Starboard

Starboard is Cobalt's porting layer and OS abstraction. It attempts to encompass all the platform-specific functionality that Cobalt actually uses, and nothing that it does not.

GN Migration Notice

Cobalt and Starboard are currently migrating from the GYP build system to the GN build system. This readme contains instructions for both systems. As of now, GN is not ready for general use for Cobalt/Starboard. If you are not a core Cobalt developer, you should probably ignore the sections titled “GN Instructions” for now.

Documentation

See src/starboard/doc for more detailed documentation.

Interesting Source Locations

All source locations are specified relative to src/starboard/ (this directory).

• . - This is the root directory for the Starboard project, and contains all the public headers that Starboard defines.
• examples/ - Example code demonstrating various aspects of Starboard API usage.
• stub/ - The home of the Stub Starboard implementation. This contains a starboard_platform.gyp file that defines a library with all the source files needed to provide a complete linkable Starboard implementation.
• nplb/ - “No Platform Left Behind,” Starboard's platform verification test suite.
• shared/ - The home of all code that can be shared between Starboard implementations. Subdirectories delimit code that can be shared between platforms that share some facet of their OS API.

Quick Guide to Starting a Port

I. Enumerate and Name Your Platform Configurations

Before starting a Cobalt/Starboard port, first you will need to define the canonical names for your set of platform configurations. These will be used when organizing the code for your platforms.

What determines what goes into one platform configuration versus another? A platform configuration has a one-to-one mapping to a production binary. So, if you will need to produce a new binary, you are going to need a new platform configuration for that.

The recommended naming convention for a <platform-configuration> is:

<family-name>-<binary-variant>

Where <family-name> is a name specific to the family of products you are porting to Starboard and <binary-variant> is one or more tokens that uniquely describe the specifics of the binary you want that configuration to produce.

For example, let's say your company is named BobCo. BobCo employs multiple different device architectures so it will need to define multiple platform configurations.

All the BobCo devices are called BobBox, so it's a reasonable choice as a product <family-name>. But they have both big- and little-endian ARM chips. So they might define two platform configurations:

1. bobbox-armeb - For big-endian ARM devices.
2. bobbox-armel - For little-endian ARM devices.

II. Choose a Location in the Source Tree for Your Starboard Port

To be perfectly compatible with the Cobalt source tree layout, any code that is written by a party that isn't the Cobalt team should be in the src/third_party/ directory. The choice is up to you, but we recommend that you follow this practice, even if, as we expect to be common, you do not plan on sharing your Starboard implementation with anyone.

Primarily, following this convention ensures that no future changes to Cobalt or Starboard will conflict with your source code additions. Starboard is intended to be a junction where new Cobalt versions or Starboard implementations can be replaced without significant (and hopefully, any) code changes.

We recommend that you place your code here in the source tree:

src/third_party/starboard/<family-name>/

With subdirectories:

• shared/ - For code shared between architectures within a product family.
• <binary-variant>/ - For any code that is specific to a specific binary variant. Each one of these must at least have configuration_public.h, atomic_public.h, thread_types_public.h, gyp_configuration.py, gyp_configuration.gypi, and starboard_platform.gyp files.

In the BobCo's BobBox example, we would see something like:

• src/third_party/starboard/bobbox/
  • shared/
  • armeb/
    • atomic_public.h
    • configuration_public.h
    • gyp_configuration.gypi
    • gyp_configuration.py
    • starboard_platform.gyp
    • thread_types_public.h
  • armel/
    • atomic_public.h
    • configuration_public.h
    • gyp_configuration.gypi
    • gyp_configuration.py
    • starboard_platform.gyp
    • thread_types_public.h

And so on.

GN Instructions

Each <binary-variant>/ directory must have at least configuration_public.h, atomic_public.h, thread_types_public.h, BUILD.gn, configuration.gni, and buildconfig.gni.

In the BobCo's BobBox example, we would see a directory tree like:

• src/third_party/starboard/bobbox/
  • shared/
  • armeb/
    • buildconfig.gni
    • configuration.gni
    • atomic_public.h
    • BUILD.gn
    • configuration_public.h
    • thread_types_public.h
  • armel/
    • buildconfig.gni
    • configuration.gni
    • atomic_public.h
    • BUILD.gn
    • configuration_public.h
    • thread_types_public.h

III. Base Your Port on a Reference Port

You can start off by copying files from a reference port to your port's location. Currently these reference ports include:

• src/starboard/stub
• src/starboard/linux
• src/starboard/raspi

The starboard_platform.gyp contains absolute paths, so the paths will still be valid if you copy it to a new directory. You can then incrementally replace files with new implementations as necessary.

The cleanest, simplest starting point is from the Stub reference implementation. Nothing will work, but you should be able to compile and link it with your toolchain. You can then replace stub implementations with implementations from src/starboard/shared or your own custom implementations module-by-module, until you have gone through all modules.

You may also choose to copy either the Desktop Linux or Raspberry Pi ports and work backwards fixing things that don't compile or work on your platform.

For example, for bobbox-armel, you might do:

mkdir -p src/third_party/starboard/bobbox
cp -R src/starboard/stub src/third_party/starboard/bobbox/armel

Modify the files in <binary-variant>/ as appropriate (you will probably be coming back to these files a lot).

Update <binary-variant>/starboard_platform.gyp to point at all the source files that you want to build as your new Starboard implementation. The '<(DEPTH)' expression in GYP expands to enough ../s to take you to the src/ directory of your source tree. Otherwise, files are assumed to be relative to the directory the .gyp or .gypi file is in.

