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
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
|
/****************************************************************************
**
** Copyright (C) 2017 The Qt Company Ltd.
** Contact: https://www.qt.io/licensing/
**
** This file is part of the documentation of the Qt Toolkit.
**
** $QT_BEGIN_LICENSE:FDL$
** Commercial License Usage
** Licensees holding valid commercial Qt licenses may use this file in
** accordance with the commercial license agreement provided with the
** Software or, alternatively, in accordance with the terms contained in
** a written agreement between you and The Qt Company. For licensing terms
** and conditions see https://www.qt.io/terms-conditions. For further
** information use the contact form at https://www.qt.io/contact-us.
**
** GNU Free Documentation License Usage
** Alternatively, this file may be used under the terms of the GNU Free
** Documentation License version 1.3 as published by the Free Software
** Foundation and appearing in the file included in the packaging of
** this file. Please review the following information to ensure
** the GNU Free Documentation License version 1.3 requirements
** will be met: https://www.gnu.org/licenses/fdl-1.3.html.
** $QT_END_LICENSE$
**
****************************************************************************/
/*!
\page wayland-and-qt.html
\title Wayland and Qt
\brief An overview of the Wayland protocol and how it fits into Qt.
\l {https://wayland.freedesktop.org/}{Wayland} is a display server protocol
that helps you to create multi-process systems. Multiple client applications
("clients") can render their own content to off-screen buffers. These buffers
are then passed to a display server, often called a compositor, using the
Wayland protocol. Finally, the compositor composites and positions the
content on a physical display.
\section1 Why Use Multi-Process
In a single-process system, all parts of the UI run in one, single process.
In a multi-process system, all clients run in their own, dedicated process.
With Qt, at any point in your development process, you can choose to switch
between single-process and multi-process.
\image ../images/wayland-multi-process.png
\caption Multi-Process Client Architecture
\image ../images/wayland-single-process-eglfs.png
\caption Single Process Client Architecture
The use of multi-process has the following benefits:
\list
\li \l{stability}{Stability}
\li \l{security}{Security}
\li \l{performance}{Performance}
\li \l{interoperability}{Interoperability}
\endlist
\target stability
\table 100%
\header
\li {2,1} Stability
\row
\li Easier to recover when clients hang or crash
\li If you have a complex UI, then multi-process is useful because if one
part of the UI crashes, it doesn't affect the entire system.
Similarly, the display won't freeze, even when one client freezes.
\note If your client is mandated by law to render safety-critical
information, consider using
\l{https://doc.qt.io/QtSafeRenderer/qtsr-overview.html}{Qt Safe Renderer Overview}.
\row
\li Protection against possible memory leaks
\li In a multi-process system, if one client has a memory leak and
consumes lots of memory, that memory is recovered when that client
exits. In contrast with single-process, the memory leak remains until
the entire system restarts.
\endtable
\target security
\table 100%
\header
\li Security
\row
\li In a single-process system, all clients can access each other's
memory. For example, there's no isolation for sensitive data
transfer; every line of code must be equally trustworthy. This
isolation is there, by design, in multi-process systems.
\endtable
\target performance
\table 100%
\header
\li Performance
\row
\li If you have a CPU with multiple cores, a multi-process system can
help distribute the load evenly across different cores, making more
efficient use of your CPU.
\endtable
\target interoperability
\table 100%
\header
\li Interoperability
\row
\li You can interface with non-Qt clients in a multi-process system, as
long as your clients understand Wayland or X11. For example, if you
use gstreamer for video or if you want to use a navigation
application built with another UI toolkit, you can run these clients
alongside your other Qt-based clients.
\endtable
\section1 Why Use Wayland Instead of X11 or Custom Solutions
X11, a desktop protocol from the 80s, no longer fits with how graphics
hardware works today. It is large, complex, and lacks customizability. In
fact, it is difficult to run a client fluidly with X11, and reach 60 fps
without tearing. Wayland, in contrast, is easier to implement, has better
performance, and contains all the necessary parts to run efficiently on
modern graphics hardware. For embedded, multi-process systems on Linux,
Wayland is the standard.
However, if you are working with old hardware or legacy applications,
then Wayland may not be a good option. The Wayland protocol is designed with
security and isolation in mind, and is strict/conservative about what
information and functionality is available to clients. While this leads to a
cleaner and more secure interface, some functionality that legacy
applications expect may no longer be available on Wayland.
Particularly, there are three common use cases where Wayland may not be the
best option:
\list 1
\li The hardware or platform is old and only supports X11; in which case
you have no choice.
