public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Gary Guo <gary@kernel.org>
To: "Miguel Ojeda" <ojeda@kernel.org>,
	"Boqun Feng" <boqun@kernel.org>, "Gary Guo" <gary@garyguo.net>,
	"Björn Roy Baron" <bjorn3_gh@protonmail.com>,
	"Benno Lossin" <lossin@kernel.org>,
	"Andreas Hindborg" <a.hindborg@kernel.org>,
	"Alice Ryhl" <aliceryhl@google.com>,
	"Trevor Gross" <tmgross@umich.edu>,
	"Danilo Krummrich" <dakr@kernel.org>
Cc: Yury Norov <ynorov@nvidia.com>,
	rust-for-linux@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: [PATCH v3 3/4] rust: rework `build_assert!` documentation
Date: Thu, 19 Mar 2026 12:16:47 +0000	[thread overview]
Message-ID: <20260319121653.2975748-4-gary@kernel.org> (raw)
In-Reply-To: <20260319121653.2975748-1-gary@kernel.org>

From: Gary Guo <gary@garyguo.net>

Add a detailed comparison and recommendation of the three types of
build-time assertion macro as module documentation (and un-hide the module
to render them).

The documentation on the macro themselves are simplified to only cover the
scenarios where they should be used; links to the module documentation is
added instead.

Reviewed-by: Yury Norov <ynorov@nvidia.com>
Signed-off-by: Gary Guo <gary@garyguo.net>
---
 rust/kernel/build_assert.rs | 119 ++++++++++++++++++++++++++++--------
 rust/kernel/lib.rs          |   1 -
 2 files changed, 92 insertions(+), 28 deletions(-)

diff --git a/rust/kernel/build_assert.rs b/rust/kernel/build_assert.rs
index 50b0fc0a80fc..726d0b76ca2b 100644
--- a/rust/kernel/build_assert.rs
+++ b/rust/kernel/build_assert.rs
@@ -1,6 +1,72 @@
 // SPDX-License-Identifier: GPL-2.0
 
 //! Various assertions that happen during build-time.
+//!
+//! There are three types of build-time assertions that you can use:
+//! - [`static_assert!`]
+//! - [`const_assert!`]
+//! - [`build_assert!`]
+//!
+//! The ones towards the bottom of the list are more expressive, while the ones towards the top of
+//! the list are more robust and trigger earlier in the compilation pipeline. Therefore, you should
+//! prefer the ones towards the top of the list wherever possible.
+//!
+//! # Choosing the correct assertion
+//!
+//! If you're asserting outside any bodies (e.g. initializers or function bodies), you should use
+//! [`static_assert!`] as it is the only assertion that can be used in that context.
+//!
+//! Inside bodies, if your assertion condition does not depend on any variable or generics, you
+//! should use [`static_assert!`]. If the condition depends on generics, but not variables
+//! (including function arguments), you should use [`const_assert!`]. Otherwise, use
+//! [`build_assert!`]. The same is true regardless if the function is `const fn`.
+//!
+//! ```
+//! // Outside any bodies
+//! static_assert!(core::mem::size_of::<u8>() == 1);
+//! // `const_assert!` and `build_assert!` cannot be used here, they will fail to compile.
+//!
+//! #[inline(always)]
+//! fn foo<const N: usize>(v: usize) {
+//!     static_assert!(core::mem::size_of::<u8>() == 1); // Preferred
+//!     const_assert!(core::mem::size_of::<u8>() == 1); // Discouraged
+//!     build_assert!(core::mem::size_of::<u8>() == 1); // Discouraged
+//!
+//!     // `static_assert!(N > 1);` is not allowed
+//!     const_assert!(N > 1); // Preferred
+//!     build_assert!(N > 1); // Discouraged
+//!
+//!     // `static_assert!(v > 1);` is not allowed
+//!     // `const_assert!(v > 1);` is not allowed
+//!     build_assert!(v > 1); // Works
+//! }
+//! ```
+//!
+//! # Detailed behavior
+//!
+//! `static_assert!()` is equivalent to `static_assert` in C. It requires `expr` to be a constant
+//! expression. This expression cannot refer to any generics. A `static_assert!(expr)` in a program
+//! is always evaluated, regardless if the function it appears in is used or not. This is also the
+//! only usable assertion outside a body.
+//!
+//! `const_assert!()` has no direct C equivalence. It is a more powerful version of
+//! `static_assert!()`, where it may refer to generics in a function. Note that due to the ability
+//! to refer to generics, the assertion is tied to a specific instance of a function. So if it is
+//! used in a generic function that is not instantiated, the assertion will not be checked. For this
+//! reason, `static_assert!()` is preferred wherever possible.
+//!
+//! `build_assert!()` is equivalent to `BUILD_BUG_ON`. It is even more powerful than
+//! `const_assert!()` because it can be used to check tautologies that depend on runtime value (this
+//! is the same as `BUILD_BUG_ON`). However, the assertion failure mechanism can possibly be
+//! undefined symbols and linker errors, it is not developer friendly to debug, so it is recommended
+//! to avoid it and prefer other two assertions where possible.
+
+pub use crate::{
+    build_assert,
+    build_error,
+    const_assert,
+    static_assert, //
+};
 
