Merge lp:~vanvugt/mir/wake-lock-on-input into lp:mir
Status: | Rejected |
---|---|
Rejected by: | Alan Griffiths |
Proposed branch: | lp:~vanvugt/mir/wake-lock-on-input |
Merge into: | lp:mir |
Diff against target: |
173 lines (+133/-3) 3 files modified
3rd_party/android-input/android/CMakeLists.txt (+6/-0) 3rd_party/android-input/android/frameworks/base/services/input/EventHub.cpp (+2/-3) 3rd_party/android-input/android/hardware/libhardware_legacy/power/power.c (+125/-0) |
To merge this branch: | bzr merge lp:~vanvugt/mir/wake-lock-on-input |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Alberto Aguirre (community) | Disapprove | ||
PS Jenkins bot (community) | continuous-integration | Needs Fixing | |
Review via email: mp+262416@code.launchpad.net |
Commit message
Re-introduce wake locks to Android input (LP: #1465514)
It appears there was no good reason for us to have ever commented this
out in Mir. Upstream Android still has it, and in theory input
responsiveness will be helped by ensuring the CPU never goes to low-power
mode in the middle of processing an input event. Which it might, as we
have to dance between server and client processes which takes system calls
and context switches.
Although I am yet to see this make a measurable difference, it doesn't
make anything worse either. And in theory in some cases it should
improve performance.
Note (i): To get wake lock functionality you need to be root (ie. a
system compositor). This is because of the restricted permissions on /sys/power/*. I don't think it matters if you fail to open them.
Note (ii): Regular Linux kernels implement the same /sys/power interface now, as first introduced in Android kernels. But most desktop kernels
don't implement the behaviour behind it by default (it's an option
apparently).
Note (iii): Yes power.c is Android code. I'm trying to avoid modifying it.
Description of the change
This is a request for comments/testing. There might be a phone where this makes a significant difference but I've only tried mako so far...
Unmerged revisions
- 2676. By Daniel van Vugt
-
Merge latest trunk and fix conflict
- 2675. By Daniel van Vugt
-
Reintroduce wake locks from the Android source tree
- 2674. By Daniel van Vugt
-
sendFinishedSignal at the correct time.
FAILED: Continuous integration, rev:2676 jenkins. qa.ubuntu. com/job/ mir-ci/ 4149/ jenkins. qa.ubuntu. com/job/ mir-android- vivid-i386- build/2926 jenkins. qa.ubuntu. com/job/ mir-clang- wily-amd64- build/441/ console jenkins. qa.ubuntu. com/job/ mir-mediumtests -vivid- touch/2874 jenkins. qa.ubuntu. com/job/ mir-wily- amd64-ci/ 305 jenkins. qa.ubuntu. com/job/ mir-wily- amd64-ci/ 305/artifact/ work/output/ *zip*/output. zip jenkins. qa.ubuntu. com/job/ mir-mediumtests -builder- vivid-armhf/ 2874 jenkins. qa.ubuntu. com/job/ mir-mediumtests -builder- vivid-armhf/ 2874/artifact/ work/output/ *zip*/output. zip jenkins. qa.ubuntu. com/job/ mir-mediumtests -runner- mako/5700 s-jenkins. ubuntu- ci:8080/ job/touch- flash-device/ 21318
http://
Executed test runs:
SUCCESS: http://
FAILURE: http://
SUCCESS: http://
SUCCESS: http://
deb: http://
SUCCESS: http://
deb: http://
SUCCESS: http://
SUCCESS: http://
Click here to trigger a rebuild: s-jenkins. ubuntu- ci:8080/ job/mir- ci/4149/ rebuild
http://