From: "Benno Lossin" <lossin@kernel.org>
To: "Suchit Karunakaran" <suchitkarunakaran@gmail.com>,
<ojeda@kernel.org>, <alex.gaynor@gmail.com>,
<boqun.feng@gmail.com>, <gary@garyguo.net>,
<bjorn3_gh@protonmail.com>, <a.hindborg@kernel.org>,
<aliceryhl@google.com>, <tmgross@umich.edu>, <dakr@kernel.org>,
<me@kloenk.dev>, <chrisi.schrefl@gmail.com>,
<rust-for-linux@vger.kernel.org>
Cc: <skhan@linuxfoundation.org>,
<linux-kernel-mentees@lists.linux.dev>,
<linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] rust/pin-init: remove workaround for type inference cycle
Date: Sun, 27 Jul 2025 19:28:15 +0200 [thread overview]
Message-ID: <DBN0JG0QP7TU.PZSZ8NZL5LD6@kernel.org> (raw)
In-Reply-To: <20250727170217.206794-1-suchitkarunakaran@gmail.com>
On Sun Jul 27, 2025 at 7:02 PM CEST, Suchit Karunakaran wrote:
> The `cast_pin_init` and `cast_init` functions previously used an
> intermediate `let` binding before returning the result expression to work
> around a Rust compiler issue causing type inference cycles. With the
> minimum Rust compiler version for the kernel now at 1.78.0, where this
> issue is fixed, the workaround is no longer needed. This patch removes the
> unnecessary `let` variables and returns the expressions directly.
>
> Signed-off-by: Suchit Karunakaran <suchitkarunakaran@gmail.com>
I still encounter the cycle when compiling with 1.78.0, which version
did you test this with?
One of The errors I see after applying the patch on top of
`pin-init-next`:
error[E0391]: cycle detected when computing type of opaque `cast_pin_init::{opaque#0}`
--> rust/pin-init/src/lib.rs:1278:73
|
1278 | pub const unsafe fn cast_pin_init<T, U, E>(init: impl PinInit<T, E>) -> impl PinInit<U, E> {
| ^^^^^^^^^^^^^^^^^^
|
note: ...which requires borrow-checking `cast_pin_init`...
--> rust/pin-init/src/lib.rs:1278:1
|
1278 | pub const unsafe fn cast_pin_init<T, U, E>(init: impl PinInit<T, E>) -> impl PinInit<U, E> {
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
note: ...which requires promoting constants in MIR for `cast_pin_init`...
--> rust/pin-init/src/lib.rs:1278:1
|
1278 | pub const unsafe fn cast_pin_init<T, U, E>(init: impl PinInit<T, E>) -> impl PinInit<U, E> {
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
note: ...which requires const checking `cast_pin_init`...
--> rust/pin-init/src/lib.rs:1278:1
|
1278 | pub const unsafe fn cast_pin_init<T, U, E>(init: impl PinInit<T, E>) -> impl PinInit<U, E> {
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
= note: ...which requires computing whether `cast_pin_init::{opaque#0}` is freeze...
= note: ...which requires evaluating trait selection obligation `cast_pin_init::{opaque#0}: core::marker::Freeze`...
= note: ...which again requires computing type of opaque `cast_pin_init::{opaque#0}`, completing the cycle
note: cycle used when computing type of `cast_pin_init::{opaque#0}`
--> rust/pin-init/src/lib.rs:1278:73
|
1278 | pub const unsafe fn cast_pin_init<T, U, E>(init: impl PinInit<T, E>) -> impl PinInit<U, E> {
| ^^^^^^^^^^^^^^^^^^
= note: see https://rustc-dev-guide.rust-lang.org/overview.html#queries and https://rustc-dev-guide.rust-lang.org/query.html for more information
> ---
> rust/pin-init/src/lib.rs | 10 ++--------
> 1 file changed, 2 insertions(+), 8 deletions(-)
>
> diff --git a/rust/pin-init/src/lib.rs b/rust/pin-init/src/lib.rs
> index 62e013a5cc20..cc244eeb19cd 100644
> --- a/rust/pin-init/src/lib.rs
> +++ b/rust/pin-init/src/lib.rs
> @@ -1278,10 +1278,7 @@ unsafe fn __pinned_init(self, slot: *mut T) -> Result<(), E> {
> pub const unsafe fn cast_pin_init<T, U, E>(init: impl PinInit<T, E>) -> impl PinInit<U, E> {
> // SAFETY: initialization delegated to a valid initializer. Cast is valid by function safety
> // requirements.
> - let res = unsafe { pin_init_from_closure(|ptr: *mut U| init.__pinned_init(ptr.cast::<T>())) };
> - // FIXME: remove the let statement once the nightly-MSRV allows it (1.78 otherwise encounters a
Do note the comment mentioning that it is needed for version 1.78 here,
so I think this patch still needs to wait until we bump the minimum.
---
Cheers,
Benno
> - // cycle when computing the type returned by this function)
> - res
> + unsafe { pin_init_from_closure(|ptr: *mut U| init.__pinned_init(ptr.cast::<T>())) }
> }
next prev parent reply other threads:[~2025-07-27 17:28 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-07-27 17:02 [PATCH] rust/pin-init: remove workaround for type inference cycle Suchit Karunakaran
2025-07-27 17:28 ` Benno Lossin [this message]
2025-07-27 17:44 ` Suchit K
2025-07-27 17:58 ` Benno Lossin
2025-07-28 4:00 ` Suchit K
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=DBN0JG0QP7TU.PZSZ8NZL5LD6@kernel.org \
--to=lossin@kernel.org \
--cc=a.hindborg@kernel.org \
--cc=alex.gaynor@gmail.com \
--cc=aliceryhl@google.com \
--cc=bjorn3_gh@protonmail.com \
--cc=boqun.feng@gmail.com \
--cc=chrisi.schrefl@gmail.com \
--cc=dakr@kernel.org \
--cc=gary@garyguo.net \
--cc=linux-kernel-mentees@lists.linux.dev \
--cc=linux-kernel@vger.kernel.org \
--cc=me@kloenk.dev \
--cc=ojeda@kernel.org \
--cc=rust-for-linux@vger.kernel.org \
--cc=skhan@linuxfoundation.org \
--cc=suchitkarunakaran@gmail.com \
--cc=tmgross@umich.edu \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox