Accelerate the build process of autoware.universe (mostly behavior_planning packages)

[soblin]

Checklist

• I've read the contribution guidelines.
• I've searched other issues and no duplicate issues were found.
• I've agreed with the maintainers that I can plan this task.

Description

Previously I found out that some packages are maybe consuming about 1/3 of its build time just for preprocessing due to huge header files.

• Analyze and improve compile time with clang #2773 (comment)

You can check how to measure the compile time of a single header file on the above thread. This table is the example:

file	time
tier4_autoware_utils/tier4_autoware_utils.hpp	9.5
motion_utils/motion_utils.hpp	9.7
route_handler/route_handler.hpp	14.0
lanelet2_extension/utility/query.hpp	4.0

To solve this issue a applied several refactorings that aims to

1. avoid include of tier4_autoware_utils.hpp and motion_utils.hpp
2. remove unnecessary includes in tier4_autoware_utils.hpp and motion_utils.hpp
3. export template functions in motion_utils.hpp as extern template
4. include lanelet2_core/lanelet2_routing/lanelet2_extension/Forward.hpp as much as possible

on this branch(https://github.com/soblin/autoware.universe/tree/build-fast/individual/behavior_velocity_planner_common) and measured how much the build time /maximum momory usage decreases.

I compiled all the files in behavior_velocity_planner_speed_bump_module in parallel and measured the process with time -v command.

Without the refactoring,

        Command being timed: "./all.sh"
        User time (seconds): 83.14
        System time (seconds): 8.26
        Percent of CPU this job got: 308%
        Elapsed (wall clock) time (h:mm:ss or m:ss): 0:29.65
        Average shared text size (kbytes): 0
        Average unshared data size (kbytes): 0
        Average stack size (kbytes): 0
        Average total size (kbytes): 0
        Maximum resident set size (kbytes): 2260592
        Average resident set size (kbytes): 0
        Major (requiring I/O) page faults: 581
        Minor (reclaiming a frame) page faults: 2463525
        Voluntary context switches: 11377
        Involuntary context switches: 970
        Swaps: 0
        File system inputs: 121160
        File system outputs: 375512
        Socket messages sent: 0
        Socket messages received: 0
        Signals delivered: 0
        Page size (bytes): 4096
        Exit status: 0

After the refactorings,

        Command being timed: "./all.sh"
        User time (seconds): 46.03
        System time (seconds): 5.06
        Percent of CPU this job got: 348%
        Elapsed (wall clock) time (h:mm:ss or m:ss): 0:14.66
        Average shared text size (kbytes): 0
        Average unshared data size (kbytes): 0
        Average stack size (kbytes): 0
        Average total size (kbytes): 0
        Maximum resident set size (kbytes): 1351576
        Average resident set size (kbytes): 0
        Major (requiring I/O) page faults: 621
        Minor (reclaiming a frame) page faults: 1449186
        Voluntary context switches: 31465
        Involuntary context switches: 625
        Swaps: 0
        File system inputs: 163024
        File system outputs: 178976
        Socket messages sent: 0
        Socket messages received: 0
        Signals delivered: 0
        Page size (bytes): 4096
        Exit status: 0

You can see the build time / maximum memory usage is reduced by about 50%.

Purpose

x

Possible approaches

This task is ready once

• feat(build)!: add forward declaration header autoware_common#203
• chore(build): include missing lanelet2 header #4819

are approved. Then I will ship each PR in several steps:

• Phase1:
  • remove tier4_autoware_utils.hpp from common/
  • remove motion_utils.hpp from common/
  • remove tier4_autoware_utils.hpp from vehicle/
• Phase2:
  • remove tier4_autoware_utils.hpp from evaluator/ and simulator/
  • remove tier4_autoware_utils.hpp from system/
  • remove tier4_autoware_utils.hpp from localization/
  • remove tier4_autoware_utils.hpp from perception/
  • remove tier4_autoware_utils.hpp and motion_utils from planning/
  • remove tier4_autoware_utils.hpp from control/
• Phase3:
  • only include forward declaration of lanelet packages in sensing
  • only include forward declaration of lanelet packages in perception
  • only include forward declaration of lanelet packages in control
  • only include forward declaration of lanelet packages in planning
• Phase4:
  • refactor tier4_autoware_utils
  • refactor motion_utils
  • refactor behavior_velocity_planner_common

Definition of done

x