Received-SPF: pass (google.com: domain of linux-kernel+bounces-164486-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
From: Thomas Gleixner <tglx@linutronix.de>
To: Benno Lossin <benno.lossin@proton.me>, Andreas Hindborg
 <nmi@metaspace.dk>, Miguel Ojeda <ojeda@kernel.org>, Alex Gaynor
 <alex.gaynor@gmail.com>, Wedson Almeida Filho <wedsonaf@gmail.com>,
 Anna-Maria Behnsen <anna-maria@linutronix.de>, Frederic Weisbecker
 <frederic@kernel.org>
Cc: Andreas Hindborg <a.hindborg@samsung.com>, Boqun Feng
 <boqun.feng@gmail.com>, Gary Guo <gary@garyguo.net>, =?utf-8?Q?Bj=C3=B6rn?=
 Roy Baron
 <bjorn3_gh@protonmail.com>, Alice Ryhl <aliceryhl@google.com>,
 rust-for-linux@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] rust: hrtimer: introduce hrtimer support
In-Reply-To: <a7a560c7-fb8c-4adf-9f46-2e272f24b335@proton.me>
References: <20240425094634.262674-1-nmi@metaspace.dk>
 <a7a560c7-fb8c-4adf-9f46-2e272f24b335@proton.me>
Date: Tue, 30 Apr 2024 19:05:37 +0200
Message-ID: <87o79qsrvi.ffs@tglx>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain

On Fri, Apr 26 2024 at 07:52, Benno Lossin wrote:
> On 25.04.24 11:46, Andreas Hindborg wrote:
>> +#[pinned_drop]
>> +impl<T> PinnedDrop for Timer<T> {
>> +    fn drop(self: Pin<&mut Self>) {
>> +        // SAFETY: By struct invariant `self.timer` was initialized by
>> +        // `hrtimer_init` so by C API contract it is safe to call
>> +        // `hrtimer_cancel`.
>> +        unsafe {
>> +            bindings::hrtimer_cancel(self.timer.get());
>> +        }
>> +    }
>> +}
>
> Why is this needed? The only way to schedule a timer using this API is
> by having an `Arc` with a timer-containing struct inside. But to
> schedule the `Arc`, you consume one refcount which is then sent to the
> timer subsystem. So it is impossible for the refcount to drop below zero
> while the timer is scheduled, but not yet running.
> Do you need to call `hrtimer_cancel` after/while a timer is running?
>
> Also is it ok to call `hrtimer_cancel` inside the timer callback? Since
> that can happen when the timer callback owns the last refcount.

You cannot invoke hrtimer_cancel() from within the callback. That
deadlocks because hrtimer_cancel() waits for the callback to complete.

Thanks,

        tglx