On Sun Jul 27, 2025 at 9:39 PM JST, Benno Lossin wrote:
On Sun Jul 27, 2025 at 12:23 PM CEST, Alice Ryhl wrote:
On Sun, Jul 27, 2025 at 08:52:00AM +0200, Alice Ryhl wrote:
On Sat, Jul 26, 2025 at 4:47 AM Alexandre Courbot <acourbot@xxxxxxxxxx> wrote:
diff --git a/rust/kernel/transmute.rs b/rust/kernel/transmute.rs
index 1c7d43771a37b90150de86699f114a2ffb84db91..69c46c19a89191d8a2abc5801564cacda232218c 100644
--- a/rust/kernel/transmute.rs
+++ b/rust/kernel/transmute.rs
@@ -47,7 +47,16 @@ macro_rules! impl_frombytes {
///
/// Values of this type may not contain any uninitialized bytes. This type must not have interior
/// mutability.
-pub unsafe trait AsBytes {}
+pub unsafe trait AsBytes {
+ /// Returns `self` as a slice of bytes.
+ fn as_bytes(&self) -> &[u8] {
+ let data = core::ptr::from_ref(self).cast::<u8>();
+ let len = size_of_val(self);
+
+ // SAFETY: `data` is non-null and valid for `len * sizeof::<u8>()` bytes.
+ unsafe { core::slice::from_raw_parts(data, len) }
+ }
+}
Let's also have an as_bytes_mut() method. I would require the type to
also implement FromBytes as it lets you replace the value with another
set of bytes.
s/I would/It would/
FromBytes is needed only for as_bytes_mut(), not for the existing
method.
I agree with your suggestion, but it can be an independent patch and
doesn't need to go in via this one, right?
Given where we are in the merge cycle, it seems like we have a couple of
months until that code gets merged anyway, so I don't see any reason to
not send a v4 with Alice's suggestion? The only drawback I see is that I
would have to reset the Reviewed-by tags.