Received-SPF: pass (google.com: domain of linux-kernel+bounces-178706-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.48.161 as permitted sender) client-ip=147.75.48.161;
Feedback-ID: iad51458e:Fastmail
Date: Tue, 14 May 2024 06:12:53 -0700
From: Boqun Feng <boqun.feng@gmail.com>
To: Miguel Ojeda <miguel.ojeda.sandonis@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>, Miguel Ojeda <ojeda@kernel.org>,
	John Stultz <jstultz@google.com>, Stephen Boyd <sboyd@kernel.org>,
	Alex Gaynor <alex.gaynor@gmail.com>,
	Wedson Almeida Filho <wedsonaf@gmail.com>,
	Gary Guo <gary@garyguo.net>, bjorn3_gh@protonmail.com,
	Benno Lossin <benno.lossin@proton.me>,
	Andreas Hindborg <a.hindborg@samsung.com>,
	Alice Ryhl <aliceryhl@google.com>, rust-for-linux@vger.kernel.org,
	linux-kernel@vger.kernel.org, Kees Cook <keescook@chromium.org>,
	Mara Bos <m-ou.se@m-ou.se>, Amanieu d'Antras <amanieu@gmail.com>
Subject: Re: [PATCH 2/2] rust: time: Use wrapping_sub() for Ktime::sub()
Message-ID: <ZkNjVbqR6gYqg4YZ@boqun-archlinux>
References: <Zigj-lY5lnBSKuib@boqun-archlinux>
 <87cypvnpwn.ffs@tglx>
 <CANiq72nGHhgRjfMOhz=ss51q2egHYexPMVvWTr0RES9WAmMF=A@mail.gmail.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <CANiq72nGHhgRjfMOhz=ss51q2egHYexPMVvWTr0RES9WAmMF=A@mail.gmail.com>

On Mon, May 13, 2024 at 05:04:43PM +0200, Miguel Ojeda wrote:
> On Thu, May 9, 2024 at 2:14 PM Thomas Gleixner <tglx@linutronix.de> wrote:
> >
> > In principle ktime_sub() should not overflow for regular use cases.
> >
> > If the binder example overflows the substraction, then something is
> > seriously wrong. Though in that case as it's only for debug purposes
> > panicing would be totally counter productive. A warning might be
> > appropriate though.
> 
> Thanks for the clarification Thomas.
> 
> Alice and I pinged about this in the RustNL Unconf, i.e. about having
> a way to customize what happens on integer overflow (and, in general,
> other panics too), which I had in:
> 
>     https://github.com/Rust-for-Linux/linux/issues/354
> 
> e.g. like UBSan that allows a "report and continue" option.
> 
> We chatted with Mara and Amanieu, who were receptive. In particular,
> Mara posted right away:
> 
>     https://github.com/rust-lang/rfcs/pull/3632
> 
> as a possible first step to eventually have that ability, mentioning
> Rust for Linux as a user.
> 
> If we get the ability to customize those, then this could simply be a
> normal arithmetic operation, i.e. like any other. That is, considered
> to be a bug if it overflows.
> 

That is nice to have, thank you guys!

> Meanwhile, one alternative is going with Boqun's approach, but I would
> just use the standard operator (i.e. what we do elsewhere) and try to
> get the customization happen as soon as possible since we will need it
> for everything else (plus we avoid to make the code uglier with
> changes that we will need to revert anyway; and it also allows us to
> easily test any new customization feature from the compiler/library).
> 

However, I must point out that it needs more than a customized panic
handler to work: we also need to change the code generation (or adding
a different flag similar to -Coverflow-checks), because the current code
generation is Rust panic when overflow happens, which means the
subsequent code is unreachable.

Regards,
Boqun

> Cheers,
> Miguel