[v2] rust: locks: Add `get_mut` method to `Lock`

Message ID 20240212-rust-locks-get-mut-v2-1-5ccd34c2b70b@gmail.com
State New
Headers
Series [v2] rust: locks: Add `get_mut` method to `Lock` |

Commit Message

Mathys-Gasnier via B4 Relay Feb. 12, 2024, 2:13 p.m. UTC
  From: Mathys-Gasnier <mathys35.gasnier@gmail.com>

Having a mutable reference guarantees that no other threads have
access to the lock, so we can take advantage of that to grant callers
access to the protected data without the the cost of acquiring and
releasing the locks. Since the lifetime of the data is tied to the
mutable reference, the borrow checker guarantees that the usage is safe.

Reviewed-by: Martin Rodriguez Reboredo <yakoyoku@gmail.com>
Signed-off-by: Mathys-Gasnier <mathys35.gasnier@gmail.com>
---
Changes in v2:
- Improved doc comment. 
- Link to v1: https://lore.kernel.org/r/20240209-rust-locks-get-mut-v1-1-ce351fc3de47@gmail.com
---
 rust/kernel/sync/lock.rs | 7 +++++++
 1 file changed, 7 insertions(+)


---
base-commit: 711cbfc717650532624ca9f56fbaf191bed56e67
change-id: 20240118-rust-locks-get-mut-c42072101d7a

Best regards,
  

Comments

Alice Ryhl Feb. 12, 2024, 2:22 p.m. UTC | #1
On Mon, Feb 12, 2024 at 3:13 PM Mathys-Gasnier via B4 Relay
<devnull+mathys35.gasnier.gmail.com@kernel.org> wrote:
> +    /// Gets the data contained in the lock
> +    /// Having a mutable reference to the lock guarantees that no other threads have access to the lock.
> +    /// Making it safe to get a mutable reference to the lock content.
> +    pub fn get_mut(&mut self) -> &mut T {
> +        self.data.get_mut()
> +    }

It's impossible to call this method. You can never have a mutable
reference to a Linux mutex because we pin our locks. At most, you can
have a Pin<&mut Self>.

Alice
  
Wedson Almeida Filho Feb. 15, 2024, 4:50 p.m. UTC | #2
On Mon, 12 Feb 2024 at 11:22, Alice Ryhl <aliceryhl@google.com> wrote:
>
> On Mon, Feb 12, 2024 at 3:13 PM Mathys-Gasnier via B4 Relay
> <devnull+mathys35.gasnier.gmail.com@kernel.org> wrote:
> > +    /// Gets the data contained in the lock
> > +    /// Having a mutable reference to the lock guarantees that no other threads have access to the lock.
> > +    /// Making it safe to get a mutable reference to the lock content.
> > +    pub fn get_mut(&mut self) -> &mut T {
> > +        self.data.get_mut()
> > +    }
>
> It's impossible to call this method. You can never have a mutable
> reference to a Linux mutex because we pin our locks. At most, you can
> have a Pin<&mut Self>.

Perhaps you meant to say that it's impossible to call this method
without unsafe blocks? From a `Pin<&mut T>`, we can call
`get_unchecked_mut` to get an `&mut T`.

This is addressing issue 924 opened by Björn some time back. The idea
here is that if there's a path where avoiding the lock/unlock calls
(which are expensive because of the memory barriers) is performance
critical, we can do it as long as we use an unsafe block.
  
Wedson Almeida Filho Feb. 15, 2024, 4:52 p.m. UTC | #3
On Mon, 12 Feb 2024 at 11:13, Mathys-Gasnier via B4 Relay
<devnull+mathys35.gasnier.gmail.com@kernel.org> wrote:
>
> From: Mathys-Gasnier <mathys35.gasnier@gmail.com>
>
> Having a mutable reference guarantees that no other threads have
> access to the lock, so we can take advantage of that to grant callers
> access to the protected data without the the cost of acquiring and
> releasing the locks. Since the lifetime of the data is tied to the
> mutable reference, the borrow checker guarantees that the usage is safe.
>
> Reviewed-by: Martin Rodriguez Reboredo <yakoyoku@gmail.com>
> Signed-off-by: Mathys-Gasnier <mathys35.gasnier@gmail.com>

Reviewed-by: Wedson Almeida Filho <walmeida@microsoft.com>
  
Benno Lossin Feb. 15, 2024, 8:55 p.m. UTC | #4
On 15.02.24 17:50, Wedson Almeida Filho wrote:
> On Mon, 12 Feb 2024 at 11:22, Alice Ryhl <aliceryhl@google.com> wrote:
>>
>> On Mon, Feb 12, 2024 at 3:13 PM Mathys-Gasnier via B4 Relay
>> <devnull+mathys35.gasnier.gmail.com@kernel.org> wrote:
>>> +    /// Gets the data contained in the lock
>>> +    /// Having a mutable reference to the lock guarantees that no other threads have access to the lock.
>>> +    /// Making it safe to get a mutable reference to the lock content.
>>> +    pub fn get_mut(&mut self) -> &mut T {
>>> +        self.data.get_mut()
>>> +    }
>>
>> It's impossible to call this method. You can never have a mutable
>> reference to a Linux mutex because we pin our locks. At most, you can
>> have a Pin<&mut Self>.
> 
> Perhaps you meant to say that it's impossible to call this method
> without unsafe blocks? From a `Pin<&mut T>`, we can call
> `get_unchecked_mut` to get an `&mut T`.

That is the wrong way to design this, since it forces an extra
`unsafe` call. Instead this function's receiver type should be
`self: Pin<&mut Self>`.
  

Patch

diff --git a/rust/kernel/sync/lock.rs b/rust/kernel/sync/lock.rs
index f12a684bc957..d15af6625d01 100644
--- a/rust/kernel/sync/lock.rs
+++ b/rust/kernel/sync/lock.rs
@@ -121,6 +121,13 @@  pub fn lock(&self) -> Guard<'_, T, B> {
         // SAFETY: The lock was just acquired.
         unsafe { Guard::new(self, state) }
     }
+
+    /// Gets the data contained in the lock
+    /// Having a mutable reference to the lock guarantees that no other threads have access to the lock.
+    /// Making it safe to get a mutable reference to the lock content.
+    pub fn get_mut(&mut self) -> &mut T {
+        self.data.get_mut()
+    }
 }
 
 /// A lock guard.