rust-for-linux.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Alice Ryhl <aliceryhl@google.com>
To: Miguel Ojeda <ojeda@kernel.org>,
	Andrew Morton <akpm@linux-foundation.org>
Cc: "Alex Gaynor" <alex.gaynor@gmail.com>,
	"Wedson Almeida Filho" <wedsonaf@gmail.com>,
	"Boqun Feng" <boqun.feng@gmail.com>,
	"Gary Guo" <gary@garyguo.net>,
	"Björn Roy Baron" <bjorn3_gh@protonmail.com>,
	"Benno Lossin" <benno.lossin@proton.me>,
	"Andreas Hindborg" <a.hindborg@samsung.com>,
	"Marco Elver" <elver@google.com>, "Coly Li" <colyli@suse.de>,
	"Paolo Abeni" <pabeni@redhat.com>,
	"Pierre Gondois" <pierre.gondois@arm.com>,
	"Ingo Molnar" <mingo@kernel.org>,
	"Jakub Kicinski" <kuba@kernel.org>,
	"Wei Yang" <richard.weiyang@gmail.com>,
	"Matthew Wilcox" <willy@infradead.org>,
	linux-kernel@vger.kernel.org, rust-for-linux@vger.kernel.org,
	"Alice Ryhl" <aliceryhl@google.com>,
	"Kees Cook" <kees@kernel.org>
Subject: [PATCH v4 09/10] rust: list: support heterogeneous lists
Date: Tue, 06 Aug 2024 13:59:01 +0000	[thread overview]
Message-ID: <20240806-linked-list-v4-9-23efc510ec92@google.com> (raw)
In-Reply-To: <20240806-linked-list-v4-0-23efc510ec92@google.com>

Support linked lists that can hold many different structs at once. This
is generally done using trait objects. The main challenge is figuring
what the struct is given only a pointer to the ListLinks.

We do this by storing a pointer to the struct next to the ListLinks
field. The container_of operation will then just read that pointer. When
the type is a trait object, that pointer will be a fat pointer whose
metadata is a vtable that tells you what kind of struct it is.

Heterogeneous lists are heavily used by Rust Binder. There are a lot of
so-called todo lists containing various events that need to be delivered
to userspace next time userspace calls into the driver. And there are
quite a few different todo item types: incoming transaction, changes to
refcounts, death notifications, and more.

Signed-off-by: Alice Ryhl <aliceryhl@google.com>
---
 rust/kernel/list.rs                    |  47 +++++++++++-
 rust/kernel/list/impl_list_item_mod.rs | 131 +++++++++++++++++++++++++++++++++
 2 files changed, 177 insertions(+), 1 deletion(-)

diff --git a/rust/kernel/list.rs b/rust/kernel/list.rs
index 904cfa454dff..cf26fec37c1e 100644
--- a/rust/kernel/list.rs
+++ b/rust/kernel/list.rs
@@ -12,7 +12,9 @@
 use core::ptr;
 
 mod impl_list_item_mod;
-pub use self::impl_list_item_mod::{impl_has_list_links, impl_list_item, HasListLinks};
+pub use self::impl_list_item_mod::{
+    impl_has_list_links, impl_has_list_links_self_ptr, impl_list_item, HasListLinks, HasSelfPtr,
+};
 
 mod arc;
 pub use self::arc::{impl_list_arc_safe, AtomicTracker, ListArc, ListArcSafe, TryNewListArc};
@@ -183,6 +185,49 @@ unsafe fn from_fields(me: *mut ListLinksFields) -> *mut Self {
     }
 }
 
