MIME-Version: 1.0
References: <20211129145732.27000-1-wangkefeng.wang@huawei.com>
 <YadiUPpJ0gADbiHQ@FVFF77S0Q05N> <811af0bc-0c99-37f6-a39a-095418b10661@huawei.com>
 <Yaic31SbYFJ4zAl0@elver.google.com> <b641f1ea-6def-0fe4-d273-03c35c4aa7d6@huawei.com>
In-Reply-To: <b641f1ea-6def-0fe4-d273-03c35c4aa7d6@huawei.com>
From:   Marco Elver <elver@google.com>
Date:   Thu, 2 Dec 2021 11:58:20 +0100
Message-ID: <CANpmjNOOVQjoczanQndUoDRMCOfV4FB6bnezEjskKh3iDzaQVg@mail.gmail.com>
Subject: Re: [PATCH v2] arm64: Enable KCSAN
To:     Kefeng Wang <wangkefeng.wang@huawei.com>
Cc:     Mark Rutland <mark.rutland@arm.com>,
        Catalin Marinas <catalin.marinas@arm.com>,
        Will Deacon <will@kernel.org>,
        linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org,
        joey.gouly@arm.com
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Thu, 2 Dec 2021 at 11:45, Kefeng Wang <wangkefeng.wang@huawei.com> wrote:
[...]
> >> for now and more explanation into changlog.
> > So what I gather arm64's final select line may look like:
> >
> >       select HAVE_ARCH_KCSAN if EXPERT && (CC_IS_GCC || CLANG_VERSION >= 120000)
> Yes,  that's what we want now.

I think we have all the pieces sorted out, so hopefully you have
everything you need to do v3.

[...]
> > I actually have a "fix" for the data race in rwsem_spin_on_owner, that
> > also shows where the other racing access comes from... which reminds me:
> > https://lkml.kernel.org/r/20211202101238.33546-1-elver@google.com
>
> There's a owner_on_cpu(),  we could reuse it,

That looks like a reasonable thing to do, but can't say if it is
actually the right thing to do. You could suggest it in reply to the
patch I just sent and see what people say.

Thanks,
-- Marco