Public bug reported: DESCRIPTION: == Summary == Repeated kernel panics on AMD Ryzen 7 4800U (Renoir/Zen2) triggered by the SRSO (Speculative Return Stack Overflow) mitigation interacting with the scheduler tick path. The system crashes every 1-11 days with no user-space trigger. Crashes occur in both linux-image-7.0.0-22-generic and linux-image-7.0.0-27-generic. Workaround: processor.max_cstate=1 (preventing deep C-states avoids the crash path). == Hardware == - System: AZW (Beelink) SER Mini PC - CPU: AMD Ryzen 7 4800U with Radeon Graphics (Zen2 / Renoir) - Microcode: 0x0860010d (updated from 0x08600106 at boot) - GPU: AMD Renoir [Radeon Vega Series] (integrated, amdgpu driver) - RAM: 16GB DDR4 - BIOS: SER_V1.14_P3C6M43_B_Link (03/24/2022) — latest available for this model - NVMe storage == Software == - Ubuntu 26.04 LTS (fresh install on 2026-05-26) - Kernel: 7.0.0-22-generic and 7.0.0-27-generic (both affected) - amd64-microcode: 3.20251202.1ubuntu2 - linux-firmware: 20260319.git217ca6e4.1ubuntu - Boot params: quiet splash crashkernel=... - Active mitigation: srso_return_thunk (confirmed via dmesg) - NOPTI active == Crash #1 — 2026-07-10 (kernel 7.0.0-27) == Oops: invalid opcode: 0000 [#1] SMP NOPTI CPU: 0 UID: 0 PID: 0 Comm: swapper/0 RIP: 0010:srso_safe_ret+0x0/0x20 Call Trace: <IRQ> sched_tick+0x48/0x280 update_process_times+0xa0/0xe0 tick_nohz_handler+0x96/0x1a0 __hrtimer_run_queues+0x10c/0x270 hrtimer_interrupt+0x104/0x250 __sysvec_apic_timer_interrupt+0x58/0x150 sysvec_apic_timer_interrupt+0x9a/0xd0 </IRQ> <TASK> asm_sysvec_apic_timer_interrupt+0x1b/0x20 RIP: 0010:cpuidle_enter_state+0xca/0x700 cpuidle_enter+0x30/0x50 cpuidle_idle_call+0x16b/0x1f0 do_idle+0x94/0xf0 </TASK> == Crash #2 — 2026-07-12 (kernel 7.0.0-27) == BUG: unable to handle page fault for address: ffffffff99053b04 #PF: supervisor write access in kernel mode #PF: error_code(0x0002) - not-present page PGD bc447067 P4D bc447067 PUD bc448063 PMD 0 Oops: Oops: 0002 [#1] SMP NOPTI CPU: 0 UID: 1000 PID: 2680076 Comm: curl RIP: 0010:kick_ilb+0x6e/0x180 Call Trace: <TASK> kick_ilb+0x52/0x180 nohz_balancer_kick+0x92/0x2d0 sched_balance_trigger+0x4c/0x90 sched_tick+0xf9/0x280 update_process_times+0xa0/0xe0 tick_nohz_handler+0x96/0x1a0 __hrtimer_run_queues+0x10c/0x270 hrtimer_interrupt+0x104/0x250 __sysvec_apic_timer_interrupt+0x58/0x150 sysvec_apic_timer_interrupt+0x44/0xd0 asm_sysvec_apic_timer_interrupt+0x1b/0x20 </TASK> == Additional context == - amdgpu driver logs "dm_irq_work_func [amdgpu] hogged CPU for >13333us" 35+ times per boot, suggesting IRQ pressure on the scheduler tick path. - Both crashes occur on CPU 0 during timer interrupt → sched_tick. - Crash #1 is an invalid opcode in the SRSO safe_ret thunk itself. - Crash #2 is a page fault in kick_ilb writing to an unmapped kernel address (PMD 0), suggesting corrupted per-cpu data or text patching issue in the SRSO path. - The system was stable for 14 days once (when amdgpu initialized without errors). - Full kdump vmcores are