+/// Similar to [`ListLinks`], but also contains a pointer to the full value.
+///
+/// This type can be used instead of [`ListLinks`] to support lists with trait objects.
+#[repr(C)]
+pub struct ListLinksSelfPtr<T: ?Sized, const ID: u64 = 0> {
+    /// The `ListLinks` field inside this value.
+    ///
+    /// This is public so that it can be used with `impl_has_list_links!`.
+    pub inner: ListLinks<ID>,
+    // UnsafeCell is not enough here because we need `Opaque::zeroed` as a dummy value, and
+    // `ptr::null()` doesn't work for `T: ?Sized`.
+    self_ptr: Opaque<*const T>,
+}
+
+// SAFETY: The fields of a ListLinksSelfPtr can be moved across thread boundaries.
+unsafe impl<T: ?Sized + Send, const ID: u64> Send for ListLinksSelfPtr<T, ID> {}
+// SAFETY: The type is opaque so immutable references to a ListLinksSelfPtr are useless. Therefore,
+// it's okay to have immutable access to a ListLinks from several threads at once.
+//
+// Note that `inner` being a public field does not prevent this type from being opaque, since
+// `inner` is a opaque type.
+unsafe impl<T: ?Sized + Sync, const ID: u64> Sync for ListLinksSelfPtr<T, ID> {}
+
+impl<T: ?Sized, const ID: u64> ListLinksSelfPtr<T, ID> {
+    /// The offset from the [`ListLinks`] to the self pointer field.
+    pub const LIST_LINKS_SELF_PTR_OFFSET: usize = core::mem::offset_of!(Self, self_ptr);
+
+    /// Creates a new initializer for this type.
+    pub fn new() -> impl PinInit<Self> {
+        // INVARIANT: Pin-init initializers can't be used on an existing `Arc`, so this value will
+        // not be constructed in an `Arc` that already has a `ListArc`.
+        Self {
+            inner: ListLinks {
+                inner: Opaque::new(ListLinksFields {
+                    prev: ptr::null_mut(),
+                    next: ptr::null_mut(),
+                }),
+            },
+            self_ptr: Opaque::uninit(),
+        }
+    }
+}
+
 impl<T: ?Sized + ListItem<ID>, const ID: u64> List<T, ID> {
     /// Creates a new empty list.
     pub const fn new() -> Self {
diff --git a/rust/kernel/list/impl_list_item_mod.rs b/rust/kernel/list/impl_list_item_mod.rs
index 30886e64b8ef..2dbbc648045f 100644
--- a/rust/kernel/list/impl_list_item_mod.rs
+++ b/rust/kernel/list/impl_list_item_mod.rs
@@ -68,6 +68,49 @@ unsafe fn raw_get_list_links(ptr: *mut Self) -> *mut $crate::list::ListLinks$(<$
 }
 pub use impl_has_list_links;
 
