pacman is the package manager used by Arch Linux and its derivates.
Creating native packages from the kernel tree has multiple advantages:

* The package triggers the correct hooks for initramfs generation and
  bootloader configuration
* Uninstallation is complete and also invokes the relevant hooks
* New UAPI headers can be installed without any manual bookkeeping

The PKGBUILD file is a modified version of the one used for the
downstream Arch Linux "linux" package.
Extra steps that should not be necessary for a development kernel have
been removed and an UAPI header package has been added.

Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Tested-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

Move array_size.h, hashtable.h, list.h, list_types.h from scripts/kconfig/
to scripts/include/.

These headers will be useful for other host programs.

Remove scripts/mod/list.h.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

Kbuild provides scripts/Makefile.host to build host programs used for
building the kernel. Unfortunately, there are two exceptions that opt
out of Kbuild. The build system under tools/ is a cheesy replica, and
cause issues. I was recently poked about a problem in the tools build
system, which I do not maintain (and nobody maintains). [1]

Without a comment, people might believe this is the right location
because that is where objtool lives, even if a more robust Kbuild
syntax satisfies their needs. [2]

[1]: https://lore.kernel.org/linux-kbuild/ZnIYWBgrJ-IJtqK8@google.com/T/#m8ece130dd0e23c6f2395ed89070161948dee8457
[2]: https://lore.kernel.org/all/20240618200501.GA1611012@google.com/



Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Acked-by: Nicolas Schier <nicolas@fjasle.eu>
Reviewed-by: Brian Norris <briannorris@chromium.org>
Reviewed-by: Sami Tolvanen <samitolvanen@google.com>

RHEL/CentOS 7, popular distributions that install GNU Make 3.82, reached
EOM/EOL on June 30, 2024. While you may get extended support, it is a
good time to raise the minimum GNU Make version.

The new requirement, GNU Make 4.0, was released in October, 2013.

I did not touch the Makefiles under tools/ because I do not know the
requirements for building tools. I do not find any GNU Make version
checks under tools/.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>

There are 11 copies of arch/*/kernel/syscalls/Makefile that all implement
the same basic logic in a somewhat awkward way.

I tried out various ways of unifying the existing copies and ended up
with something that hooks into the logic for generating the redirections
to asm-generic headers. This gives a nicer syntax of being able to list
the generated files in $(syscall-y) inside of arch/*/include/asm/Kbuild
instead of both $(generated-y) in that place and also in another
Makefile.

The configuration for which syscall.tbl file to use and which ABIs to
enable is now done in arch/*/kernel/Makefile.syscalls. I have done
patches for all architectures and made sure that the new generic
rules implement a superset of all the architecture specific corner
cases.

ince the header file is not specific to asm-generic/*.h redirects
now, I ended up renaming the file to scripts/Makefile.asm-headers.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>

All Clippy lint groups that we enable, except `correctness`, have a
default `warn` level, thus they may be removed now that we relaxed all
lints to `warn`.

Moreover, Clippy provides an `all` lint group that covers the groups
we enable by default. Thus just use `all` instead -- the only change is
that, if Clippy introduces a new lint group or splits an existing one,
we will cover that one automatically.

In addition, `let_unit_value` is in `style` since Rust 1.62.0, thus it
does not need to be enabled manually.

Reviewed-by: Finn Behrens <me@kloenk.dev>
Tested-by: Benno Lossin <benno.lossin@proton.me>
Tested-by: Andreas Hindborg <a.hindborg@samsung.com>
Link: https://lore.kernel.org/r/20240709160615.998336-6-ojeda@kernel.org


Signed-off-by: Miguel Ojeda <ojeda@kernel.org>

Since we are starting to support several Rust toolchains, lints (including
Clippy ones) now may behave differently and lint groups may include
new lints.

Therefore, to maximize the chances a given version works, relax some
deny-level lints to warnings. It may also make our lives a bit easier
while developing new code or refactoring.

