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 v2 0/2] Replace and improve "mcsafe" with copy_safe()
Date:   Thu, 30 Apr 2020 17:23:31 -0700
Message-Id: <1962EE67-8AD1-409D-963A-4F1E1AB3B865@amacapital.net>
References: <CAHk-=wiMs=A90np0Hv5WjHY8HXQWpgtuq-xrrJvyk7_pNB4meg@mail.gmail.com>
Cc:     Dan Williams <dan.j.williams@intel.com>,
        "Luck, Tony" <tony.luck@intel.com>,
        Andy Lutomirski <luto@kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Borislav Petkov <bp@alien8.de>,
        stable <stable@vger.kernel.org>,
        the arch/x86 maintainers <x86@kernel.org>,
        "H. Peter Anvin" <hpa@zytor.com>,
        Paul Mackerras <paulus@samba.org>,
        Benjamin Herrenschmidt <benh@kernel.crashing.org>,
        Erwin Tsaur <erwin.tsaur@intel.com>,
        Michael Ellerman <mpe@ellerman.id.au>,
        Arnaldo Carvalho de Melo <acme@kernel.org>,
        linux-nvdimm <linux-nvdimm@lists.01.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
In-Reply-To: <CAHk-=wiMs=A90np0Hv5WjHY8HXQWpgtuq-xrrJvyk7_pNB4meg@mail.gmail.com>
To:     Linus Torvalds <torvalds@linux-foundation.org>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


> On Apr 30, 2020, at 5:10 PM, Linus Torvalds <torvalds@linux-foundation.org=
> wrote:
>=20
> =EF=BB=BFOn Thu, Apr 30, 2020 at 4:52 PM Dan Williams <dan.j.williams@inte=
l.com> wrote:
>>=20
>> You had me until here. Up to this point I was grokking that Andy's
>> "_fallible" suggestion does help explain better than "_safe", because
>> the copy is doing extra safety checks. copy_to_user() and
>> copy_to_user_fallible() mean *something* where copy_to_user_safe()
>> does not.
>=20
> It's a horrible word, btw. The word doesn't actually mean what Andy
> means it to mean. "fallible" means "can make mistakes", not "can
> fault".
>=20
> So "fallible" is a horrible name.

What I was trying to get at was not =E2=80=9Ccan fault=E2=80=9D but =E2=80=9C=
can malfunction=E2=80=9D.  Maybe =E2=80=9Cunreliable=E2=80=9D?  Better words=
 welcome.

>=20
> But anyway, I don't hate something like "copy_to_user_fallible()"
> conceptually. The naming needs to be fixed, in that "user" can always
> take a fault, so it's the _source_ that can fault, not the "user"
> part.

I don=E2=80=99t like this.  =E2=80=9Cuser=E2=80=9D already implied that basi=
cally anything can be wrong with the memory =E2=80=94 it can be unmapped ent=
irely, it can have the wrong permissions, it can have the wrong protection k=
ey, it can have an ECC error, etc.  If the operation you want is =E2=80=9Cco=
py from unreliable kernel memory (but with a definitely valid pointer) to us=
er memory=E2=80=9D, you want copy_unreliable_to_user().

Now maybe copy_to_user() should *always* work this way, but I=E2=80=99m not c=
onvinced. Certainly put_user() shouldn=E2=80=99t =E2=80=94 the result wouldn=
=E2=80=99t even be well defined. And I=E2=80=99m unconvinced that it makes m=
uch sense for the majority of copy_to_user() callers that are also directly a=
ccessing the source structure.

I also tend to think that the probe_kernel stuff should just stay separate. T=
hose are really for two totally separate types of use: either the kernel is t=
rying its best to print an errr message without exploding worse, or it=E2=80=
=99s involved in eBPF or trading hacks in which address is arbitrary and ess=
entially untrusted.