0.2% ( 0.5) <kernel core> : neigh_table_init_no_netlink
(neigh_periodic_timer)
0.1% ( 0.3) <kernel module> : neigh_table_init_no_netlink
(neigh_periodic_timer)
0.1% ( 0.2) <kernel core> : __netdev_watchdog_up (dev_watchdog)
0.1% ( 0.2) rpc.idmapd : schedule_timeout (process_timeout)
0.0% ( 0.1) sshd : __neigh_event_send (neigh_timer_handler)
0.0% ( 0.1) <kernel core> : dst_run_gc (dst_run_gc)
0.0% ( 0.1) rpc.nfsd : nfs4_state_start (delayed_work_timer_fn)
0.0% ( 0.1) syslogd : do_setitimer (it_real_fn)
------------------------------------------------------------------------
Q1. I know my BIOS only supports C1. It has the C1E thing that linux
doesn't seem to notice. I'm assuming C1E is working properly though.
Why isn't any "Avg residency" information shown? I'd have thought linux
could track this on any computer, not just "Mobile CPUs".
Q2. Any ideas about hibernate.sh? It's the Ubuntu hibernation script
which I run manually. But it's definitely not running now - I checked
"ps -ax|grep hibernate". Maybe hibernation is spawning a periodic timer
that won't go away? Looks very buggy to me.
Q3. My computer has a HPET. I have to patch the kernel to get the HPET
acpi table. It seems to be working:
[EMAIL PROTECTED]:~/pkg/pm/work$ dmesg|grep hpet
[ 19.093421] hpet0: at MMIO 0xfed00000, IRQs 2, 8, 0
[ 19.093541] hpet0: 3 64-bit timers, 14318180 Hz
[ 21.583081] PM: Adding info for No Bus:hpet
[ 21.583277] hpet_resources: 0xfed00000 is busy
[ 29.811683] Time: hpet clocksource has been installed.
but the tsc clocksource is preferred, since I have fixed tscs (so
/proc/cpuinfo tells me, yay). Is this OK, or - once I have NO_HZ for my
shiny x86_64 kernel, will I need to force HPET in order to get the full
benefit?
Alan
_______________________________________________
Power mailing list
[email protected]
http://www.bughost.org/mailman/listinfo/power
