A locked screen (blank screensaver) has compiz generating ca. 40 wakeups/second

[Timo Jyrinki]

Following a fix to bug #917210, a new problem was uncovered. When the screen is locked (blank screensaver is shown), the compiz generates about 40 wakeups / second for the duration of screen being kept blank. Unlocking the screen brings the wakeups back to under 10 that the #917210 fix made possible.

[Colin Ian King (colin-king)]

Confirmed, a locked blank screen is generating 19-20 wakeups/second on my idle machine.

[Alan Pope 🍺🐧🐱 🦄 (popey)]

Confirmed. Sandybridge laptop is waking up lots whilst screensaver is on.

PowerTOP 1.97 Overview Idle stats Frequency stats Device stats Tunables

Summary: 146.0 wakeups/second, 0.0 GPU ops/second and 0.0 VFS ops/sec

                Usage Events/s Category Description
             2972 rpm Device Laptop fan
            100.0% Device Audio codec hwC0D0: Conexant
            100.0% Device Audio codec hwC0D3: Intel
              3.4 ms/s 35.5 Process compiz
              2.8 ms/s 21.7 Process /usr/bin/X :0 -auth /var/run/lightdm/root/:0 -nolisten tcp vt7 -novtswitch
            510.8 µs/s 16.8 Timer hrtimer_wakeup
              0.8 ms/s 12.8 Interrupt [6] tasklet(softirq)
            513.6 µs/s 10.8 Process nautilus -n
            439.5 µs/s 9.9 Process syndaemon -i 0.5 -K -R
             58.5 µs/s 8.9 kWork ieee80211_iface_work
            209.6 µs/s 4.9 Interrupt [7] sched(softirq)
              8.6 ms/s 1.0 Process powertop
             97.3 µs/s 3.9 Timer tick_sched_timer
             73.6 µs/s 3.9 Interrupt [47] i915
             75.8 µs/s 2.0 Process /usr/lib/rtkit/rtkit-daemon
              0.9 ms/s 1.0 Process /usr/lib/gnome-settings-daemon/gnome-settings-daemon
            412.4 µs/s 1.0 kWork e1000_watchdog_task
            162.5 µs/s 1.0 Process sshd: alan@pts/1
            127.1 µs/s 1.0 Process /usr/lib/indicator-datetime/indicator-datetime-service
             79.0 µs/s 1.0 Process /usr/lib/accountsservice/accounts-daemon
             76.3 µs/s 1.0 Process zeitgeist-datahub
             75.7 µs/s 1.0 Process /usr/lib/gvfs/gvfsd-trash --spawner :1.6 /org/gtk/gvfs/exec_spaw/0
             65.2 µs/s 1.0 Process /usr/lib/gvfs/gvfs-gdu-volume-monitor
             64.0 µs/s 1.0 Process /usr/lib/x86_64-linux-gnu/gconf/gconfd-2
             59.6 µs/s 1.0 Process /usr/lib/x86_64-linux-gnu/colord/colord
             53.4 µs/s 1.0 Process /usr/lib/gvfs/gvfs-afc-volume-monitor
             45.1 µs/s 1.0 Process /sbin/wpa_supplicant -B -P /run/sendsigs.omit.d/wpasupplicant.pid -u -s -O /var/run/wpa_supplicant
              7.0 µs/s 1.0 kWork gen6_pm_rps_work
              4.1 µs/s 1.0 kWork flush_to_ldisc
              0.8 ms/s 0.0 Interrupt [50] iwlwifi
            174.5 µs/s 0.0 Process [kworker/u:2]
            131.7 µs/s 0.0 Process [kworker/u:1]
            125.8 µs/s 0.0 Interrupt [9] RCU(softirq)
             84.5 µs/s 0.0 Timer delayed_work_timer_fn
             81.5 µs/s 0.0 kWork ...

[MaxNegro (maxnegro)]

I did a couple of tests on my setup. First of all, problem seems to disappear if i use the radeon driver instead of fglrx one. Then i ssh-ed in from another machine. This is the output of powertop with screen blank (only first few lines copied, compiz using 100% of a core):

            993,3 ms/s 0,0 Process compiz
            100,0% Device Codec audio hwC0D1: LSI
            100,0% Device Codec audio hwC0D0: IDT
              0,9 ms/s 81,6 Timer tick_sched_timer
              4,9 ms/s 15,7 Process /usr/lib/indicator-application/indicator-application-service
            108,0 µs/s 11,8 Timer hrtimer_wakeup

This the same, with screen active:

            100,0% Device Codec audio hwC0D1: LSI
            100,0% Device Codec audio hwC0D0: IDT
              3,5 ms/s 114,9 Timer tick_sched_timer
             69,7 ms/s 27,5 Process compiz
              1,6 ms/s 44,2 Interrupt [58] fglrx[0]@PCI:1:0:0
              0,9 ms/s 37,3 Interrupt [6] tasklet(softirq)
             34,8 ms/s 19,6 Process /usr/lib/unity/unity-panel-service
              9,4 ms/s 20,6 Process /usr/lib/indicator-application/indicator-application-service
              9,0 ms/s 20,6 Process //bin/dbus-daemon --fork --print-pid 5 --print-address 7 --session
            620,4 µs/s 16,7 Timer hrtimer_wakeup

fglrx interrupts are absent in the first lot. Is this expected?

[Daniel van Vugt (vanvugt)]

MaxNegro, you have bug 969860, which is different.

[MC Return (mc-return)]

Is this a duplicate ?