https://bugs.kde.org/show_bug.cgi?id=525886
--- Comment #4 from Sergey <[email protected]> --- Created attachment 197445 --> https://bugs.kde.org/attachment.cgi?id=197445&action=edit drm_info output, nvidia-drm / GB206M (RTX 5060 Max-Q), open module 615.71.09, kernel 7.2.4-3-cachyos-bore-lto Attaching drm_info output as requested (drm_info 2.10.0, node /dev/dri/card1, nvidia-drm, PCI 10de:2d19 GB206M [GeForce RTX 5060 Max-Q], NVIDIA open kernel module 615.71.09, kernel 7.2.4-3-cachyos-bore-lto, Wayland session). The dump is from a normal running session, not from a broken one. The freeze is not deterministic on this machine: I disabled the restart-on-resume workaround from the description, suspended, resumed, and the resume came back clean. No "Context is lost" anywhere in plasmashell's journal, no scenegraph teardown, the panel kept rendering. So there is nothing post-resume to attach for this cycle. Two things that may matter for the driver theory: 1. Video memory preservation is already on here. PreserveVideoMemoryAllocations=2 in /proc/driver/nvidia/params, and nvidia-suspend.service / nvidia-resume.service are enabled (both from the nvidia package). This machine still got the context loss in the original report, so preservation being enabled does not prevent it. 2. The resume that did break was a generally bad one at the driver level, not only in plasmashell: kwin logged "atomic commit failed: Permission denied" at 08:28 and nvidia_drm logged a sync FD semaphore registration error earlier in the same boot. kwin and other applications recovered their GL contexts afterwards, plasmashell did not. I have a journal hook installed that captures drm_info plus both journals on every resume, so if the freeze shows up again I will attach that dump from the actual failure. (In reply to kdedev from comment #2) > The crash needs a separate report. That one is not mine to file, my plasmashell never got quit manually while frozen. I have not seen a DrKonqi dialog here. Max Nomad can file it, or I can try to reproduce with the workaround disabled for a few cycles if that helps. Setting the status back to REPORTED since the requested information is attached. -- You are receiving this mail because: You are watching all bug changes.
