[PATCH] rust: less allocation in CString::try_from

[PATCH] rust: less allocation in CString::try_from

[Onur Özkan]

Allocates buffer with the correct capacity upfront by
using the length of the `CStr` to avoid extra and unnecessary
re-allocation when converting from `CStr` to `CString`.

Signed-off-by: Onur Özkan <work@onurozkan.dev>
---
 rust/kernel/str.rs | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/rust/kernel/str.rs b/rust/kernel/str.rs
index 6c892550c0ba..98d41d995e45 100644
--- a/rust/kernel/str.rs
+++ b/rust/kernel/str.rs
@@ -946,9 +946,10 @@ impl<'a> TryFrom<&'a CStr> for CString {
     type Error = AllocError;

     fn try_from(cstr: &'a CStr) -> Result<CString, AllocError> {
-        let mut buf = KVec::new();
+        let bytes = cstr.to_bytes_with_nul();

-        buf.extend_from_slice(cstr.to_bytes_with_nul(), GFP_KERNEL)?;
+        let mut buf = KVec::with_capacity(bytes.len(), GFP_KERNEL)?;
+        buf.extend_from_slice(bytes, GFP_KERNEL)?;

         // INVARIANT: The `CStr` and `CString` types have the same invariants for
         // the string data, and we copied it over without changes.
--
2.51.0

Re: [PATCH] rust: less allocation in CString::try_from

[Alice Ryhl]

On Mon, Sep 15, 2025 at 8:54 AM Onur Özkan <work@onurozkan.dev> wrote:
>
> Allocates buffer with the correct capacity upfront by
> using the length of the `CStr` to avoid extra and unnecessary
> re-allocation when converting from `CStr` to `CString`.
>
> Signed-off-by: Onur Özkan <work@onurozkan.dev>
> ---
>  rust/kernel/str.rs | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/rust/kernel/str.rs b/rust/kernel/str.rs
> index 6c892550c0ba..98d41d995e45 100644
> --- a/rust/kernel/str.rs
> +++ b/rust/kernel/str.rs
> @@ -946,9 +946,10 @@ impl<'a> TryFrom<&'a CStr> for CString {
>      type Error = AllocError;
>
>      fn try_from(cstr: &'a CStr) -> Result<CString, AllocError> {
> -        let mut buf = KVec::new();
> +        let bytes = cstr.to_bytes_with_nul();
>
> -        buf.extend_from_slice(cstr.to_bytes_with_nul(), GFP_KERNEL)?;
> +        let mut buf = KVec::with_capacity(bytes.len(), GFP_KERNEL)?;
> +        buf.extend_from_slice(bytes, GFP_KERNEL)?;

I don't think this changes the number of allocations. KVec::new() does
not allocate.

Alice

Re: [PATCH] rust: less allocation in CString::try_from

[Onur]

On Mon, 15 Sep 2025 09:26:00 +0200
Alice Ryhl <aliceryhl@google.com> wrote:

> On Mon, Sep 15, 2025 at 8:54 AM Onur Özkan <work@onurozkan.dev> wrote:
> >
> > Allocates buffer with the correct capacity upfront by
> > using the length of the `CStr` to avoid extra and unnecessary
> > re-allocation when converting from `CStr` to `CString`.
> >
> > Signed-off-by: Onur Özkan <work@onurozkan.dev>
> > ---
> >  rust/kernel/str.rs | 5 +++--
> >  1 file changed, 3 insertions(+), 2 deletions(-)
> >
> > diff --git a/rust/kernel/str.rs b/rust/kernel/str.rs
> > index 6c892550c0ba..98d41d995e45 100644
> > --- a/rust/kernel/str.rs
> > +++ b/rust/kernel/str.rs
> > @@ -946,9 +946,10 @@ impl<'a> TryFrom<&'a CStr> for CString {
> >      type Error = AllocError;
> >
> >      fn try_from(cstr: &'a CStr) -> Result<CString, AllocError> {
> > -        let mut buf = KVec::new();
> > +        let bytes = cstr.to_bytes_with_nul();
> >
> > -        buf.extend_from_slice(cstr.to_bytes_with_nul(),
> > GFP_KERNEL)?;
> > +        let mut buf = KVec::with_capacity(bytes.len(),
> > GFP_KERNEL)?;
> > +        buf.extend_from_slice(bytes, GFP_KERNEL)?;
> 
> I don't think this changes the number of allocations. KVec::new() does
> not allocate.
> 
> Alice

Good to know, thanks!

Onur

Re: [PATCH] rust: less allocation in CString::try_from

[Daniel Almeida]

Hi Onur, thanks for working on this :)

The commit title doesn’t parse very well, can you rework it?

— Daniel

Re: [PATCH] rust: less allocation in CString::try_from

[Onur Özkan]

On Mon, 15 Sep 2025 17:38:32 +0200
Daniel Almeida <daniel.almeida@collabora.com> wrote:

> Hi Onur, thanks for working on this :)
> 
> The commit title doesn’t parse very well, can you rework it?
> 
> — Daniel
> 
> 

Hi Daniel,

As Alice mentioned in [0], my assumption about reducing allocations was
wrong. Looking at the KVec implementation details this patch doesn't
actually change anything other than how the code looks. We can safely
ignore it. I should have checked that, sorry.

[0]:
https://lore.kernel.org/all/CAH5fLgi-VrhDSEjJA4Bg+5Wmr2JDAmmOc1FkfZAZ29SMqkX++w@mail.gmail.com/

Thanks,
Onur