1 .. _vkms: 1 .. _vkms:
2 2
3 ========================================== 3 ==========================================
4 drm/vkms Virtual Kernel Modesetting 4 drm/vkms Virtual Kernel Modesetting
5 ========================================== 5 ==========================================
6 6
7 .. kernel-doc:: drivers/gpu/drm/vkms/vkms_drv. 7 .. kernel-doc:: drivers/gpu/drm/vkms/vkms_drv.c
8 :doc: vkms (Virtual Kernel Modesetting) 8 :doc: vkms (Virtual Kernel Modesetting)
9 9
10 Setup !! 10 TODO
11 ===== !! 11 ====
12 <<
13 The VKMS driver can be setup with the followin <<
14 <<
15 To check if VKMS is loaded, run:: <<
16 <<
17 lsmod | grep vkms <<
18 <<
19 This should list the VKMS driver. If no output <<
20 you need to enable and/or load the VKMS driver <<
21 Ensure that the VKMS driver has been set as a <<
22 kernel config file. Do:: <<
23 <<
24 make nconfig <<
25 <<
26 Go to `Device Drivers> Graphics support` <<
27 <<
28 Enable `Virtual KMS (EXPERIMENTAL)` <<
29 <<
30 Compile and build the kernel for the changes t <<
31 Now, to load the driver, use:: <<
32 <<
33 sudo modprobe vkms <<
34 <<
35 On running the lsmod command now, the VKMS dri <<
36 You can also observe the driver being loaded i <<
37 <<
38 The VKMS driver has optional features to simul <<
39 which are exposed as module options. You can u <<
40 to see the module options for vkms:: <<
41 <<
42 modinfo vkms <<
43 <<
44 Module options are helpful when testing, and e <<
45 can be done while loading vkms. For example, t <<
46 use:: <<
47 <<
48 sudo modprobe vkms enable_cursor=1 <<
49 <<
50 To disable the driver, use :: <<
51 <<
52 sudo modprobe -r vkms <<
53 <<
54 Testing With IGT <<
55 ================ <<
56 <<
57 The IGT GPU Tools is a test suite used specifi <<
58 development of the DRM drivers. <<
59 The IGT Tools can be installed from <<
60 `here <https://gitlab.freedesktop.org/drm/igt- <<
61 <<
62 The tests need to be run without a compositor, <<
63 only mode. You can do this by:: <<
64 <<
65 sudo systemctl isolate multi-user.target <<
66 <<
67 To return to graphical mode, do:: <<
68 <<
69 sudo systemctl isolate graphical.target <<
70 <<
71 Once you are in text only mode, you can run te <<
72 or IGT_DEVICE variable to specify the device f <<
73 to test. IGT_DEVICE can also be used with the <<
74 tests for a specific driver:: <<
75 <<
76 sudo ./build/tests/<name of test> --device " <<
77 sudo IGT_DEVICE="sys:/sys/devices/platform/v <<
78 sudo IGT_DEVICE="sys:/sys/devices/platform/v <<
79 12
80 For example, to test the functionality of the !! 13 CRC API Improvements
81 we can run the kms_writeback test:: !! 14 --------------------
82 15
83 sudo ./build/tests/kms_writeback --device "s !! 16 - Optimize CRC computation ``compute_crc()`` and plane blending ``blend()``
84 sudo IGT_DEVICE="sys:/sys/devices/platform/v <<
85 sudo IGT_DEVICE="sys:/sys/devices/platform/v <<
86 17
87 You can also run subtests if you do not want t !! 18 - Use the alpha value to blend vaddr_src with vaddr_dst instead of
>> 19 overwriting it in ``blend()``.
88 20
89 sudo ./build/tests/kms_flip --run-subtest ba !! 21 - Add igt test to check cleared alpha value for XRGB plane format.
90 sudo IGT_DEVICE="sys:/sys/devices/platform/v <<
91 22
92 TODO !! 23 - Add igt test to check extreme alpha values i.e. fully opaque and fully
93 ==== !! 24 transparent (intermediate values are affected by hw-specific rounding modes).
94 25
95 If you want to do any of the items listed belo !! 26 Runtime Configuration
96 with VKMS maintainers. !! 27 ---------------------
97 28
98 IGT better support !! 29 We want to be able to reconfigure vkms instance without having to reload the
99 ------------------ !! 30 module. Use/Test-cases:
100 31
101 Debugging: !! 32 - Hotplug/hotremove connectors on the fly (to be able to test DP MST handling of
>> 33 compositors).
102 34
103 - kms_plane: some test cases are failing due t !! 35 - Configure planes/crtcs/connectors (we'd need some code to have more than 1 of
>> 36 them first).
104 37
105 Virtual hardware (vblank-less) mode: !! 38 - Change output configuration: Plug/unplug screens, change EDID, allow changing
>> 39 the refresh rate.
106 40
107 - VKMS already has support for vblanks simulat !! 41 The currently proposed solution is to expose vkms configuration through
108 tested with kms_flip test; in some way, we c !! 42 configfs. All existing module options should be supported through configfs too.
109 the real hardware vblank. However, we also h <<
110 not support vblank interrupt and completes p <<
111 this case, compositor developers may end up <<
112 hardware. It would be useful to support Virt <<
113 because this can help compositor developers <<
114 multiple scenarios. <<
115 43
116 Add Plane Features 44 Add Plane Features
117 ------------------ 45 ------------------
118 46
119 There's lots of plane features we could add su 47 There's lots of plane features we could add support for:
120 48
121 - Add background color KMS property[Good to ge !! 49 - Real overlay planes, not just cursor.
>> 50
>> 51 - Full alpha blending on all planes.
122 52
123 - Scaling. !! 53 - Rotation, scaling.
124 54
125 - Additional buffer formats, especially YUV fo 55 - Additional buffer formats, especially YUV formats for video like NV12.
126 Low/high bpp RGB formats would also be inter 56 Low/high bpp RGB formats would also be interesting.
127 57
128 - Async updates (currently only possible on cu !! 58 - Async updates (currently only possible on cursor plane using the legacy cursor
129 cursor api). !! 59 api).
130 <<
131 For all of these, we also want to review the i <<
132 all relevant igt testcases work on vkms. They <<
133 project. <<
134 <<
135 Runtime Configuration <<
136 --------------------- <<
137 <<
138 We want to be able to reconfigure vkms instanc <<
139 module. Use/Test-cases: <<
140 <<
141 - Hotplug/hotremove connectors on the fly (to <<
142 of compositors). <<
143 <<
144 - Configure planes/crtcs/connectors (we'd need <<
145 them first). <<
146 <<
147 - Change output configuration: Plug/unplug scr <<
148 the refresh rate. <<
149 60
150 The currently proposed solution is to expose v !! 61 For all of these, we also want to review the igt test coverage and make sure all
151 configfs. All existing module options should b !! 62 relevant igt testcases work on vkms.
152 too. <<
153 63
154 Writeback support 64 Writeback support
155 ----------------- 65 -----------------
156 66
157 - The writeback and CRC capture operations sha !! 67 Currently vkms only computes a CRC for each frame. Once we have additional plane
158 boolean to ensure vblanks. Probably, when th !! 68 features, we could write back the entire composited frame, and expose it as:
159 composer_enabled needs to refcounting the co <<
160 [Good to get started] <<
161 69
162 - Add support for cloned writeback outputs and !! 70 - Writeback connector. This is useful for testing compositors if you don't have
163 cloned output in the IGT kms_writeback. !! 71 hardware with writeback support.
164 72
165 - As a v4l device. This is useful for debuggin 73 - As a v4l device. This is useful for debugging compositors on special vkms
166 configurations, so that developers see what' 74 configurations, so that developers see what's really going on.
167 75
>> 76 Prime Buffer Sharing
>> 77 --------------------
>> 78
>> 79 We already have vgem, which is a gem driver for testing rendering, similar to
>> 80 how vkms is for testing the modeset side. Adding buffer sharing support to vkms
>> 81 allows us to test them together, to test synchronization and lots of other
>> 82 features. Also, this allows compositors to test whether they work correctly on
>> 83 SoC chips, where the display and rendering is very often split between 2
>> 84 drivers.
>> 85
168 Output Features 86 Output Features
169 --------------- 87 ---------------
170 88
171 - Variable refresh rate/freesync support. This 89 - Variable refresh rate/freesync support. This probably needs prime buffer
172 sharing support, so that we can use vgem fen 90 sharing support, so that we can use vgem fences to simulate rendering in
173 testing. Also needs support to specify the E 91 testing. Also needs support to specify the EDID.
174 92
175 - Add support for link status, so that composi 93 - Add support for link status, so that compositors can validate their runtime
176 fallbacks when e.g. a Display Port link goes 94 fallbacks when e.g. a Display Port link goes bad.
177 95
178 CRC API Improvements !! 96 - All the hotplug handling describe under "Runtime Configuration".
179 -------------------- <<
180 <<
181 - Optimize CRC computation ``compute_crc()`` a <<
182 97
183 Atomic Check using eBPF 98 Atomic Check using eBPF
184 ----------------------- 99 -----------------------
185 100
186 Atomic drivers have lots of restrictions which 101 Atomic drivers have lots of restrictions which are not exposed to userspace in
187 any explicit form through e.g. possible proper 102 any explicit form through e.g. possible property values. Userspace can only
188 inquiry about these limits through the atomic 103 inquiry about these limits through the atomic IOCTL, possibly using the
189 TEST_ONLY flag. Trying to add configurable cod 104 TEST_ONLY flag. Trying to add configurable code for all these limits, to allow
190 compositors to be tested against them, would b 105 compositors to be tested against them, would be rather futile exercise. Instead
191 we could add support for eBPF to validate any 106 we could add support for eBPF to validate any kind of atomic state, and
192 implement a library of different restrictions. 107 implement a library of different restrictions.
193 108
194 This needs a bunch of features (plane composit 109 This needs a bunch of features (plane compositing, multiple outputs, ...)
195 enabled already to make sense. 110 enabled already to make sense.
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.