linux-kernel - Re: [PATCH v2 2/4] rust: debugfs: Bind file creation for long-lived Display

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <aBOwLxagw4NwoQwX@pollux>
Date: Thu, 1 May 2025 19:32:31 +0200
From: Danilo Krummrich <dakr@...nel.org>
To: Matthew Maurer <mmaurer@...gle.com>
Cc: Miguel Ojeda <ojeda@...nel.org>, Alex Gaynor <alex.gaynor@...il.com>,
	Boqun Feng <boqun.feng@...il.com>, Gary Guo <gary@...yguo.net>,
	Björn Roy Baron <bjorn3_gh@...tonmail.com>,
	Benno Lossin <benno.lossin@...ton.me>,
	Andreas Hindborg <a.hindborg@...nel.org>,
	Alice Ryhl <aliceryhl@...gle.com>, Trevor Gross <tmgross@...ch.edu>,
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
	"Rafael J. Wysocki" <rafael@...nel.org>,
	Sami Tolvanen <samitolvanen@...gle.com>,
	Timur Tabi <ttabi@...dia.com>, linux-kernel@...r.kernel.org,
	rust-for-linux@...r.kernel.org
Subject: Re: [PATCH v2 2/4] rust: debugfs: Bind file creation for long-lived
 Display

On Thu, May 01, 2025 at 09:09:27AM -0700, Matthew Maurer wrote:
> On Thu, May 1, 2025 at 3:37 AM Danilo Krummrich <dakr@...nel.org> wrote:
> > On Wed, Apr 30, 2025 at 11:31:57PM +0000, Matthew Maurer wrote:
> > Besides that, don't we also need a separate type for a file to be able to attach
> > non-static data anyways? I.e. something like:
> >
> >         #[cfg(CONFIG_DEBUG_FS)]
> >         struct File<T> {
> >            dentry: *mut bindings::dentry,
> >            data: T,
> >         }
> >
> >         #[cfg(not(CONFIG_DEBUG_FS))]
> >         struct File<T> {
> >            _p: PhantomData<T>,
> >         }
> >
> > I'm not exactly sure how v1 did this; I haven't had time to look at v1 before v2
> > was posted. I seems like v1 relied on a separate structure storing the data,
> > which also held a reference to the corresponding dentry or something along those
> > lines?
> In v1, this was done via
> ```
> #[pin_data]
> struct Values<T> {
>   dir: /* ignore this type */,
>   #[pin]
>   backing: T
> }
> ```
> Then, there was an interface that let the user provide a building
> function which had to have a fully polymorphic lifetime, which would
> be passed a backing reference that it was allowed to attach to
> subdirectory files. Since the dir would be cleaned up before the
> backing went away, we could know that it successfully outlived it.
> It'll probably look a little different when I send the follow-up
> series on top of this one.
> 
> Attaching to the root directory rather than each individual file made
> sense to me because this meant that if you had
> ```
> struct Foo {
>   prop_a: u32,
>   prop_b: u32
> }
> ```
> it would not be as tricky to attach `prop_a` to one file and `prop_b`
> to another, because the directory would own `Foo`.

Thanks for the explanation! I need to think a bit more about this approach.

> This'll probably be
> clearer when I send up a dependent series on top of v3 later today.

At least from my side, no need to rush. :) I'll have quite limited bandwidth for
the next ~10 days anyways.