6 This chapter explains the security requirements that QEMU is designed to meet
7 and principles for securely deploying QEMU.
9 @section Security Requirements
11 QEMU supports many different use cases, some of which have stricter security
12 requirements than others. The community has agreed on the overall security
13 requirements that users may depend on. These requirements define what is
14 considered supported from a security perspective.
16 @subsection Virtualization Use Case
18 The virtualization use case covers cloud and virtual private server (VPS)
19 hosting, as well as traditional data center and desktop virtualization. These
20 use cases rely on hardware virtualization extensions to execute guest code
21 safely on the physical CPU at close-to-native speed.
23 The following entities are untrusted, meaning that they may be buggy or
28 @item User-facing interfaces (e.g. VNC, SPICE, WebSocket)
29 @item Network protocols (e.g. NBD, live migration)
30 @item User-supplied files (e.g. disk images, kernels, device trees)
31 @item Passthrough devices (e.g. PCI, USB)
34 Bugs affecting these entities are evaluated on whether they can cause damage in
35 real-world use cases and treated as security bugs if this is the case.
37 @subsection Non-virtualization Use Case
39 The non-virtualization use case covers emulation using the Tiny Code Generator
40 (TCG). In principle the TCG and device emulation code used in conjunction with
41 the non-virtualization use case should meet the same security requirements as
42 the virtualization use case. However, for historical reasons much of the
43 non-virtualization use case code was not written with these security
46 Bugs affecting the non-virtualization use case are not considered security
47 bugs at this time. Users with non-virtualization use cases must not rely on
48 QEMU to provide guest isolation or any security guarantees.
52 This section describes the design principles that ensure the security
55 @subsection Guest Isolation
57 Guest isolation is the confinement of guest code to the virtual machine. When
58 guest code gains control of execution on the host this is called escaping the
59 virtual machine. Isolation also includes resource limits such as throttling of
60 CPU, memory, disk, or network. Guests must be unable to exceed their resource
63 QEMU presents an attack surface to the guest in the form of emulated devices.
64 The guest must not be able to gain control of QEMU. Bugs in emulated devices
65 could allow malicious guests to gain code execution in QEMU. At this point the
66 guest has escaped the virtual machine and is able to act in the context of the
67 QEMU process on the host.
69 Guests often interact with other guests and share resources with them. A
70 malicious guest must not gain control of other guests or access their data.
71 Disk image files and network traffic must be protected from other guests unless
72 explicitly shared between them by the user.
74 @subsection Principle of Least Privilege
76 The principle of least privilege states that each component only has access to
77 the privileges necessary for its function. In the case of QEMU this means that
78 each process only has access to resources belonging to the guest.
80 The QEMU process should not have access to any resources that are inaccessible
81 to the guest. This way the guest does not gain anything by escaping into the
82 QEMU process since it already has access to those same resources from within
85 Following the principle of least privilege immediately fulfills guest isolation
86 requirements. For example, guest A only has access to its own disk image file
87 @code{a.img} and not guest B's disk image file @code{b.img}.
89 In reality certain resources are inaccessible to the guest but must be
90 available to QEMU to perform its function. For example, host system calls are
91 necessary for QEMU but are not exposed to guests. A guest that escapes into
92 the QEMU process can then begin invoking host system calls.
94 New features must be designed to follow the principle of least privilege.
95 Should this not be possible for technical reasons, the security risk must be
96 clearly documented so users are aware of the trade-off of enabling the feature.
98 @subsection Isolation mechanisms
100 Several isolation mechanisms are available to realize this architecture of
101 guest isolation and the principle of least privilege. With the exception of
102 Linux seccomp, these mechanisms are all deployed by management tools that
103 launch QEMU, such as libvirt. They are also platform-specific so they are only
104 described briefly for Linux here.
106 The fundamental isolation mechanism is that QEMU processes must run as
107 unprivileged users. Sometimes it seems more convenient to launch QEMU as
108 root to give it access to host devices (e.g. @code{/dev/net/tun}) but this poses a
109 huge security risk. File descriptor passing can be used to give an otherwise
110 unprivileged QEMU process access to host devices without running QEMU as root.
111 It is also possible to launch QEMU as a non-root user and configure UNIX groups
112 for access to @code{/dev/kvm}, @code{/dev/net/tun}, and other device nodes.
113 Some Linux distros already ship with UNIX groups for these devices by default.
116 @item SELinux and AppArmor make it possible to confine processes beyond the
117 traditional UNIX process and file permissions model. They restrict the QEMU
118 process from accessing processes and files on the host system that are not
121 @item Resource limits and cgroup controllers provide throughput and utilization
122 limits on key resources such as CPU time, memory, and I/O bandwidth.
124 @item Linux namespaces can be used to make process, file system, and other system
125 resources unavailable to QEMU. A namespaced QEMU process is restricted to only
126 those resources that were granted to it.
128 @item Linux seccomp is available via the QEMU @option{--sandbox} option. It disables
129 system calls that are not needed by QEMU, thereby reducing the host kernel
133 @section Sensitive configurations
135 There are aspects of QEMU that can have security implications which users &
136 management applications must be aware of.
138 @subsection Monitor console (QMP and HMP)
140 The monitor console (whether used with QMP or HMP) provides an interface
141 to dynamically control many aspects of QEMU's runtime operation. Many of the
142 commands exposed will instruct QEMU to access content on the host file system
143 and/or trigger spawning of external processes.
145 For example, the @code{migrate} command allows for the spawning of arbitrary
146 processes for the purpose of tunnelling the migration data stream. The
147 @code{blockdev-add} command instructs QEMU to open arbitrary files, exposing
148 their content to the guest as a virtual disk.
150 Unless QEMU is otherwise confined using technologies such as SELinux, AppArmor,
151 or Linux namespaces, the monitor console should be considered to have privileges
152 equivalent to those of the user account QEMU is running under.
154 It is further important to consider the security of the character device backend
155 over which the monitor console is exposed. It needs to have protection against
156 malicious third parties which might try to make unauthorized connections, or
157 perform man-in-the-middle attacks. Many of the character device backends do not
158 satisfy this requirement and so must not be used for the monitor console.
160 The general recommendation is that the monitor console should be exposed over
161 a UNIX domain socket backend to the local host only. Use of the TCP based
162 character device backend is inappropriate unless configured to use both TLS
163 encryption and authorization control policy on client connections.
165 In summary, the monitor console is considered a privileged control interface to
166 QEMU and as such should only be made accessible to a trusted management