\li You have to support legacy applications that depend on features that
are absent in the Wayland protocol for security and simplicity.
\li You have to support legacy applications that use a UI toolkit that
doesn't run on Wayland at all. In some cases, you may be able to work
around this by running those applications on
\l{https://wayland.freedesktop.org/xserver.html}{XWayland} instead.
\endlist
Back when X11 was very popular, developers wrote their own custom solutions
to circumvent X11 issues. Older Qt versions had the Qt Windowing System
(QWS), which is now discontinued. Today, most of these use cases are covered
by Wayland, and custom solutions are becoming less and less common.
\section1 Possible Trade-Offs with Multi-Process
Use of multi-process systems do bring about the following trade-offs:
\list
\li \l{increased-video-memory}{Increased video memory consumption}
\li \l{increased-main-memory}{Increased main memory consumption}
\li \l{repeated-storage}{Repeated storage of graphical resources}
\endlist
\target increased-video-memory
\table 100%
\header
\li Increased video memory consumption
\row
\li This can be a constraint for embedded devices. In multi-process, each
client needs to have its own graphics buffer, which it sends to the
compositor. Consequently, you use more video memory compared to the
single-process case: where everything is drawn at once and there is
no need to store the different parts in intermediary buffers.
\endtable
\target increased-main-memory
\table 100%
\header
\li Increased main memory consumption
\row
\li Apart from some extra overhead at the OS level, running multiple
clients may also use more main memory as some parts need to be
duplicated once per client. For example, if you run QML, each client
requires a separate QML engine. Consequently, if you run a single
client that uses Qt Quick Controls, it's loaded once. If you then
split this client into multiple clients, you're loading Qt Quick
Controls multiple times, resulting in a higher startup cost to
initialize your clients.
\endtable
\target repeated-storage
\table 100%
\header
\li Repeated storage of graphical resources
\row
\li In a single-process system, if you're using the same textures,
background, or icons in many places, those images are only stored
once. In contrast, if you use these images in a multi-process system,
then you have to store them multiple times. In this case, one
solution is to share graphical resource between clients. Qt already
allows sharing image resources in main memory across processes
without involving Wayland. Sharing GPU textures across processes, on
the other hand, requires more intricate solutions. Such solutions are
currently in development for the Qt Wayland Compositor.
\endtable
\section1 What Qt Wayland Offers
\b{For Clients}
\br
Qt clients can run on any Wayland compositor, including Weston, the reference
compositor developed as part of the Wayland project.
Any Qt program can run as a Wayland client (as part of a multi-process system)
or a standalone client (single-process). This is determined on startup, where
you can choose between the different backends. During the development process,
you can develop the client on the desktop first, then test it on the target
hardware later. You don't need to run your clients on the actual target
hardware all the time.
\image ../images/wayland-single-process-develop.png
\caption Single-Process Client Development
If you develop on a Linux machine, you can also run the compositor within a
window on your development machine. This lets you run clients in an
environment that closely resembles the target device. Without rebuilding the
client, you can also run it with \c{-platform wayland} to run it inside the
compositor. If you use \c{-platform xcb} (for X11), you can run the client on
the desktop. In other words, you can start developing your clients before the
compositor is ready for use.
\b{For Servers}
\br
The server, or compositor, connects to the display and shows the contents of
each client on the screen. The compositor handles input and sends input
events to the corresponding client. In turn, each client connects to the
compositor and sends the content of its windows. It's up to the compositor to
decide:
\list
\li How and where to show the content
\li Which content to show
\li What to do with the different client graphics buffers
\endlist
This means, it's up to the compositor to decide what a multi-process system
is. For instance, the clients could be part of a 3D scene with windows on the
walls, on a VR system, mapped to a sphere, and so on.
The Qt Wayland Compositor is an API for building your own compositor. It
gives you full freedom to build a custom compositor UI and manage the windows
of various clients. You can combine both Qt Quick and QML with the Qt Wayland
Compositor to create impressive, imaginative UIs. For more information, see
\l{Qt Wayland Compositor}.
\section2 Related Content
\list
\li \l{https://resources.qt.io/videos/qtws17-qt-wayland-compositor-creating-multi-process-user-interface-johan-helsing-the-qt-company}
{QtWS17 - Qt Wayland Compositor: Creating multi-process user interface}
\li \l{https://doc.qt.io/QtApplicationManager/introduction.html}{Qt Application Manager}
\endlist
*/
|
