Replace the explicit bounds comparisons with a single checked_add()-based
range check. This avoids redundant comparisons, makes the overflow case
explicit, and results in simpler generated code (checked with godbolt
for x86).

Option::is_none_or() would be nicer, but it requires Rust 1.82; the
kernel currently targets 1.78.

No functional change intended.

Signed-off-by: Kari Argillander <kari.argillander@gmail.com>
---
 rust/kernel/page.rs | 18 ++++++++----------
 1 file changed, 8 insertions(+), 10 deletions(-)

diff --git a/rust/kernel/page.rs b/rust/kernel/page.rs
index 432fc0297d4a..a07e6d256860 100644
--- a/rust/kernel/page.rs
+++ b/rust/kernel/page.rs
@@ -239,17 +239,15 @@ fn with_pointer_into_page<T>(
         len: usize,
         f: impl FnOnce(*mut u8) -> Result<T>,
     ) -> Result<T> {
-        let bounds_ok = off <= PAGE_SIZE && len <= PAGE_SIZE && (off + len) <= PAGE_SIZE;
-
-        if bounds_ok {
-            self.with_page_mapped(move |page_addr| {
-                // SAFETY: The `off` integer is at most `PAGE_SIZE`, so this pointer offset will
-                // result in a pointer that is in bounds or one off the end of the page.
-                f(unsafe { page_addr.add(off) })
-            })
-        } else {
-            Err(EINVAL)
+        if off.checked_add(len).map_or(true, |end| end > PAGE_SIZE) {
+            return Err(EINVAL);
         }
+
+        self.with_page_mapped(move |page_addr| {
+            // SAFETY: The `off` integer is at most `PAGE_SIZE`, so this pointer offset will
+            // result in a pointer that is in bounds or one off the end of the page.
+            f(unsafe { page_addr.add(off) })
+        })
     }
 
     /// Maps the page and reads from it into the given buffer.

---
base-commit: cc3aa43b44bdb43dfbac0fcb51c56594a11338a8
change-id: 20251219-rust-page-check-819ccc39c53a

Best regards,
-- 
Kari Argillander <kari.argillander@gmail.com>

On 19.12.25 22:29, Kari Argillander wrote:
> Replace the explicit bounds comparisons with a single checked_add()-based
> range check. This avoids redundant comparisons, makes the overflow case
> explicit, and results in simpler generated code (checked with godbolt
> for x86).
> 
> Option::is_none_or() would be nicer, but it requires Rust 1.82; the
> kernel currently targets 1.78.
> 
> No functional change intended.
> 
> Signed-off-by: Kari Argillander <kari.argillander@gmail.com>
> ---
>  rust/kernel/page.rs | 18 ++++++++----------
>  1 file changed, 8 insertions(+), 10 deletions(-)
> 
> diff --git a/rust/kernel/page.rs b/rust/kernel/page.rs
> index 432fc0297d4a..a07e6d256860 100644
> --- a/rust/kernel/page.rs
> +++ b/rust/kernel/page.rs
> @@ -239,17 +239,15 @@ fn with_pointer_into_page<T>(
>          len: usize,
>          f: impl FnOnce(*mut u8) -> Result<T>,
>      ) -> Result<T> {
> -        let bounds_ok = off <= PAGE_SIZE && len <= PAGE_SIZE && (off + len) <= PAGE_SIZE;
> -
> -        if bounds_ok {
> -            self.with_page_mapped(move |page_addr| {
> -                // SAFETY: The `off` integer is at most `PAGE_SIZE`, so this pointer offset will
> -                // result in a pointer that is in bounds or one off the end of the page.
> -                f(unsafe { page_addr.add(off) })
> -            })
> -        } else {
> -            Err(EINVAL)
> +        if off.checked_add(len).map_or(true, |end| end > PAGE_SIZE) {
> +            return Err(EINVAL);
>          }
> +
> +        self.with_page_mapped(move |page_addr| {
> +            // SAFETY: The `off` integer is at most `PAGE_SIZE`, so this pointer offset will
> +            // result in a pointer that is in bounds or one off the end of the page.
> +            f(unsafe { page_addr.add(off) })
> +        })
>      }
>  
>      /// Maps the page and reads from it into the given buffer.
> 
> ---
> base-commit: cc3aa43b44bdb43dfbac0fcb51c56594a11338a8
> change-id: 20251219-rust-page-check-819ccc39c53a


Thanks, this switches to the early return style we recently discussed
in [1].

I just wonder if we would like to add a comment like

// TODO: replace `map_or` with `is_none_or` once the MSRV is >= 1.82.

like we do in e.g. transmute.rs [2]?

Anyway

Reviewed-by: Dirk Behme <dirk.behme@de.bosch.com>

Thanks

Dirk

[1]
https://lore.kernel.org/rust-for-linux/CANiq72kiscT5euAUjcSzvxMzM9Hdj8aQGeUN_pVF-vHf3DhBuQ@mail.gmail.com/

[2]
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/rust/kernel/transmute.rs#n75

On 20.12.2025 10.41 Dirk Behme (dirk.behme@gmail.com) wrote:
>
> On 19.12.25 22:29, Kari Argillander wrote:
> > Option::is_none_or() would be nicer, but it requires Rust 1.82; the
> > kernel currently targets 1.78.
> >
>
> I just wonder if we would like to add a comment like
>
> // TODO: replace `map_or` with `is_none_or` once the MSRV is >= 1.82.
>
> like we do in e.g. transmute.rs [2]?

Sure, I can do that. I’ll send a respin maybe next Tuesday.

> [2]
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/rust/kernel/transmute.rs#n75

On Sat Dec 20, 2025 at 6:29 AM JST, Kari Argillander wrote:
> Replace the explicit bounds comparisons with a single checked_add()-based
> range check. This avoids redundant comparisons, makes the overflow case
> explicit, and results in simpler generated code (checked with godbolt
> for x86).
>
> Option::is_none_or() would be nicer, but it requires Rust 1.82; the
> kernel currently targets 1.78.
>
> No functional change intended.
>
> Signed-off-by: Kari Argillander <kari.argillander@gmail.com>
> ---
>  rust/kernel/page.rs | 18 ++++++++----------
>  1 file changed, 8 insertions(+), 10 deletions(-)
>
> diff --git a/rust/kernel/page.rs b/rust/kernel/page.rs
> index 432fc0297d4a..a07e6d256860 100644
> --- a/rust/kernel/page.rs
> +++ b/rust/kernel/page.rs
> @@ -239,17 +239,15 @@ fn with_pointer_into_page<T>(
>          len: usize,
>          f: impl FnOnce(*mut u8) -> Result<T>,
>      ) -> Result<T> {
> -        let bounds_ok = off <= PAGE_SIZE && len <= PAGE_SIZE && (off + len) <= PAGE_SIZE;
> -
> -        if bounds_ok {
> -            self.with_page_mapped(move |page_addr| {
> -                // SAFETY: The `off` integer is at most `PAGE_SIZE`, so this pointer offset will
> -                // result in a pointer that is in bounds or one off the end of the page.
> -                f(unsafe { page_addr.add(off) })
> -            })
> -        } else {
> -            Err(EINVAL)
> +        if off.checked_add(len).map_or(true, |end| end > PAGE_SIZE) {

I find this line a bit heavy with the two-steps check, how about
something like:

    if !matches!(off.checked_add(len), Some(e) if e <= PAGE_SIZE) {

Just for your consideration - the original is still valid and an
improvement over the original code, so regardless:

Reviewed-by: Alexandre Courbot <acourbot@nvidia.com>