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;
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
From:   Andy Lutomirski <luto@amacapital.net>
Mime-Version: 1.0 (1.0)
Subject: Re: [PATCH v3] x86/fault: Send a SIGBUS to user process always for hwpoison page access.
Date:   Mon, 8 Mar 2021 11:00:28 -0800
Message-Id: <047D5B49-FDBB-494C-81E9-DA811476747D@amacapital.net>
References: <4fc1b4e8f1fb4c8c81f280db09178797@intel.com>
Cc:     Aili Yao <yaoaili@kingsoft.com>, Andy Lutomirski <luto@kernel.org>,
        HORIGUCHI NAOYA <naoya.horiguchi@nec.com>,
        Dave Hansen <dave.hansen@linux.intel.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        "H. Peter Anvin" <hpa@zytor.com>, X86 ML <x86@kernel.org>,
        yangfeng1@kingsoft.com, Linux-MM <linux-mm@kvack.org>,
        LKML <linux-kernel@vger.kernel.org>
In-Reply-To: <4fc1b4e8f1fb4c8c81f280db09178797@intel.com>
To:     "Luck, Tony" <tony.luck@intel.com>,
        Oleg Nesterov <oleg@redhat.com>,
        Linux API <linux-api@vger.kernel.org>
Precedence: bulk


> On Mar 8, 2021, at 10:31 AM, Luck, Tony <tony.luck@intel.com> wrote:
>=20
> =EF=BB=BF
>>=20
>> Can you point me at that SIGBUS code in a current kernel?
>=20
> It is in kill_me_maybe().  mce_vaddr is setup when we disassemble whatever=
 get_user()
> or copy from user variant was in use in the kernel when the poison memory w=
as consumed.
>=20
>        if (p->mce_vaddr !=3D (void __user *)-1l) {
>                force_sig_mceerr(BUS_MCEERR_AR, p->mce_vaddr, PAGE_SHIFT);

Hmm. On the one hand, no one has complained yet. On the other hand, hardware=
 that supports this isn=E2=80=99t exactly common.

We may need some actual ABI design here. We also need to make sure that thin=
gs like io_uring accesses or, more generally, anything using the use_mm / us=
e_temporary_mm ends up either sending no signal or sending a signal to the r=
ight target.

>=20
> Would it be any better if we used the BUS_MCEERR_AO code that goes into si=
ginfo?

Dunno.

>=20
> That would make it match up better with what happens when poison is found
> asynchronously by the patrol scrubber. I.e. the semantics are:
>=20
> AR: You just touched poison at this address and need to do something about=
 that.
> AO: Just letting you know that you have some poison at the address in sigi=
nfo.
>=20
> -Tony