Public bug reported: On the ASUS Vivobook S 15 (Snapdragon X / X1E-78-100), when the internal eDP panel is powered off by the screensaver (DRM DPMS off) and then woken up, only half of the panel displays content — the other half stays black. The session keeps running normally; only the panel output is broken. A full display re-initialization restores the panel completely: - Switch to a text VT (Ctrl+Alt+F3) and back (Ctrl+Alt+F2), or - `chvt 3 && chvt <previous>` from a root shell After the VT switch the panel is fully functional again — no reboot or logout needed. This proves the failure is in the display pipeline's power-on path (encoder/panel re-init after DPMS), not in the userspace session. Affected hardware / software ---------------------------- - ASUS Vivobook S 15 S5507QA (Snapdragon X X1E-78-100), internal eDP panel Samsung ATNA33XC20 (driver: panel_samsung_atna33xc20) - Display controller: msm_dpu ae01000.display-controller (drm/msm, DPU 9.0.2: "dpu hardware revision:0x90020000") - Ubuntu 26.04.1 "Resolute", arm64 - Kernel: 7.0.0-30-generic (latest at time of report, resolute-updates) - GNOME Shell on Wayland Reproduction ------------ 1. Boot normally, log in (Wayland session) 2. Let the screensaver blank the panel (GNOME Screen Blank / DPMS power-off) 3. Wake the display (keypad/touch) 4. Only half of the panel shows output; the other half remains black 5. Ctrl+Alt+F3 / Ctrl+Alt+F2 → panel fully restored Notes / observations -------------------- - No drm/dpu error messages are logged at the moment the half-screen state appears — the failure is silent (journalctl -k shows only the boot-time "dpu hardware revision:0x90020000" line). - Unrelated-but-possibly-relevant error seen earlier this boot under heavy software-rendering load: [drm:dpu_crtc_frame_event_cb [msm]] *ERROR* crtc109 event 1 overflow (observed while the GPU was unaccelerated due to missing firmware — see separate linux-firmware report; the half-screen issue persists independently of GPU acceleration state.) - The panel is driven with DSC; a plausible cause is incorrect DSC/tile reconfiguration on the DPMS power-up path in dpu1 encoder code, since a full modeset (VT switch) reprograms everything and recovers. - Not the same as the known x1e80100 eDP HPD pinctrl issue (Stephan Gerhold's Aug 2025 series, display never comes up at all) — here the display works fine from boot and only breaks on DPMS resume. Suggested next steps for maintainers ------------------------------------ - Reproduce with dpms off/on (e.g. `sleep 5 &&swaymsg ...` equivalent: `modetest -M msm_dpu -w` DPMS cycles, or GNOME screen-blank timer) - Instrument dpu1 encoder enable/power-on path around DSC config restore - Compare against the DPMS handling for the same panel on other x1e80100 devices (Lenovo T14s Gen 6 uses the same ATNA33XC20 panel and may be affected identically) Environment (from affected machine) ----------------------------------- DistroRelease: Ubuntu 26.04 Architecture: arm64 MachineType: ASUSTeK COMPUTER INC. Vivobook S 15 S5507QA_S5507QAD Kernel: 7.0.0-30-generic SourcePackage: linux (version 7.0.0-30.30) Tags: arm64 wayland-session kernel-daily-bug Happy to test patches, capture mode dumps (modetest), or edid on request. ** Affects: linux (Ubuntu) Importance: Undecided Status: New -- You received this bug notification because you are subscribed to linux in Ubuntu. Matching subscriptions: Bgg, Bmail, Nb https://bugs.launchpad.net/bugs/2165828 Title: ASUS Vivobook S 15 (x1e80100, Samsung ATNA33XC20 eDP): only half of internal panel displays after DPMS off/on; VT switch (full modeset) recovers Status in linux package in Ubuntu: New Bug description: On the ASUS Vivobook S 15 (Snapdragon X / X1E-78-100), when the internal eDP panel is powered off by the screensaver (DRM DPMS off) and then woken up, only half of the panel displays content — the other half stays black. The session keeps running normally; only the panel output is broken. A full display re-initialization restores the panel completely: - Switch to a text VT (Ctrl+Alt+F3) and back (Ctrl+Alt+F2), or - `chvt 3 && chvt <previous>` from a root shell After the VT switch the panel is fully functional again — no reboot or logout needed. This proves the failure is in the display pipeline's power-on path (encoder/panel re-init after DPMS), not in the userspace session. Affected hardware / software ---------------------------- - ASUS Vivobook S 15 S5507QA (Snapdragon X X1E-78-100), internal eDP panel Samsung ATNA33XC20 (driver: panel_samsung_atna33xc20) - Display controller: msm_dpu ae01000.display-controller (drm/msm, DPU 9.0.2: "dpu hardware revision:0x90020000") - Ubuntu 26.04.1 "Resolute", arm64 - Kernel: 7.0.0-30-generic (latest at time of report, resolute-updates) - GNOME Shell on Wayland Reproduction ------------ 1. Boot normally, log in (Wayland session) 2. Let the screensaver blank the panel (GNOME Screen Blank / DPMS power-off) 3. Wake the display (keypad/touch) 4. Only half of the panel shows output; the other half remains black 5. Ctrl+Alt+F3 / Ctrl+Alt+F2 → panel fully restored Notes / observations -------------------- - No drm/dpu error messages are logged at the moment the half-screen state appears — the failure is silent (journalctl -k shows only the boot-time "dpu hardware revision:0x90020000" line). - Unrelated-but-possibly-relevant error seen earlier this boot under heavy software-rendering load: [drm:dpu_crtc_frame_event_cb [msm]] *ERROR* crtc109 event 1 overflow (observed while the GPU was unaccelerated due to missing firmware — see separate linux-firmware report; the half-screen issue persists independently of GPU acceleration state.) - The panel is driven with DSC; a plausible cause is incorrect DSC/tile reconfiguration on the DPMS power-up path in dpu1 encoder code, since a full modeset (VT switch) reprograms everything and recovers. - Not the same as the known x1e80100 eDP HPD pinctrl issue (Stephan Gerhold's Aug 2025 series, display never comes up at all) — here the display works fine from boot and only breaks on DPMS resume. Suggested next steps for maintainers ------------------------------------ - Reproduce with dpms off/on (e.g. `sleep 5 &&swaymsg ...` equivalent: `modetest -M msm_dpu -w` DPMS cycles, or GNOME screen-blank timer) - Instrument dpu1 encoder enable/power-on path around DSC config restore - Compare against the DPMS handling for the same panel on other x1e80100 devices (Lenovo T14s Gen 6 uses the same ATNA33XC20 panel and may be affected identically) Environment (from affected machine) ----------------------------------- DistroRelease: Ubuntu 26.04 Architecture: arm64 MachineType: ASUSTeK COMPUTER INC. Vivobook S 15 S5507QA_S5507QAD Kernel: 7.0.0-30-generic SourcePackage: linux (version 7.0.0-30.30) Tags: arm64 wayland-session kernel-daily-bug Happy to test patches, capture mode dumps (modetest), or edid on request. To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2165828/+subscriptions
Комментариев нет:
Отправить комментарий