In order to use a new platform configuration in a build, you need to ensure that you have a gyp_configuration.py, gyp_configuration.gypi, and starboard_platform.gyp in their own directory for each binary variant, plus the header files configuration_public.h, atomic_public.h, and thread_types_public.h. gyp_cobalt will scan your directories for these files, and then calculate a port name based on the directories between src/third_party/starboard and your gyp_configuration.* files. (e.g. for src/third_party/starboard/bobbox/armeb/gyp_configuration.py, it would choose the platform configuration name bobbox-armeb.)

GN Instructions

Update <binary-variant>/BUILD.gn to point at all the source files that you want to build as your new Starboard implementation. The // expression in GN refers to the src/ directory of your source tree. Otherwise, files are assumed to be relative to the directory the BUILD.gn or .gni file is in. The BUILD.gn file contains absolute paths, so the paths will still be valid if you copy it to a new directory. You can then incrementally replace files with new implementations as necessary.

In order to use a new platform configuration in a build, you need to ensure that you have a BUILD.gn, configuration.gni, and buildconfig.gni in their own directory for each binary variant, plus the header files configuration_public.h, atomic_public.h, and thread_types_public.h. The GN build will scan your directories for these files, and then calculate a port name based on the directories between src/third_party/starboard and your configuration.gni files. (e.g. for src/third_party/starboard/bobbox/armeb/configuration.gni, it would choose the platform configuration name bobbox-armeb.)

IV. A New Port, Step-by-Step

1. Recursively copy src/starboard/stub to src/third_party/starboard/<family-name>/<binary-variant>. You may also consider copying from another reference platform, like raspi-2 or linux-x64x11.
2. In gyp_configuration.py
   1. In the CreatePlatformConfig() function, pass your <platform-configuration> as the parameter to the PlatformConfig constructor, like return PlatformConfig('bobbox-armeb').
   2. In GetVariables
      1. Set 'clang': 1 if your toolchain is clang.
      2. Delete other variables in that function that are not needed for your platform.
   3. In GetEnvironmentVariables, set the dictionary values to point to the toolchain analogs for the toolchain for your platform.
3. In gyp_configuration.gypi
   1. Update the names of the configurations and the default_configuration to be <platform-configuration>_<build-type> for your platform configuration name, where <build-type> is one of debug, devel, qa, gold.
   2. Update your platform variables.
      1. Set 'target_arch' to your architecture: 'arm', 'x64', 'x86'
      2. Set 'target_os': 'linux' if your platform is Linux-based.
      3. Set 'gl_type': 'system_gles2' if you are using the system EGL + GLES2 implementation.
      4. Set 'in_app_dial' to 1 or 0. This enables or disables the DIAL server that runs inside Cobalt, only when Coblat is running. You do not want in-app DIAL if you already have system-wide DIAL support.
   3. Update your toolchain command-line flags and libraries. Make sure you don't assume a particular workstation layout, as it is likely to be different for someone else.
   4. Update the global defines in 'target_defaults'.'defines', if necessary.
4. Go through configuration_public.h and adjust all the configuration values as appropriate for your platform.
5. Update starboard_platform.gyp to point at all the source files you want to build as part of your new Starboard implementation (as mentioned above).
6. Update atomic_public.h and thread_types_public.h as necessary to point at the appropriate shared or custom implementations.

You should now be able to run gyp with your new port. From your src/ directory:

$ cobalt/build/gyp_cobalt -C debug bobbox-armeb
$ ninja -C out/bobbox-armeb_debug nplb

This will attempt to build the “No Platform Left Behind” test suite with your new Starboard implementation, and you are ready to start porting!

GN Instructions

Follow the above list, except:

1. Ignore the steps about gyp_configuration.py and gyp_configuration.gypi.
2. Update BUILD.gn instead of starboard_platform.gyp.
3. Also in BUILD.gn:
   1. Update your toolchain command-line flags and libraries, for all configurations as well as for each individual configuration.
   2. Implement generic compiler configs such as sb_pedantic_warnings and rtti.
   3. If you're not using a predefined toolchain, define one.
4. In buildconfig.gni:
   1. If your platform is Linux-based, set target_os_ = "linux".
   2. Set target_cpu_ to your target architecture (e.g. arm, x64, x86).
   3. Set the target and host toolchains. If you defined a target toolchain in BUILD.gn, you'll want to set it to that.
5. In configuration.gni, set platform-specific defaults for any variables that should be overridden for your Starboard platform. You can find a list of such variables at //cobalt/build/config/base.gni and //starboard/build/config/base.gni.

You should now be able to run GN with your new port. From your src/ directory:

$ gn args out/bobbox-armeb_debug

An editor will open up. Type into the editor:

cobalt_config = "debug"
target_platform = "bobbox-armeb"

Save and close the editor. Then run

$ ninja -C out/bobbox-armeb_debug nplb

This will attempt to build the “No Platform Left Behind” test suite with your new Starboard implementation, and you are ready to start porting!

Suggested Implementation Order

When bringing up a new Starboard platform, it is suggested that you try to get the NPLB tests passing module-by-module. Because of dependencies between modules, you will find it easier to get some modules passing sooner than other modules.

Here's a recommended module implementation order in which to get things going (still significantly subject to change based on feedback):

1. Configuration
2. main(), Application, & Event Pump (i.e. the call into SbEventHandle)
3. Memory
4. Byte Swap
5. Time
6. String/Character/Double
7. Log
8. File
9. Directory
10. System
11. Atomic
12. Thread & Thread Types
13. Mutex
14. Condition Variable
15. Once
16. Socket
17. SocketWaiter
18. Window
19. Input
20. Blitter (if applicable)
21. Audio Sink
22. Media & Player
23. DRM
24. TimeZone
25. User
26. Storage