available in /var/crash/ for all crashes. == Crash frequency == 6+ crashes since 2026-05-29, intervals ranging from 1.5 to 11 days. No specific user-space trigger identified — crashes happen during idle or light workloads (cron jobs, curl). == Workaround == Adding processor.max_cstate=1 to kernel boot parameters prevents the CPU from entering deep C-states where the crash occurs. This confirms the interaction between cpuidle exit path, SRSO return thunks, and sched_tick. == Possible root cause == The SRSO mitigation (srso_return_thunk / srso_safe_ret) appears to have a bug in the text patching or in the interaction with the scheduler's nohz/idle balancer on AMD Zen2 processors when exiting deep C-states. The amdgpu driver hogging the CPU in interrupt context likely increases the probability of hitting the race condition. == Kdump files == /var/crash/202607101059/ — Crash #1 (srso_safe_ret) /var/crash/202607120124/ — Crash #2 (kick_ilb) Both contain full vmcore dumps and dmesg extracts. ProblemType: Bug DistroRelease: Ubuntu 26.04 Package: linux-image-7.0.0-27-generic 7.0.0-27.27 ProcVersionSignature: Ubuntu 7.0.0-27.27-generic 7.0.6 Uname: Linux 7.0.0-27-generic x86_64 ApportVersion: 2.34.0-0ubuntu2 Architecture: amd64 AudioDevicesInUse: USER PID ACCESS COMMAND /dev/snd/controlC1: gdm-greeter 2490 F.... wireplumber /dev/snd/controlC0: gdm-greeter 2490 F.... wireplumber /dev/snd/seq: gdm-greeter 2474 F.... pipewire CasperMD5CheckResult: unknown Date: Sun Jul 12 12:34:59 2026 InstallationDate: Installed on 2026-05-26 (47 days ago) InstallationMedia: Ubuntu 26.04 "Resolute Raccoon" - Release amd64 (20260423.1) IwDevWlp2s0Link: Not connected. MachineType: AZW SER ProcEnviron: LANG=en_US.UTF-8 PATH=(custom, no user) SHELL=/bin/bash TERM=screen-256color ProcFB: 0 amdgpudrmfb ProcKernelCmdLine: BOOT_IMAGE=/boot/vmlinuz-7.0.0-27-generic root=UUID=3cc618bf-7287-4c56-86b6-9ddbfe59a14a ro quiet splash crashkernel=2G-4G:320M,4G-32G:512M,32G-64G:1024M,64G-128G:2048M,128G-:4096M SourcePackage: linux UpgradeStatus: No upgrade log present (probably fresh install) dmi.bios.date: 03/24/2022 dmi.bios.release: 5.16 dmi.bios.vendor: American Megatrends Inc. dmi.bios.version: SER_V1.14_P3C6M43_B_Link dmi.board.asset.tag: Default string dmi.board.name: SER dmi.board.vendor: AZW dmi.board.version: V01 dmi.chassis.asset.tag: Default string dmi.chassis.type: 35 dmi.chassis.vendor: AZW dmi.chassis.version: Default string dmi.modalias: dmi:bvnAmericanMegatrendsInc.:bvrSER_V1.14_P3C6M43_B_Link:bd03/24/2022:br5.16:svnAZW:pnSER:pvrDefaultstring:rvnAZW:rnSER:rvrV01:cvnAZW:ct35:cvrDefaultstring:skuSKU4:pfaSER: dmi.product.family: SER dmi.product.name: SER dmi.product.sku: SKU4 dmi.product.version: Default string dmi.sys.vendor: AZW ** Affects: linux (Ubuntu) Importance: Undecided Status: New ** Tags: amd64 apport-bug resolute -- You received this bug notification because you are subscribed to linux in Ubuntu. Matching subscriptions: Bgg, Bmail, Nb https://bugs.launchpad.net/bugs/2160457 