Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 03 Nov 2022 09:38:10 +0000
To:     Dennis Dai <dzy.0424thu@gmail.com>
From:   =?utf-8?Q?Bj=C3=B6rn_Roy_Baron?= <bjorn3_gh@protonmail.com>
Cc:     Miguel Ojeda <ojeda@kernel.org>,
        Alex Gaynor <alex.gaynor@gmail.com>,
        Wedson Almeida Filho <wedsonaf@google.com>,
        Boqun Feng <boqun.feng@gmail.com>, Gary Guo <gary@garyguo.net>,
        rust-for-linux@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: rust nvme driver: potential sleep-in-atomic-context
Message-ID: <nyfz0H8A-5nGCgsAynqZyYtFlEcvPHDecY6c8MC840zNNdEWz2NSPPuZgE_VS7RnrTwFhd6aCDi2TXWh6YtqfUeUEWZvFCT7HjpO-Z_UPCk=@protonmail.com>
In-Reply-To: <CACMswuMVERF_+R3Qxb0xGHCBQUCuxtBLA+eDo__Coed3BoVC3A@mail.gmail.com>
References: <CACMswuMVERF_+R3Qxb0xGHCBQUCuxtBLA+eDo__Coed3BoVC3A@mail.gmail.com>
Feedback-ID: 27884398:user:proton
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Thursday, November 3rd, 2022 at 07:12, Dennis Dai <dzy.0424thu@gmail.com=
> wrote:


> The rust nvme driver [1] (which is still pending to be merged into
> mainline [2]) has a potential sleep-in-atomic-context bug.
>=20
> The potential buggy code is below
>=20
> // drivers/block/nvme.rs:192
> dev.queues.lock().io.try_reserve(nr_io_queues as _)?;
> // drivers/block/nvme.rs:227
> dev.queues.lock().io.try_push(io_queue.clone())?;
>=20
> The queues field is wrapped in SpinLock, which means that we cannot
> sleep (or indirectly call any function that may sleep) when the lock
> is held.
> However try_reserve function may indirectly call krealloc with a
> sleepable flag GFP_KERNEL (that's default behaviour of the global rust
> allocator).
> The the case is similar for try_push.
>=20
> I wonder if the bug could be confirmed.
>=20
>=20
> [1] https://github.com/metaspace/rust-linux/commit/d88c3744d6cbdf11767e08=
bad56cbfb67c4c96d0
> [2] https://lore.kernel.org/lkml/202210010816.1317F2C@keescook/

setup_io_queues is only called by dev_add which in turn is only called Nvme=
Device::probe. This last function is responsible for creating the &Ref<Devi=
ceData> that ends up being passed to setup_io_queues. It doesn't seem like =
any reference is passed to another thread between &Ref<DeviceData>. As such=
 no other thread can block on the current thread due to holding the lock. A=
s far as I understand this means that sleeping while the lock is held is ha=
rmless. I think it would be possible to replace the &Ref<DeviceData> argume=
nt with an Pin<&mut DeviceData> argument by moving the add_dev call to befo=
re Ref::<DeviceData>::from(data). This would make it clear that only the cu=
rrent thread holds a reference and would also allow using a method like get=
_mut [1] to get a reference to the protected data without actually locking =
the spinlock as it is statically enforced that nobody can else can hold the=
 lock. It seems that get_mut is missing from all of the locks offered in th=
e kernel crate. I opened an issue for this. [2]

[1]: https://doc.rust-lang.org/stable/std/sync/struct.Mutex.html#method.get=
_mut
[2]: https://github.com/Rust-for-Linux/linux/issues/924

Cheers,
Bj=C3=B6rn