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;
Date:   Fri, 07 Jun 2019 13:22:31 -0700
User-Agent: K-9 Mail for Android
In-Reply-To: <3DA961AB-950B-4886-9656-C0D268D521F1@amacapital.net>
References: <20190605130753.327195108@infradead.org> <20190605131945.005681046@infradead.org> <20190608004708.7646b287151cf613838ce05f@kernel.org> <20190607173427.GK3436@hirez.programming.kicks-ass.net> <3DA961AB-950B-4886-9656-C0D268D521F1@amacapital.net>
MIME-Version: 1.0
Content-Type: text/plain;
 charset=utf-8
Content-Transfer-Encoding: quoted-printable
Subject: Re: [PATCH 08/15] x86/alternatives: Teach text_poke_bp() to emulate instructions
To:     Andy Lutomirski <luto@amacapital.net>,
        Peter Zijlstra <peterz@infradead.org>
CC:     Masami Hiramatsu <mhiramat@kernel.org>, x86@kernel.org,
        linux-kernel@vger.kernel.org,
        Ard Biesheuvel <ard.biesheuvel@linaro.org>,
        Andy Lutomirski <luto@kernel.org>,
        Steven Rostedt <rostedt@goodmis.org>,
        Ingo Molnar <mingo@kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        Jason Baron <jbaron@akamai.com>, Jiri Kosina <jkosina@suse.cz>,
        David Laight <David.Laight@aculab.com>,
        Borislav Petkov <bp@alien8.de>,
        Julia Cartwright <julia@ni.com>, Jessica Yu <jeyu@kernel.org>,
        Nadav Amit <namit@vmware.com>,
        Rasmus Villemoes <linux@rasmusvillemoes.dk>,
        Edward Cree <ecree@solarflare.com>,
        Daniel Bristot de Oliveira <bristot@redhat.com>
From:   hpa@zytor.com
Message-ID: <E4AE85BE-FB3B-45BE-8E1E-DD27E2520A4C@zytor.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On June 7, 2019 11:10:19 AM PDT, Andy Lutomirski <luto@amacapital=2Enet> wr=
ote:
>
>
>> On Jun 7, 2019, at 10:34 AM, Peter Zijlstra <peterz@infradead=2Eorg>
>wrote:
>>=20
>> On Sat, Jun 08, 2019 at 12:47:08AM +0900, Masami Hiramatsu wrote:
>>=20
>>>> This fits almost all text_poke_bp() users, except
>>>> arch_unoptimize_kprobe() which restores random text, and for that
>site
>>>> we have to build an explicit emulate instruction=2E
>>>=20
>>> Hm, actually it doesn't restores randome text, since the first byte
>>> must always be int3=2E As the function name means, it just unoptimizes
>>> (jump based optprobe -> int3 based kprobe)=2E
>>> Anyway, that is not an issue=2E With this patch, optprobe must still
>work=2E
>>=20
>> I thought it basically restored 5 bytes of original text (with no
>> guarantee it is a single instruction, or even a complete
>instruction),
>> with the first byte replaced with INT3=2E
>>=20
>
>I am surely missing some kprobe context, but is it really safe to use
>this mechanism to replace more than one instruction?

I don't see how it could be, except *perhaps* inside an NMI have, because =
you could have a preempted or interrupted now having an in-memory IP pointi=
ng inside the middle of the region you are intending to patch=2E


--=20
Sent from my Android device with K-9 Mail=2E Please excuse my brevity=2E