Title: Kernel panic in srso_safe_ret / kick_ilb during sched_tick on AMD Ryzen 7 4800U (Zen2) Status in linux package in Ubuntu: New Bug description: DESCRIPTION: == Summary == Repeated kernel panics on AMD Ryzen 7 4800U (Renoir/Zen2) triggered by the SRSO (Speculative Return Stack Overflow) mitigation interacting with the scheduler tick path. The system crashes every 1-11 days with no user-space trigger. Crashes occur in both linux-image-7.0.0-22-generic and linux-image-7.0.0-27-generic. Workaround: processor.max_cstate=1 (preventing deep C-states avoids the crash path). == Hardware == - System: AZW (Beelink) SER Mini PC - CPU: AMD Ryzen 7 4800U with Radeon Graphics (Zen2 / Renoir) - Microcode: 0x0860010d (updated from 0x08600106 at boot) - GPU: AMD Renoir [Radeon Vega Series] (integrated, amdgpu driver) - RAM: 16GB DDR4 - BIOS: SER_V1.14_P3C6M43_B_Link (03/24/2022) — latest available for this model - NVMe storage == Software == - Ubuntu 26.04 LTS (fresh install on 2026-05-26) - Kernel: 7.0.0-22-generic and 7.0.0-27-generic (both affected) - amd64-microcode: 3.20251202.1ubuntu2 - linux-firmware: 20260319.git217ca6e4.1ubuntu - Boot params: quiet splash crashkernel=... - Active mitigation: srso_return_thunk (confirmed via dmesg) - NOPTI active == Crash #1 — 2026-07-10 (kernel 7.0.0-27) == Oops: invalid opcode: 0000 [#1] SMP NOPTI CPU: 0 UID: 0 PID: 0 Comm: swapper/0 RIP: 0010:srso_safe_ret+0x0/0x20 Call Trace: <IRQ> sched_tick+0x48/0x280 update_process_times+0xa0/0xe0 tick_nohz_handler+0x96/0x1a0 __hrtimer_run_queues+0x10c/0x270 hrtimer_interrupt+0x104/0x250 __sysvec_apic_timer_interrupt+0x58/0x150 sysvec_apic_timer_interrupt+0x9a/0xd0 </IRQ> <TASK> asm_sysvec_apic_timer_interrupt+0x1b/0x20 RIP: 0010:cpuidle_enter_state+0xca/0x700 cpuidle_enter+0x30/0x50 cpuidle_idle_call+0x16b/0x1f0 do_idle+0x94/0xf0 </TASK> == Crash #2 — 2026-07-12 (kernel 7.0.0-27) == BUG: unable to handle page fault for address: ffffffff99053b04 #PF: supervisor write access in kernel mode #PF: error_code(0x0002) - not-present page PGD bc447067 P4D bc447067 PUD bc448063 PMD 0 Oops: Oops: 0002 [#1] SMP NOPTI CPU: 0 UID: 1000 PID: 2680076 Comm: curl RIP: 0010:kick_ilb+0x6e/0x180 Call Trace: <TASK> kick_ilb+0x52/0x180 nohz_balancer_kick+0x92/0x2d0 sched_balance_trigger+0x4c/0x90 sched_tick+0xf9/0x280 update_process_times+0xa0/0xe0 tick_nohz_handler+0x96/0x1a0 __hrtimer_run_queues+0x10c/0x270 hrtimer_interrupt+0x104/0x250 __sysvec_apic_timer_interrupt+0x58/0x150 sysvec_apic_timer_interrupt+0x44/0xd0 asm_sysvec_apic_timer_interrupt+0x1b/0x20 </TASK> == Additional context == - amdgpu driver logs "dm_irq_work_func [amdgpu] hogged CPU for >13333us" 35+ times per boot, suggesting IRQ pressure on the scheduler tick path. - Both crashes occur on CPU 0 during timer interrupt → sched_tick. - Crash #1 is an invalid opcode in the SRSO safe_ret thunk itself. - Crash #2 is a page fault in kick_ilb writing to an unmapped kernel address (PMD 0), suggesting corrupted per-cpu data or text