1 |
On Sun, Nov 7, 2021 at 2:04 PM Wol <antlists@××××××××××××.uk> wrote: |
2 |
> |
3 |
> On 07/11/2021 20:39, Mark Knecht wrote: |
4 |
> > Hi, |
5 |
> > I've been playing with a Raspberry Pi 4 as a utility divice for a |
6 |
> > couple of purposes - a backup server for here at home as well as an |
7 |
> > astrophotography manager to run my camera and mount. So far everything |
8 |
> > is fine when I have a monitor attached. I can ssh in and run X apps, |
9 |
> > and I can run the VNC client and control the desktop. |
10 |
> > |
11 |
> > What I find, however, is that when I boot the RP4 without the |
12 |
> > monitor attached, which I would be doing most all the time, I lose the |
13 |
> > desktop functionality through VNC. I can still ssh X apps, and both |
14 |
> > xdpyinfo and xrandr still say the X resolution is 1920 x 1280, but |
15 |
> > there doesn't appear to be a real desktop for VNC to latch onto. |
16 |
> > |
17 |
> > I get that this must be possible as essentially VirtualBox is able |
18 |
> > to do this. I wonder if there's a way for mere mortals to do something |
19 |
> > similar? |
20 |
> > |
21 |
> I'm guessing the reason there isn't a virtual desktop is because |
22 |
> nobody's logged in AT THE PI and started one! |
23 |
> |
24 |
> Just start Xorg over ssh, or plasma-wayland, or whatever. Sorry I don't |
25 |
> (yet) know how to do it, I need to find out because I want to do exactly |
26 |
> the same :-) |
27 |
> |
28 |
> Cheers, |
29 |
> Wol |
30 |
> |
31 |
|
32 |
It seems so far that it's even a little more onerous than that: |
33 |
|
34 |
In the case where the monitor is detached at boot two modules are not |
35 |
loaded - fuse & nvmem_rmem. Other than that the module list appears |
36 |
identical. |
37 |
|
38 |
If I ssh into the RP4 when it's running RPi OS (previously called |
39 |
Raspbian apparently) but with the monitor DISCONNECTED when I boot |
40 |
then screenfetch says there's no GPU. Plugging in the monitor after |
41 |
it's booted doesn't change anything: |
42 |
|
43 |
mark@science:~$ ssh -X pi@192.168.86.84 |
44 |
pi@192.168.86.84's password: |
45 |
Linux raspberrypi 5.10.63-v7l+ #1459 SMP Wed Oct 6 16:41:57 BST 2021 armv7l |
46 |
|
47 |
<SNIP> |
48 |
pi@raspberrypi:~ $ screenfetch -nN |
49 |
pi@raspberrypi |
50 |
OS: Raspbian 10 buster |
51 |
Kernel: armv7l Linux 5.10.63-v7l+ |
52 |
Uptime: 0m |
53 |
Packages: 2038 |
54 |
Shell: 723 |
55 |
Resolution: 5760x1080 |
56 |
DE: KDE5 |
57 |
WM: KWin |
58 |
CPU: ARMv7 rev 3 (v7l) @ 4x 1.5GHz [29.6°C] |
59 |
GPU: |
60 |
RAM: 61MiB / 7898MiB |
61 |
|
62 |
pi@raspberrypi:~ $ lspci -vm -k |
63 |
Device: 00:00.0 |
64 |
Class: PCI bridge |
65 |
Vendor: Broadcom Limited |
66 |
Device: Device 2711 |
67 |
Rev: 10 |
68 |
Driver: pcieport |
69 |
|
70 |
Device: 01:00.0 |
71 |
Class: USB controller |
72 |
Vendor: VIA Technologies, Inc. |
73 |
Device: VL805 USB 3.0 Host Controller |
74 |
SVendor: VIA Technologies, Inc. |
75 |
SDevice: VL805 USB 3.0 Host Controller |
76 |
Rev: 01 |
77 |
ProgIf: 30 |
78 |
Driver: xhci_hcd |
79 |
|
80 |
pi@raspberrypi:~ $ |
81 |
|
82 |
However, booting with the monitor ATTACHED now shows the GPU, or at |
83 |
least a DRM framebuffer possibly, in screenfetch. The PCI devices are |
84 |
the same: |
85 |
|
86 |
pi@raspberrypi:~ $ screenfetch -nN |
87 |
pi@raspberrypi |
88 |
OS: Raspbian 10 buster |
89 |
Kernel: armv7l Linux 5.10.63-v7l+ |
90 |
Uptime: 0m |
91 |
Packages: 2038 |
92 |
Shell: 935 |
93 |
Resolution: 3840x1080 |
94 |
DE: KDE5 |
95 |
WM: OpenBox |
96 |
CPU: ARMv7 rev 3 (v7l) @ 4x 1.5GHz [37.9°C] |
97 |
GPU: vc4drmfb |
98 |
RAM: 144MiB / 7898MiB |
99 |
pi@raspberrypi:~ $ |
100 |
|
101 |
pi@raspberrypi:~ $ lspci -vm -k |
102 |
Device: 00:00.0 |
103 |
Class: PCI bridge |
104 |
Vendor: Broadcom Limited |
105 |
Device: Device 2711 |
106 |
Rev: 10 |
107 |
Driver: pcieport |
108 |
|
109 |
Device: 01:00.0 |
110 |
Class: USB controller |
111 |
Vendor: VIA Technologies, Inc. |
112 |
Device: VL805 USB 3.0 Host Controller |
113 |
SVendor: VIA Technologies, Inc. |
114 |
SDevice: VL805 USB 3.0 Host Controller |
115 |
Rev: 01 |
116 |
ProgIf: 30 |
117 |
Driver: xhci_hcd |
118 |
|
119 |
pi@raspberrypi:~ $ |
120 |
|
121 |
I will repeat this experiment using Ubuntu in a little while to see if |
122 |
the missing GPU is an OS or hardware thing. In this hardware it looks |
123 |
like the GPU hardware is behind the PCI bridge. |
124 |
|
125 |
Still to look into Neil's x2go suggestion. |
126 |
|
127 |
Mark |