skia2/modules/canvaskit/README.md
Kevin Lubick 9cb74e9079 [bazel] Compile gms for wasm and WebGL
PS 1 is re-generating existing BUILD.bazel files
PS 2 is generating BUILD.bazel files for tests/gms
PS 3+ makes modifications to build all of the gms and tests.

It is recommended to view this CL with just a diff between
PS 2 and the end, due to the large amount of generated changes
in PS 1 and 2.

We make a filegroup for the gms and tests because they need
to be compiled as one large blob in order for the registries
to work. Maybe in the future we will break these up, but at least
for WASM/JS, the overhead of starting a browser for each new
test would likely grind things to a halt, so we just group them
all together for now. It's also the most similar to what we
currently do.

In gm/BUILD.bazel and tests/BUILD.bazel, we add a cc_library
that encapsulates all of the deps of the tests, so we can
easily include that the build. These were discovered via
trial and error, not anything automatic or systematic.

The is_skia_dev_build config_setting is very similar to the
GN equivalent from which it was based.

The list of gms and tests to skip (e.g. which are incompatible
with WASM) was determined by building the wasm bundle:

modules/canvaskit$ make bazel_gms_release
tools/run-wasm-gm-tests$ make run_local_debug
# Don't forget to click the button on the screen after the
# browser loads

This way of invoking the tests will be replace soon with
`bazel test <something>`. As such, I didn't bother fully
documenting the current way.

Suggested review order:
 - modules/canvaskit/BUILD.bazel taking note that we always
   use profiling-funcs to make the stacktraces human readable.
 - gm/BUILD.bazel and tests/BUILD.bazel to see the lists of
   gms/tests. Notice the tests are roughly partitioned because
   we don't support things like vulkan/PDF in the wasm build
   and we will want a way to not build certain tests for
   certain configurations
 - tools/* noting some of the cc_libraries added to make
   dependencies easier to add when needed.
 - All other files.

Change-Id: I43059cd93c28af1c4c12b93d6ebd9c46a12d381f
Bug: skia:12541
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/506256
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Kevin Lubick <kjlubick@google.com>
2022-02-09 18:56:17 +00:00

7.4 KiB

Prerequisites

Node v14 or later is required to run tests. We use npm (the Node Package Manager) to install test dependencies. Recent installations of Node have npm as well. CanvasKit has no other external source dependencies.

To compile CanvasKit, you will first need to install emscripten. This will set the environment EMSDK (among others) which is required for compilation. Which version should you use? /infra/wasm-common/docker/emsdk-base/Dockerfile shows the version we build and test with. We try to use as recent a version of emscripten as is reasonable.

Be sure to both install and activate the correct version. For example:

    ./emsdk install 3.1.3
    ./emsdk activate 3.1.3

This document also assumes you have followed the instructions to download Skia and its deps https://skia.org/user/download.

MacOS specific notes

Make sure you have Python3 installed, otherwise the downloading emscripten toolchain can fail with errors about SSL certificates. https://github.com/emscripten-core/emsdk/pull/273

See also https://github.com/emscripten-core/emscripten/issues/9036#issuecomment-532092743 for a solution to Python3 using the wrong certificates.

Compile and Run Local Example

# The following installs all npm dependencies and only needs to be when setting up
# or if our npm dependencies have changed (rarely).
npm ci

make release  # make debug is much faster and has better error messages
make local-example

This will print a local endpoint for viewing the example. You can experiment with the CanvasKit API by modifying ./npm_build/example.html and refreshing the page. For some more experimental APIs, there's also ./npm_build/extra.html.

For other available build targets, see Makefile and compile.sh. For example, building a stripped-down version of CanvasKit with no text support or any of the "extras", one might run:

./compile.sh no_skottie no_particles no_font

Such a stripped-down version is about half the size of the default release build.

Unit tests, performance tests, and coverage.

To run unit tests and compute test coverage on a debug gpu build

make debug
make test-continuous

This reads karma.conf.js, and opens a Chrome browser and begins running all the test in test/ it will detect changes to the tests in that directory and automatically run again, however it will automatically rebuild and reload CanvasKit. Closing the chrome window will just cause it to re-opened. Kill the karma process to stop continuous monitoring for changes.

The tests are run with whichever build of CanvasKit you last made. be sure to also test with release, debug_cpu, and release_cpu. testing with release builds will expose problems in closure compilation and usually forgotten externs.