 #[doc(hidden)]
 pub use build_error::build_error;
@@ -15,6 +81,10 @@
 ///
 /// The feature may be added to Rust in the future: see [RFC 2790].
 ///
+/// You cannot refer to generics or variables with [`static_assert!`]. If you need to refer to
+/// generics, use [`const_assert!`]; if you need to refer to variables, use [`build_assert!`]. See
+/// the [module documentation](self).
+///
 /// [`_Static_assert`]: https://en.cppreference.com/w/c/language/_Static_assert
 /// [`static_assert`]: https://en.cppreference.com/w/cpp/language/static_assert
 /// [RFC 2790]: https://github.com/rust-lang/rfcs/issues/2790
@@ -47,6 +117,10 @@ macro_rules! static_assert {
 /// functions or implementation blocks. However, it also has a limitation where it can only appear
 /// in places where statements can appear; for example, you cannot use it as an item in the module.
 ///
+/// [`static_assert!`] should be preferred if no generics are referred to in the condition. You
+/// cannot refer to variables with [`const_assert!`] (even inside `const fn`); if you need the
+/// capability, use [`build_assert!`]. See the [module documentation](self).
+///
 /// # Examples
 ///
 /// ```
@@ -98,41 +172,32 @@ macro_rules! build_error {
 /// will panic. If the compiler or optimizer cannot guarantee the condition will
 /// be evaluated to `true`, a build error will be triggered.
 ///
-/// [`static_assert!`] should be preferred to `build_assert!` whenever possible.
+/// When a condition depends on a function argument, the function must be annotated with
+/// `#[inline(always)]`. Without this attribute, the compiler may choose to not inline the
+/// function, preventing it from optimizing out the error path.
+///
+/// If the assertion condition does not depend on any variables or generics, you should use
+/// [`static_assert!`]. If the assertion condition does not depend on variables, but does depend on
+/// generics, you should use [`const_assert!`]. See the [module documentation](self).
 ///
 /// # Examples
 ///
-/// These examples show that different types of [`assert!`] will trigger errors
-/// at different stage of compilation. It is preferred to err as early as
-/// possible, so [`static_assert!`] should be used whenever possible.
-/// ```ignore
-/// fn foo() {
-///     static_assert!(1 > 1); // Compile-time error
-///     build_assert!(1 > 1); // Build-time error
-///     assert!(1 > 1); // Run-time error
-/// }
 /// ```
+/// #[inline(always)] // Important
+/// fn bar(n: usize) {
+///     build_assert!(n > 1);
+/// }
 ///
-/// When the condition refers to generic parameters or parameters of an inline function,
-/// [`static_assert!`] cannot be used. Use `build_assert!` in this scenario.
-/// ```
-/// fn foo<const N: usize>() {
-///     // `static_assert!(N > 1);` is not allowed
-///     build_assert!(N > 1); // Build-time check
-///     assert!(N > 1); // Run-time check
+/// fn foo() {
+///     bar(2);
 /// }
-/// ```
 ///
-/// When a condition depends on a function argument, the function must be annotated with
-/// `#[inline(always)]`. Without this attribute, the compiler may choose to not inline the
-/// function, preventing it from optimizing out the error path.
-/// ```
-/// #[inline(always)]
-/// fn bar(n: usize) {
-///     // `static_assert!(n > 1);` is not allowed
-///     build_assert!(n > 1); // Build-time check
-///     assert!(n > 1); // Run-time check
+/// #[inline(always)] // Important
+/// const fn const_bar(n: usize) {
+///     build_assert!(n > 1);
 /// }
+///
+/// const _: () = const_bar(2);
 /// ```
 #[macro_export]
 macro_rules! build_assert {
diff --git a/rust/kernel/lib.rs b/rust/kernel/lib.rs
index f427cd3c8cce..d772713459a3 100644
--- a/rust/kernel/lib.rs
+++ b/rust/kernel/lib.rs
@@ -80,7 +80,6 @@
 #[cfg(CONFIG_BLOCK)]
 pub mod block;
 pub mod bug;
-#[doc(hidden)]
 pub mod build_assert;
 pub mod clk;
 #[cfg(CONFIG_CONFIGFS_FS)]
-- 
2.51.2


