[PATCH] rust/kernel/faux: mark Registration methods inline

[PATCH] rust/kernel/faux: mark Registration methods inline

[Ethan Carter Edwards]

When building the kernel on Arch Linux using on x86_64 with tools:
$ rustc --version
rustc 1.84.0 (9fc6b4312 2025-01-07)
$ cargo --version
cargo 1.84.0 (66221abde 2024-11-19)
$ clang --version
clang version 19.1.7
Target: x86_64-pc-linux-gnu

The following symbols are generated:
$ nm vmlinux | rg ' _R' | rustfilt | rg faux
ffffffff81959ae0 T <kernel::faux::Registration>::new
ffffffff81959b40 T <kernel::faux::Registration as core::ops::drop::Drop>::drop

However, these Rust symbols are wrappers around bindings in the C faux
code. Inlining these functions removes the middle-man wrapper function
After applying this patch, the above function signatures disappear.

Link: https://github.com/Rust-for-Linux/linux/issues/1145
Signed-off-by: Ethan Carter Edwards <ethan@ethancedwards.com>
---
 rust/kernel/faux.rs | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/rust/kernel/faux.rs b/rust/kernel/faux.rs
index 5acc0c02d451f6d5a26b837d509374d508f26368..5fdd85ea64398130066d38e42f7c7485673f290c 100644
--- a/rust/kernel/faux.rs
+++ b/rust/kernel/faux.rs
@@ -24,6 +24,7 @@
 
 impl Registration {
     /// Create and register a new faux device with the given name.
+    #[inline]
     pub fn new(name: &CStr) -> Result<Self> {
         // SAFETY:
         // - `name` is copied by this function into its own storage
@@ -50,6 +51,7 @@ fn as_ref(&self) -> &device::Device {
 }
 
 impl Drop for Registration {
+    #[inline]
     fn drop(&mut self) {
         // SAFETY: `self.0` is a valid registered faux_device via our type invariants.
         unsafe { bindings::faux_device_destroy(self.as_raw()) }

---
base-commit: fc2f191f850d9a2fb1b78c51d49076e60fb42c49
change-id: 20250309-faux-inline-6c8eead1bcd0

Best regards,
-- 
Ethan Carter Edwards <ethan@ethancedwards.com>

Re: [PATCH] rust/kernel/faux: mark Registration methods inline

[Charalampos Mitrodimas]

Ethan Carter Edwards <ethan@ethancedwards.com> writes:

> When building the kernel on Arch Linux using on x86_64 with tools:
> $ rustc --version
> rustc 1.84.0 (9fc6b4312 2025-01-07)
> $ cargo --version
> cargo 1.84.0 (66221abde 2024-11-19)

Hi Ethan,

I think stating your Cargo version is unnecessary, we don't use cargo
(atleast not anymore).

> $ clang --version
> clang version 19.1.7
> Target: x86_64-pc-linux-gnu
>
> The following symbols are generated:
> $ nm vmlinux | rg ' _R' | rustfilt | rg faux
> ffffffff81959ae0 T <kernel::faux::Registration>::new
> ffffffff81959b40 T <kernel::faux::Registration as core::ops::drop::Drop>::drop
>
> However, these Rust symbols are wrappers around bindings in the C faux
> code. Inlining these functions removes the middle-man wrapper function
> After applying this patch, the above function signatures disappear.
>
> Link: https://github.com/Rust-for-Linux/linux/issues/1145
> Signed-off-by: Ethan Carter Edwards <ethan@ethancedwards.com>
> ---
>  rust/kernel/faux.rs | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/rust/kernel/faux.rs b/rust/kernel/faux.rs
> index 5acc0c02d451f6d5a26b837d509374d508f26368..5fdd85ea64398130066d38e42f7c7485673f290c 100644
> --- a/rust/kernel/faux.rs
> +++ b/rust/kernel/faux.rs
> @@ -24,6 +24,7 @@
>  
>  impl Registration {
>      /// Create and register a new faux device with the given name.
> +    #[inline]
>      pub fn new(name: &CStr) -> Result<Self> {
>          // SAFETY:
>          // - `name` is copied by this function into its own storage
> @@ -50,6 +51,7 @@ fn as_ref(&self) -> &device::Device {
>  }
>  
>  impl Drop for Registration {
> +    #[inline]
>      fn drop(&mut self) {
>          // SAFETY: `self.0` is a valid registered faux_device via our type invariants.
>          unsafe { bindings::faux_device_destroy(self.as_raw()) }
>
> ---
> base-commit: fc2f191f850d9a2fb1b78c51d49076e60fb42c49
> change-id: 20250309-faux-inline-6c8eead1bcd0
>
> Best regards,

Re: [PATCH] rust/kernel/faux: mark Registration methods inline

[Alice Ryhl]

