Bloaty McBloatface GitHub Action

GitHub Action to run Google's Bloaty McBloatface - a size profiler for binaries: https://github.com/google/bloaty/

- name: Run Bloaty McBloatface on an ELF file
  uses: carlosperate/bloaty-action@v1
  with:
    bloaty-args: <path_to_your_file_and_any_bloaty_flags>

• 🧑‍💻 Additional examples, including Job Summaries and PR comments, can be found in the "Additional Action Examples" section.
• 🐳 A Bloaty Docker image (ghcr.io/carlosperate/bloaty) is also provided, more info in the "Using the Docker Image to run Bloaty directly" section.

Action Inputs/Outputs

Inputs:

• bloaty-args: (Required) All arguments to pass to Bloaty McBloatface.
• output-to-summary: (Optional, default false) Boolean (true or false) to include the bloaty output in the GitHub Actions Job Summary.
• summary-title: (Optional, default "bloaty output") If output-to-summary is enabled, this is the title on top of the bloaty output.

Outputs:

• bloaty-output: A string with the output from Bloaty McBloatface
• bloaty-output-encoded: The bloaty output string with escaped characters (so you'll get things like \n). It can be easier to pass this to other action steps.

Using the Docker image to run Bloaty directly

This repository contains two Dockerfiles and the two Docker images are hosted in the GitHub Docker Container registry.

The ghcr.io/carlosperate/bloaty Docker image contains the bloaty application on its own, and can be used used to easily run bloaty directly in your own environment or applications.

For example, to diff two ELF files contained in this repo, you can run the following command from this repository root directory:

docker run --rm -v $(pwd):/home ghcr.io/carlosperate/bloaty:latest test-elf-files/example-after.elf -- test-elf-files/example-before.elf

    FILE SIZE        VM SIZE    
 --------------  -------------- 
   +14%  +221Ki  [ = ]       0    .debug_info
   +12% +52.4Ki  [ = ]       0    .debug_line
   +13% +46.0Ki  [ = ]       0    .debug_loc
   +13% +24.5Ki  [ = ]       0    .debug_abbrev
   +18% +20.8Ki   +18% +20.8Ki    .text
  +9.8% +16.4Ki  [ = ]       0    .debug_str
   +17% +11.2Ki  [ = ]       0    .symtab
   +21% +10.9Ki  [ = ]       0    .strtab
   +11% +7.89Ki  [ = ]       0    .debug_ranges
  +9.5% +5.57Ki  [ = ]       0    .debug_frame
  +9.3% +1.71Ki  [ = ]       0    .debug_aranges
  [ = ]       0  +9.6%    +792    .bss
 -24.4%    -120 -26.5%    -120    .data
  [ = ]       0  -0.7%    -792    .heap
 -15.2% -20.7Ki  [ = ]       0    [Unmapped]
   +12%  +397Ki  +5.6% +20.7Ki    TOTAL

The other ghcr.io/carlosperate/bloaty-action Docker image is built on top of the base bloaty image to run with GitHub Actions, and includes a custom script adding GitHub Actions specific features.

Additional Action Examples

To better understand what arguments to use with Bloaty, the documentation is not very long and a recommended read: google/bloaty/doc/using.md

Personally, I like to use the following flags to analyse where the data/memory is going:

bloaty -d compileunits,symbols --domain=vm <path_to_file>

To add a the Bloaty output to the GitHub Actions Job Summary simply set the output-to-summary input to true:

- name: Run Bloaty & add output to the run summary
  uses: carlosperate/bloaty-action@v1
  with:
    bloaty-args: -d compileunits,symbols test-elf-files/example-before.elf
    output-to-summary: true
    summary-title: "Size profile of `example-before.elf` largest components"

To create a PR comment, add an id to the carlosperate/bloaty-action step, and then use its output with the the actions/github-script action to post a markdown comment to the PR:

- name: Run Bloaty McBloatface on an ELF file
  uses: carlosperate/bloaty-action@v1
  id: bloaty-step
  with:
    bloaty-args: test-elf-files/example-before.elf
- name: Add a PR comment with the bloaty output
  uses: actions/github-script@v6
  with:
    script: |
      github.rest.issues.createComment({
        issue_number: context.issue.number,
        owner: context.repo.owner,
        repo: context.repo.repo,
        body: '## Bloaty output\n```\n${{ steps.bloaty-step.outputs.bloaty-output-encoded }}```\n'
      })

The following example shows how to build your project before and after the PR commits, and how to post the size diff as a PR comment:

steps:
  - name: Check out the repo with the full git history
    uses: actions/checkout@v3
    with:
      fetch-depth: '0'
  - name: Build your project (example for a standard Makefile, change as required)
    run: make
  - name: Save the built ELF/Mach-O/PE/COFF file to a different directory where it doesn't get cleaned out
    run: mv <path_to_your_elf> ../original.elf
  - name: If it's a PR checkout the base commit
    if: ${{ github.event.pull_request }}
    run: git checkout ${{ github.event.pull_request.base.sha }}
  - name: Clean the build and rebuild with the base commit
    if: ${{ github.event.pull_request }}
    run: |
      make clean
      make
  - name: Run Bloaty to compare both output files
    if: ${{ github.event.pull_request }}
    id: bloaty-comparison
    uses: carlosperate/bloaty-action@v1
    with:
      bloaty-args: ../original.elf -- <path_to_the_base_commit_elf>
      output-to-summary: true
  - name: Add a PR comment with the bloaty diff
    if: ${{ github.event.pull_request }}
    continue-on-error: true
    uses: actions/github-script@v6
    with:
      script: |
        github.rest.issues.createComment({
          issue_number: context.issue.number,
          owner: context.repo.owner,
          repo: context.repo.repo,
          body: '## PR build size diff\n```\n${{ steps.bloaty-comparison.outputs.bloaty-output-encoded }}```\n'
        })