Keyboard shortcuts

Press ← or → to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Phase 3: panel-edp knows the mica panel (2026-11-03)

Attempt 2026-11-03-9. Image c360e4af47255421, kernel #9 7.3.0-ga804c5742bff-dirty (2026-11-03 4:17 PM ET), commit c6ef957. Sources: the netlog dumps/firstboot/2026-11-03-netlog.txt and the AP console log dumps/firstboot/2026-11-03-ap.log (both gitignored).

Why

Every boot since the display came up (attempt -4, display-2026-11-03.md) logged a WARNING: drivers/gpu/drm/panel/panel-edp.c:829 at generic_edp_panel_probe backtrace and Unknown panel LGD 0x0804, using conservative timings. The fallback is unprepare 2000 ms and enable 200 ms (panel-edp.c:755-772 at a804c57).

  • No mica source for the timings: the stock tree has no panel node. Its vendor eDP node only names GPIOs and backlight limits (entry01-overlay.dts:119-140). A web search (2026-11-03) found no upstream entry for LGD 0x0804 or “LP134WQ”.
  • Owner’s decisions (2026-11-03, “1a 2a 3b”): carry kernel patches in the repo (ADR 0007), use the LG Display delays, and keep the patch local.

Change (commit c6ef957)

  • kernel/patches/0001-drm-panel-edp-add-lgd-0x0804.patch: EDP_PANEL_ENTRY('L', 'G', 'D', 0x0804, &delay_200_500_e200_d200, "LP134WQ"), after 0x07fe.
    • delay_200_500_e200_d200 (hpd_absent 200, unprepare 500, enable 200, disable 200 ms) is what all 12 LG Display entries use at a804c57. Not from a datasheet.
    • The name does not match the EDID’s cut-off text ("RNYV4?LP134WQ"), so find_edp_panel() matches by panel ID, its second loop (panel-edp.c:2211-2232).
  • scripts/build-kernel.sh applies kernel/patches/ before each build. check.sh check 20 fails when a patch stops applying (red-tested with a broken copy).

Result: pass (panel)

  • panel-simple-dp-aux aux-aea0000.displayport-controller: Detected LGD LP134WQ (0x0804) at 1.675 s. No WARNING and no generic_edp_panel_probe backtrace anywhere in the log.
  • card2-eDP-1 connected at 2560x1600, VREG_EDP_3P3 and VBL9 enabled, backlight 2048/4095.
  • At shutdown (189.5 s) the panel driver logs Skipping disable of already disabled panel and Skipping unprepare of already unprepared panel. Inference: msm had already turned the panel off; not new information about the entry.

Regression seen in the same boot: SCMI timed out

Not caused by this change (the patch touches only panel-edp.c), and attempt -8 with the same config had working SCMI. Details and the open question are in scmi-2026-11-03.md, “Update: attempt -9”.