To be clear, the requirements for in-tree code are still the same, since
Rust code still needs to be warning-free (patches should be clean under
`WERROR=y`) and the set of lints is not changed.

`unsafe_op_in_unsafe_fn` is left unmodified, i.e. as an error, since it is
becoming the default in the language (warn-by-default in Rust 2024 [1] and
ideally an error later on) and thus it should also be very well tested. In
addition, it is simple enough that it should not have false positives
(unlike e.g. `rust_2018_idioms`'s `explicit_outlives_requirements`).

`non_ascii_idents` is left unmodified as well, i.e. as an error, since
it is unlikely one gains any productivity during development if it
were a warning (in fact, it may be worse, since it is likely one made
a typo). In addition, it should not have false positives.

Finally, put the two `-D` ones at the top and take the chance to do one
per line.

Link: https://github.com/rust-lang/rust/pull/112038

 [1]
Reviewed-by: Finn Behrens <me@kloenk.dev>
Tested-by: Benno Lossin <benno.lossin@proton.me>
Tested-by: Andreas Hindborg <a.hindborg@samsung.com>
Link: https://lore.kernel.org/r/20240709160615.998336-5-ojeda@kernel.org


Signed-off-by: Miguel Ojeda <ojeda@kernel.org>

Replace the cryptic phrase ("IDE support targets") that initially
appears to be about how to support old hard drives, with a few sentences
that explain what "make rust-analyzer" provides.

Signed-off-by: John Hubbard <jhubbard@nvidia.com>
Reviewed-by: Finn Behrens <me@kloenk.dev>
Reviewed-by: Alice Ryhl <aliceryhl@google.com>
Link: https://lore.kernel.org/r/20240628004356.1384486-3-jhubbard@nvidia.com


[ Reworded title. - Miguel ]
Signed-off-by: Miguel Ojeda <ojeda@kernel.org>

1) Provide a better error message for the "Rust not available" case.
Without this patch, one gets various misleading messages, such as:

    "No rule to make target 'rust-analyzer'"

Instead, run scripts/rust_is_available.sh directly, as a prerequisite,
and let that script report the cause of any problems, as well as
providing a link to the documentation. Thanks to Miguel Ojeda for the
idea of just letting rust_is_available.sh report its results directly.

The new output in the failure case looks like this:

$ make rust-analyzer
***
*** Rust compiler 'rustc' could not be found.
***
***
*** Please see Documentation/rust/quick-start.rst for details
*** on how to set up the Rust support.
***
make[1]: *** [/kernel_work/linux-github/Makefile:1975: rust-analyzer] Error 1
make: *** [Makefile:240: __sub-make] Error 2

Reviewed-by: Finn Behrens <me@kloenk.dev>
Reviewed-by: Alice Ryhl <aliceryhl@google.com>
Tested-by: Alice Ryhl <aliceryhl@google.com>
Signed-off-by: John Hubbard <jhubbard@nvidia.com>
Link: https://lore.kernel.org/r/20240628004356.1384486-2-jhubbard@nvidia.com


[ Reworded title. - Miguel ]
Signed-off-by: Miguel Ojeda <ojeda@kernel.org>

Since we dropped our custom `alloc` in commit 9d0441ba ("rust: alloc:
remove our fork of the `alloc` crate"), there is no need anymore to keep
the custom sysroot hack.

Thus delete it, which makes the target way simpler and faster too.

This also means we are not using Cargo for anything at the moment,
and that no download is required anymore, so update the main `Makefile`
and the documentation accordingly.

