Thanks for following up. I performed controlled A/B tests with the same dual-display setup (both the internal eDP panel and external HDMI monitor are driven by nvidia-drm) using S3 deep sleep.
7.0.0-29.29-generic: - This version also reproduces the same failure. I therefore do not have evidence that -29 was a known-good deep-suspend kernel. 7.0.0-28 was not installed or tested. - Resume logged: xhci_hcd 0000:00:0d.0: xHC error in resume, USBSTS 0x401, Reinit usb usb1: root hub lost power or was reset usb usb2: root hub lost power or was reset - The kernel exited suspend, but systemd-suspend.service did not finish until about 25 seconds later (27.655 seconds of reported wall-clock service time). - GNOME Shell was marked unresponsive during display recovery. - After resume, Mutter exposed only a 640x480 mode for eDP-1 (normal state is 2560x1600 at 165 Hz). The internal display stayed black and the external display showed wallpaper only. A forced reboot was required. 7.0.0-31.31-generic from resolute-proposed: - I booted Ubuntu 7.0.0-31.31-generic (upstream 7.0.14) with the Canonical-signed precompiled NVIDIA 595.84 open module. Secure Boot remained enabled and the NVIDIA module/display path was active. - Before suspend, eDP-1 was 2560x1600 at 165 Hz and HDMI-1 was 2560x1440 at 100 Hz. - The first deep resume failed in the same way and logged the same xHCI USBSTS 0x401/root-hub reset messages. - systemd-suspend.service again took about 25 seconds after the kernel returned, reporting 27.355 seconds of wall-clock service time. - GNOME Shell was marked unresponsive. The internal panel remained black and the external monitor remained wallpaper-only for more than 60 seconds, so I had to force reboot. - After the first failed resume, a later short power-button press caused logind to initiate a second suspend. NVIDIA then emitted NV_ERR_INVALID_STATE/MMU unmap assertions. These messages occurred after the initial failed resume and are not the initial trigger. For comparison, 7.0.0-30 with MemorySleepMode=s2idle resumes correctly on the same hardware and display setup, without the xHCI 0x401 error or broken display topology. Therefore: - 7.0.0-29 is not a known-good deep-suspend version on this machine. - 7.0.0-31 from proposed does not fix this report. - The failure was already present by 7.0.0-29; I have not established an older known-good deep-suspend kernel. - Bug 2161309 appears technically different: its cited fix changes the i915 eDP backlight path, while this system's eDP and HDMI connectors are both driven by nvidia-drm. The -31 test also confirms that fix did not resolve this failure. Please let me know which older kernel or additional diagnostic you would like me to test next. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2164645 Title: [Lenovo 82YA][NVIDIA RTX 4060] S3 deep resume leaves internal display black and external display wallpaper-only; s2idle works To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2164645/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
