[vulkan snapshot]: handle vkResetCommandPool

When this is called, the command buffers allocated
from this command pool should all reset their recording.

This call is equivalent to ask each command buffer from
this pool to do vkResetCommandBuffer.

Bug: 415349936
Change-Id: I7a9c76ab16463c16f48336fcef34f532410aac41
5 files changed
tree: b43ddb99a1f58b3a617616fec1ef8983d2483cf5
  1. cmake/
  2. codegen/
  3. common/
  4. docs/
  5. guest/
  6. host/
  7. qnx/
  8. scripts/
  9. subprojects/
  10. third-party/
  11. utils/
  12. .clang-format
  13. .gitignore
  14. Android.bp
  15. BUILD.bazel
  16. CMakeLists.txt
  17. LICENSE
  18. meson.build
  19. meson_options.txt
  20. MODULE.bazel
  21. MODULE_LICENSE_APACHE2
  22. OWNERS
  23. README.md
README.md

Graphics Streaming Kit (formerly: Vulkan Cereal)

Graphics Streaming Kit (colloquially known as Gfxstream) is a code generator that makes it easier to serialize and forward graphics API calls from one place to another:

  • From a virtual machine guest to host for virtualized graphics
  • From one process to another for IPC graphics
  • From one computer to another via network sockets

Building

Linux

Bazel

The Bazel build used for building the host backend for virtual device host tooling.

The Bazel build currently requires an Android repo but a standalone build is a WIP.

cd <aosp/emu-dev>

prebuilts/bazel/linux-x86_64/bazel test \
    --platforms=@//build/bazel/platforms:linux_x64 \
    @gfxstream//...

CMake

The CMake build has historically been used for building the host backend for Goldfish.

The CMake build can be used from either a standalone Gfxstream checkout or from inside an Android repo.

Then,

mkdir build && cd build
cmake .. -G Ninja
ninja

For validating a Goldfish build,

cd <aosp/emu-master-dev repo>

cd external/qemu

python android/build/python/cmake.py --gfxstream

Meson

The Meson build has historically been used for building the backend for Linux guest on Linux host use cases.

cd <aosp/main repo>

meson setup \
    -Ddefault_library=static \
    -Dgfxstream-build=host \
    build

meson compile -C build

Soong

The Android Soong build is used for building the guest components for virtual device (Cuttlefish, Goldfish, etc) images and was previously used for building the host backend for virtual device host tools.

Please follow the instructions here for getting started with Android development and setting up a repo.

Then,

m libgfxstream_backend

and libgfxstream_backend.so can be found in out/host.

For validating changes, consider running

cd hardware/google/gfxstream
mma

to build everything inside of the Gfxstream directory.

Windows

Make sure the latest CMake is installed. Make sure Visual Studio 2019 is installed on your system along with all the Clang C++ toolchain components. Then:

mkdir build
cd build
cmake . ../ -A x64 -T ClangCL

A solution file should be generated. Then open the solution file in Visual studio and build the gfxstream_backend target.

Codegen

Regenerating GLES/RenderControl code

First, build build/gfxstream-generic-apigen. Then run:

scripts/generate-apigen-source.sh

Regenerating Vulkan code

To re-generate both guest and Vulkan code, please run:

scripts/generate-gfxstream-vulkan.sh

Testing

Android Host Tests

There are Android mock tests available, runnable on Linux. To build these tests, run:

m GfxstreamEnd2EndTests

Windows Tests

There are a bunch of test executables generated. They require libEGL.dll and libGLESv2.dll and vulkan-1.dll to be available, possibly from your GPU vendor or ANGLE, in the %PATH%.

Features

Tracing

The host renderer has optional support for Perfetto tracing which can be enabled by defining GFXSTREAM_BUILD_WITH_TRACING (enabled by default on Android builds).

The perfetto and traced tools from Perfetto should be installed. Please see the Perfetto Quickstart or follow these short form instructions:

cd <your Android repo>/external/perfetto

./tools/install-build-deps

./tools/gn gen --args='is_debug=false' out/linux

./tools/ninja -C out/linux traced perfetto

To capture a trace on Linux, start the Perfetto daemon:

./out/linux/traced

Then, run Gfxstream with Cuttlefish:

cvd start --gpu_mode=gfxstream_guest_angle_host_swiftshader

Next, start a trace capture with:

./out/linux/perfetto --txt -c gfxstream_trace.cfg -o gfxstream_trace.perfetto

with gfxstream_trace.cfg containing the following or similar:

buffers {
  size_kb: 4096
}
data_sources {
  config {
    name: "track_event"
    track_event_config {
    }
  }
}

Next, end the trace capture with Ctrl + C.

Finally, open https://ui.perfetto.dev/ in your webbrowser and use “Open trace file” to view the trace.

Design Notes

Guest Vulkan

gfxstream vulkan is the most actively developed component. Some key commponents of the current design include:

  • 1:1 threading model - each guest Vulkan encoder thread gets host side decoding thread
  • Support for both virtio-gpu, goldish and testing transports.
  • Support for Android, Fuchsia, and Linux guests.
  • Ring Buffer to stream commands, in the style of io_uring.
  • Mesa embedded to provide dispatch and objects.
  • Currently, there are a set of Mesa objects and gfxstream objects. For example, struct gfxstream_vk_device and the gfxstream object goldfish_device both are internal representations of Vulkan opaque handle VkDevice. The Mesa object is used first, since Mesa provides dispatch. The Mesa object contains a key to the hash table to get a gfxstream internal object (for example, gfxstream_vk_device::internal_object). Eventually, gfxstream objects will be phased out and Mesa objects used exclusively.

Future Project Ideas

Gfxstream is a first class open source project, and welcomes new contributors. There are many interesting projects available, for new and experienced software enthusiasts. Some ideas include:

  1. New OS support (Windows, Haiku, MacOS) support for gfxstream guest
  2. Rewriting the gfxstream protocol using python templates and working with other FOSS projects to de-duplicate
  3. Guided performance optimizations
  4. KVM or hypervisor integration to close gap between HW GPU virtualization
  5. Improving rutabaga integrations
  6. Improving display virtualization

Please reach out to your local gfxstreamist today if you are interested!