On Sun, Mar 09, 2025 at 10:14:36PM -0400, Ethan Carter Edwards wrote:
> When building the kernel on Arch Linux using on x86_64 with tools:
> $ rustc --version
> rustc 1.84.0 (9fc6b4312 2025-01-07)
> $ cargo --version
> cargo 1.84.0 (66221abde 2024-11-19)
> $ clang --version
> clang version 19.1.7
> Target: x86_64-pc-linux-gnu
> 
> The following symbols are generated:
> $ nm vmlinux | rg ' _R' | rustfilt | rg faux
> ffffffff81959ae0 T <kernel::faux::Registration>::new
> ffffffff81959b40 T <kernel::faux::Registration as core::ops::drop::Drop>::drop
> 
> However, these Rust symbols are wrappers around bindings in the C faux
> code. Inlining these functions removes the middle-man wrapper function
> After applying this patch, the above function signatures disappear.
> 
> Link: https://github.com/Rust-for-Linux/linux/issues/1145
> Signed-off-by: Ethan Carter Edwards <ethan@ethancedwards.com>

LGTM to me.

Reviewed-by: Alice Ryhl <aliceryhl@google.com>

Re: [PATCH] rust/kernel/faux: mark Registration methods inline

[Greg Kroah-Hartman]

On Sun, Mar 09, 2025 at 10:14:36PM -0400, Ethan Carter Edwards wrote:
> When building the kernel on Arch Linux using on x86_64 with tools:
> $ rustc --version
> rustc 1.84.0 (9fc6b4312 2025-01-07)
> $ cargo --version
> cargo 1.84.0 (66221abde 2024-11-19)
> $ clang --version
> clang version 19.1.7
> Target: x86_64-pc-linux-gnu
> 
> The following symbols are generated:
> $ nm vmlinux | rg ' _R' | rustfilt | rg faux
> ffffffff81959ae0 T <kernel::faux::Registration>::new
> ffffffff81959b40 T <kernel::faux::Registration as core::ops::drop::Drop>::drop
> 
> However, these Rust symbols are wrappers around bindings in the C faux
> code. Inlining these functions removes the middle-man wrapper function
> After applying this patch, the above function signatures disappear.
> 
> Link: https://github.com/Rust-for-Linux/linux/issues/1145
> Signed-off-by: Ethan Carter Edwards <ethan@ethancedwards.com>
> ---
>  rust/kernel/faux.rs | 2 ++
>  1 file changed, 2 insertions(+)

This does not apply to my tree.  Can you regenerate it against either
the driver-core.git driver-core-next branch, or the linux-next tree and
resend it?

What release/branch did you make this against?

thanks,

greg k-h

Re: [PATCH] rust/kernel/faux: mark Registration methods inline

[Miguel Ojeda]

On Mon, Mar 10, 2025 at 7:24 PM Greg Kroah-Hartman
<gregkh@linuxfoundation.org> wrote:
>
> What release/branch did you make this against?

That hash comes from `rust-next`.

Cheers,
Miguel

Re: [PATCH] rust/kernel/faux: mark Registration methods inline

[Ethan Carter Edwards]

On 25/03/10 07:19PM, Greg Kroah-Hartman wrote:
> On Sun, Mar 09, 2025 at 10:14:36PM -0400, Ethan Carter Edwards wrote:
> > When building the kernel on Arch Linux using on x86_64 with tools:
> > $ rustc --version
> > rustc 1.84.0 (9fc6b4312 2025-01-07)
> > $ cargo --version
> > cargo 1.84.0 (66221abde 2024-11-19)
> > $ clang --version
> > clang version 19.1.7
> > Target: x86_64-pc-linux-gnu
> > 
> > The following symbols are generated:
> > $ nm vmlinux | rg ' _R' | rustfilt | rg faux
> > ffffffff81959ae0 T <kernel::faux::Registration>::new
> > ffffffff81959b40 T <kernel::faux::Registration as core::ops::drop::Drop>::drop
> > 
> > However, these Rust symbols are wrappers around bindings in the C faux
> > code. Inlining these functions removes the middle-man wrapper function
> > After applying this patch, the above function signatures disappear.
> > 
> > Link: https://github.com/Rust-for-Linux/linux/issues/1145
> > Signed-off-by: Ethan Carter Edwards <ethan@ethancedwards.com>
> > ---
> >  rust/kernel/faux.rs | 2 ++
> >  1 file changed, 2 insertions(+)
> 
> This does not apply to my tree.  Can you regenerate it against either
> the driver-core.git driver-core-next branch, or the linux-next tree and
> resend it?

Will do. My apologies, I based this patch off linux-next. I thought it
was customary to use the -next tree for whatever subsystem one is
developing against. I'll do it off of linux-next and send a v2.

> 
> What release/branch did you make this against?

Tree: https://github.com/rust-for-linux/linux
Branch: rust-next

> 
> thanks,
> 
> greg k-h

Thanks,
Ethan