Repository navigation
Description
Summary
On a 14" MacBook Pro M1 Pro (J314s), Aurora 2026.10.11.15 (7.1.12-2-20261011.15-sep-ARCH), the internal panel's DRM connector reports a physical size of 0x0 mm, although the device tree has it:
$ modetest -M apple -c
47 46 connected eDP-1 0x0 6 46
61 60 connected HDMI-A-1 600x340 29 60
74 76 connected DP-1 600x340 6 76
/proc/device-tree/soc/dcp@38bc00000/panel: width-mm = 302, height-mm = 196
Hyprland shows the same (hyprctl monitors: physicalWidth 0, physicalHeight 0 for eDP-1). Anything that works out real size from it falls back or skips the panel. I ran into this testing omacom/omarchy-mac#638, whose "Linked displays" scaling can't include the built-in panel because it has no size (omacom/omarchy-mac#638 (comment)).
Why
The driver does read the size: drivers/gpu/drm/apple/dcp.c takes width-mm/height-mm from the apple,panel node into dcp->panel at probe, dcp_attributes_replace() puts them into the modes, and dcp_set_dimensions() writes them to connector->display_info. But every probe goes through dcp_get_modes() (drivers/gpu/drm/apple/iomfb.c), which calls drm_edid_connector_update(connector, edid). The internal panel has no EDID, so edid is NULL, and drm_edid_connector_update() → drm_reset_display_info() sets width_mm/height_mm back to 0. The comment right after that call already says the EDID update "resets display_info", and puts back the VRR range for that reason, but not the size. The modes still carry 302x196; only the connector loses it.
Fix (tested)
After the EDID update, put the size back when the connector has none, the same way the VRR range is put back:
--- a/drivers/gpu/drm/apple/iomfb.c
+++ b/drivers/gpu/drm/apple/iomfb.c
@@ -721,6 +721,20 @@
}
/*
+ * Without an EDID (an internal panel), the EDID update above also reset
+ * the physical size. Put back what the modes carry: the size DCP reported
+ * or, failing that, the panel node's width-mm/height-mm.
+ */
+ if (count && !connector->display_info.width_mm) {
+ mutex_lock(&dcp->modes_lock);
+ if (dcp->nr_modes) {
+ connector->display_info.width_mm = dcp->modes[0].mode.width_mm;
+ connector->display_info.height_mm = dcp->modes[0].mode.height_mm;
+ }
+ mutex_unlock(&dcp->modes_lock);
+ }
+
+ /*
* An internal panel has no EDID, so nothing fills in the refresh range
* that userspace needs before it will drive VRR. Supply the range DCP
* reported for the mode. This has to follow the EDID update, whichTested on the J314s as an out-of-tree appledrm.ko built from the aurora-2026.10.11.15 source with only this change, on the release kernel. After a reboot:
$ modetest -M apple -c
47 46 connected eDP-1 302x196 6 46
61 60 connected HDMI-A-1 600x340 29 60
74 76 connected DP-1 600x340 6 76
hyprctl monitors now shows physicalWidth 302, physicalHeight 196 for eDP-1. The LG UltraFine 5K (Thunderbolt) and the Dell U2715H (HDMI) came up as before, and the external sizes are unchanged because they come from their EDIDs. modes[0] already holds the size DCP reported, or the panel node's width-mm/height-mm as a fallback, so this reuses the same choice dcp_set_dimensions() makes.
Activity
cb2206 commentedon Nov 10, 2026
The fix for this landed upstream in b7c75a9 ("drm/apple: restore panel dimensions after mode probes", 2026-10-26). It restores width_mm/height_mm after
drm_edid_connector_update()clears them, which is the same root cause I found here. It should be in the next linux-aurora release. Once that's on my j314s, I can confirm that the panel reports its real size again (instead of 0 mm). Until then, this could be closed as fixed, or kept open until that check.cb2206 commentedon Nov 11, 2026
Confirmed on my j314s: with the packaged
linux-aurora 7.2.11.aurora16-1(7.2.11-16-1-ARCH, no local appledrm module), the built-in panel reports 302 × 196 mm to userspace again (hyprctl monitors: eDP-1physicalWidth 302,physicalHeight 196). From my side this can be closed.