src/third_party/llvm-libc/src/docs/porting.rst
.. _porting:
If you are starting to bring up LLVM's libc on a new operating system, the first
step is to add a directory for that OS in the libc/config directory. Both
Linux <https://github.com/llvm/llvm-project/tree/main/libc/config/linux>_ and
Windows <https://github.com/llvm/llvm-project/tree/main/libc/config/windows>_,
the two operating systems on which LLVM's libc is being actively developed,
have their own config directory.
.. note:: Windows development is not as active as the development on Linux.
There is a
Darwin <https://github.com/llvm/llvm-project/tree/main/libc/config/darwin>_
config also which is in a similar state as Windows.
.. note:: LLVM's libc is being brought up on the
Fuchsia <https://fuchsia.dev/>_ operating system also. However, there is no
config directory for Fuchsia as the bring up is being done in the Fuchsia
source tree.
There are parts of the libc which are implemented differently for different
architectures. The simplest example of this is the syscall function and
its internal implementation - its Linux implementation differs for different
architectures. Since a large part of the libc makes use of syscalls (or an
equivalent on non-Linux like platforms), it might be simpler and convenient to
bring up the libc for one architecture at a time. In such cases, wherein the
support surface of LLVM's libc differs for each target architecture, one will
have to add a subdirectory (within the config directory of the operating
system) for each target architecture, and list the relevant config information
separately in those subdirectories. For example, for Linux, the x86_64 and
aarch64 configs are in separate directories, named
x86_64 <https://github.com/llvm/llvm-project/tree/main/libc/config/linux/x86_64>_
and aarch64 <https://github.com/llvm/llvm-project/tree/main/libc/config/linux/aarch64>_.
The libc CMake machinery looks for subdirectories named after the target
architecture.
The entrypoints.txt file lists the targets for the entrypoints to be
included in the build for a specific platform. For more technical details on
what entrypoints are and how they are registered as targets, see the
:ref:entrypoints documentation.
If you are doing an architecture specific bring-up, then an entrypoints.txt
file should be created in the architecture subdirectory for each architecture.
Else, having a single entrypoints.txt in the operating system directory is
sufficient.
A typical bring-up procedure will normally involve progressively adding targets to this file as they are implemented and tested.
Another important piece of config information is listed in a file named
headers.txt. It lists the targets for the set of public headers that are
provided by the libc. This is relevant only if the libc is to be used in the
:ref:full_host_build on the target operating system and architecture. As with
the entrypoints.txt file, one headers.txt file should be listed for
each individual target architecture if you are doing an architecture specific
bring up. The Linux config has headers.txt file listed separately for the
aarch64 <https://github.com/llvm/llvm-project/tree/main/libc/config/linux/aarch64>_
config and the
x86_64 <https://github.com/llvm/llvm-project/tree/main/libc/config/linux/x86_64>_
config.
Adding a target to the main LLVM-libc has some requirements to ensure that the targets stay in usable condition. LLVM-libc is under active development and without active maintenance targets will become stale and may be sunset.
To add a target there must be one or more people whose responsibility it is to keep the target up to date or push it forwards if it's not complete. Those people are the maintainers, and they are responsible for keeping their target in good shape. This means fixing their target when it breaks, reviewing patches related to their target, and keeping the target's CI running.
Maintainers are listed in libc/maintainers.rst and must follow
LLVM's maintainer policy <https://llvm.org/docs/DeveloperPolicy.html#maintainers>_.
Every target needs at least one CI builder. These are used to check when the
target breaks, and to help people who don't have access to the specific
architecture fix their bugs. LLVM-libc has both presubmit CI on github
and postsubmit CI on the LLVM buildbot <https://lab.llvm.org/buildbot>. For
instructions on contributing a postsubmit buildbot read
the LLVM documentation <https://llvm.org/docs/HowToAddABuilder.html> and for
presubmit tests read
the best practices <https://llvm.org/docs/CIBestPractices.html>_
The test configurations are at these links:
Linux Postsubmit <https://github.com/llvm/llvm-zorg/blob/main/zorg/buildbot/builders/annotated/libc-linux.py>_Windows Postsubmit <https://github.com/llvm/llvm-zorg/blob/main/zorg/buildbot/builders/annotated/libc-windows.py>_Fullbuild Presubmit <https://github.com/llvm/llvm-project/blob/main/.github/workflows/libc-fullbuild-tests.yml>_Overlay Presubmit <https://github.com/llvm/llvm-project/blob/main/.github/workflows/libc-overlay-tests.yml>_Sunsetting is the process through which targets can be removed from LLVM-libc. If a target is broken or stale it may be sunset. It is the responsibility of the target's maintainers to keep it from being sunset. The target's maintainers may be marked inactive if their target is sunset.
Broken targets are ones where the target's CI has been failing for at least 30 days. After 30 days have passed the CI should be changed so it does not block commits and does not notify people when it fails. If the target remains broken for another 90 days it may be sunset.
Stale targets are ones with no active maintainers or no contributions. If a target has 0 specific contributions between two major LLVM releases, or if it has no active maintainers, then it may be marked "deprecated" in the next major release. If there are still no contributions or no maintainers after the next major release it may be sunset.
To sunset a target, all specific references to that target in the code and build system should be removed. All buildbots for that target should be shut down, or otherwise removed.
To restart a target that was previously sunset, the new maintainers are encouraged to look at the commit(s) removing the target to provide a starting point.