[regression] mir_demo_client_eglplasma results in 100% cpu and hardly any updates in dragonboard
Bug #1665802 reported by
Alberto Aguirre
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Mir |
Opinion
|
High
|
Daniel van Vugt | ||
0.26 |
Opinion
|
High
|
Daniel van Vugt | ||
mesa (Ubuntu) |
New
|
High
|
Unassigned |
Bug Description
On a dragonboard 410c,
With Mir 0.25 mir_demo_
With mir 0.26.1, the process takes 100% cpu (according to top) and frames hardly update on screen, but the client performance is 60-27FPS.
To replicate:
mir_demo_server --arw-file --launch-client mir_demo_
Related branches
lp://staging/~vanvugt/mir/testfix-1665802
- Mir CI Bot: Needs Fixing (continuous-integration)
- Daniel van Vugt: Disapprove
-
Diff: 136 lines (+96/-1)3 files modifiedsrc/client/frame_clock.cpp (+7/-1)
src/client/frame_clock.h (+1/-0)
tests/unit-tests/client/test_frame_clock.cpp (+88/-0)
lp://staging/~vanvugt/mir/fix-1665802
- Daniel van Vugt: Disapprove
-
Diff: 163 lines (+105/-5)3 files modifiedsrc/client/frame_clock.cpp (+10/-1)
src/client/frame_clock.h (+1/-0)
tests/unit-tests/client/test_frame_clock.cpp (+94/-4)
lp://staging/~vanvugt/mir/schedule-nonblocking
- Alan Griffiths: Approve
- Mir CI Bot: Approve (continuous-integration)
-
Diff: 254 lines (+110/-3)11 files modifiedsrc/server/compositor/dropping_schedule.cpp (+17/-1)
src/server/compositor/dropping_schedule.h (+2/-0)
src/server/compositor/queueing_schedule.cpp (+7/-0)
src/server/compositor/queueing_schedule.h (+2/-0)
src/server/compositor/schedule.h (+3/-0)
src/server/compositor/stream.cpp (+11/-1)
src/server/compositor/stream.h (+1/-1)
tests/unit-tests/compositor/test_dropping_schedule.cpp (+25/-0)
tests/unit-tests/compositor/test_multi_monitor_arbiter.cpp (+6/-0)
tests/unit-tests/compositor/test_queueing_schedule.cpp (+19/-0)
tests/unit-tests/compositor/test_stream.cpp (+17/-0)
lp://staging/~vanvugt/mir/workaround-1665802
- Mir development team: Pending requested
-
Diff: 142 lines (+71/-6)3 files modifiedsrc/client/buffer_stream.cpp (+20/-5)
src/client/buffer_stream.h (+1/-0)
tests/unit-tests/client/test_client_buffer_stream.cpp (+50/-1)
lp://staging/~vanvugt/mir/testfix-1211700
- Mir CI Bot: Needs Fixing (continuous-integration)
- Daniel van Vugt: Disapprove
-
Diff: 82 lines (+18/-1)2 files modifiedsrc/server/compositor/multi_threaded_compositor.cpp (+10/-0)
src/server/compositor/stream.cpp (+8/-1)
Changed in mir: | |
importance: | Undecided → Critical |
description: | updated |
Changed in mir: | |
status: | Fix Committed → In Progress |
Changed in mir: | |
milestone: | 1.0.0 → none |
status: | In Progress → Won't Fix |
To post a comment you must log in.
Sounds like it's spinning (experiencing swap interval 0 behaviour by accident), similar to bug 1661072 and bug 1661128.
1. Can you provide the output from 'mirout' or just the display config part of the server log?
2. Also please provide output from 'env MIR_CLIENT_ PERF_REPORT= log mir_demo_ client_ eglplasma' ...
3. Do other clients like egltriangle suffer?