Link: https://lore.kernel.org/r/20240528163502.411600-1-ojeda@kernel.org


Signed-off-by: Miguel Ojeda <ojeda@kernel.org>

Several architectures provide an API to enable the FPU and run
floating-point SIMD code in kernel space.  However, the function names,
header locations, and semantics are inconsistent across architectures, and
FPU support may be gated behind other Kconfig options.

provide a standard way for architectures to declare that kernel space
FPU support is available. Architectures selecting this option must
implement what is currently the most common API (kernel_fpu_begin() and
kernel_fpu_end(), plus a new function kernel_fpu_available()) and
provide the appropriate CFLAGS for compiling floating-point C code.

Link: https://lkml.kernel.org/r/20240329072441.591471-2-samuel.holland@sifive.com


Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
Suggested-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Acked-by: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Borislav Petkov (AMD) <bp@alien8.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Huacai Chen <chenhuacai@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Masahiro Yamada <masahiroy@kernel.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nicolas Schier <nicolas@fjasle.eu>
Cc: Palmer Dabbelt <palmer@rivosinc.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: WANG Xuerui <git@xen0n.name>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

Currently, Kbuild produces inconsistent results in some cases.

You can do an interesting experiment using the --shuffle option, which
is supported by GNU Make 4.4 or later.

Set CONFIG_KVM_INTEL=y and CONFIG_KVM_AMD=m (or vice versa), and repeat
incremental builds w/wo --shuffle=reverse.

  $ make
    [ snip ]
    CC      arch/x86/kvm/kvm-asm-offsets.s

  $ make --shuffle=reverse
    [ snip ]
    CC [M]  arch/x86/kvm/kvm-asm-offsets.s

  $ make
    [ snip ]
    CC      arch/x86/kvm/kvm-asm-offsets.s

arch/x86/kvm/kvm-asm-offsets.s is rebuilt every time w/wo the [M] marker.

arch/x86/kvm/kvm-asm-offsets.s is built as built-in when it is built as
a prerequisite of arch/x86/kvm/kvm-intel.o, which is built-in.

arch/x86/kvm/kvm-asm-offsets.s is built as modular when it is built as
a prerequisite of arch/x86/kvm/kvm-amd.o, which is a module.

Another odd example is single target builds.

When CONFIG_LKDTM=m, drivers/misc/lkdtm/rodata.o can be built as
built-in or modular, depending on how it is built.

  $ make drivers/misc/lkdtm/lkdtm.o
    [ snip ]
    CC [M]  drivers/misc/lkdtm/rodata.o

  $ make drivers/misc/lkdtm/rodata.o
    [ snip ]
    CC      drivers/misc/lkdtm/rodata.o

drivers/misc/lkdtm/rodata.o is built as modular when it is built as a
prerequisite of another, but built as built-in when it is a final
target.

The same thing happens to drivers/memory/emif-asm-offsets.s when
CONFIG_TI_EMIF_SRAM=m.

  $ make drivers/memory/ti-emif-sram.o
    [ snip ]
    CC [M]  drivers/memory/emif-asm-offsets.s

  $ make drivers/memory/emif-asm-offsets.s
    [ snip ]
    CC      drivers/memory/emif-asm-offsets.s

This is because the part-of-module=y flag defined for the modules is
inherited by its prerequisites.

Target-specific variables are likely intended only for local use.
This commit adds 'private' to them.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <n.schier@avm.de>

The $(wildcard ) is called in quiet_cmd_rmfiles.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <n.schier@avm.de>

Kbuild conventionally uses $(obj)/ for generated files, and $(src)/ for
checked-in source files. It is merely a convention without any functional
difference. In fact, $(obj) and $(src) are exactly the same, as defined
in scripts/Makefile.build:

    src := $(obj)

When the kernel is built in a separate output directory, $(src) does
not accurately reflect the source directory location. While Kbuild
resolves this discrepancy by specifying VPATH=$(srctree) to search for
source files, it does not cover all cases. For example, when adding a
header search path for local headers, -I$(srctree)/$(src) is typically
passed to the compiler.

This introduces inconsistency between upstream and downstream Makefiles
because $(src) is used instead of $(srctree)/$(src) for the latter.

