[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAGSQo03g8bjRosYVPzxhK9FpSW1ZfW52unPaoxpcn=pXaDUriQ@mail.gmail.com>
Date: Tue, 3 Feb 2026 10:04:03 -0800
From: Matthew Maurer <mmaurer@...gle.com>
To: Gary Guo <gary@...yguo.net>
Cc: Bjorn Andersson <andersson@...nel.org>, Konrad Dybcio <konradybcio@...nel.org>,
Satya Durga Srinivasu Prabhala <satyap@...cinc.com>, Miguel Ojeda <ojeda@...nel.org>, Boqun Feng <boqun.feng@...il.com>,
Björn Roy Baron <bjorn3_gh@...tonmail.com>,
Benno Lossin <lossin@...nel.org>, Andreas Hindborg <a.hindborg@...nel.org>,
Alice Ryhl <aliceryhl@...gle.com>, Trevor Gross <tmgross@...ch.edu>,
Danilo Krummrich <dakr@...nel.org>, Daniel Almeida <daniel.almeida@...labora.com>,
Greg Kroah-Hartman <gregkh@...uxfoundation.org>, "Rafael J. Wysocki" <rafael@...nel.org>,
David Airlie <airlied@...il.com>, Simona Vetter <simona@...ll.ch>,
Michal Wilczynski <m.wilczynski@...sung.com>, Dave Ertman <david.m.ertman@...el.com>,
Ira Weiny <ira.weiny@...el.com>, Leon Romanovsky <leon@...nel.org>, Trilok Soni <tsoni@...cinc.com>,
linux-kernel@...r.kernel.org, linux-arm-msm@...r.kernel.org,
rust-for-linux@...r.kernel.org, driver-core@...ts.linux.dev,
dri-devel@...ts.freedesktop.org, linux-pwm@...r.kernel.org
Subject: Re: [PATCH v2 5/6] rust: debugfs: Allow access to device in
Devres-wrapped scopes
On Tue, Feb 3, 2026 at 8:48 AM Gary Guo <gary@...yguo.net> wrote:
>
> On Tue Feb 3, 2026 at 3:46 PM GMT, Matthew Maurer wrote:
> > This adds support for creating a DebugFS directory which is aware that
> > it is bound to a device. As a result, callbacks under that directory
> > have access to a bound device which gives them efficient access to other
> > Devres, ability to use dev_err! and friends, etc.
> >
> > Signed-off-by: Matthew Maurer <mmaurer@...gle.com>
> > ---
> > rust/kernel/debugfs.rs | 40 ++++++++++++++++++++++++++++++++++++++++
> > 1 file changed, 40 insertions(+)
> >
> > diff --git a/rust/kernel/debugfs.rs b/rust/kernel/debugfs.rs
> > index d7b8014a6474698235203f2b7d8fec96f2bb43f8..ac614d693fa73929d095b669e9ba61958bec609e 100644
> > --- a/rust/kernel/debugfs.rs
> > +++ b/rust/kernel/debugfs.rs
> > @@ -11,6 +11,11 @@
> > #[cfg(CONFIG_DEBUG_FS)]
> > use crate::sync::Arc;
> > use crate::{
> > + device::{
> > + Bound,
> > + Device, //
> > + },
> > + devres::Devres,
> > fmt,
> > prelude::*,
> > str::CStr,
> > @@ -722,3 +727,38 @@ fn new(name: &CStr) -> ScopedDir<'data, 'static> {
> > }
> > }
> > }
> > +
> > +impl<'a, T: 'a + Send> Devres<Scope<T>> {
> > + /// Creates a new scope, which is a directory at the root of the debugfs filesystem,
> > + /// associated with some data `T`, enclosed in a [`Devres`] for the provided device.
> > + ///
> > + /// The `init` closure is called to populate the directory with files and subdirectories. These
> > + /// files can reference the data stored in the scope. Because it is stored inside a `Devres`,
> > + /// the init method is granted access to a `&Device<Bound>`.
> > + ///
> > + /// This can be used for cheaply accessing device-protected data inside DebugFS methods or
> > + /// accessing device-specific methods (e.g. [`dev_err!`]).
> > + ///
> > + /// The entire directory tree created within the scope will be removed when the returned
> > + /// `Scope` handle is dropped.
> > + pub fn dir<E: 'a, F>(
> > + dev: &'a Device<Bound>,
> > + data: impl PinInit<T, E> + 'a,
> > + name: &'a CStr,
> > + init: F,
> > + ) -> impl PinInit<Self, Error> + 'a
> > + where
> > + F: for<'data, 'dir> FnOnce(&'data T, &'data Device<Bound>, &'dir ScopedDir<'data, 'dir>)
> > + + 'a,
> > + Error: From<E>,
> > + {
> > + Devres::new(
> > + dev,
> > + Scope::new(data, |data| {
> > + let scoped = ScopedDir::new(name);
> > + init(data, dev, &scoped);
> > + scoped.into_entry()
> > + }),
> > + )
> > + }
> > +}
>
> I think it is a big strange to have this on `Devres` (in patch v6 it has `Devres::dir` doesn't make
> too much sense). I would suggest that we domsomething like
>
> impl<'a, T: 'a + Send> Scope<T> {
> pub fn devres_dir(
> ...
> ) -> impl PinInit<Devres<Self>, Error> + 'a;
> }
>
> To me `Devres` is just a generic container type, just like `Arc` and `ARef`, so
> the assoc functions should be defined on the concrete type.
>
> Also: is there a reason that this needs a special API, and by
>
> Devres::new(device, Scope::dir(data, c"name", |data| {
> // use data and device
> });
>
> ?
Yes - that won't work, because the function being provided to
`Scope::dir` is `for<'data, 'dir> FnOnce(&'data T, &'dir
ScopedDir<'data, 'dir>)` - this means that *intentionally*, if you
capture any non-static-lifetime variable from outside the closure, you
won't be able to use it with the methods on `ScopedDir`, because the
`'data` lifetime bound should stop you. In the general case, we
wouldn't want a reference with the same lifetime as `device` in that
example to be usable inside the debugfs callbacks. The device of a
Devres wrapped scope is a special case because we know that it will
outlive it.
>
> Best,
> Gary
>
>
Powered by blists - more mailing lists