8a96c78931
CloneContext::Replace(T* what, T* with) is bug-prone, as complex transforms may want to clone `what` multiple times, or not at all. In both cases, this will likely result in an ICE as either the replacement will be reachable multiple times, or not at all. The CTS test: webgpu:shader,execution,robust_access:linear_memory:storageClass="storage";storageMode="read_write";access="read";atomic=true;baseType="i32" Was triggering this brokenness with DecomposeMemoryAccess's use of CloneContext::Replace(T*, T*). Switch the usage of CloneContext::Replace(T*, T*) to the new function form. As std::function is copyable, it cannot hold a captured std::unique_ptr. This prevented the Replace() lambdas from capturing the necessary `BufferAccess` data, as this held a `std::unique_ptr<Offset>`. To fix this, use a `BlockAllocator` for Offsets, and use raw pointers instead. Because the function passed to Replace() is called just before the node is cloned, insertion of new functions will occur just before the currently evaluated module-scope entity. This allows us to remove the "insert_after" arguments to LoadFunc(), StoreFunc(), and AtomicFunc(). We can also kill the icky InsertGlobal() and TypeDeclOf() helpers. Bug: tint:993 Change-Id: I60972bc13a2fa819a163ee2671f61e82d0e68d2a Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/58222 Kokoro: Kokoro <noreply+kokoro@google.com> Reviewed-by: James Price <jrprice@google.com> Commit-Queue: Ben Clayton <bclayton@google.com> |
||
|---|---|---|
| .vscode | ||
| build_overrides | ||
| docs | ||
| fuzzers | ||
| include/tint | ||
| infra/config | ||
| kokoro | ||
| samples | ||
| src | ||
| test | ||
| third_party | ||
| tools | ||
| .clang-format | ||
| .gitignore | ||
| .gn | ||
| AUTHORS | ||
| BUILD.gn | ||
| CMakeLists.txt | ||
| CMakeSettings.json | ||
| CODE_OF_CONDUCT.md | ||
| CONTRIBUTING.md | ||
| CPPLINT.cfg | ||
| DEPS | ||
| Doxyfile | ||
| LICENSE | ||
| OWNERS | ||
| PRESUBMIT.py | ||
| README.md | ||
| standalone.gclient | ||
| tint_overrides_with_defaults.gni | ||
README.md
Tint
Tint is a compiler for the WebGPU Shader Language (WGSL).
This is not an officially supported Google product.
Requirements
- Git
- CMake (3.10.2 or later)
- Ninja (or other build tool)
- Python, for fetching dependencies
- depot_tools in your path
Build options
TINT_BUILD_SPV_READER: enable the SPIR-V input reader (off by default)TINT_BUILD_WGSL_READER: enable the WGSL input reader (on by default)TINT_BUILD_SPV_WRITER: enable the SPIR-V output writer (on by default)TINT_BUILD_WGSL_WRITER: enable the WGSL output writer (on by default)TINT_BUILD_FUZZERS: enable building fuzzzers (off by default)
Building
Tint uses Chromium dependency management so you need to install depot_tools and add it to your PATH.
Getting source & dependencies
# Clone the repo as "tint"
git clone https://dawn.googlesource.com/tint tint
cd tint
# Bootstrap the gclient configuration
cp standalone.gclient .gclient
# Fetch external dependencies and toolchains with gclient
gclient sync
Compiling using CMake + Ninja
mkdir -p out/Debug
cd out/Debug
cmake -GNinja ../..
ninja # or autoninja
Compiling using CMake + make
mkdir -p out/Debug
cd out/Debug
cmake ../..
make # -j N for N-way parallel build
Compiling using gn + ninja
mkdir -p out/Debug
gn gen out/Debug
autoninja -C out/Debug
Fuzzers on MacOS
If you are attempting fuzz, using TINT_BUILD_FUZZERS=ON, the version of llvm
in the XCode SDK does not have the needed libfuzzer functionality included.
The build error that you will see from using the XCode SDK will look something like this:
ld: file not found:/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/lib/clang/11.0.0/lib/darwin/libclang_rt.fuzzer_osx.a
The solution to this problem is to use a full version llvm, like what you would
get via homebrew, brew install llvm, and use something like CC=<path to full clang> cmake .. to setup a build using that toolchain.
Checking [chromium-style] issues in CMake builds
The gn based work flow uses the Chromium toolchain for building in anticipation of integration of Tint into Chromium based projects. This toolchain has additional plugins for checking for style issues, which are marked with [chromium-style] in log messages. This means that this toolchain is more strict then the default clang toolchain.
In the future we will have a CQ that will build this work flow and flag issues
automatically. Until that is in place, to avoid causing breakages you can run
the [chromium-style] checks using the CMake based work flows. This requires
setting CC to the version of clang checked out by gclient sync and setting
the TINT_CHECK_CHROMIUM_STYLE to ON.
mkdir -p out/style
cd out/style
cmake ../..
CC=../../third_party/llvm-build/Release+Asserts/bin/clang cmake -DTINT_CHECK_CHROMIUM_STYLE=ON ../../ # add -GNinja for ninja builds
Issues
Please file any issues or feature requests at https://bugs.chromium.org/p/tint/issues/entry
Contributing
Please see the CONTRIBUTING and CODE_OF_CONDUCT files on how to contribute to Tint.
Tint has a process for supporting experimental extensions.