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;
MIME-Version: 1.0
References: <a770d45d-b147-a8c5-b7f8-30d668cbed84@intel.com>
 <4BDFD364-798C-4537-A88E-F94F101F524B@amacapital.net> <CAMe9rOoTjSwRSPuqP6RKkDzPA_VPh5gVYRVFJ-ezAD4Et-FUng@mail.gmail.com>
 <CALCETrW=-ahC7GUCCyX7nPjCHfG3tiyDespud2Z7UbB6yWWWAA@mail.gmail.com>
In-Reply-To: <CALCETrW=-ahC7GUCCyX7nPjCHfG3tiyDespud2Z7UbB6yWWWAA@mail.gmail.com>
From:   "H.J. Lu" <hjl.tools@gmail.com>
Date:   Thu, 27 Aug 2020 18:44:27 -0700
Message-ID: <CAMe9rOrt5hz6qsNAxPgdKCOhRcKKESv-D3rxdSfraeJ-LFHM4w@mail.gmail.com>
Subject: Re: [PATCH v11 25/25] x86/cet/shstk: Add arch_prctl functions for
 shadow stack
To:     Andy Lutomirski <luto@kernel.org>
Cc:     "Yu, Yu-cheng" <yu-cheng.yu@intel.com>,
        Florian Weimer <fweimer@redhat.com>,
        Dave Martin <Dave.Martin@arm.com>,
        Dave Hansen <dave.hansen@intel.com>, X86 ML <x86@kernel.org>,
        "H. Peter Anvin" <hpa@zytor.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>,
        LKML <linux-kernel@vger.kernel.org>,
        "open list:DOCUMENTATION" <linux-doc@vger.kernel.org>,
        Linux-MM <linux-mm@kvack.org>,
        linux-arch <linux-arch@vger.kernel.org>,
        Linux API <linux-api@vger.kernel.org>,
        Arnd Bergmann <arnd@arndb.de>,
        Balbir Singh <bsingharora@gmail.com>,
        Borislav Petkov <bp@alien8.de>,
        Cyrill Gorcunov <gorcunov@gmail.com>,
        Dave Hansen <dave.hansen@linux.intel.com>,
        Eugene Syromiatnikov <esyr@redhat.com>,
        Jann Horn <jannh@google.com>, Jonathan Corbet <corbet@lwn.net>,
        Kees Cook <keescook@chromium.org>,
        Mike Kravetz <mike.kravetz@oracle.com>,
        Nadav Amit <nadav.amit@gmail.com>,
        Oleg Nesterov <oleg@redhat.com>, Pavel Machek <pavel@ucw.cz>,
        Peter Zijlstra <peterz@infradead.org>,
        Randy Dunlap <rdunlap@infradead.org>,
        "Ravi V. Shankar" <ravi.v.shankar@intel.com>,
        Vedvyas Shanbhogue <vedvyas.shanbhogue@intel.com>,
        Weijiang Yang <weijiang.yang@intel.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, Aug 27, 2020 at 6:35 PM Andy Lutomirski <luto@kernel.org> wrote:
>
> On Thu, Aug 27, 2020 at 12:38 PM H.J. Lu <hjl.tools@gmail.com> wrote:
> >
> > On Thu, Aug 27, 2020 at 11:56 AM Andy Lutomirski <luto@amacapital.net> =
wrote:
> > >
> > >
> > >
> > > > On Aug 27, 2020, at 11:13 AM, Yu, Yu-cheng <yu-cheng.yu@intel.com> =
wrote:
> > > >
> > > > =EF=BB=BFOn 8/27/2020 6:36 AM, Florian Weimer wrote:
> > > >> * H. J. Lu:
> > > >>>> On Thu, Aug 27, 2020 at 6:19 AM Florian Weimer <fweimer@redhat.c=
om> wrote:
> > > >>>>>
> > > >>>>> * Dave Martin:
> > > >>>>>
> > > >>>>>> You're right that this has implications: for i386, libc probab=
ly pulls
> > > >>>>>> more arguments off the stack than are really there in some sit=
uations.
> > > >>>>>> This isn't a new problem though.  There are already generic pr=
ctls with
> > > >>>>>> fewer than 4 args that are used on x86.
> > > >>>>>
> > > >>>>> As originally posted, glibc prctl would have to know that it ha=
s to pull
> > > >>>>> an u64 argument off the argument list for ARCH_X86_CET_DISABLE.=
  But
> > > >>>>> then the u64 argument is a problem for arch_prctl as well.
> > > >>>>>
> > > >>>
> > > >>> Argument of ARCH_X86_CET_DISABLE is int and passed in register.
> > > >> The commit message and the C source say otherwise, I think (not su=
re
> > > >> about the C source, not a kernel hacker).
> > > >
> > > > H.J. Lu suggested that we fix x86 arch_prctl() to take four argumen=
ts, and then keep MMAP_SHSTK as an arch_prctl().  Because now the map flags=
 and size are all in registers, this also solves problems being pointed out=
 earlier.  Without a wrapper, the shadow stack mmap call (from user space) =
will be:
> > > >
> > > > syscall(_NR_arch_prctl, ARCH_X86_CET_MMAP_SHSTK, size, MAP_32BIT).
> > >
> > > I admit I don=E2=80=99t see a show stopping technical reason we can=
=E2=80=99t add arguments to an existing syscall, but I=E2=80=99m pretty sur=
e it=E2=80=99s unprecedented, and it doesn=E2=80=99t seem like a good idea.
> >
> > prctl prototype is:
> >
> > extern int prctl (int __option, ...)
> >
> > and implemented in kernel as:
> >
> >       int prctl(int option, unsigned long arg2, unsigned long arg3,
> >                  unsigned long arg4, unsigned long arg5);
> >
> > Not all prctl operations take all 5 arguments.   It also applies
> > to arch_prctl.  It is quite normal for different operations of
> > arch_prctl to take different numbers of arguments.
>
> If by "quite normal" you mean "does not happen", then I agree.
>
> In any event, I will not have anything to do with a patch that changes
> an existing syscall signature unless Linus personally acks it.  So if
> you want to email him and linux-abi, be my guest.

Can you think of ANY issues of passing more arguments to arch_prctl?
syscall () provided by glibc always passes 6 arguments to the kernel.
Arguments are already in the registers.  What kind of problems do
you see?

--=20
H.J.