Re: [PATCH 7/7] xen/evtchn: read producer index only once

From: Julien Grall <julien@xen.org>
To: "Jürgen Groß" <jgross@suse.com>, "Jan Beulich" <jbeulich@suse.com>
Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com>,
	Stefano Stabellini <sstabellini@kernel.org>,
	linux-kernel@vger.kernel.org, xen-devel@lists.xenproject.org
Subject: Re: [PATCH 7/7] xen/evtchn: read producer index only once
Date: Mon, 8 Feb 2021 11:50:58 +0000	[thread overview]
Message-ID: <42e15cc4-56d1-b34b-d97e-d579e771788a@xen.org> (raw)
In-Reply-To: <d0ca217c-ecc9-55f7-abb1-30a687a46b31@suse.com>

On 08/02/2021 10:59, Jürgen Groß wrote:
> On 08.02.21 11:51, Jan Beulich wrote:
> Yes, but I don't see an urgent need to fix that, as there would
> be thousands of accesses in the kernel needing a fix. A compiler
> tearing a naturally aligned access into multiple memory accesses
> would be rejected as buggy from the kernel community IMO.

I would not be so sure. From lwn [1]:

"In the Linux kernel, tearing of plain C-language loads has been 
observed even given properly aligned and machine-word-sized loads.)"

And for store tearing:

"Note that this tearing can happen even on properly aligned and 
machine-word-sized accesses, and in this particular case, even for 
volatile stores. Some might argue that this behavior constitutes a bug 
in the compiler, but either way it illustrates the perceived value of 
store tearing from a compiler-writer viewpoint. [...] But for properly 
aligned machine-sized stores, WRITE_ONCE() will prevent store tearing."

Cheers,

[1] https://lwn.net/Articles/793253/#Load%20Tearing

> 
> 
> Juergen

-- 
Julien Grall