NVIDIA GPU driver infrastructure
 help / color / mirror / Atom feed
From: Gary Guo <gary@garyguo.net>
To: "Miguel Ojeda" <ojeda@kernel.org>,
	"Boqun Feng" <boqun@kernel.org>,
	"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>,
	"Will Deacon" <will@kernel.org>,
	"Peter Zijlstra" <peterz@infradead.org>,
	"Mark Rutland" <mark.rutland@arm.com>,
	"Alexandre Courbot" <acourbot@nvidia.com>,
	"David Airlie" <airlied@gmail.com>,
	"Simona Vetter" <simona@ffwll.ch>
Cc: Gary Guo <gary@garyguo.net>,
	Alan Stern <stern@rowland.harvard.edu>,
	 Andrea Parri <parri.andrea@gmail.com>,
	Nicholas Piggin <npiggin@gmail.com>,
	 David Howells <dhowells@redhat.com>,
	Jade Alglave <j.alglave@ucl.ac.uk>,
	 Luc Maranget <luc.maranget@inria.fr>,
	 "Paul E. McKenney" <paulmck@kernel.org>,
	Akira Yokosawa <akiyks@gmail.com>,
	 Daniel Lustig <dlustig@nvidia.com>,
	Joel Fernandes <joelagnelf@nvidia.com>,
	 rust-for-linux@vger.kernel.org, linux-kernel@vger.kernel.org,
	 nova-gpu@lists.linux.dev, dri-devel@lists.freedesktop.org,
	 lkmm@lists.linux.dev
Subject: [PATCH v2 2/4] rust: sync: generic memory barriers
Date: Tue, 09 Jun 2026 16:38:38 +0100	[thread overview]
Message-ID: <20260609-rust-barrier-v2-2-30fcc48e1cd0@garyguo.net> (raw)
In-Reply-To: <20260609-rust-barrier-v2-0-30fcc48e1cd0@garyguo.net>

Implement a generic interface for memory barriers (full system/DMA/SMP).
The interface uses a parameter to force user to specify their intent with
barriers.

Provide `Read`, `Write`, `Full` orderings which map to the existing
`rmb()`, `wmb()` and `mb()`. Generic is used here instead of providing
individual standalone functions to reduce code duplication; for example,
the `CONFIG_SMP` check in `smp_mb` is uniformly implemented for all SMP
barriers. This could extend to `virt_mb`'s if they're introduced in the
future. It would also make it easier if new ordering types are introduced
in the future (e.g. `Acquire`, `Release`).

Signed-off-by: Gary Guo <gary@garyguo.net>
---
 rust/kernel/sync/atomic/ordering.rs |   2 +-
 rust/kernel/sync/barrier.rs         | 123 ++++++++++++++++++++++++++++--------
 2 files changed, 96 insertions(+), 29 deletions(-)

diff --git a/rust/kernel/sync/atomic/ordering.rs b/rust/kernel/sync/atomic/ordering.rs
index 3f103aa8db99..c4e732e7212f 100644
--- a/rust/kernel/sync/atomic/ordering.rs
+++ b/rust/kernel/sync/atomic/ordering.rs
@@ -15,7 +15,7 @@
 //!   - It provides ordering between the annotated operation and all the following memory accesses.
 //!   - It provides ordering between all the preceding memory accesses and all the following memory
 //!     accesses.
-//!   - All the orderings are the same strength as a full memory barrier (i.e. `smp_mb()`).
+//!   - All the orderings are the same strength as a full memory barrier (i.e. `smp_mb(Full)`).
 //! - [`Relaxed`] provides no ordering except the dependency orderings. Dependency orderings are
 //!   described in "DEPENDENCY RELATIONS" in [`LKMM`]'s [`explanation`].
 //!
diff --git a/rust/kernel/sync/barrier.rs b/rust/kernel/sync/barrier.rs
index 8f2d435fcd94..54c527fdb760 100644
--- a/rust/kernel/sync/barrier.rs
+++ b/rust/kernel/sync/barrier.rs
@@ -7,6 +7,34 @@
 //!
 //! [`LKMM`]: srctree/tools/memory-model/
 
+#![expect(private_bounds, reason = "sealed implementation")]
+
+/// Memory barrier orderings.
+///
+/// The semantics of these orderings follows the [`LKMM`] definitions and rules.
+///
+/// - [`Read`] provides ordering between preceding load operations and succeeding load operations.
+/// - [`Write`] provides ordering between preceding store operations and succeeding store
+///   operations.
+/// - [`Full`] provides ordering between all the preceding memory accesses and succeeding memory
+///   accesses.
+///
+/// [`LKMM`]: srctree/tools/memory-model/
+pub mod ordering {
+    pub use crate::sync::atomic::ordering::Full;
+
+    /// The annotation type for read-read barrier ordering.
+    pub struct Read;
+
+    /// The annotation type for write-write barrier ordering.
+    pub struct Write;
+}
+
+pub use ordering::{Full, Read, Write};
+
+struct Smp;
+struct Dma;
+
 /// A compiler barrier.
 ///
 /// A barrier that prevents compiler from reordering memory accesses across the barrier.
@@ -19,43 +47,82 @@ pub(crate) fn barrier() {
     unsafe { core::arch::asm!("") };
 }
 
