Different Tonga hardware and a different trigger, same newer-kernel breakage.
Hardware: Apple iMac17,1 (late 2015), AMD R9 M395/M395X Mac Edition (Tonga) [1002:6920], subsystem Apple [106b:014c], internal eDP 5K panel, no external displays. GNOME/Mutter Wayland session (Ubuntu defaults) in all tests. Results on identical hardware: - Ubuntu 22.04 and 24.04 GA kernels (through 6.8): stable — daily driver for years, zero amdgpu errors. - 6.17.0-35 (24.04 HWE): hard GPU hang under sustained CPU load (no video decode active): ring uvd timeout → "atombios stuck in loop for more than 20secs aborting" (tables E57E, B916) → GPU Recovery Failed: -22 → ring sdma1 timeout → kworkers blocked >122s..>737s → power-off required. - 7.0 (26.04): freezes within seconds of launching GPU-accelerated apps; TTY unresponsive. amdgpu.dc=0 helps but freezes still occur, with call traces in the legacy path: dce_v10_0_crtc_disable → amdgpu_dpm_compute_clocks → smu7_send_msg_to_smc / phm_wait_for_register_unequal, repeating at ~5 s intervals. This looks like a different mechanism than the slow-wake EDID issue discussed in drm/amd#5123: no suspend or DPMS involved, no external monitor, and no "No EDID read" anywhere in my logs. Full details and logs upstream: https://gitlab.freedesktop.org/drm/amd/-/work_items/5680. Daily-driver machine, so I can't run development patches, but I can provide further logs and will test any fix once it lands in a release/stable kernel. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2142389 Title: amdgpu (R9 380) fails to resume from suspend (deep sleep) – black screen, requires hard reboot To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2142389/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