+/// Declares that the `ListLinks<ID>` field in this struct is inside a `ListLinksSelfPtr<T, ID>`.
+///
+/// # Safety
+///
+/// The `ListLinks<ID>` field of this struct at the offset `HasListLinks<ID>::OFFSET` must be
+/// inside a `ListLinksSelfPtr<T, ID>`.
+pub unsafe trait HasSelfPtr<T: ?Sized, const ID: u64 = 0>
+where
+    Self: HasListLinks<ID>,
+{
+}
+
+/// Implements the [`HasListLinks`] and [`HasSelfPtr`] traits for the given type.
+#[macro_export]
+macro_rules! impl_has_list_links_self_ptr {
+    ($(impl$({$($implarg:tt)*})?
+       HasSelfPtr<$item_type:ty $(, $id:tt)?>
+       for $self:ident $(<$($selfarg:ty),*>)?
+       { self.$field:ident }
+    )*) => {$(
+        // SAFETY: The implementation of `raw_get_list_links` only compiles if the field has the
+        // right type.
+        unsafe impl$(<$($implarg)*>)? $crate::list::HasSelfPtr<$item_type $(, $id)?> for
+            $self $(<$($selfarg),*>)?
+        {}
+
+        unsafe impl$(<$($implarg)*>)? $crate::list::HasListLinks$(<$id>)? for
+            $self $(<$($selfarg),*>)?
+        {
+            const OFFSET: usize = ::core::mem::offset_of!(Self, $field) as usize;
+
+            #[inline]
+            unsafe fn raw_get_list_links(ptr: *mut Self) -> *mut $crate::list::ListLinks$(<$id>)? {
+                // SAFETY: The caller promises that the pointer is not dangling.
+                let ptr: *mut $crate::list::ListLinksSelfPtr<$item_type $(, $id)?> =
+                    unsafe { ::core::ptr::addr_of_mut!((*ptr).$field) };
+                ptr.cast()
+            }
+        }
+    )*};
+}
+pub use impl_has_list_links_self_ptr;
+
 /// Implements the [`ListItem`] trait for the given type.
 ///
 /// Requires that the type implements [`HasListLinks`]. Use the [`impl_has_list_links!`] macro to
@@ -139,5 +182,93 @@ unsafe fn post_remove(me: *mut $crate::list::ListLinks<$num>) -> *const Self {
             }
         }
     )*};
+
+    (
+        $(impl$({$($generics:tt)*})? ListItem<$num:tt> for $t:ty {
+            using ListLinksSelfPtr;
+        })*
+    ) => {$(
+        // SAFETY: See GUARANTEES comment on each method.
+        unsafe impl$(<$($generics)*>)? $crate::list::ListItem<$num> for $t {
+            // GUARANTEES:
+            // This implementation of `ListItem` will not give out exclusive access to the same
+            // `ListLinks` several times because calls to `prepare_to_insert` and `post_remove`
+            // must alternate and exclusive access is given up when `post_remove` is called.
+            //
+            // Other invocations of `impl_list_item!` also cannot give out exclusive access to the
+            // same `ListLinks` because you can only implement `ListItem` once for each value of
+            // `ID`, and the `ListLinks` fields only work with the specified `ID`.
+            unsafe fn prepare_to_insert(me: *const Self) -> *mut $crate::list::ListLinks<$num> {
+                // SAFETY: The caller promises that `me` points at a valid value of type `Self`.
+                let links_field = unsafe { <Self as $crate::list::ListItem<$num>>::view_links(me) };
+
+                let spoff = $crate::list::ListLinksSelfPtr::<Self, $num>::LIST_LINKS_SELF_PTR_OFFSET;
+                // Goes via the offset as the field is private.
+                //
+                // SAFETY: The constant is equal to `offset_of!(ListLinksSelfPtr, self_ptr)`, so
+                // the pointer stays in bounds of the allocation.
+                let self_ptr = unsafe { (links_field as *const u8).add(spoff) }
+                    as *const $crate::types::Opaque<*const Self>;
+                let cell_inner = $crate::types::Opaque::raw_get(self_ptr);
+
+                // SAFETY: This value is not accessed in any other places than `prepare_to_insert`,
+                // `post_remove`, or `view_value`. By the safety requirements of those methods,
+                // none of these three methods may be called in parallel with this call to
+                // `prepare_to_insert`, so this write will not race with any other access to the
+                // value.
+                unsafe { ::core::ptr::write(cell_inner, me) };
+
+                links_field
+            }
+
+            // GUARANTEES:
+            // * This returns the same pointer as `prepare_to_insert` because `prepare_to_insert`
+            //   returns the return value of `view_links`.
+            // * By the type invariants of `ListLinks`, the `ListLinks` has two null pointers when
+            //   this value is not in a list.
+            unsafe fn view_links(me: *const Self) -> *mut $crate::list::ListLinks<$num> {
+                // SAFETY: The caller promises that `me` points at a valid value of type `Self`.
+                unsafe { <Self as HasListLinks<$num>>::raw_get_list_links(me.cast_mut()) }
+            }
+
+            // This function is also used as the implementation of `post_remove`, so the caller
+            // may choose to satisfy the safety requirements of `post_remove` instead of the safety
+            // requirements for `view_value`.
+            //
+            // GUARANTEES: (always)
+            // * This returns the same pointer as the one passed to the most recent call to
+            //   `prepare_to_insert` since that call wrote that pointer to this location. The value
+            //   is only modified in `prepare_to_insert`, so it has not been modified since the
+            //   most recent call.
+            //
+            // GUARANTEES: (only when using the `view_value` safety requirements)
+            // * The pointer remains valid until the next call to `post_remove` because the caller
+            //   of the most recent call to `prepare_to_insert` promised to retain ownership of the
+            //   `ListArc` containing `Self` until the next call to `post_remove`. The value cannot
+            //   be destroyed while a `ListArc` reference exists.
+            unsafe fn view_value(links_field: *mut $crate::list::ListLinks<$num>) -> *const Self {
+                let spoff = $crate::list::ListLinksSelfPtr::<Self, $num>::LIST_LINKS_SELF_PTR_OFFSET;
+                // SAFETY: The constant is equal to `offset_of!(ListLinksSelfPtr, self_ptr)`, so
+                // the pointer stays in bounds of the allocation.
+                let self_ptr = unsafe { (links_field as *const u8).add(spoff) }
+                    as *const ::core::cell::UnsafeCell<*const Self>;
+                let cell_inner = ::core::cell::UnsafeCell::raw_get(self_ptr);
+                // SAFETY: This is not a data race, because the only function that writes to this
+                // value is `prepare_to_insert`, but by the safety requirements the
+                // `prepare_to_insert` method may not be called in parallel with `view_value` or
+                // `post_remove`.
+                unsafe { ::core::ptr::read(cell_inner) }
+            }
+
+            // GUARANTEES:
+            // The first guarantee of `view_value` is exactly what `post_remove` guarantees.
+            unsafe fn post_remove(me: *mut $crate::list::ListLinks<$num>) -> *const Self {
+                // SAFETY: This specific implementation of `view_value` allows the caller to
+                // promise the safety requirements of `post_remove` instead of the safety
+                // requirements for `view_value`.
+                unsafe { <Self as $crate::list::ListItem<$num>>::view_value(me) }
+            }
+        }
+    )*};
 }
 pub use impl_list_item;

