Skip to contentSkip to conversation

tests: Introduce kms_wm_stress

PATCH In review Matt Roper 2 participants 2 messages Jun 21, 2016

lore
Share & tools
lore

Matt Roper

author
#0 lore raw

PATCH 1/1 tests: Introduce kms_wm_stress

Stress test watermark programming by programming lots of random pipe/plane configurations. The test itself should always return success, but the presence of underrun messages in dmesg will indicate watermark programming failures.

There's a lot of room for future improvement here --- we should exercise random plane scaling, random pipe scaling, random modesets, etc. But this should at least provide a starting point upon which additional testcases can be written.

---
 tests/Makefile.sources |   1 +
 tests/kms_wm_stress.c  | 185 +++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 186 insertions(+)
 create mode 100644 tests/kms_wm_stress.c
tests/Makefile.sources +1 -0
index 8d8ede6..db2d5af 100644
@@ -106,6 +106,7 @@ TESTS_progs_M = \
 	kms_setmode \
 	kms_universal_plane \
 	kms_vblank \
+	kms_wm_stress \
 	kms_crtc_background_color \
 	kms_plane_scaling \
 	kms_panel_fitting \
tests/kms_wm_stress.c new file +185 -0

185 lines not on this page · raw message

Signature · 6 lines
2.1.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#1 lore raw
Quoted 4 lines · Matt Roper

Stress test watermark programming by programming lots of random pipe/plane configurations. The test itself should always return success, but the presence of underrun messages in dmesg will indicate watermark programming failures.

Could we expose a per-pipe underrun counter to debugfs?
(And then we could just emit a warning message on the first, but keep on counting for debug.)
-Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx