All of lore.kernel.org
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Danilo Krummrich" <dakr@kernel.org>
Cc: linux-pci@vger.kernel.org, ojeda@kernel.org
Subject: Re: [PATCH v4 14/27] rust: usb: make Driver trait lifetime-parameterized
Date: Fri, 22 May 2026 00:23:56 +0000	[thread overview]
Message-ID: <20260522002357.01BA91F000E9@smtp.kernel.org> (raw)
In-Reply-To: <20260521233501.1191842-15-dakr@kernel.org>

Thank you for your contribution! Sashiko AI review found 4 potential issue(s) to consider:
- [Critical] The new `'bound` lifetime parameter incorrectly allows drivers to capture and store the dynamic `usb_device_id` pointer (`id: &'bound DeviceId`), which can be freed by userspace at any time, leading to a Use-After-Free.
- [Critical] The patch relies on unsound lifetime transmutation when storing and retrieving the lifetime-parameterized driver data via `drvdata`.
- [Low] The commit message incorrectly refers to the `unbind()` callback instead of `disconnect()`.
- [Low] Stale comments refer to `T::Data` without its new lifetime parameter.
--

commit ca8bfff774a157ce3cf21956820d617a046e6a08
Author: Danilo Krummrich <dakr@kernel.org>

rust: usb: make Driver trait lifetime-parameterized

Add a 'bound lifetime to the associated Data, changing type Data to type
Data<'bound>. This allows the driver's bus device private data to capture
the device / driver bound lifetime.

> The probe() and unbind() callbacks thus gain a 'bound lifetime parameter
> on the methods themselves; avoiding a global lifetime on the trait impl.

[Severity: Low]
The commit message states "unbind()", but the usb::Driver trait defines
a disconnect() method. Should this say disconnect() instead?

> diff --git a/rust/kernel/usb.rs b/rust/kernel/usb.rs
> index 1dbb8387b4636..616e22e34c6fd 100644
> --- a/rust/kernel/usb.rs
> +++ b/rust/kernel/usb.rs

[ ... ]

> @@ -110,7 +110,7 @@ extern "C" fn disconnect_callback(intf: *mut bindings::usb_interface) {
>          // SAFETY: `disconnect_callback` is only ever called after a successful call to
>          // `probe_callback`, hence it's guaranteed that `Device::set_drvdata()` has been called
>          // and stored a `Pin<KBox<T::Data>>`.

[Severity: Low]
This comment refers to T::Data without the new lifetime parameter.
Should it be updated to Pin<KBox<T::Data<'_>>>?

> -        let data = unsafe { dev.drvdata_borrow::<T::Data>() };
> +        let data = unsafe { dev.drvdata_borrow::<T::Data<'_>>() };

[Severity: Critical]
Does this create an unsound lifetime transmutation?

The driver's data T::Data<'a> is instantiated with an ephemeral lifetime 'a
in probe_callback and passed to dev.set_drvdata(), erasing it to a raw
pointer.

Here, it is retrieved using a new, disjoint ephemeral lifetime. If T::Data
is invariant over 'bound (for example, using Mutex<Cell<&'bound T>>), these
represent completely different types.

ForeignOwnable::borrow requires an exact type match. Violating this might
allow safe Rust code to write a short-lived reference into the driver data
during disconnect, which would then dangle when disconnect returns.

Could this lead to a use-after-free when the Drop implementation runs during
post_unbind_callback?

[ ... ]

> @@ -315,16 +319,19 @@ pub trait Driver {
>      ///
>      /// Called when a new USB interface is bound to this driver.
>      /// Implementers should attempt to initialize the interface here.
> -    fn probe(
> -        interface: &Interface<device::Core<'_>>,
> -        id: &DeviceId,
> -        id_info: &Self::IdInfo,
> -    ) -> impl PinInit<Self::Data, Error>;
> +    fn probe<'bound>(
> +        interface: &'bound Interface<device::Core<'_>>,
> +        id: &'bound DeviceId,
> +        id_info: &'bound Self::IdInfo,
> +    ) -> impl PinInit<Self::Data<'bound>, Error> + 'bound;

[Severity: Critical]
Can this lead to a use-after-free if userspace removes a dynamic ID?

The probe method receives id: &'bound DeviceId, allowing the driver to
capture this reference into its T::Data<'bound>.

However, this id pointer can come from a dynamically allocated struct
usb_dynid via usb_match_dynamic_id(). Userspace can trigger
remove_id_store() via sysfs, which immediately frees the usb_dynid
structure without unbinding existing devices.

If the driver's private data holds this reference, it seems it would be
left with a dangling pointer.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260521233501.1191842-1-dakr@kernel.org?part=14

  reply	other threads:[~2026-05-22  0:23 UTC|newest]

Thread overview: 63+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-05-21 23:34 [PATCH v4 00/27] rust: device: Higher-Ranked Lifetime Types for device drivers Danilo Krummrich
2026-05-21 23:34 ` [PATCH v4 01/27] rust: alloc: remove `'static` bound on `ForeignOwnable` Danilo Krummrich
2026-05-22  0:13   ` sashiko-bot
2026-05-21 23:34 ` [PATCH v4 02/27] rust: driver: move 'static bounds to constructor Danilo Krummrich
2026-05-21 23:34 ` [PATCH v4 03/27] rust: driver: decouple driver private data from driver type Danilo Krummrich
2026-05-21 23:56   ` sashiko-bot
2026-05-21 23:34 ` [PATCH v4 04/27] rust: driver core: drop drvdata before devres release Danilo Krummrich
2026-05-22  0:10   ` sashiko-bot
2026-05-21 23:34 ` [PATCH v4 05/27] rust: pci: implement Sync for Device<Bound> Danilo Krummrich
2026-05-21 23:34 ` [PATCH v4 06/27] rust: platform: " Danilo Krummrich
2026-05-21 23:34 ` [PATCH v4 07/27] rust: auxiliary: " Danilo Krummrich
2026-05-21 23:34 ` [PATCH v4 08/27] rust: usb: " Danilo Krummrich
2026-05-22  0:16   ` sashiko-bot
2026-05-21 23:34 ` [PATCH v4 09/27] rust: device: " Danilo Krummrich
2026-05-21 23:34 ` [PATCH v4 10/27] rust: device: make Core and CoreInternal lifetime-parameterized Danilo Krummrich
2026-05-25  4:21   ` Eliot Courtney
2026-05-25 11:02   ` Alexandre Courbot
2026-05-21 23:34 ` [PATCH v4 11/27] rust: pci: make Driver trait lifetime-parameterized Danilo Krummrich
2026-05-22  0:14   ` sashiko-bot
2026-05-21 23:34 ` [PATCH v4 12/27] rust: platform: " Danilo Krummrich
2026-05-21 23:34 ` [PATCH v4 13/27] rust: auxiliary: " Danilo Krummrich
2026-05-21 23:34 ` [PATCH v4 14/27] rust: usb: " Danilo Krummrich
2026-05-22  0:23   ` sashiko-bot [this message]
2026-05-25  4:31   ` Eliot Courtney
2026-05-21 23:34 ` [PATCH v4 15/27] rust: i2c: " Danilo Krummrich
2026-05-21 23:34 ` [PATCH v4 16/27] rust: driver: update module documentation for GAT-based Data type Danilo Krummrich
2026-05-21 23:34 ` [PATCH v4 17/27] rust: pci: make Bar lifetime-parameterized Danilo Krummrich
2026-05-22  0:49   ` sashiko-bot
2026-05-25  4:37   ` Eliot Courtney
2026-05-25 11:40     ` Gary Guo
2026-05-25 12:05       ` Danilo Krummrich
2026-05-25 11:10   ` Alexandre Courbot
2026-05-25 11:12     ` Alexandre Courbot
2026-05-21 23:34 ` [PATCH v4 18/27] rust: io: make IoMem and ExclusiveIoMem lifetime-parameterized Danilo Krummrich
2026-05-22  0:45   ` sashiko-bot
2026-05-25 13:10   ` Alexandre Courbot
2026-05-21 23:34 ` [PATCH v4 19/27] samples: rust: rust_driver_pci: use HRT lifetime for Bar Danilo Krummrich
2026-05-22  1:27   ` sashiko-bot
2026-05-25 13:55   ` Alexandre Courbot
2026-05-21 23:34 ` [PATCH v4 20/27] gpu: nova-core: separate driver type from driver data Danilo Krummrich
2026-05-25  4:40   ` Eliot Courtney
2026-05-25 14:11   ` Alexandre Courbot
2026-05-21 23:34 ` [PATCH v4 21/27] rust: types: add `ForLt` trait for higher-ranked lifetime support Danilo Krummrich
2026-05-22  0:31   ` sashiko-bot
2026-05-23 15:46   ` Danilo Krummrich
2026-05-25 12:31     ` Eliot Courtney
2026-05-21 23:34 ` [PATCH v4 22/27] rust: auxiliary: generalize Registration over ForLt Danilo Krummrich
2026-05-22  0:49   ` sashiko-bot
2026-05-25  6:03   ` Eliot Courtney
2026-05-25 14:42   ` Alexandre Courbot
2026-05-21 23:34 ` [PATCH v4 23/27] samples: rust: rust_driver_auxiliary: showcase lifetime-bound registration data Danilo Krummrich
2026-05-25 14:48   ` Alexandre Courbot
2026-05-21 23:34 ` [PATCH REF v4 24/27] gpu: nova-core: use lifetime for Bar Danilo Krummrich
2026-05-22  1:28   ` sashiko-bot
2026-05-26  2:10   ` Alexandre Courbot
2026-05-26  5:48     ` Alexandre Courbot
2026-05-21 23:34 ` [PATCH REF v4 25/27] gpu: nova-core: unregister sysmem flush page from Drop Danilo Krummrich
2026-05-22  0:47   ` sashiko-bot
2026-05-21 23:34 ` [PATCH REF v4 26/27] gpu: nova-core: replace ARef<Device> with &'bound Device in SysmemFlush Danilo Krummrich
2026-05-22  0:46   ` sashiko-bot
2026-05-21 23:34 ` [PATCH REF v4 27/27] gpu: drm: tyr: use lifetime for IoMem Danilo Krummrich
2026-05-22  0:42   ` sashiko-bot
2026-05-22 10:14 ` [PATCH v4 00/27] rust: device: Higher-Ranked Lifetime Types for device drivers Greg KH

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=20260522002357.01BA91F000E9@smtp.kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=dakr@kernel.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=ojeda@kernel.org \
    --cc=sashiko-reviews@lists.linux.dev \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.