[Bug 2158978] Re: System freeze on startup: nvidia-modeset/DifrPrefetchEventDeferredWork deadlock with KMS thread (driver 595.71.05, kernel 7.0.0-27)

After upgrading to version 595.84, the issue still persists. Specifically: approximately 1–2 hours after the system suspends, the system completely freezes, and the kernel log still shows the "KMS thread blocked" warning. I have attached the full nvidia-bug- report.log.gz report file for your reference. ** Attachment added: "nvidia-bug-report.log.gz" https://bugs.launchpad.net/ubuntu/+source/nvidia-graphics-drivers-595/+bug/2158978/+attachment/5986278/+files/nvidia-bug-report.log.gz -- You received this bug notification because you are subscribed to linux in Ubuntu. Matching subscriptions: Bgg, Bmail, Nb https://bugs.launchpad.net/bugs/2158978 Title: System freeze on startup: nvidia-modeset/DifrPrefetchEventDeferredWork deadlock with KMS thread (driver 595.71.05, kernel 7.0.0-27) Status in linux package in Ubuntu: Invalid Status in nvidia-graphics-drivers-595 package in Ubuntu: Confirmed Bug description: System freezes completely shortly after login. A deadlock occurs between the nvidia-modeset kernel thread and the KMS thread. Requires hard reboot to recover. Hardware: Lenovo 83JM (BIOS PZCN49WW 2025-06-06) GPU: NVIDIA GeForce RTX 5050 laptop (hybrid GPU with Intel i915) Driver: nvidia 595.71.05 (Open Kernel Module) Kernel: 7.0.0-27-generic (Ubuntu 26.04) --- Kernel log (Jul 02 ~03:58) --- INFO: task KMS thread:5195 blocked for more than 122 seconds. task:KMS thread state:D pid:5195 nvkms_ioctl_from_kapi_try_pmlock+0x55/0xd0 [nvidia_modeset] ApplyModeSetConfig+0x357/0xba0 [nvidia_modeset] nv_drm_atomic_apply_modeset_config+0x80b/0x9c0 [nvidia_drm] drm_atomic_commit+0x6c/0xf0 nv_drm_ioctl+0x38/0x3f0 [nvidia_drm] INFO: task KMS thread:5195 blocked on a semaphore likely last held by task nvidia-modeset/:511 task:nvidia-modeset/ state:R (running but stuck) pid:511 RIP: nvWriteGpEntry+0xa4/0x370 [nvidia_modeset] nvPushKickoff+0x28/0x50 [nvidia_modeset] PrefetchHelperSurfaceEvo+0x45c/0x650 [nvidia_modeset] nvDIFRPrefetchSurfaces+0x135/0x1f0 [nvidia_modeset] DifrPrefetchEventDeferredWork+0x16/0x30 [nvidia_modeset] nvkms_kthread_q_callback+0xe9/0x1a0 [nvidia_modeset] (Repeated every ~2 min for 20+ minutes, from 122s to 1228s blocked) --- Additional errors at driver load --- NVRM: NV_ERR_INVALID_DATA from SBIOS for platform power mode and target temperature nvidia_init_module: "Unpatched return thunk in use" (kernel taint OE) To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2158978/+subscriptions