-- 
2.46.0.rc2.264.g509ed76dc8-goog


  parent reply	other threads:[~2024-08-06 13:59 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-08-06 13:58 [PATCH v4 00/10] Add Rust linked list for reference counted values Alice Ryhl
2024-08-06 13:58 ` [PATCH v4 01/10] rust: init: add `assert_pinned` macro Alice Ryhl
2024-08-06 13:58 ` [PATCH v4 02/10] rust: list: add ListArc Alice Ryhl
2024-08-06 13:58 ` [PATCH v4 03/10] rust: list: add tracking for ListArc Alice Ryhl
2024-08-06 14:30   ` Benno Lossin
2024-08-06 13:58 ` [PATCH v4 04/10] rust: list: add struct with prev/next pointers Alice Ryhl
2024-08-06 14:32   ` Benno Lossin
2024-08-06 13:58 ` [PATCH v4 05/10] rust: list: add macro for implementing ListItem Alice Ryhl
2024-08-06 14:47   ` Benno Lossin
2024-08-06 13:58 ` [PATCH v4 06/10] rust: list: add List Alice Ryhl
2024-08-06 14:44   ` Benno Lossin
2024-08-06 13:58 ` [PATCH v4 07/10] rust: list: add iterators Alice Ryhl
2024-08-06 13:59 ` [PATCH v4 08/10] rust: list: add cursor Alice Ryhl
2024-08-06 13:59 ` Alice Ryhl [this message]
2024-08-06 14:47   ` [PATCH v4 09/10] rust: list: support heterogeneous lists Benno Lossin
2024-08-06 13:59 ` [PATCH v4 10/10] rust: list: add ListArcField Alice Ryhl
2024-08-06 14:00 ` [PATCH v4 00/10] Add Rust linked list for reference counted values Alice Ryhl

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=20240806-linked-list-v4-9-23efc510ec92@google.com \
    --to=aliceryhl@google.com \
    --cc=a.hindborg@samsung.com \
    --cc=akpm@linux-foundation.org \
    --cc=alex.gaynor@gmail.com \
    --cc=benno.lossin@proton.me \
    --cc=bjorn3_gh@protonmail.com \
    --cc=boqun.feng@gmail.com \
    --cc=colyli@suse.de \
    --cc=elver@google.com \
    --cc=gary@garyguo.net \
    --cc=kees@kernel.org \
    --cc=kuba@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@kernel.org \
    --cc=ojeda@kernel.org \
    --cc=pabeni@redhat.com \
    --cc=pierre.gondois@arm.com \
    --cc=richard.weiyang@gmail.com \
    --cc=rust-for-linux@vger.kernel.org \
    --cc=wedsonaf@gmail.com \
    --cc=willy@infradead.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;
as well as URLs for NNTP newsgroup(s).