-
Notifications
You must be signed in to change notification settings - Fork 5
/
charter.html
149 lines (149 loc) · 7.72 KB
/
charter.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
<html>
<title>OASIS Virtual I/O Device (VIRTIO) Technical Committee TC charter</title>
<body>
<ol id="charter-outline" class="bold-lower-alpha">
<li>
TC Name
<p>OASIS Virtual I/O Device (VIRTIO) Technical Committee</p>
</li>
<li>Statement of Purpose
<p>
Background:
</p>
<p>
Hardware virtualization allows multiple operating systems ("guests") to share the same hardware ("host") managed by host software ("hypervisor").
</p>
<p>
These guests need networks, storage, consoles and similar but non-virtualization-aware standard devices cannot be shared, or guests may not be permitted to access host devices at all. The simplest solution is to emulate a device expected by the guest operating system, but this can be slow and/or complicated. As most operating systems have facilities for adding drivers for new physical hardware, we can use the same facilities to add drivers for devices which are easier and/or more efficient to implement in software.
</p>
<p>
As every hypervisor is different, they tend to implement hypervisor-specific devices, requiring every guest to support a new device for that environment. For example, in 2013 Linux supported completely separate drivers for eight different virtualization platforms, with most drivers being sub-optimal. In 2007, an attempt was made to implement a hypervisor and OS-agnostic device model in Linux guests and the KVM hypervisor over the standard PCI bus. This is now supported by multiple other hypervisors.
</p>
<p>
A Draft Specification
</p>
<p>
In 2009, as interest accumulated, the "Virtio PCI Card Specification" was published, with appendices for network, block storage and console devices. The emphasis was that virtual devices should be simple, look like driver authors expect physical devices to look, should be extensible, and that they should perform well.
</p>
<p>
Even at the time, there were implementations of virtio devices over non-PCI transports, and in 2011 the simplified "mmio" transport was added as an Appendix, as well as a "remote processor message" device which is actually used to communicate to a separate, physical CPU, rather than a virtual guest.
</p>
<p>
The years of experience have highlighted some of the implementation and design mistakes: enhancements have worked around many of them, but at cost of simplicity. Implementation bugs have also caused occasional anguish.
</p>
<p>
A 1.0 Specification
</p>
<p>
With the 1.0 Specification, the goal of the OASIS Virtual I/O Device (VIRTIO) Technical Committee was to keep the good, discard the bad, and make the ugly optional. The "Virtio PCI Card Specification" 0.9.5, was used as a starting point (referred to as "legacy"). In particular, the Committee tried not to break too much, and (with the exception of the mmio transport) the specification is designed to make it possible for devices to support both 1.0 compliant and legacy guest drivers.
</p>
<p>
1.X Specifications
</p>
<p>
Support for new technologies often calls for extensions to the standard. For example,
as technologies such as nested virtualization and hardware-based
implementations of VIRTIO devices became popular, these devices are no longer
necessarily part of a hypervisor. This in turn requires a strong commitment to
driver and device compatibility, as well as to driver and device security.
With the 1.1, 1.2 and future revisions of the Specification, we aim to evolve the VIRTIO standard further, addressing such new requirements while both continuing to honor the goals of the 1.0 Specification and including new objectives.
</p>
</li>
<li>
Scope
<p>
The TC will develop and produce versions 1.X of the OASIS Virtual I/O Device (VIRTIO) OASIS Standard by refining and documenting existing implementations and practice.
</p>
<p>
The TC will also act as a registrar for the list of reserved device numbers for non-specified device types, as well as the list of reserved feature numbers for non-specified features for each device type.
</p>
<p>
Starting from version 1.1, each version of the specification shall be used as a starting point for the development of the next version of the specification.
</p>
<p>
Note that we expect the output of this TC to be compatible with all previous versions of the specification, such that
a compliant devices and drivers remain compliant with future versions of the specification.
In particular, special care will be taken such that drivers compliant with a certain version of the specification can also support devices compliant with all past versions of the specification, and vice versa.
</p>
<p>
When correcting defects in the specification, e.g. by adding new normative statements, the TC will
also take special care to avoid declaring existing working implementations non-compliant
as far as possible.
The TC will take into account factors such as existance and prevalence of implementations with the
defective behaviour and whether these implementations are working and robust.
When in doubt, the TC will err on the conservative side, such as by adding new feature bits
or by making the corrected behaviour recommended but not mandatory.
</p>
<p>
The TC will consider one or more buses for virtual devices, including PCI. It will consider various kinds of devices, including network devices. Each part of the OASIS Standard will be considered in terms of portability, simplicity, least-surprise for driver authors, security, compatibility, extensibility and performance. In particular, the effect of future extensions (such as layout changes) will be considered.
</p>
</li>
<li>
Deliverables
<ol class="spaced decimal">
<li>
Specification of feature negotiation, configuration and queues, from both driver and device points of view.
<ol class="vanilla-lower-alpha">
<li>
Specifically for virtio over PCI.
</li>
<li>
Specifically for virtio over mmio.
</li>
<li>
Specifically for virtio over channel I/O.
</li>
<li>
Specifically for other transports as decided by TC.
</li>
</ol>
</li>
<li>
Specification of device-specific configuration.
<ol class="vanilla-lower-alpha">
<li>
Normative requirements for specified device types.
</li>
<li>
Non-normative advice for designing new device types.
</li>
<li>
List of reserved device numbers for non-specified device types. (This section is likely to see ongoing maintenance).
</li>
<li>
List of reserved feature numbers for non-specified features for each device type. (This section is likely to see ongoing maintenance).
</li>
</ol>
</li>
<li>
Non-normative code examples for operation of driver/device side of buffers.
</li>
<li>
Non-normative guide for creating devices and drivers which also support a legacy mode.
</li>
</ol>
<p>
The projected delivery dates for the specifications are each 12 to 16 months.
</p>
</li>
<li>
IPR Mode
<p>
Non-Assertion Mode
</p>
</li>
<li>
Audience
<p>
Developers of devices, hypervisors and device drivers.
</p>
</li>
<li>
Language
<p>
US English.
</p>
</li>
</ol>
</body>
</html>