I appear to be seeing the same issue. Below is output from Claude after long debugging session. Please let me know if I can provide additional debug info: Reproducing this same TTM bulk-move list corruption / NULL pointer crash on different hardware — Framework Laptop 13 (AMD Ryzen AI 300 Series), Ryzen AI 7 350, Radeon 860M (gfx_v11 / DCN 3.5), same kernel build 7.0.0-29-generic on Ubuntu 26.04. Linux Freya 7.0.0-29-generic #29-Ubuntu SMP PREEMPT_DYNAMIC Fri Jul 17 20:52:35 UTC 2026 x86_64 GNU/Linux Description: Ubuntu 26.04 LTS ii linux-image-7.0.0-29-generic 7.0.0-29.29 amd64 Signed kernel image generic Multiple independent crashes so far, both NULL derefs in ttm_lru_bulk_move_tail via the same amdgpu command-submission path: Crash 1 (triggered by Papers/PDF viewer during normal use): RIP: ttm_lru_bulk_move_tail+0x1a5/0x360 [ttm] CR2: 0000000000000008 Call Trace: amdgpu_vm_move_to_lru_tail -> amdgpu_cs_submit -> amdgpu_cs_ioctl -> amdgpu_drm_ioctl Crash 2 (triggered by a terminal app (ptyxis), immediately after resuming from suspend-then-hibernate): RIP: ttm_lru_bulk_move_tail+0x15c/0x360 [ttm] CR2: 0000000000000180 Call Trace: amdgpu_vm_move_to_lru_tail -> amdgpu_cs_submit -> amdgpu_cs_ioctl -> amdgpu_drm_ioctl Same crash site, different offset/address each time, different triggering app both times — looks like a race rather than a deterministic bug. Notably, in my case the crash appears strongly correlated with resume from suspend/hibernate: the first GPU command submitted by whatever app happens to run right after resume is what crashes, which fits ttm_lru_bulk_move_pos_tail/ttm_lru_bulk_move_tail sharing the same underlying bulk-move list corruption this bug describes, just hit via a different sibling function. Happy to test a kernel with Thomas Hellstrom's candidate patch (https://lore.kernel.org/all/20260714141400.20904-1-thomas.hellstrom@linux.intel.com/) if that would help move this forward — let me know how to build/obtain a test kernel. Full crash reports (apport) attached: linux-image-7.0.0-29-generic-202608161036.crash, linux-image-7.0.0-29-generic-202608182014.crash ** Attachment added: "linux-image-7.0.0-29-generic-202608112328.crash" https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2163363/+attachment/5993399/+files/linux-image-7.0.0-29-generic-202608112328.crash -- You received this bug notification because you are subscribed to linux in Ubuntu. Matching subscriptions: Bgg, Bmail, Nb https://bugs.launchpad.net/bugs/2163363 Title: [amdgpu/ttm] TTM list corruption and NULL dereference under GPU memory pressure on ASUS ProArt PX13 (7.0.0-29) Status in linux package in Ubuntu: Incomplete Bug description: An ASUS ProArt PX13 HN7306EAC with an AMD integrated GPU (PCI ID 1002:1586) repeatedly experienced kernel-side TTM/AMDGPU corruption and complete graphical failure while running Ubuntu 26.04 kernel 7.0.0-29-generic. The failures occurred with different visible userspace processes. They are therefore reported as possible triggers, not assumed root causes. FIRST HARD FREEZE / NULL DEREFERENCE A complete desktop freeze required a hard reset. The first fatal event was: BUG: kernel NULL pointer dereference, address: 0000000000000019 Comm: ptyxis RIP: ttm_lru_bulk_move_pos_tail+0x53/0xd0 [ttm] Relevant call path: ttm_resource_add_bulk_move ttm_bo_populate ttm_bo_vm_fault_reserved amdgpu_gem_fault A previous occurrence involved firefox:gl0 and failed in the same TTM/AMDGPU resource-management area. LATER TTM LIST CORRUPTION Immediately before the warnings, the kernel repeatedly logged: amdgpu_vm_validate() failed. Not enough memory for command submission! This was followed