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;
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
From:   Andy Lutomirski <luto@amacapital.net>
Mime-Version: 1.0 (1.0)
Date:   Tue, 22 Oct 2019 15:45:26 -0700
Message-Id: <7364B113-DD65-423D-BED3-FF90C4DF8334@amacapital.net>
Subject: Re: [PATCH 3/3] x86/ftrace: Use text_poke()
In-Reply-To: <20191022215841.2qsmhd6vxi4mwade@ast-mbp.dhcp.thefacebook.com>
References: <20191022215841.2qsmhd6vxi4mwade@ast-mbp.dhcp.thefacebook.com>
To:     Alexei Starovoitov <alexei.starovoitov@gmail.com>
Cc:     Steven Rostedt <rostedt@goodmis.org>,
        Peter Zijlstra <peterz@infradead.org>,
        Daniel Bristot de Oliveira <bristot@redhat.com>,
        LKML <linux-kernel@vger.kernel.org>, X86 ML <x86@kernel.org>,
        Nadav Amit <nadav.amit@gmail.com>,
        Andy Lutomirski <luto@kernel.org>,
        Dave Hansen <dave.hansen@linux.intel.com>,
        Song Liu <songliubraving@fb.com>,
        Masami Hiramatsu <mhiramat@kernel.org>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


>> On Oct 22, 2019, at 2:58 PM, Alexei Starovoitov <alexei.starovoitov@gmail=
.com> wrote:
>>=20
>> =EF=BB=BFOn Tue, Oct 22, 2019 at 05:04:30PM -0400, Steven Rostedt wrote:
>> I gave a solution for this. And that is to add another flag to allow
>> for just the minimum to change the ip. And we can even add another flag
>> to allow for changing the stack if needed (to emulate a call with the
>> same parameters).
>=20
> your solution is to reduce the overhead.
> my solution is to remove it competely. See the difference?
>=20
>> By doing this work, live kernel patching will also benefit. Because it
>> is also dealing with the unnecessary overhead of saving regs.
>> And we could possibly even have kprobes benefit from this if a kprobe
>> doesn't need full regs.
>=20
> Neither of two statements are true. The per-function generated trampoline
> I'm talking about is bpf specific. For a function with two arguments it's j=
ust:
> push rbp=20
> mov rbp, rsp
> push rdi
> push rsi
> lea  rdi,[rbp-0x10]
> call jited_bpf_prog
> pop rsi
> pop rdi
> leave
> ret

Why are you saving rsi?  You said upthread that you=E2=80=99re saving the ar=
gs, but rsi is already available in rsi.

Just how custom is this bpf program?  It seems to clobber no regs (except ar=
gs), and it doesn=E2=80=99t return anything. Is it entirely specific to the p=
robed function?  If so, why not just call it directly?

In any event, I think you can=E2=80=99t *just* use text_poke.  Something nee=
ds to coordinate to ensure that, if you bpf trace an already-kprobed functio=
n, the right thing happens.

FWIW, if you are going to use a trampoline like this, consider using r11 for=
 the caller frame instead of rsi.  You won=E2=80=99t need to restore it. But=
 I=E2=80=99m wondering whether the bpf jitted code could just directly acces=
s the frame instead of indirecting through a register. (Or am I entirely mis=
understanding what rdi is for in your example?)=