Project

General

Profile

Actions

Bug #10468

closed

Improve environment models for runtime power management callbacks

Added by Ilja Zakharov over 4 years ago. Updated over 4 years ago.

Status:
Closed
Priority:
Urgent
Assignee:
Category:
Environment models
Target version:
Start date:
08/19/2020
Due date:
% Done:

0%

Estimated time:
Detected in build:
git
Platform:
Published in build:

Description

There is a PM specification for general PM callbacks that are available for implementation in several kinds of bus drivers. The latest runs show that there are many false alarms caused by an incorrect order of invocation of runtime callbacks. One needs to investigate reasons and improve existing specifications.

Examples (clk2)
drivers/media/platform/stm32/stm32-dcmi.ko
drivers/media/platform/sti/delta/st-delta.ko
drivers/media/platform/qcom/venus/venus-enc.ko
drivers/media/platform/rockchip/rga/rockchip-rga.ko
drivers/media/platform/sunxi/sun4i-csi/sun4i-csi.ko
drivers/media/platform/sunxi/sun8i-di/sun8i-di.ko
drivers/media/platform/mtk-jpeg/mtk_jpeg.ko
drivers/media/platform/s5p-cec/s5p-cec.ko

Actions #1

Updated by Ilja Zakharov over 4 years ago

  • Status changed from New to Feedback

Implemented in runtime-pm-fixes.

Note, that some FP are not gone because of rule-related problems.

Actions #2

Updated by Evgeny Novikov over 4 years ago

  • Status changed from Feedback to Resolved

Ilja fixed the model one more time, so, it seems, that now everything is okay, but let's wait for new testing results.

Actions #3

Updated by Evgeny Novikov over 4 years ago

  • Subject changed from Improve environment models for runtime power management callbacks. to Improve environment models for runtime power management callbacks
  • Status changed from Resolved to Closed

Tested passed. Moreover, CI showed that 2 negative regressions introduced by branch extra-specifications have gone. Testing on real-world drivers demonstrate that corresponding false alarms disappear, but the environment model became a slightly more complicated, so, there are several new timeouts. So, I merged the branch to master in 30baa063e.

Actions

Also available in: Atom PDF