Received-SPF: pass (google.com: domain of linux-kernel+bounces-110092-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Precedence: bulk
MIME-Version: 1.0
References: <20240209-alice-file-v5-0-a37886783025@google.com>
 <20240209-alice-file-v5-8-a37886783025@google.com> <20240320-massieren-lackschaden-9b30825babec@brauner>
In-Reply-To: <20240320-massieren-lackschaden-9b30825babec@brauner>
From: Alice Ryhl <aliceryhl@google.com>
Date: Thu, 21 Mar 2024 14:28:15 +0100
Message-ID: <CAH5fLgjeet1nhycCqdaKE9CSY02XW85jn302zNEjJNQaJ1czGQ@mail.gmail.com>
Subject: Re: [PATCH v5 8/9] rust: file: add `DeferredFdCloser`
To: Christian Brauner <brauner@kernel.org>
Cc: Miguel Ojeda <ojeda@kernel.org>, Alex Gaynor <alex.gaynor@gmail.com>, 
	Wedson Almeida Filho <wedsonaf@gmail.com>, Boqun Feng <boqun.feng@gmail.com>, Gary Guo <gary@garyguo.net>, 
	=?UTF-8?Q?Bj=C3=B6rn_Roy_Baron?= <bjorn3_gh@protonmail.com>, 
	Benno Lossin <benno.lossin@proton.me>, Andreas Hindborg <a.hindborg@samsung.com>, 
	Peter Zijlstra <peterz@infradead.org>, Alexander Viro <viro@zeniv.linux.org.uk>, 
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>, =?UTF-8?B?QXJ2ZSBIasO4bm5ldsOlZw==?= <arve@android.com>, 
	Todd Kjos <tkjos@android.com>, Martijn Coenen <maco@android.com>, 
	Joel Fernandes <joel@joelfernandes.org>, Carlos Llamas <cmllamas@google.com>, 
	Suren Baghdasaryan <surenb@google.com>, Dan Williams <dan.j.williams@intel.com>, 
	Kees Cook <keescook@chromium.org>, Matthew Wilcox <willy@infradead.org>, 
	Thomas Gleixner <tglx@linutronix.de>, Daniel Xu <dxu@dxuuu.xyz>, linux-kernel@vger.kernel.org, 
	rust-for-linux@vger.kernel.org, linux-fsdevel@vger.kernel.org, 
	Martin Rodriguez Reboredo <yakoyoku@gmail.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Wed, Mar 20, 2024 at 3:22=E2=80=AFPM Christian Brauner <brauner@kernel.o=
rg> wrote:
>
> On Fri, Feb 09, 2024 at 11:18:21AM +0000, Alice Ryhl wrote:
> >
> > +/// Helper used for closing file descriptors in a way that is safe eve=
n if the file is currently
> > +/// held using `fdget`.
> > +///
> > +/// Additional motivation can be found in commit 80cd795630d6 ("binder=
: fix use-after-free due to
> > +/// ksys_close() during fdget()") and in the comments on `binder_do_fd=
_close`.
> > +pub struct DeferredFdCloser {
> > +    inner: Box<DeferredFdCloserInner>,
> > +}
> > +
> > +/// SAFETY: This just holds an allocation with no real content, so the=
re's no safety issue with
> > +/// moving it across threads.
> > +unsafe impl Send for DeferredFdCloser {}
> > +unsafe impl Sync for DeferredFdCloser {}
> > +
> > +/// # Invariants
> > +///
> > +/// If the `file` pointer is non-null, then it points at a `struct fil=
e` and owns a refcount to
> > +/// that file.
> > +#[repr(C)]
> > +struct DeferredFdCloserInner {
> > +    twork: mem::MaybeUninit<bindings::callback_head>,
> > +    file: *mut bindings::file,
> > +}
> > +
> > +impl DeferredFdCloser {
>
> So the explicitly deferred close is due to how binder works so it's not
> much of a general purpose interface as I don't recall having other
> codepaths with similar problems. So this should live in the binder
> specific rust code imo.

Hmm. Are there really no other ioctls that call ksys_close on a
user-provided fd?

As far as I can tell, this kind of deferred API is the only way for us
to provide a fully safe Rust api for closing an fd. Directly calling
ksys_close must unsafely assert that the fd does not have an active
fdget call. So it makes sense to me as an API that others might want
to use.

Still I can move it elsewhere if necessary.

Alice