2021-11-10 14:37:47 +03:00
---
title: Porting to New Architectures
category: Contributing
layout: default
SPDX-License-Identifier: LGPL-2.1-or-later
---
# Porting systemd to New Architectures
Here's a brief checklist of things to implement when porting systemd to a new
architecture.
1. Patch
[src/basic/architecture.h ](https://github.com/systemd/systemd/blob/main/src/basic/architecture.h )
and
[src/basic/architecture.c ](https://github.com/systemd/systemd/blob/main/src/basic/architecture.c )
to make your architecture known to systemd. Besides an `ARCHITECTURE_XYZ`
enumeration entry you need to provide an implementation of
`native_architecture()` and `uname_architecture()` .
2. Patch
[src/shared/gpt.h ](https://github.com/systemd/systemd/blob/main/src/shared/gpt.h )
and
[src/shared/gpt.c ](https://github.com/systemd/systemd/blob/main/src/shared/gpt.c )
and define a new set of GPT partition type UUIDs for the root file system,
`/usr/` file system, and the matching Verity and Verity signature
partitions. Use `systemd-id128 new -p` to generate new suitable UUIDs you
can use for this. Make sure to register your new types in the various
functions in `gpt.c` . Also make sure to update the tables in
2022-11-14 11:44:39 +03:00
[Discoverable Partitions Specification ](https://uapi-group.org/specifications/specs/discoverable_partitions_specification )
and `man/systemd-gpt-auto-generator.xml` accordingly.
2021-11-10 14:37:47 +03:00
3. If your architecture supports UEFI, make sure to update the `efi_arch`
variable logic in `meson.build` to be set to the right architecture string
as defined by the UEFI specification. (This ensures that `systemd-boot` will
be built as the appropriately named `BOOT<arch>.EFI` binary.) Also, if your
architecture uses a special boot protocol for the Linux kernel make sure to
implement it in `src/boot/efi/linux*.c` , so that the `systemd-stub` EFI stub
can work.
4. Make sure to register the right system call numbers for your architecture in
`src/basic/missing_syscall_def.h` . systemd uses various system calls the
Linux kernel provides that are currently not wrapped by glibc (or are only
in very new glibc), and we need to know the right numbers for them. It might
also be necessary to tweak `src/basic/raw-clone.h` .
5. Make sure the code in `src/shared/seccomp-util.c` properly understands the
local architecture and its system call quirks.
6. If your architecture uses a `/lib64/` library directory, then make sure that
the `BaseFilesystem` table in `src/shared/base-filesystem.c` has an entry
for it so that it can be set up automatically if missing. This is useful to
support booting into OS trees that have an empty root directory with only
`/usr/` mounted in.
random-util: remove RDRAND usage
/dev/urandom is seeded with RDRAND. Calling genuine_random_bytes(...,
..., 0) will use /dev/urandom as a last resort. Hence, we gain nothing
here by having our own RDRAND wrapper, because /dev/urandom already is
based on RDRAND output, even before /dev/urandom has fully initialized.
Furthermore, RDRAND is not actually fast! And on each successive
generation of new x86 CPUs, from both AMD and Intel, it just gets
slower.
This commit simplifies things by just using /dev/urandom in cases where
we before might use RDRAND, since /dev/urandom will always have RDRAND
mixed in as part of it.
And above where I say "/dev/urandom", what I actually mean is
GRND_INSECURE, which is the same thing but won't generate warnings in
dmesg.
2022-03-07 08:15:44 +03:00
7. If your architecture supports VM virtualization and provides CPU opcodes
2021-11-10 14:37:47 +03:00
similar to x86' CPUID consider adding native support for detecting VMs this
way to `src/basic/virt.c` .