To address this inconsistency, this commit changes the semantics of
$(src) so that it always points to the directory in the source tree.

Going forward, the variables used in Makefiles will have the following
meanings:

  $(obj)     - directory in the object tree
  $(src)     - directory in the source tree  (changed by this commit)
  $(objtree) - the top of the kernel object tree
  $(srctree) - the top of the kernel source tree

Consequently, $(srctree)/$(src) in upstream Makefiles need to be replaced
with $(src).

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <nicolas@fjasle.eu>

There is an issue in clang's ThinLTO caching (enabled for the kernel via
'--thinlto-cache-dir') with .incbin, which the kernel occasionally uses
to include data within the kernel, such as the .config file for
/proc/config.gz. For example, when changing the .config and rebuilding
vmlinux, the copy of .config in vmlinux does not match the copy of
.config in the build folder:

  $ echo 'CONFIG_LTO_NONE=n
  CONFIG_LTO_CLANG_THIN=y
  CONFIG_IKCONFIG=y
  CONFIG_HEADERS_INSTALL=y' >kernel/configs/repro.config

  $ make -skj"$(nproc)" ARCH=x86_64 LLVM=1 clean defconfig repro.config vmlinux
  ...

  $ grep CONFIG_HEADERS_INSTALL .config
  CONFIG_HEADERS_INSTALL=y

  $ scripts/extract-ikconfig vmlinux | grep CONFIG_HEADERS_INSTALL
  CONFIG_HEADERS_INSTALL=y

  $ scripts/config -d HEADERS_INSTALL

  $ make -kj"$(nproc)" ARCH=x86_64 LLVM=1 vmlinux
  ...
    UPD     kernel/config_data
    GZIP    kernel/config_data.gz
    CC      kernel/configs.o
  ...
    LD      vmlinux
  ...

  $ grep CONFIG_HEADERS_INSTALL .config
  # CONFIG_HEADERS_INSTALL is not set

  $ scripts/extract-ikconfig vmlinux | grep CONFIG_HEADERS_INSTALL
  CONFIG_HEADERS_INSTALL=y

Without '--thinlto-cache-dir' or when using full LTO, this issue does
not occur.

Benchmarking incremental builds on a few different machines with and
without the cache shows a 20% increase in incremental build time without
the cache when measured by touching init/main.c and running 'make all'.

ARCH=arm64 defconfig + CONFIG_LTO_CLANG_THIN=y on an arm64 host:

  Benchmark 1: With ThinLTO cache
    Time (mean ± σ):     56.347 s ±  0.163 s    [User: 83.768 s, System: 24.661 s]
    Range (min … max):   56.109 s … 56.594 s    10 runs

  Benchmark 2: Without ThinLTO cache
    Time (mean ± σ):     67.740 s ±  0.479 s    [User: 718.458 s, System: 31.797 s]
    Range (min … max):   67.059 s … 68.556 s    10 runs

  Summary
    With ThinLTO cache ran
      1.20 ± 0.01 times faster than Without ThinLTO cache

ARCH=x86_64 defconfig + CONFIG_LTO_CLANG_THIN=y on an x86_64 host:

  Benchmark 1: With ThinLTO cache
    Time (mean ± σ):     85.772 s ±  0.252 s    [User: 91.505 s, System: 8.408 s]
    Range (min … max):   85.447 s … 86.244 s    10 runs

  Benchmark 2: Without ThinLTO cache
    Time (mean ± σ):     103.833 s ±  0.288 s    [User: 232.058 s, System: 8.569 s]
    Range (min … max):   103.286 s … 104.124 s    10 runs

  Summary
    With ThinLTO cache ran
      1.21 ± 0.00 times faster than Without ThinLTO cache

While it is unfortunate to take this performance improvement off the
table, correctness is more important. If/when this is fixed in LLVM, it
can potentially be brought back in a conditional manner. Alternatively,
a developer can just disable LTO if doing incremental compiles quickly
is important, as a full compile cycle can still take over a minute even
with the cache and it is unlikely that LTO will result in functional
differences for a kernel change.