-/// A full memory barrier.
+trait MemoryBarrier<Flavour = ()> {
+    fn run();
+}
+
+macro_rules! define_barrier {
+    ($([$flavour:ident])? $ordering:ident, $binding:ident) => {
+        impl MemoryBarrier$(<$flavour>)? for $ordering {
+            #[inline]
+            fn run() {
+                // SAFETY: barrier methods are safe to call.
+                unsafe { bindings::$binding() };
+            }
+        }
+    };
+}
+
+define_barrier!(Full, mb);
+define_barrier!(Read, rmb);
+define_barrier!(Write, wmb);
+define_barrier!([Dma] Full, dma_mb);
+define_barrier!([Dma] Read, dma_rmb);
+define_barrier!([Dma] Write, dma_wmb);
+define_barrier!([Smp] Full, smp_mb);
+define_barrier!([Smp] Read, smp_rmb);
+define_barrier!([Smp] Write, smp_wmb);
+
+/// Memory barrier.
 ///
 /// A barrier that prevents compiler and CPU from reordering memory accesses across the barrier.
-#[inline(always)]
-pub fn smp_mb() {
-    if cfg!(CONFIG_SMP) {
-        // SAFETY: `smp_mb()` is safe to call.
-        unsafe { bindings::smp_mb() };
-    } else {
-        barrier();
-    }
+///
+/// The specific forms of reordering can be specified using the parameter.
+/// - `mb(Read)` provides a read-read barrier.
+/// - `mb(Write)` provides a write-write barrier.
+/// - `mb(Full)` provides a full barrier.
+///
+/// # Examples
+///
+/// ```
+/// # use kernel::sync::barrier::*;
+/// mb(Read);
+/// mb(Write);
+/// mb(Full);
+/// ```
+#[inline]
+#[doc(alias = "rmb")]
+#[doc(alias = "wmb")]
+pub fn mb<T: MemoryBarrier>(_: T) {
+    T::run()
 }
 
-/// A write-write memory barrier.
+/// Memory barrier between CPUs.
 ///
-/// A barrier that prevents compiler and CPU from reordering memory write accesses across the
-/// barrier.
-#[inline(always)]
-pub fn smp_wmb() {
+/// A barrier that prevents compiler and CPU from reordering memory accesses across the barrier.
+/// Does not prevent re-ordering with respect to other bus-mastering devices.
+///
+/// See [`mb`] for usage.
+#[inline]
+#[doc(alias = "smp_rmb")]
+#[doc(alias = "smp_wmb")]
+pub fn smp_mb<T: MemoryBarrier<Smp>>(_: T) {
     if cfg!(CONFIG_SMP) {
-        // SAFETY: `smp_wmb()` is safe to call.
-        unsafe { bindings::smp_wmb() };
+        T::run()
     } else {
-        barrier();
+        barrier()
     }
 }
 
-/// A read-read memory barrier.
+/// Memory barrier between local CPU and bus-mastering devices.
 ///
-/// A barrier that prevents compiler and CPU from reordering memory read accesses across the
-/// barrier.
-#[inline(always)]
-pub fn smp_rmb() {
-    if cfg!(CONFIG_SMP) {
-        // SAFETY: `smp_rmb()` is safe to call.
-        unsafe { bindings::smp_rmb() };
-    } else {
-        barrier();
-    }
+/// A barrier that prevents compiler and CPU from reordering memory accesses across the barrier.
+/// Does not prevent re-ordering with respect to other CPUs.
+///
+/// See [`mb`] for usage.
+#[inline]
+#[doc(alias = "dma_rmb")]
+#[doc(alias = "dma_wmb")]
+pub fn dma_mb<T: MemoryBarrier<Dma>>(_: T) {
+    T::run()
 }

-- 
2.54.0


  parent reply	other threads:[~2026-06-09 15:39 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-06-09 15:38 [PATCH v2 0/4] rust: more memory barriers bindings Gary Guo
2026-06-09 15:38 ` [PATCH v2 1/4] rust: sync: add helpers for mb, dma_mb and friends Gary Guo
2026-07-02  4:35   ` Boqun Feng
2026-06-09 15:38 ` Gary Guo [this message]
2026-07-02  4:56   ` [PATCH v2 2/4] rust: sync: generic memory barriers Boqun Feng
2026-06-09 15:38 ` [PATCH v2 3/4] gpu: nova-core: fix barrier usage in CPU->GSP messaging path Gary Guo
2026-06-09 15:38 ` [PATCH v2 4/4] gpu: nova-core: fix barrier usage in GSP->CPU " Gary Guo

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=20260609-rust-barrier-v2-2-30fcc48e1cd0@garyguo.net \
    --to=gary@garyguo.net \
    --cc=a.hindborg@kernel.org \
    --cc=acourbot@nvidia.com \
    --cc=airlied@gmail.com \
    --cc=akiyks@gmail.com \
    --cc=aliceryhl@google.com \
    --cc=bjorn3_gh@protonmail.com \
    --cc=boqun@kernel.org \
    --cc=dakr@kernel.org \
    --cc=dhowells@redhat.com \
    --cc=dlustig@nvidia.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=j.alglave@ucl.ac.uk \
    --cc=joelagnelf@nvidia.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=lkmm@lists.linux.dev \
    --cc=lossin@kernel.org \
    --cc=luc.maranget@inria.fr \
    --cc=mark.rutland@arm.com \
    --cc=nova-gpu@lists.linux.dev \
    --cc=npiggin@gmail.com \
    --cc=ojeda@kernel.org \
    --cc=parri.andrea@gmail.com \
    --cc=paulmck@kernel.org \
    --cc=peterz@infradead.org \
    --cc=rust-for-linux@vger.kernel.org \
    --cc=simona@ffwll.ch \
    --cc=stern@rowland.harvard.edu \
    --cc=tmgross@umich.edu \
    --cc=will@kernel.org \
    /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