Comment 8 for bug 1421455

I was monitoring my syslog while pressing the button.

Apr 29 14:37:35 ubuntu-phablet powerd[949]: handle_requestSysState from :1.13 - ACTIVE (1)
Apr 29 14:37:35 ubuntu-phablet powerd[949]: name_watch_add: looking for :1.13
Apr 29 14:37:35 ubuntu-phablet powerd[949]: watching :1.13 to see when it disappears on dbus
Apr 29 14:37:35 ubuntu-phablet powerd[949]: libsuspend: acquire_wake_lock: powerd_power_request
Apr 29 14:37:35 ubuntu-phablet powerd[949]: handle_requestSysState - SUCCESS
Apr 29 14:37:35 ubuntu-phablet powerd[949]: Enqueue state change to ACTIVE
Apr 29 14:37:35 ubuntu-phablet powerd[949]: exiting suspend
Apr 29 14:37:35 ubuntu-phablet powerd[949]: libsuspend: exit_suspend.
Apr 29 14:37:35 ubuntu-phablet kernel: [21759.723201][Ker_PM][request_suspend_state]wakeup (3->0) at 21760134073325 (2015-04-29 12:37:35.882532317 UTC)

< exactly here there is a short delay>

Apr 29 14:37:36 ubuntu-phablet powerd[949]: Emitting signal for transition to state ACTIVE (1)
Apr 29 14:37:36 ubuntu-phablet powerd[949]: Transition to ACTIVE complete
Apr 29 14:37:36 ubuntu-phablet powerd[949]: libsuspend: release_wake_lock: powerd_power_request
Apr 29 14:37:36 ubuntu-phablet kernel: [21760.273491]mtk-tpd: TPD wake up
Apr 29 14:37:36 ubuntu-phablet powerd[949]: light_dev: setting brightness to 120
Apr 29 14:37:36 ubuntu-phablet kernel: [21760.310044]mtk-tpd: TPD wake up done
Apr 29 14:37:36 ubuntu-phablet kernel: [21760.310066]<MAGNETIC> mag_context_obj ok------->hwm_obj->early_suspend=0
Apr 29 14:37:36 ubuntu-phablet powerd[949]: we get signal from :1.13: DisplayPowerStateChange
Apr 29 14:37:36 ubuntu-phablet powerd[949]: Received DisplayPowerStateChange: state=1 flags=2

Since some of those messages are from the kernel ringbuffer, I looked at the corresponding dmesg output:

[21606.412808] [Ker_PM][request_suspend_state]wakeup (3->0) at 21606823677239 (2015-04-29 12:35:02.572136231 UTC)
[21606.953480] mtk-tpd: TPD wake up
[21606.980027] mtk-tpd: TPD wake up done

So there seems to be at least half a second of delay between the kernel registering the button press and the display actually being requested to power on. Since display power-on is not instant, the total time from button press to visible display content is close to a second on my device in best case, sometimes up to three seconds and in rare cases the display does not turn on at all.