Coverage

Coverage will be automatically computed when running test-continuous locally. Note that the results will only be useful when testing a debug build. Open coverage/<browser version>/index.html For a summary and detailed line-by-line result.

Measuring Performance

We use puppeteer to run a Chrome browser to gather performance data in a consistent way. See //tools/perf-canvaskit-puppeteer for more.

Adding tests

The tests in tests/ are grouped into files by topic. Within each file there are describe blocks further organizing the tests, and within those it() functions which test particular behaviors. describe and it are jasmine methods which can both be temporarily renamed fdescribe and fit. Which causes jasmine to only those.

We have also defined gm which is a method for defining a test which draws something to a canvas that is shapshotted and reported to gold.skia.org, where you can compare it with the snapshot at head.

Testing from Gerrit

When submitting a CL in gerrit, click "choose tryjobs" and type CanvasKit to filter them. select all of them, which at the time of this writing is four jobs, for each combination of perf/test gpu/cpu.

The performance results are reported to perf.skia.org gold results are reported to gold.skia.org

Coverage is not measured while running tests this way.

Inspecting output WASM

The wasm2wat tool from the WebAssembly Binary Toolkit can be used to produce a human-readable text version of a .wasm file.

The output of wasm2wat --version should be 1.0.13 (1.0.17). This version has been checked to work with the tools in wasm_tools/SIMD/. These tools programmatically inspect the .wasm output of a CanvasKit build to detect the presence of wasm SIMD operations.

Infrastructure Playbook

When dealing with CanvasKit (or PathKit) on our bots, we use Docker. Check out $SKIA_ROOT/infra/wasm-common/docker/README.md for more on building/editing the images used for building and testing.

Updating the version of Emscripten we build/test with

This presumes you have updated emscripten locally to a newer version of the sdk and verified/fixed any build issues that have arisen.

  1. Edit $SKIA_ROOT/infra/wasm-common/docker/emsdk-base/Dockerfile to install and activate the desired version of Emscripten.
  2. Edit $SKIA_ROOT/infra/wasm-common/docker/Makefile to have EMSDK_VERSION be set to that desired version. If there is a suffix that is not _v1, reset it to be _v1. If testing the image later does not work and edits are made to the emsdk-base Dockerfile to correct that, increment to _v2,_v3, etc to force the bots to pick up the new image.
  3. In $SKIA_ROOT/infra/wasm-common/docker/, run make publish_emsdk_base
  4. Edit $SKIA_ROOT/infra/canvaskit/docker/canvaskit-emsdk/Dockerfile to be based off the new version from step 2. CanvasKit has its own docker image because it needs a few extra dependencies to build with font support.
  5. Edit $SKIA_ROOT/infra/canvaskit/docker/Makefile to have the same version from step 2. It's easiest to keep the emsdk-base and canvaskit-emsdk versions be in lock-step.
  6. In $SKIA_ROOT/infra/canvaskit/docker/, run make publish_canvaskit_emsdk.
  7. In $SKIA_ROOT/infra/bots/recipe_modules/build/, update canvaskit.py and pathkit.py to have DOCKER_IMAGE point to the desired tagged Docker containers from steps 2 and 5 (which should be the same).
  8. In $SKIA_ROOT/infra/bots/task_drivers/compile_wasm_gm_tests.go, update dockerImage to refer to the desired Docker containers from steps 2 and 5.
  9. In $SKIA_ROOT/infra/bots/, run make train to re-train the recipes.
  10. Optional: Run something like git grep 1\\.38\\. in $SKIA_ROOT to see if there are any other references that need updating.
  11. Upload a CL with all the changes. Run all Test.+CanvasKit, Perf.+Puppeteer, Test.+PathKit, Perf.+PathKit jobs to make sure the new builds pass all tests and don't crash the perf harnesses.
  12. Send out CL for review. Feel free to point the reviewer at these steps.

Running Skia's GMs and Unit Tests against wasm+WebGL

TODO(kjlubick)

General Tips:

  • Make use of the skip lists and start indexes in the run-wasm-gm-tests.html to focus in on problematic tests.
  • Uncaught (in promise) RuntimeError: function signature mismatch tends to mean null was dereferenced somewhere. Add SkASSERT to verify.