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: pt_regs->ax == -ENOSYS
Date:   Tue, 27 Apr 2021 14:28:29 -0700
Message-Id: <F9F5E9D4-C1EE-455A-A6B1-4DF9D349BBAA@amacapital.net>
References: <f0240e15-223a-7600-4494-7a0a75155bdb@zytor.com>
Cc:     Linus Torvalds <torvalds@linux-foundation.org>,
        Ingo Molnar <mingo@kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Andrew Lutomirski <luto@kernel.org>,
        Borislav Petkov <bp@alien8.de>, linux-kernel@vger.kernel.org,
        oleg@redhat.com, Kees Cook <keescook@chromium.org>,
        Will Drewry <wad@chromium.org>
In-Reply-To: <f0240e15-223a-7600-4494-7a0a75155bdb@zytor.com>
To:     "H. Peter Anvin" <hpa@zytor.com>
Precedence: bulk


> On Apr 27, 2021, at 2:15 PM, H. Peter Anvin <hpa@zytor.com> wrote:
>=20
> =EF=BB=BFTrying to stomp out some possible cargo cult programming?
>=20
> In the process of going through the various entry code paths, I have to ad=
mit to being a bit confused why pt_regs->ax is set to -ENOSYS very early in t=
he system call path.
>=20

It has to get set to _something_, and copying orig_ax seems perhaps silly.  T=
here could also be code that relies on ptrace poking -1 into the nr resultin=
g in -ENOSYS.

> What is perhaps even more confusing is:
>=20
> __visible noinstr void do_syscall_64(struct pt_regs *regs, unsigned long n=
r)
> {
>        nr =3D syscall_enter_from_user_mode(regs, nr);
>=20
>        instrumentation_begin();
>        if (likely(nr < NR_syscalls)) {
>                nr =3D array_index_nospec(nr, NR_syscalls);
>                regs->ax =3D sys_call_table[nr](regs);
> #ifdef CONFIG_X86_X32_ABI
>        } else if (likely((nr & __X32_SYSCALL_BIT) &&
>                          (nr & ~__X32_SYSCALL_BIT) < X32_NR_syscalls)) {
>                nr =3D array_index_nospec(nr & ~__X32_SYSCALL_BIT,
>                                        X32_NR_syscalls);
>                regs->ax =3D x32_sys_call_table[nr](regs);
> #endif
>        }
>        instrumentation_end();
>        syscall_exit_to_user_mode(regs);
> }
> #endif
>=20
> Now, unless I'm completely out to sea, it seems to me that if syscall_ente=
r_from_user_mode() changes the system call number to an invalid number and p=
t_regs->ax to !-ENOSYS then the system call will return a different value(!)=
 depending on if it is out of range for the table (whatever was poked into p=
t_regs->ax) or if it corresponds to a hole in the table. This seems to me at=
 least to be The Wrong Thing.

I think you=E2=80=99re right.

>=20
> Calling regs->ax =3D sys_ni_syscall() in an else clause would arguably be t=
he right thing here, except possibly in the case where nr (or (int)nr, see b=
elow) =3D=3D -1 or < 0.

I think the check should be -1 for 64 bit but (u32)nr =3D=3D (u32)-1 for the=
 32-bit path. Does that seem reasonable?

>=20
> Now, syscall_get_nr() returns the low 32 bits of the system call number un=
conditionally. There are places where we look at the sign of this number, wh=
ich means that 0xffffffff7fffffff is "positive" and 0x7fffffffffffffff is "n=
egative". We have gone back and forth more than once on if we should look at=
 %rax or just %eax on a system call... I have to admit that the current desi=
gn makes me a bit nervous.
>=20
> Finally, can anything bad happen in some weird corner case inside one of t=
he syscall_*_mode() calls or after an interrupt if someone tries to call sys=
call(-1) or another negative number?
>=20
> Food for thought or just my not being up to date?
>=20
> Thanks,
>=20
>    -hpa
>=20