Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
1.. SPDX-License-Identifier: GPL-2.0
2
3=============
4SoC Subsystem
5=============
6
7Overview
8--------
9
10The SoC subsystem is a place of aggregation for SoC-specific code.
11The main components of the subsystem are:
12
13* devicetrees (DTS) for 32- & 64-bit ARM and RISC-V
14* 32-bit ARM board files (arch/arm/mach*)
15* 32- & 64-bit ARM defconfigs
16* SoC-specific drivers across architectures, in particular for 32- & 64-bit
17 ARM, RISC-V and Loongarch
18
19These "SoC-specific drivers" do not include clock, GPIO etc drivers that have
20other top-level maintainers. The drivers/soc/ directory is generally meant
21for kernel-internal drivers that are used by other drivers to provide SoC-
22specific functionality like identifying an SoC revision or interfacing with
23power domains.
24
25The SoC subsystem also serves as an intermediate location for changes to
26drivers/bus, drivers/firmware, drivers/reset and drivers/memory. The addition
27of new platforms, or the removal of existing ones, often go through the SoC
28tree as a dedicated branch covering multiple subsystems.
29
30The main SoC tree is housed on git.kernel.org:
31 https://git.kernel.org/pub/scm/linux/kernel/git/soc/soc.git/
32
33Maintainers
34-----------
35
36Clearly this is quite a wide range of topics, which no one person, or even
37small group of people are capable of maintaining. Instead, the SoC subsystem
38is comprised of many submaintainers (platform maintainers), each taking care of
39individual platforms and driver subdirectories.
40In this regard, "platform" usually refers to a series of SoCs from a given
41vendor, for example, Nvidia's series of Tegra SoCs. Many submaintainers operate
42on a vendor level, responsible for multiple product lines. For several reasons,
43including acquisitions/different business units in a company, things vary
44significantly here. The various submaintainers are documented in the
45MAINTAINERS file.
46
47Most of these submaintainers have their own trees where they stage patches,
48sending pull requests to the main SoC tree. These trees are usually, but not
49always, listed in MAINTAINERS.
50
51What the SoC tree is not, however, is a location for architecture-specific code
52changes. Each architecture has its own maintainers that are responsible for
53architectural details, CPU errata and the like.
54
55Submitting Patches for Given SoC
56~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
57
58All typical platform related patches should be sent via SoC submaintainers
59(platform-specific maintainers). This includes also changes to per-platform or
60shared defconfigs. Note that scripts/get_maintainer.pl might not provide
61correct addresses for the shared defconfig, so ignore its output and manually
62create CC-list based on MAINTAINERS file or use something like
63``scripts/get_maintainer.pl -f drivers/soc/FOO/``).
64
65Submitting Patches to the Main SoC Maintainers
66~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
67
68The main SoC maintainers can be reached via the alias soc@kernel.org only in
69following cases:
70
711. There are no platform-specific maintainers.
72
732. Platform-specific maintainers are unresponsive.
74
753. Introducing a completely new SoC platform. Such new SoC work should be sent
76 first to common mailing lists, pointed out by scripts/get_maintainer.pl, for
77 community review. After positive community review, work should be sent to
78 soc@kernel.org in one patchset containing new arch/foo/Kconfig entry, DTS
79 files, MAINTAINERS file entry and optionally initial drivers with their
80 Devicetree bindings. The MAINTAINERS file entry should list new
81 platform-specific maintainers, who are going to be responsible for handling
82 patches for the platform from now on.
83
84Note that the soc@kernel.org is usually not the place to discuss the patches,
85thus work sent to this address should be already considered as acceptable by
86the community.
87
88Information for (new) Submaintainers
89------------------------------------
90
91As new platforms spring up, they often bring with them new submaintainers,
92many of whom work for the silicon vendor, and may not be familiar with the
93process.
94
95Devicetree ABI Stability
96~~~~~~~~~~~~~~~~~~~~~~~~
97
98Perhaps one of the most important things to highlight is that dt-bindings
99document the ABI between the devicetree and the kernel.
100Please read Documentation/devicetree/bindings/ABI.rst.
101
102If changes are being made to a DTS that are incompatible with old
103kernels, the DTS patch should not be applied until the driver is, or an
104appropriate time later. Most importantly, any incompatible changes should be
105clearly pointed out in the patch description and pull request, along with the
106expected impact on existing users, such as bootloaders or other operating
107systems.
108
109Driver Branch Dependencies
110~~~~~~~~~~~~~~~~~~~~~~~~~~
111
112A common problem is synchronizing changes between device drivers and devicetree
113files. Even if a change is compatible in both directions, this may require
114coordinating how the changes get merged through different maintainer trees.