  parent reply	other threads:[~2026-03-19 12:17 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <20260319121653.2975748-1-gary@kernel.org>
2026-03-19 12:16 ` [PATCH v3 1/4] rust: move `static_assert` into `build_assert` Gary Guo
2026-03-19 14:09   ` Alice Ryhl
2026-03-19 12:16 ` [PATCH v3 2/4] rust: add `const_assert!` macro Gary Guo
2026-03-19 14:12   ` Alice Ryhl
2026-03-19 14:26     ` Gary Guo
2026-03-19 14:34       ` Alice Ryhl
2026-03-21 13:05   ` Alexandre Courbot
2026-03-19 12:16 ` Gary Guo [this message]
2026-03-19 14:17   ` [PATCH v3 3/4] rust: rework `build_assert!` documentation Alice Ryhl
2026-03-21 13:05   ` Alexandre Courbot
2026-03-19 12:16 ` [PATCH v3 4/4] rust: make `build_assert` module the home of related macros Gary Guo
2026-03-19 14:14   ` Danilo Krummrich
2026-03-19 14:33   ` Alice Ryhl
2026-03-21 13:05   ` Alexandre Courbot
2026-03-21 13:32     ` Gary Guo
2026-03-21 13:41       ` Miguel Ojeda
2026-03-22 23:36   ` Miguel Ojeda
2026-03-23  1:08     ` Alexandre Courbot
2026-03-23  1:38       ` Miguel Ojeda
2026-03-23  1:50         ` Alexandre Courbot
2026-03-26 12:09     ` FUJITA Tomonori
2026-03-23  0:15   ` Tamir Duberstein

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260319121653.2975748-4-gary@kernel.org \
    --to=gary@kernel.org \
    --cc=a.hindborg@kernel.org \
    --cc=aliceryhl@google.com \
    --cc=bjorn3_gh@protonmail.com \
    --cc=boqun@kernel.org \
    --cc=dakr@kernel.org \
    --cc=gary@garyguo.net \
    --cc=linux-kernel@vger.kernel.org \
    --cc=lossin@kernel.org \
    --cc=ojeda@kernel.org \
    --cc=rust-for-linux@vger.kernel.org \
    --cc=tmgross@umich.edu \
    --cc=ynorov@nvidia.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox