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: <20200421171552.28393-1-luke.r.nels@gmail.com> <6f1130b3-eaea-cc5e-716f-5d6be77101b9@zytor.com>
 <CAKU6vyb38-XcFeAiP7OW0j++0jS-J4gZP6S2E21dpQwvcEFpKQ@mail.gmail.com>
In-Reply-To: <CAKU6vyb38-XcFeAiP7OW0j++0jS-J4gZP6S2E21dpQwvcEFpKQ@mail.gmail.com>
From:   Brian Gerst <brgerst@gmail.com>
Date:   Tue, 21 Apr 2020 23:22:36 -0400
Message-ID: <CAMzpN2hpwK00duVmrzuhDeZY+H7doJ+C-O6=SWrzy+KvAsupqw@mail.gmail.com>
Subject: Re: [PATCH bpf 1/2] bpf, x32: Fix invalid instruction in BPF_LDX zero-extension
To:     Xi Wang <xi.wang@gmail.com>
Cc:     "H. Peter Anvin" <hpa@zytor.com>,
        Luke Nelson <lukenels@cs.washington.edu>, bpf@vger.kernel.org,
        Luke Nelson <luke.r.nels@gmail.com>,
        Wang YanQing <udknight@gmail.com>,
        "David S. Miller" <davem@davemloft.net>,
        Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>,
        Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        "the arch/x86 maintainers" <x86@kernel.org>,
        Alexei Starovoitov <ast@kernel.org>,
        Daniel Borkmann <daniel@iogearbox.net>,
        Martin KaFai Lau <kafai@fb.com>,
        Song Liu <songliubraving@fb.com>, Yonghong Song <yhs@fb.com>,
        Andrii Nakryiko <andriin@fb.com>,
        John Fastabend <john.fastabend@gmail.com>,
        KP Singh <kpsingh@chromium.org>, netdev@vger.kernel.org,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Tue, Apr 21, 2020 at 3:32 PM Xi Wang <xi.wang@gmail.com> wrote:
>
> On Tue, Apr 21, 2020 at 10:39 AM H. Peter Anvin <hpa@zytor.com> wrote:
> > x32 is not x86-32.  In Linux we generally call the latter "i386".
>
> Agreed.  Most of the previous patches to this file use "x32" and this
> one just wanted to be consistent.
>
> > C7 /0 imm32 is a valid instruction on i386. However, it is also
> > inefficient when the destination is a register, because B8+r imm32 is
> > equivalent, and when the value is zero, XOR is indeed more efficient.
> >
> > The real error is using EMIT3() instead of EMIT2_off32(), but XOR is
> > more efficient. However, let's make the bug statement *correct*, or it
> > is going to confuse the Hades out of people in the future.
>
> I don't see how the bug statement is incorrect, which merely points
> out that "C7 C0 0" is an invalid instruction, regardless of whether
> the JIT intended to emit C7 /0 imm32, B8+r imm32, 31 /r, 33 /r, or any
> other equivalent form.

You should explain the reason it is invalid, ie. the instruction
encoding needs a 32-bit immediate but the current code only emits an
8-bit immediate.

--
Brian Gerst