by: list_add corruption list_del corruption Relevant call paths included: ttm_resource_move_to_lru_tail ttm_bo_populate ttm_bo_vm_fault_reserved amdgpu_gem_fault and: ttm_resource_fini ttm_resource_free amdgpu_bo_move amdgpu_cs_bo_validate amdgpu_cs_ioctl LATER GRAPHICAL-SESSION FAILURE AMDGPU again rejected command submissions with -12. GNOME Shell then aborted with SIGABRT in Mesa Gallium, closing the Wayland session and all terminal applications. This event was a graphical-session restart, not a full machine reboot. ENVIRONMENT Hardware: ASUS ProArt PX13 HN7306EAC CPU/GPU: AMD Ryzen AI Max+ 395 / Radeon 8060S GPU PCI ID: 1002:1586 BIOS: HN7306EAC.307 (2026-01-27) RAM: 128 GB shared system/GPU memory Distribution: Ubuntu 26.04 Kernel: 7.0.0-29-generic, package 7.0.0-29.29 (upstream base 7.0.12) Desktop: GNOME on Wayland WORKLOAD / POSSIBLE TRIGGER Local llama.cpp/Vulkan language-model workloads and desktop applications were active. They use shared system/GPU memory and may increase memory pressure. This is reported as a possible trigger only, not as a demonstrated root cause. NEGATIVE EVIDENCE The inspected logs before the first Oops did not show an OOM kill, MCE, thermal shutdown, PCIe AER error, GPU ring timeout, or GPU reset. The kernel was not tainted before the first Oops. EXPECTED RESULT Userspace GPU activity or memory pressure must not corrupt TTM lists, trigger a kernel NULL pointer dereference, or terminate the graphical session. PRIVACY No vmcore, dump-incomplete, or full memory image is attached because those artifacts may contain credentials, documents, medical data, and browser contents. RELATED REPORT A public report with a similar ttm_lru_bulk_move_pos_tail signature and hard desktop freeze is available at: https://github.com/basecamp/omarchy/issues/6443 --- ProblemType: Bug ApportVersion: 2.34.1-0ubuntu0.1 Architecture: amd64 CasperMD5CheckResult: pass CurrentDesktop: ubuntu:GNOME DistroRelease: Ubuntu 26.04 InstallationDate: Installed on 2026-08-01 (15 days ago) InstallationMedia: Ubuntu 26.04 "Resolute Raccoon" - Release amd64 (20260423.1) MachineType: ASUS ProArt PX13 HN7306EAC Package: linux (not installed) ProcEnviron: LANG=es_ES.UTF-8 PATH=(custom, no user) SHELL=/bin/bash TERM=xterm-256color XDG_RUNTIME_DIR=<set> ProcFB: 0 amdgpudrmfb ProcKernelCmdLine: BOOT_IMAGE=/boot/vmlinuz-7.0.0-29-generic root=UUID=80141fa6-a50a-4bc8-816b-cc4d43d1fa74 ro quiet splash crashkernel=2G-4G:320M,4G-32G:512M,32G-64G:1024M,64G-128G:2048M,128G-:4096M ProcVersionSignature: Ubuntu 7.0.0-29.29-generic 7.0.12 Tags: resolute wayland-session Uname: Linux 7.0.0-29-generic x86_64 UpgradeStatus: No upgrade log present (probably fresh install) UserGroups: adm cdrom dip lpadmin lxd plugdev sudo users _MarkForUpload: True dmi.bios.date: 01/27/2026 dmi.bios.release: 5.36 dmi.bios.vendor: American Megatrends International, LLC. dmi.bios.version: HN7306EAC.307 dmi.board.asset.tag: ATN12345678901234567 dmi.board.name: HN7306EAC dmi.board.vendor: ASUSTeK COMPUTER INC. dmi.board.version: 1.0 dmi.chassis.asset.tag: No Asset Tag dmi.chassis.type: 31 dmi.chassis.vendor: ASUSTeK COMPUTER INC. dmi.chassis.version: 1.0 dmi.ec.firmware.release: 3.6 dmi.modalias: dmi:bvnAmericanMegatrendsInternational,LLC.:bvrHN7306EAC.307:bd01/27/2026:br5.36:efr3.6:svnASUS:pnProArtPX13HN7306EAC:pvr1.0:rvnASUSTeKCOMPUTERINC.:rnHN7306EAC:rvr1.0:cvnASUSTeKCOMPUTERINC.:ct31:cvr1.0:sku:pfaProArtPX13: dmi.product.family: ProArt PX13 dmi.product.name: ProArt PX13 HN7306EAC dmi.product.version: 1.0 dmi.sys.vendor: ASUS To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2163363/+subscriptions
Комментариев нет:
Отправить комментарий