[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20250727130257.3549ea3c.gary@garyguo.net>
Date: Sun, 27 Jul 2025 13:02:57 +0100
From: Gary Guo <gary@...yguo.net>
To: Alice Ryhl <aliceryhl@...gle.com>
Cc: Andrew Morton <akpm@...ux-foundation.org>, "Liam R. Howlett"
<Liam.Howlett@...cle.com>, Lorenzo Stoakes <lorenzo.stoakes@...cle.com>,
Miguel Ojeda <ojeda@...nel.org>, Andrew Ballance
<andrewjballance@...il.com>, Boqun Feng <boqun.feng@...il.com>,
Björn Roy Baron <bjorn3_gh@...tonmail.com>, Benno Lossin
<lossin@...nel.org>, Andreas Hindborg <a.hindborg@...nel.org>, Trevor Gross
<tmgross@...ch.edu>, Danilo Krummrich <dakr@...nel.org>,
linux-kernel@...r.kernel.org, maple-tree@...ts.infradead.org,
rust-for-linux@...r.kernel.org, linux-mm@...ck.org
Subject: Re: [PATCH 2/3] rust: maple_tree: add MapleTree::lock() and load()
On Sat, 26 Jul 2025 18:18:02 +0200
Alice Ryhl <aliceryhl@...gle.com> wrote:
> On Sat, Jul 26, 2025 at 6:15 PM Alice Ryhl <aliceryhl@...gle.com> wrote:
> >
> > On Sat, Jul 26, 2025 at 5:50 PM Gary Guo <gary@...yguo.net> wrote:
> > >
> > > On Sat, 26 Jul 2025 13:23:23 +0000
> > > Alice Ryhl <aliceryhl@...gle.com> wrote:
> > >
> > > > To load a value, one must be careful to hold the lock while accessing
> > > > it. To enable this, we add a lock() method so that you can perform
> > > > operations on the value before the spinlock is released.
> > > >
> > > > Co-developed-by: Andrew Ballance <andrewjballance@...il.com>
> > > > Signed-off-by: Andrew Ballance <andrewjballance@...il.com>
> > > > Signed-off-by: Alice Ryhl <aliceryhl@...gle.com>
> > > > ---
> > > > rust/kernel/maple_tree.rs | 94 +++++++++++++++++++++++++++++++++++++++++++++++
> > > > 1 file changed, 94 insertions(+)
> > > >
> > > > diff --git a/rust/kernel/maple_tree.rs b/rust/kernel/maple_tree.rs
> > > > index 0f26c173eedc7c79bb8e2b56fe85e8a266b3ae0c..c7ef504a9c78065b3d5752b4f5337fb6277182d1 100644
> > > > --- a/rust/kernel/maple_tree.rs
> > > > +++ b/rust/kernel/maple_tree.rs
> > > > @@ -206,6 +206,23 @@ pub fn erase(&self, index: usize) -> Option<T> {
> > > > unsafe { T::try_from_foreign(ret) }
> > > > }
> > > >
> > > > + /// Lock the internal spinlock.
> > > > + #[inline]
> > > > + pub fn lock(&self) -> MapleLock<'_, T> {
> > > > + // SAFETY: It's safe to lock the spinlock in a maple tree.
> > > > + unsafe { bindings::spin_lock(self.ma_lock()) };
> > > > +
> > > > + // INVARIANT: We just took the spinlock.
> > > > + MapleLock(self)
> > > > + }
> > > > +
> > > > + #[inline]
> > > > + fn ma_lock(&self) -> *mut bindings::spinlock_t {
> > > > + // SAFETY: This pointer offset operation stays in-bounds.
> > > > + let lock = unsafe { &raw mut (*self.tree.get()).__bindgen_anon_1.ma_lock };
> > > > + lock.cast()
> > > > + }
> > >
> > > Could this return `&SpinLock<()>` using `Lock::from_raw`?
> > >
> > > I guess it has the drawback of having `MapleLock` needing to store
> > > `ma_lock` pointer but the guard is usually just all on stack and
> > > inlined so it probably won't make a difference?
> > >
> > > This way you remove `unsafe` from `lock` and gets a free `drop`.
> >
> > I ended up going this way to avoid the extra field in MapleLock, like
> > you mention.
>
> Oh, and it also avoids assuming anything about the layout of SpinLock<()>
>
> Alice
Well, `Lock::from_raw` is designed to interact with a C-side lock:
> Construct a Lock from a raw pointer
>
> This can be useful for interacting with a lock which was initialised outside of Rust.
- Gary
Powered by blists - more mailing lists