Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <20190502204958.7868-1-joel@joelfernandes.org>
In-Reply-To: <20190502204958.7868-1-joel@joelfernandes.org>
From:   Alexei Starovoitov <alexei.starovoitov@gmail.com>
Date:   Sun, 5 May 2019 00:19:42 -0700
Message-ID: <CAADnVQ+CPXKv_fryOxnDXjCLmWxor2j+WBFvPtG-Tcyr=hzRpQ@mail.gmail.com>
Subject: Re: [PATCH RFC] bpf: Add support for reading user pointers
To:     "Joel Fernandes (Google)" <joel@joelfernandes.org>
Cc:     LKML <linux-kernel@vger.kernel.org>,
        Michal Gregorczyk <michalgr@live.com>,
        Adrian Ratiu <adrian.ratiu@collabora.com>,
        Mohammad Husain <russoue@gmail.com>,
        Qais Yousef <qais.yousef@arm.com>,
        Srinivas Ramana <sramana@codeaurora.org>,
        duyuchao <yuchao.du@unisoc.com>,
        Manjo Raja Rao <linux@manojrajarao.com>,
        Karim Yaghmour <karim.yaghmour@opersys.com>,
        Tamir Carmeli <carmeli.tamir@gmail.com>,
        Yonghong Song <yhs@fb.com>,
        Alexei Starovoitov <ast@kernel.org>,
        Brendan Gregg <brendan.d.gregg@gmail.com>,
        Masami Hiramatsu <mhiramat@kernel.org>,
        Peter Ziljstra <peterz@infradead.org>,
        Steven Rostedt <rostedt@goodmis.org>,
        Kees Cook <keescook@chromium.org>,
        Android Kernel Team <kernel-team@android.com>,
        Daniel Borkmann <daniel@iogearbox.net>,
        Ingo Molnar <mingo@redhat.com>,
        Network Development <netdev@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, May 2, 2019 at 1:50 PM Joel Fernandes (Google)
<joel@joelfernandes.org> wrote:
>
> The eBPF based opensnoop tool fails to read the file path string passed
> to the do_sys_open function. This is because it is a pointer to
> userspace address and causes an -EFAULT when read with
> probe_kernel_read. This is not an issue when running the tool on x86 but
> is an issue on arm64. This patch adds a new bpf function call based
> which calls the recently proposed probe_user_read function [1].
> Using this function call from opensnoop fixes the issue on arm64.
>
> [1] https://lore.kernel.org/patchwork/patch/1051588/
...
> +BPF_CALL_3(bpf_probe_read_user, void *, dst, u32, size, const void *, unsafe_ptr)
> +{
> +       int ret;
> +
> +       ret = probe_user_read(dst, unsafe_ptr, size);
> +       if (unlikely(ret < 0))
> +               memset(dst, 0, size);
> +
> +       return ret;
> +}

probe_user_read() doesn't exist in bpf-next
therefore this patch has to wait for the next merge window.
At the same time we would need to introduce
bpf_probe_read_kernel() and introduce a load time warning
for existing bpf_probe_read(), so we can deprecate it eventually.