Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Thomas Gleixner <tglx@linutronix.de>
To:     Catalin Marinas <catalin.marinas@arm.com>
Cc:     Andrei Vagin <avagin@gmail.com>, Will Deacon <will@kernel.org>,
        linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org,
        Vincenzo Frascino <vincenzo.frascino@arm.com>,
        Mark Rutland <mark.rutland@arm.com>,
        Dmitry Safonov <dima@arista.com>,
        Christian Brauner <christian.brauner@ubuntu.com>
Subject: Re: [PATCH v5 0/6] arm64: add the time namespace support
In-Reply-To: <20200724110131.GC23388@gaia>
References: <20200624083321.144975-1-avagin@gmail.com> <20200705064055.GA28894@gmail.com> <20200714015743.GA843937@gmail.com> <20200722181506.GA4517@gaia> <20200723174140.GA3991167@gmail.com> <87d04mvv0g.fsf@nanos.tec.linutronix.de> <20200724110131.GC23388@gaia>
Date:   Fri, 24 Jul 2020 15:22:34 +0200
Message-ID: <874kpxujxh.fsf@nanos.tec.linutronix.de>
MIME-Version: 1.0
Content-Type: text/plain
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Catalin Marinas <catalin.marinas@arm.com> writes:
> On Thu, Jul 23, 2020 at 10:25:35PM +0200, Thomas Gleixner wrote:
>> Andrei Vagin <avagin@gmail.com> writes:
>> > On Wed, Jul 22, 2020 at 07:15:06PM +0100, Catalin Marinas wrote:
>> >
>> > I don't think that we need to handle this case in the kernel. Users
>> > must understand what they are doing and have to write code so that avoid
>> > these sort of situations. In general, I would say that in most cases it
>> > is a bad idea to call setns from a signal handler.
>> 
>> This should not be supported in the first place and just let the
>> offender die right there.
>
> It would have been nice if we caught the offender easily but since
> signal handling doesn't have to be paired with sigreturn(), the kernel
> can't tell whether setns() is called in the wrong context. I guess we
> just have to live with this (maybe document the restriction in
> time_namespaces(7) or setns(2)).

Yes, I know that it's more or less impossible. The 'should' was just
wishful thinking :)

But yes, proper documentation is required.

Thanks,

        tglx