Skip to contentSkip to conversation

rust: cpu: use `unsafe_precondition_assert!`

PATCH In review Ritvik Gupta 2 participants 3 messages May 18

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 rust: cpu: use `unsafe_precondition_assert!`

Replace debug_assert! invocations with unsafe_precondition_assert! in from_i32_unchecked and from_u32_unchecked unsafe functions.

---
 rust/kernel/cpu.rs | 22 +++++++++++++++++-----
 1 file changed, 17 insertions(+), 5 deletions(-)
rust/kernel/cpu.rs +17 -5
index cb6c0338ef5a..e33a87304c59 100644
@@ -4,7 +4,7 @@
 //!
 //! C header: [`include/linux/cpu.h`](srctree/include/linux/cpu.h)
 
-use crate::{bindings, device::Device, error::Result, prelude::ENODEV};
+use crate::{bindings, device::Device, error::Result, prelude::ENODEV, unsafe_precondition_assert};
 
 /// Returns the maximum number of possible CPUs in the current system configuration.
 #[inline]
@@ -54,8 +54,14 @@ impl CpuId {
     /// The caller must ensure that `id` is a valid CPU ID (i.e., `0 <= id < nr_cpu_ids()`).
     #[inline]
     pub unsafe fn from_i32_unchecked(id: i32) -> Self {
-        debug_assert!(id >= 0);
-        debug_assert!((id as u32) < nr_cpu_ids());
+        unsafe_precondition_assert!(
+            id >= 0,
+            "CpuId::from_i32_unchecked requires a non-negative 'id'"
+        );
+        unsafe_precondition_assert!(
+            (id as u32) < nr_cpu_ids(),
+            "CpuId::from_i32_unchecked requires a valid 'id' (i.e., `0 <= id < nr_cpu_ids()`)"
+        );
 
         // INVARIANT: The function safety guarantees `id` is a valid CPU id.
         Self(id as u32)
@@ -78,10 +84,16 @@ pub fn from_i32(id: i32) -> Option<Self> {
     /// The caller must ensure that `id` is a valid CPU ID (i.e., `0 <= id < nr_cpu_ids()`).
     #[inline]
     pub unsafe fn from_u32_unchecked(id: u32) -> Self {
-        debug_assert!(id < nr_cpu_ids());
+        unsafe_precondition_assert!(
+            id < nr_cpu_ids(),
+            "CpuId::from_i32_unchecked requires a valid 'id' (i.e., `0 <= id < nr_cpu_ids()`)"
+        );
 
         // Ensure the `id` fits in an [`i32`] as it's also representable that way.
-        debug_assert!(id <= i32::MAX as u32);
+        unsafe_precondition_assert!(
+            id <= i32::MAX as u32,
+            "CpuId::from_u32_unchecked requires the 'id' to not overflow i32::MAX"
+        );
 
         // INVARIANT: The function safety guarantees `id` is a valid CPU id.
         Self(id)
Signature · 1 line
2.54.0
#1 lore raw
Quoted 4 lines · Ritvik Gupta

Replace debug_assert! invocations with unsafe_precondition_assert! in from_i32_unchecked and from_u32_unchecked unsafe functions.

Thanks for the patch!

This should mention that we are adding messages too. That could actually be a second patch in the series to make it even clearer, but it is not a big a deal either way (and perhaps rustfmt makes that first patch not that clean due to the line wrapping anyway).

rust/kernel/cpu.rs line 7

Quoted 2 lines
rust/kernel/cpu.rs +1 -1
-use crate::{bindings, device::Device, error::Result, prelude::ENODEV};
+use crate::{bindings, device::Device, error::Result, prelude::ENODEV, unsafe_precondition_assert};

We should probably put it in the prelude along the others -- the easier it is to type, the better, to encourage its use.

Do you want to add a first patch in the series doing that?

Quoted 1 line

+ "CpuId::from_i32_unchecked requires a valid 'id' (i.e., 0 <= id < nr_cpu_ids())"

This seems to mix Markdown backquotes with single quotes, and then doesn't use backquotes in the first part. Could we make it consistent if we are adding the messages? e.g.

    "`CpuId::from_i32_unchecked` requires a valid `id` (i.e., `0 <= id

< nr_cpu_ids()`)"

Sashiko has also spotted a copy-paste error in the one below.

Cheers,
Miguel

#2 lore raw
Quoted 16 lines · Miguel Ojeda
Quoted 4 lines · Ritvik Gupta

Replace debug_assert! invocations with unsafe_precondition_assert! in from_i32_unchecked and from_u32_unchecked unsafe functions.

Thanks for the patch!

This should mention that we are adding messages too. That could actually be a second patch in the series to make it even clearer, but it is not a big a deal either way (and perhaps rustfmt makes that first patch not that clean due to the line wrapping anyway).

Quoted 8 lines

rust/kernel/cpu.rs line 7

Quoted 2 lines
rust/kernel/cpu.rs +1 -1
-use crate::{bindings, device::Device, error::Result, prelude::ENODEV};
+use crate::{bindings, device::Device, error::Result, prelude::ENODEV, unsafe_precondition_assert};

We should probably put it in the prelude along the others -- the easier it is to type, the better, to encourage its use.

Do you want to add a first patch in the series doing that?

Sure! In v2 series, I'll add the first patch that adds unsafe_precondition_assert! to the prelude and second patch that uses it.

Quoted 9 lines
Quoted 1 line

+ "CpuId::from_i32_unchecked requires a valid 'id' (i.e., 0 <= id < nr_cpu_ids())"

This seems to mix Markdown backquotes with single quotes, and then doesn't use backquotes in the first part. Could we make it consistent if we are adding the messages? e.g.

    "`CpuId::from_i32_unchecked` requires a valid `id` (i.e., `0 <= id

< nr_cpu_ids()`)"

Sure, will fix :)

Quoted 4 lines

Sashiko has also spotted a copy-paste error in the one below.

Cheers,
Miguel