Google Git
Sign in
android / platform / hardware / google / gfxstream / e81891a3affd40437c0f39f885b55ef800b57d5c
commite81891a3affd40437c0f39f885b55ef800b57d5c[log] [tgz]
authorGurchetan Singh <[email protected]>Tue Jun 11 10:31:17 2024 -0700
committerGurchetan Singh <[email protected]>Mon Jul 01 10:39:11 2024 -0700
tree3878423d264c5c369e499a32d6a389fa3c350ff0
parent59f581f2a15524fd46a5dc0c0bf8d620589a97ab [diff]
gfxstream: guest: enable virtio-gpu kumquat

This uses libvirtgpu_kumquat_ffi.so to send commands to the
Kumquat Media Server.  The implementation of Kumquat is found
at crrev.com/c/5645904.

This is a more complex, but more fully featured end-to-end testing
framework.  This biggest benefit it is can run complex apps (vkcube,
gfxbench + ANGLE/Zink) on Linux, without a VM or a full Android tree.
The rutabaga FFI path -- which relied on a nested Vulkan loader -- didn't
work for complex apps and funkiness was observed with the nested
Vulkan loader.

Plus, app 1 + app 2 could connect Kumquat at the same time.

This may also benefit snapshot tests, particular since end to end
flow relies on external blob.  For example, a snapshot save command
could actually call stream_renderer_teardown(..) on the host-side,
while the guest retains it's state (ASG mappings + vulkano mappings).

A snapshot restore would actually call stream_renderer_init(..)
+ stream_renderer_restore(..), which an actual upstream VMM would
do.

Another additional benefit is multi-context testing.  Since
libvirtgpu_kumquat_ffi.so is portable, other libraries (minigbm)
can also call it.  We can mimic the cross-domain context allocating
via libminigbm.so, while context libgfxstream_vulkan.so imports the
file descriptor.

This change only transitions the meson build since the main goal
is once again is Linux on Linux testing w/o an Android tree.

BUG=300140266
TEST=run apps via Kumquat

Change-Id: I37494a6146a969f5f21ac208141b640d2272692e
10 files changed
tree: 3878423d264c5c369e499a32d6a389fa3c350ff0
  1. cmake/
  2. codegen/
  3. common/
  4. docs/
  5. guest/
  6. host/
  7. include/
  8. qnx/
  9. scripts/
  10. subprojects/
  11. third-party/
  12. utils/
  13. .clang-format
  14. .gitignore
  15. Android.bp
  16. BUILD.bazel
  17. CMakeLists.txt
  18. LICENSE
  19. meson.build
  20. meson_options.txt
  21. MODULE_LICENSE_APACHE2
  22. OWNERS
  23. README.md
README.md

Graphics Streaming Kit (formerly: Vulkan Cereal)

Graphics Streaming Kit 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

Build: Linux

The latest directions for the standalone Linux build are provided here.

Build: 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.

Build: Android for host

Be in the Android build system. Then:

m libgfxstream_backend

It then ends up in out/host

This also builds for Android on-device.

Output artifacts

libgfxstream_backend.(dll|so|dylib)

Regenerating Vulkan code

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

scripts/generate-gfxstream-vulkan.sh

Regenerating GLES/RenderControl code

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

scripts/generate-apigen-source.sh

Tests

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%.

Android Host Tests

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

m GfxstreamEnd2EndTests

Structure

  • CMakeLists.txt: specifies all host-side build targets. This includes all backends along with client/server setups that live only on the host. Some
    • Backend implementations
    • Implementations of the host side of various transports
    • Frontends used for host-side testing with a mock implementation of guest graphics stack (mainly Android)
    • Frontends that result in actual Linux/macOS/Windows gles/vk libraries (isolation / fault tolerance use case)
  • Android.bp: specifies all guest-side build targets for Android:
    • Implementations of the guest side of various transports (above the kernel)
    • Frontends
  • BUILD.gn: specifies all guest-side build targets for Fuchsia
    • Implementations of the guest side of various transports (above the kernel)
    • Frontends
  • base/: common libraries that are built for both the guest and host. Contains utility code related to synchronization, threading, and suballocation.
  • protocols/: implementations of protocols for various graphics APIs. May contain code generators to make it easy to regen the protocol based on certain things.
  • host-common/: implementations of host-side support code that makes it easier to run the server in a variety of virtual device environments. Contains concrete implementations of auxiliary virtual devices such as Address Space Device and Goldfish Pipe.
  • stream-servers/: implementations of various backends for various graphics APIs that consume protocol. gfxstream-virtio-gpu-renderer.cpp contains a virtio-gpu backend implementation.

Guest Vulkan design

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.