115
116Usually the branch that includes a driver change will also include the
117corresponding change to the devicetree binding description, to ensure they are
118in fact compatible. This means that the devicetree branch can end up causing
119warnings in the ``make dtbs_check`` step. If a devicetree change depends on
120missing additions to a header file in include/dt-bindings/, it will fail the
121``make dtbs`` step and not get merged.
122
123There are multiple ways to deal with this:
124
125* Avoid defining custom macros in include/dt-bindings/ for hardware constants
126 that can be derived from a datasheet -- binding macros in header files should
127 only be used as a last resort if there is no natural way to define a binding
128
129* Use literal values in the devicetree file in place of macros even when a
130 header is required, and change them to the named representation in a
131 following release
132
133* Defer the devicetree changes to a release after the binding and driver have
134 already been merged
135
136* Change the bindings in a shared immutable branch that is used as the base for
137 both the driver change and the devicetree changes
138
139* Add duplicate defines in the devicetree file guarded by an #ifndef section,
140 removing them in a later release
141
142Devicetree Naming Convention
143~~~~~~~~~~~~~~~~~~~~~~~~~~~~
144
145The general naming scheme for devicetree files is as follows. The aspects of a
146platform that are set at the SoC level, like CPU cores, are contained in a file
147named $soc.dtsi, for example, jh7100.dtsi. Integration details, that will vary
148from board to board, are described in $soc-$board.dts. An example of this is
149jh7100-beaglev-starlight.dts. Often many boards are variations on a theme, and
150frequently there are intermediate files, such as jh7100-common.dtsi, which sit
151between the $soc.dtsi and $soc-$board.dts files, containing the descriptions of
152common hardware.
153
154Some platforms also have System on Modules, containing an SoC, which are then
155integrated into several different boards. For these platforms, $soc-$som.dtsi
156and $soc-$som-$board.dts are typical.
157
158Directories are usually named after the vendor of the SoC at the time of its
159inclusion, leading to some historical directory names in the tree.
160
161Validating Devicetree Files
162~~~~~~~~~~~~~~~~~~~~~~~~~~~
163
164``make dtbs_check`` can be used to validate that devicetree files are compliant
165with the dt-bindings that describe the ABI. Please read the section
166"Running checks" of Documentation/devicetree/bindings/writing-schema.rst for
167more information on the validation of devicetrees.
168
169For new platforms, or additions to existing ones, ``make dtbs_check`` should not
170add any new warnings. For RISC-V and Samsung SoC, ``make dtbs_check W=1`` is
171required to not add any new warnings.
172
173Branches and Pull Requests
174~~~~~~~~~~~~~~~~~~~~~~~~~~
175
176Just as the main SoC tree has several branches, it is expected that
177submaintainers will do the same. Driver, defconfig and devicetree changes should
178all be split into separate branches and appear in separate pull requests to the
179SoC maintainers. Each branch should be usable by itself and avoid
180regressions that originate from dependencies on other branches.
181
182Small sets of patches can also be sent as separate emails to soc@kernel.org,
183grouped into the same categories.
184
185If changes do not fit into the normal patterns, there can be additional
186top-level branches, e.g. for a treewide rework, or the addition of new SoC
187platforms including dts files and drivers.
188
189Branches with a lot of changes can benefit from getting split up into separate
190topics branches, even if they end up getting merged into the same branch of the
191SoC tree. An example here would be one branch for devicetree warning fixes, one
192for a rework and one for newly added boards.
193
194Another common way to split up changes is to send an early pull request with the
195majority of the changes at some point between rc1 and rc4, following up with one
196or more smaller pull requests towards the end of the cycle that can add late
197changes or address problems identified while testing the first set.
198
199While there is no cut-off time for late pull requests, it helps to only send
200small branches as time gets closer to the merge window.
201
202Pull requests for bugfixes for the current release can be sent at any time, but
203again having multiple smaller branches is better than trying to combine too many
204patches into one pull request.
205
206The subject line of a pull request should begin with "[GIT PULL]" and made using
207a signed tag, rather than a branch. This tag should contain a short description
208summarising the changes in the pull request. For more detail on sending pull
209requests, please see Documentation/maintainer/pull-requests.rst.
210
211Defconfigs purpose
212~~~~~~~~~~~~~~~~~~
213
214Defconfigs are primarily used by the kernel developers, because distros have
215their own configs. A change adding new CONFIG options to a defconfig should
216explain why the kernel developers in general would want such option, e.g. by
217providing a name of an upstream-supported machine/board using that new option.
218This implies that enabling options in defconfig for non-upstream machines shall
219not be accepted.