On Mon, Jul 14, 2025 at 01:06:08PM +0200, Benno Lossin wrote:
On Mon Jul 14, 2025 at 7:36 AM CEST, Boqun Feng wrote:
quoted
+// Defines an internal type that always maps to the integer type which has the same size alignment
+// as `isize` and `usize`, and `isize` and `usize` are always bi-directional transmutable to
+// `isize_atomic_repr`, which also always implements `AtomicImpl`.
+#[allow(non_camel_case_types)]
+#[cfg(not(CONFIG_64BIT))]
+type isize_atomic_repr = i32;
+#[allow(non_camel_case_types)]
+#[cfg(CONFIG_64BIT)]
+type isize_atomic_repr = i64;
+
+// Ensure size and alignment requirements are checked.
+crate::static_assert!(core::mem::size_of::<isize>() == core::mem::size_of::<isize_atomic_repr>());
+crate::static_assert!(core::mem::align_of::<isize>() == core::mem::align_of::<isize_atomic_repr>());
+crate::static_assert!(core::mem::size_of::<usize>() == core::mem::size_of::<isize_atomic_repr>());
+crate::static_assert!(core::mem::align_of::<usize>() == core::mem::align_of::<isize_atomic_repr>());
This is fine for now, but I would prefer for this to go into an
`assumptions` module like Miguel proposed some time ago.
Well, sure. Also if `assumptions` or other core kernel mod can provide
a definition similar to `isize_atomic_repr`, I'm happy to use it.
Regards,
Boqun
---
Cheers,
Benno