From: Beata Michalska <beata.michalska@arm.com>
To: Danilo Krummrich <dakr@kernel.org>
Cc: ojeda@kernel.org, alex.gaynor@gmail.com, aliceryhl@google.com,
daniel.almeida@collabora.com, boqun.feng@gmail.com,
gary@garyguo.net, bjorn3_gh@protonmail.com, lossin@kernel.org,
a.hindborg@kernel.org, tmgross@umich.edu, alyssa@rosenzweig.io,
lyude@redhat.com, rust-for-linux@vger.kernel.org,
dri-devel@lists.freedesktop.org
Subject: Re: [PATCH] rust: drm: Drop the use of Opaque for ioctl arguments
Date: Fri, 20 Jun 2025 14:17:36 +0200 [thread overview]
Message-ID: <aFVQxX70wCZv7A82@arm.com> (raw)
In-Reply-To: <aFPsjIFIC1J2qtdc@pollux>
On Thu, Jun 19, 2025 at 12:55:08PM +0200, Danilo Krummrich wrote:
> On Thu, Jun 19, 2025 at 12:21:02PM +0200, Beata Michalska wrote:
> > With the Opaque<T>, the expectations are that Rust should not make any
> > assumptions on the layout or invariants of the wrapped C types.
> > That runs rather counter to ioctl arguments, which must adhere to
> > certain data-layout constraints. By using Opaque<T>, ioctl handlers
> > end up doing unsound castings,
>
> Which unsound casts? Please see [1] and [2] for how nova implements those IOCTL
> handlers.
>
That's a bad warding on my side.
> Speaking of which, this patch breaks the build, since it doesn't adjust the
> users of the API, i.e. nova.
>
True, totally slipped my mind that there are already users of those.
Apologies for that.
> If you want I can post a diff to fix up nova accordingly for you to add to this
> patch.
I think you already did - so thank you.
>
> [1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/nova/uapi.rs
> [2] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/nova/file.rs
>
> > which adds needless complexity and
> > maintenance overhead, brining no safety benefits.
> > Drop the use of Opaque for ioctl arguments as that is not the best
> > fit here.
> >
> > Signed-off-by: Beata Michalska <beata.michalska@arm.com>
> > ---
> >
> > Additional comments:
> > - UAPI types already automatically derive MaybeZeroable,
> > so it probably makes little sense to add any verification for that
> > - FromBytes is pending, but due to the orphan rule, adding verification
> > of it being implemented for IOCTL args here is pointless
> > - Verifying pointer alignment could make use of strict_provenance,
> > but that one is unstable and I am not sure what are the exact rules
> > here for using those. Without that one though, verifying alignment in
> > some cases (i.e. pointer tagging) might require more extensive changes.
> > Happy to deal with either.
> >
> > rust/kernel/drm/ioctl.rs | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/rust/kernel/drm/ioctl.rs b/rust/kernel/drm/ioctl.rs
> > index 445639404fb7..12b296131672 100644
> > --- a/rust/kernel/drm/ioctl.rs
> > +++ b/rust/kernel/drm/ioctl.rs
> > @@ -139,7 +139,7 @@ pub mod internal {
> > // asserted above matches the size of this type, and all bit patterns of
> > // UAPI structs must be valid.
> > let data = unsafe {
> > - &*(raw_data as *const $crate::types::Opaque<$crate::uapi::$struct>)
> > + &mut *(raw_data as *mut $crate::uapi::$struct)
>
> I think we have to document the guarantees we rely on to create this mutable
> reference.
>
Will do.
---
BR
Beata
> > };
>
> This should be formatted as one single line and also adjust the doc-comment of
> the macro accordingly, i.e.:
>
> diff --git a/rust/kernel/drm/ioctl.rs b/rust/kernel/drm/ioctl.rs
> index 12b296131672..f0c599f15a41 100644
> --- a/rust/kernel/drm/ioctl.rs
> +++ b/rust/kernel/drm/ioctl.rs
> @@ -83,7 +83,7 @@ pub mod internal {
> ///
> /// ```ignore
> /// fn foo(device: &kernel::drm::Device<Self>,
> -/// data: &Opaque<uapi::argument_type>,
> +/// data: &mut uapi::argument_type,
> /// file: &kernel::drm::File<Self::File>,
> /// ) -> Result<u32>
> /// ```
> @@ -138,9 +138,7 @@ macro_rules! declare_drm_ioctls {
> // SAFETY: The ioctl argument has size `_IOC_SIZE(cmd)`, which we
> // asserted above matches the size of this type, and all bit patterns of
> // UAPI structs must be valid.
> - let data = unsafe {
> - &mut *(raw_data as *mut $crate::uapi::$struct)
> - };
> + let data = unsafe { &mut *(raw_data as *mut $crate::uapi::$struct) };
> // SAFETY: This is just the DRM file structure
> let file = unsafe { $crate::drm::File::as_ref(raw_file) };
next prev parent reply other threads:[~2025-06-20 12:17 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-06-19 10:21 [PATCH] rust: drm: Drop the use of Opaque for ioctl arguments Beata Michalska
2025-06-19 10:55 ` Danilo Krummrich
2025-06-19 11:01 ` Benno Lossin
2025-06-19 12:26 ` Daniel Almeida
2025-06-19 12:31 ` Daniel Almeida
2025-06-19 13:17 ` Benno Lossin
2025-06-19 18:51 ` Boqun Feng
2025-06-20 12:25 ` Beata Michalska
2025-06-20 13:42 ` Daniel Almeida
2025-06-23 8:14 ` Beata Michalska
2025-06-19 12:34 ` Alice Ryhl
2025-06-19 13:01 ` Danilo Krummrich
2025-06-20 12:17 ` Beata Michalska [this message]
2025-06-19 12:30 ` Daniel Almeida
2025-06-20 12:17 ` Beata Michalska
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=aFVQxX70wCZv7A82@arm.com \
--to=beata.michalska@arm.com \
--cc=a.hindborg@kernel.org \
--cc=alex.gaynor@gmail.com \
--cc=aliceryhl@google.com \
--cc=alyssa@rosenzweig.io \
--cc=bjorn3_gh@protonmail.com \
--cc=boqun.feng@gmail.com \
--cc=dakr@kernel.org \
--cc=daniel.almeida@collabora.com \
--cc=dri-devel@lists.freedesktop.org \
--cc=gary@garyguo.net \
--cc=lossin@kernel.org \
--cc=lyude@redhat.com \
--cc=ojeda@kernel.org \
--cc=rust-for-linux@vger.kernel.org \
--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;
as well as URLs for NNTP newsgroup(s).