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
|
From 0cdc68b17ed3290a233e699dc31e973650b216ee Mon Sep 17 00:00:00 2001
From: Daniel Golle <daniel@makrotopia.org>
Date: Fri, 25 Sep 2026 11:22:40 +0100
Subject: [PATCH] drm: Probe connectors while holding DRM master
The DRM renderer drops master immediately after opening the device and
takes it back only when the splash is activated, so the connector
enumeration in query_device () runs as a client that is not DRM master.
The kernel refuses a forced probe from such a client and returns the
cached connection status instead:
User-space requested a forced probe on [CONNECTOR:84:HDMI-A-1] but is
not the DRM master, demoting to read-only probe
On a display nothing has driven yet that status is unknown, so no output
counts as connected, no controller is assigned, and the renderer gives
up with "Could not initialize heads". The splash therefore appears only
on a screen that some other client has already lit, which on a system
without fbdev emulation never happens.
This commit holds master for the duration of the connector enumeration,
dropping it again unless the backend was already active. The probe then
detects for real, reads the EDID and brings up a sink that had gone to
low power. On an rk3588 board with no other DRM client running, this
takes the HDMI connector from unknown to connected at 1920x1080.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
---
src/plugins/renderers/drm/plugin.c | 57 ++++++++++++++++++++++--------
1 file changed, 43 insertions(+), 14 deletions(-)
--- a/src/plugins/renderers/drm/plugin.c
+++ b/src/plugins/renderers/drm/plugin.c
@@ -1421,6 +1421,46 @@ check_simpledrm_resolution (ply_renderer
return true;
}
+/* Probe every connector, holding DRM master so the kernel performs a real
+ * detect rather than handing back the cached connection status.
+ */
+static bool
+probe_outputs (ply_renderer_backend_t *backend,
+ ply_output_t *outputs,
+ int outputs_len,
+ bool force,
+ bool *changed)
+{
+ bool was_master;
+ bool ret = true;
+ int i;
+
+ was_master = backend->is_active;
+ if (!was_master)
+ drmSetMaster (backend->device_fd);
+
+ backend->connected_count = 0;
+ for (i = 0; i < outputs_len; i++) {
+ get_output_info (backend, backend->resources->connectors[i], &outputs[i]);
+
+ if (!force && !check_simpledrm_resolution (backend, &outputs[i])) {
+ ret = false;
+ break;
+ }
+
+ if (check_if_output_has_changed (backend, &outputs[i]))
+ *changed = true;
+
+ if (outputs[i].connected)
+ backend->connected_count++;
+ }
+
+ if (!was_master)
+ drmDropMaster (backend->device_fd);
+
+ return ret;
+}
+
/* Update our outputs array to match the hardware state and
* create and/or remove heads as necessary.
* Returns true if any heads were modified.
@@ -1447,20 +1487,9 @@ create_heads_for_active_connectors (ply_
outputs = calloc (backend->resources->count_connectors, sizeof(*outputs));
outputs_len = backend->resources->count_connectors;
- backend->connected_count = 0;
- for (i = 0; i < outputs_len; i++) {
- get_output_info (backend, backend->resources->connectors[i], &outputs[i]);
-
- if (!force && !check_simpledrm_resolution (backend, &outputs[i])) {
- free (outputs);
- return false;
- }
-
- if (check_if_output_has_changed (backend, &outputs[i]))
- changed = true;
-
- if (outputs[i].connected)
- backend->connected_count++;
+ if (!probe_outputs (backend, outputs, outputs_len, force, &changed)) {
+ free (outputs);
+ return false;
}
/* Step 2:
|
