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: [PATCH RFC] seccomp: Implement syscall isolation based on memory areas
Date:   Sat, 30 May 2020 15:09:47 -0700
Message-Id: <AF65147C-15DB-4BA4-A08B-55676B489BA5@amacapital.net>
References: <20200530055953.817666-1-krisman@collabora.com>
Cc:     linux-mm@kvack.org, linux-kernel@vger.kernel.org,
        kernel@collabora.com, Thomas Gleixner <tglx@linutronix.de>,
        Kees Cook <keescook@chromium.org>,
        Will Drewry <wad@chromium.org>,
        "H . Peter Anvin" <hpa@zytor.com>, Paul Gofman <gofmanp@gmail.com>
In-Reply-To: <20200530055953.817666-1-krisman@collabora.com>
To:     Gabriel Krisman Bertazi <krisman@collabora.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


> On May 29, 2020, at 11:00 PM, Gabriel Krisman Bertazi <krisman@collabora.c=
om> wrote:
>=20
> =EF=BB=BFModern Windows applications are executing system call instruction=
s
> directly from the application's code without going through the WinAPI.
> This breaks Wine emulation, because it doesn't have a chance to
> intercept and emulate these syscalls before they are submitted to Linux.
>=20
> In addition, we cannot simply trap every system call of the application
> to userspace using PTRACE_SYSEMU, because performance would suffer,
> since our main use case is to run Windows games over Linux.  Therefore,
> we need some in-kernel filtering to decide whether the syscall was
> issued by the wine code or by the windows application.

Do you really need in-kernel filtering?  What if you could have efficient us=
erspace filtering instead?  That is, set something up so that all syscalls, e=
xcept those from a special address, are translated to CALL thunk where the t=
hunk is configured per task.  Then the thunk can do whatever emulation is ne=
eded.

Getting the details and especially the interaction with any seccomp filters t=
hat may be installed right could be tricky, but the performance should be de=
cent, at least on non-PTI systems.

(If we go this route, I suspect that the correct interaction with seccomp is=
 that this type of redirection takes precedence over seccomp and seccomp fil=
ters are not invoked for redirected syscalls. After all, a redirected syscal=
l is, functionally, not a syscall at all.)

>=20