Re: [PATCH 1/3] rust: page: replace the page pointer wrapper with Opaque

From: Boqun Feng
Date: Tue Oct 08 2024 - 04:01:19 EST


On Tue, Oct 08, 2024 at 08:58:56AM +0200, Alice Ryhl wrote:
> On Mon, Oct 7, 2024 at 10:28 PM Abdiel Janulgue
> <abdiel.janulgue@xxxxxxxxx> wrote:
> >
> > Replace NonNull with Opaque to make it possible to cast to a Page pointer
> > from a raw struct page pointer.
> >
> > Signed-off-by: Abdiel Janulgue <abdiel.janulgue@xxxxxxxxx>
> > ---
> > rust/kernel/page.rs | 19 +++++++++++++------
> > 1 file changed, 13 insertions(+), 6 deletions(-)
> >
> > diff --git a/rust/kernel/page.rs b/rust/kernel/page.rs
> > index 208a006d587c..08ff09a25223 100644
> > --- a/rust/kernel/page.rs
> > +++ b/rust/kernel/page.rs
> > @@ -8,8 +8,9 @@
> > error::code::*,
> > error::Result,
> > uaccess::UserSliceReader,
> > + types::Opaque,
> > };
> > -use core::ptr::{self, NonNull};
> > +use core::ptr::{self};
> >
> > /// A bitwise shift for the page size.
> > pub const PAGE_SHIFT: usize = bindings::PAGE_SHIFT as usize;
> > @@ -25,8 +26,9 @@
> > /// # Invariants
> > ///
> > /// The pointer is valid, and has ownership over the page.
> > +#[repr(transparent)]
> > pub struct Page {
> > - page: NonNull<bindings::page>,
> > + page: Opaque<bindings::page>,
> > }
> >
> > // SAFETY: Pages have no logic that relies on them staying on a given thread, so moving them across
> > @@ -65,15 +67,20 @@ pub fn alloc_page(flags: Flags) -> Result<Self, AllocError> {
> > // SAFETY: Depending on the value of `gfp_flags`, this call may sleep. Other than that, it
> > // is always safe to call this method.
> > let page = unsafe { bindings::alloc_pages(flags.as_raw(), 0) };
> > - let page = NonNull::new(page).ok_or(AllocError)?;
> > + if page.is_null() {
> > + return Err(AllocError);
> > + }
> > + // CAST: Self` is a `repr(transparent)` wrapper around `bindings::page`.
> > + let ptr = page.cast::<Self>();
> > // INVARIANT: We just successfully allocated a page, so we now have ownership of the newly
> > // allocated page. We transfer that ownership to the new `Page` object.
> > - Ok(Self { page })
> > + // SAFETY: According to invariant above ptr is valid.
> > + Ok(unsafe { ptr::read(ptr) })
>
> Using `ptr::read` on the page is definitely not okay. That duplicates
> the contents of the `struct page`. You'll need some sort of pointer
> type around `Page` instead.
>

Agreed. So may I suggest we introduce `Owned` type and `Ownable` trait
[1]? `alloc_page()` can be refactor to return a `Result<Owned<Self>,
AllocError>`.

[1]: https://lore.kernel.org/rust-for-linux/ZnCzLIly3DRK2eab@boqun-archlinux/

Regards,
Boqun

> Alice