This is a follow-up to bug #2085434 which tracks the same TLB invalidation crash on Meteor Lake. My report focuses on the linux- firmware aspect (missing GuC 70.44.1 for MTL), while #2085434 covers the kernel/driver crash itself.
Additional data from my system (Lenovo ThinkPad E16 Gen 2, Core Ultra 7 155H, same motherboard family as E14 Gen 6 from #2085434 comment #11): Kernel 6.17.0-1028-oem, xe driver, linux-firmware 20240318.git3b128b60-0ubuntu2.27. I have now documented 7 TLB timeout events over 17 days: | Date | GT | Type | Outcome | |------|----|------|---------| | Jul 2 | GT1 | Single TLB timeout | Auto-recovered | | Jul 7 | GT1 | Single TLB timeout | Auto-recovered | | Jul 12 | GT0 (x2), GT1 | 3 TLB timeouts | Auto-recovered | | Jul 13 | GT1 | Single TLB timeout | Auto-recovered | | Jul 18 | GT0 | TLB timeout → reset storm | 1801 resets in 11 min, hard reboot | The July 18 reset storm: ``` Jul 18 16:25:27 kernel: *ERROR* GT0: TLB invalidation fence timeout, seqno=1014994 recv=1014993 Jul 18 16:25:31 kernel: *ERROR* GT0: TLB invalidation fence timeout, seqno=1014995 recv=1014993 Jul 18 16:25:34 → 16:36:55 kernel: GT0: reset queued/started/done (1801×) ``` Massive WARNING spam at xe_guc_submit.c:1311 guc_exec_queue_timedout_job across CPUs 1, 6, 8, 16, 18, 19. What I have tried: - VT-d disabled in BIOS (eliminated hard freezes but TLB timeouts still occur as reset storms) - intel_iommu=igfx_off (removed as redundant after VT-d disable) - xe vs i915 driver: both produce the same TLB invalidation timeout (consistent with #2085434) What I have NOT tried: - kobuk-team/intel-graphics PPA (would provide GuC 70.44.1 but also ships custom kernel + MESA — too invasive) - i915.enable_guc=0 (known to cause boot failures on some MTL systems per #2085434 comment #8) The crash affects at least 4 Lenovo Meteor Lake models confirmed in #2085434: - ThinkPad E14 Gen 6 (comment #11) - ThinkPad X1 Carbon Gen 12 (comments #2, #4) - ThinkBook 14 G7 IML (comment #8) - ThinkPad E16 Gen 2 (this report) Request: please include i915/mtl_guc_70.bin version 70.44.1 (or newer) from upstream linux-firmware in the noble package. While I acknowledge this may not fully resolve the TLB crashes (Arch Linux ships 70.44.1 and crashes still occur per bbs.archlinux.org #308313), having the recommended firmware version is a prerequisite for any further debugging and eliminates the kernel warning at every boot. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2161220 Title: linux-firmware: missing GuC firmware 70.44.1 for Intel Meteor Lake (i915/mtl_guc_70.bin) To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux-firmware/+bug/2161220/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