patching issue in the SRSO path. - The system was stable for 14 days once (when amdgpu initialized without errors). - Full kdump vmcores are available in /var/crash/ for all crashes. == Crash frequency == 6+ crashes since 2026-05-29, intervals ranging from 1.5 to 11 days. No specific user-space trigger identified — crashes happen during idle or light workloads (cron jobs, curl). == Workaround == Adding processor.max_cstate=1 to kernel boot parameters prevents the CPU from entering deep C-states where the crash occurs. This confirms the interaction between cpuidle exit path, SRSO return thunks, and sched_tick. == Possible root cause == The SRSO mitigation (srso_return_thunk / srso_safe_ret) appears to have a bug in the text patching or in the interaction with the scheduler's nohz/idle balancer on AMD Zen2 processors when exiting deep C-states. The amdgpu driver hogging the CPU in interrupt context likely increases the probability of hitting the race condition. == Kdump files == /var/crash/202607101059/ — Crash #1 (srso_safe_ret) /var/crash/202607120124/ — Crash #2 (kick_ilb) Both contain full vmcore dumps and dmesg extracts. ProblemType: Bug DistroRelease: Ubuntu 26.04 Package: linux-image-7.0.0-27-generic 7.0.0-27.27 ProcVersionSignature: Ubuntu 7.0.0-27.27-generic 7.0.6 Uname: Linux 7.0.0-27-generic x86_64 ApportVersion: 2.34.0-0ubuntu2 Architecture: amd64 AudioDevicesInUse: USER PID ACCESS COMMAND /dev/snd/controlC1: gdm-greeter 2490 F.... wireplumber /dev/snd/controlC0: gdm-greeter 2490 F.... wireplumber /dev/snd/seq: gdm-greeter 2474 F.... pipewire CasperMD5CheckResult: unknown Date: Sun Jul 12 12:34:59 2026 InstallationDate: Installed on 2026-05-26 (47 days ago) InstallationMedia: Ubuntu 26.04 "Resolute Raccoon" - Release amd64 (20260423.1) IwDevWlp2s0Link: Not connected. MachineType: AZW SER ProcEnviron: LANG=en_US.UTF-8 PATH=(custom, no user) SHELL=/bin/bash TERM=screen-256color ProcFB: 0 amdgpudrmfb ProcKernelCmdLine: BOOT_IMAGE=/boot/vmlinuz-7.0.0-27-generic root=UUID=3cc618bf-7287-4c56-86b6-9ddbfe59a14a ro quiet splash crashkernel=2G-4G:320M,4G-32G:512M,32G-64G:1024M,64G-128G:2048M,128G-:4096M SourcePackage: linux UpgradeStatus: No upgrade log present (probably fresh install) dmi.bios.date: 03/24/2022 dmi.bios.release: 5.16 dmi.bios.vendor: American Megatrends Inc. dmi.bios.version: SER_V1.14_P3C6M43_B_Link dmi.board.asset.tag: Default string dmi.board.name: SER dmi.board.vendor: AZW dmi.board.version: V01 dmi.chassis.asset.tag: Default string dmi.chassis.type: 35 dmi.chassis.vendor: AZW dmi.chassis.version: Default string dmi.modalias: dmi:bvnAmericanMegatrendsInc.:bvrSER_V1.14_P3C6M43_B_Link:bd03/24/2022:br5.16:svnAZW:pnSER:pvrDefaultstring:rvnAZW:rnSER:rvrV01:cvnAZW:ct35:cvrDefaultstring:skuSKU4:pfaSER: dmi.product.family: SER dmi.product.name: SER dmi.product.sku: SKU4 dmi.product.version: Default string dmi.sys.vendor: AZW To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2160457/+subscriptions
Комментариев нет:
Отправить комментарий