Cc: stable@vger.kernel.org
Fixes: dc5723b0 ("kbuild: add support for Clang LTO")
Reported-by: Yifan Hong <elsk@google.com>
Closes: https://github.com/ClangBuiltLinux/linux/issues/2021


Reported-by: Masami Hiramatsu <mhiramat@kernel.org>
Closes: https://lore.kernel.org/r/20220327115526.cc4b0ff55fc53c97683c3e4d@kernel.org/


Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

Running dtbs_check and dt_compatible_check targets really only depend
on processed-schema.json, but the dependency is 'dt_binding_check'. That
was sort worked around with the CHECK_DT_BINDING variable in order to
skip some of the work that 'dt_binding_check' does. It still runs the
full checks of the schemas which is not necessary and adds 10s of
seconds to the build time. That's significant when checking only a few
DTBs and with recent changes that have improved the validation time by
6-7x.

Add a new target, dt_binding_schema, which just builds
processed-schema.json and can be used as the dependency for other
targets. The scripts_dtc dependency isn't needed either as the examples
aren't built for it.

Signed-off-by: Rob Herring <robh@kernel.org>
Tested-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

Commit 25b146c5 ("kbuild: allow Kbuild to start from any directory")
exported abs_srctree and abs_objtree to avoid recomputation after the
sub-make. However, this approach turned out to be fragile.

Commit 5fa94ceb ("kbuild: set correct abs_srctree and abs_objtree
for package builds") moved them above "ifneq ($(sub_make_done),1)",
eliminating the need for exporting them.

These are only needed in the top Makefile. If an absolute path is
required in sub-directories, you can use $(abspath ) or $(realpath )
as needed.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <nicolas@fjasle.eu>

Commit 3b9ab248 ("kbuild: use 4-space indentation when followed
by conditionals") introduced inconsistent indentation because it
deliberately touched only the conditional directives to minimize the
change set.

This commit reformats some blocks in the top Makefile so they are
consistently indented with 4 spaces.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>

`rustc` (like Cargo) may take advantage of the jobserver at any time
(e.g. for backend parallelism, or eventually frontend too). In the kernel,
we call `rustc` with `-Ccodegen-units=1` (and `-Zthreads` is 1 so far),
so we do not expect parallelism. However, in the upcoming Rust 1.76.0, a
warning is emitted by `rustc` [1] when it cannot connect to the jobserver
it was passed (in many cases, but not all: compiling and `--print sysroot`
do, but `--version` does not). And given GNU Make always passes
the jobserver in the environment variable (even when a line is deemed
non-recursive), `rustc` will end up complaining about it (in particular
in Make 4.3 where there is only the simple pipe jobserver style).

One solution is to remove the jobserver from `MAKEFLAGS`. However, we
can mark the lines with calls to `rustc` (and Cargo) as recursive, which
looks simpler. This is being documented as a recommendation in `rustc`
[2] and allows us to be ready for the time we may use parallelism inside
`rustc` (potentially now, if a user passes `-Zthreads`). Thus do so.

Similarly, do the same for `rustdoc` and `cargo` calls.

Finally, there is one case that the solution does not cover, which is the
`$(shell ...)` call we have. Thus, for that one, set an empty `MAKEFLAGS`
environment variable.

Link: https://github.com/rust-lang/rust/issues/120515

 [1]
Acked-by: Masahiro Yamada <masahiroy@kernel.org>
Link: https://github.com/rust-lang/rust/pull/121564 [2]
Link: https://lore.kernel.org/r/20240217002638.57373-1-ojeda@kernel.org


[ Reworded to add link to PR documenting the recommendation. ]
Signed-off-by: Miguel Ojeda <ojeda@kernel.org>