> Thank you for doing your research into the issue but I am confused by > your results. > > Are you claiming that removing mentioned Ubuntu specific commit resolves > the issue?
Specifically it's the first commit where I experience the bug, so I would assume that reverting it will resolve the issue. The commit enables the following compile time kernel config options: CONFIG_INTEL_IOMMU_DEFAULT_ON=y CONFIG_INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON=y These enable the Intel IOMMU by default (although it can still be disabled by the linux command line at runtime via Grub). For completeness here is the corresponding commit in the noble kernel: https://git.launchpad.net/~ubuntu- kernel/ubuntu/+source/linux/+git/noble/commit/?id=77e530c1a864c601b96622db03bc1f38e51155f1 > What about the fact that the issue also affect upstream/mainline > kernels? I don't know how the kernel config values for the mainline builds that I linked to were derived, but in the build that I tried (built July 25 2024) the above config options were also enabled. I don't know if this is because the upstream kernel has now enabled these config options by default or if the builds that I linked to still use custom Ubuntu config options. > Are you trying to say that the issue is just related to the conditions > described in the commit and intel_iommu=igfx_off? Specifying intel_iommu= on the linux command line is a way of overriding at runtime whatever is specified by the CONFIG_INTEL_IOMMU_DEFAULT_ON config option when the kernel is compiled. Specifying intel_iommu=off would be the way to completely undo the config changed by the problematic commit, but since the bug that we are experiencing only (to my knowledge) affects graphics, I figured that intel_iommu=igfx_off would be more conservative workaround, as it only disables Intel's integrated graphics from going through the IOMMU but otherwise keeps the IOMMU enabled. It's not exactly clear to me what the benefits of having graphics going through the Intel IOMMU are, but if Intel's IOMMU was only enabled recently in Ubuntu kernels then I just assume that I won't miss whatever those benefits are. -- You received this bug notification because you are a member of Kernel Packages, which is subscribed to linux in Ubuntu. https://bugs.launchpad.net/bugs/2062951 Title: Random flickering with Intel i915 (Gen9 GPUs) on Linux 6.8 (Ubuntu 24.04) Status in linux package in Ubuntu: Confirmed Bug description: After migrating Mantic 23.10 install to Linux 6.8.0, I am experiencing heavy flickering at random. It happens after a moderate to heavy on screen activity like window switching, scrolling etc. It never happens when playing video or when there is no interaction using keyboard/pointer. It tend to happen when the mouse cursor is in the bottom quarter of the screen and stops immediately when the cursor leaves that screen region. Around the same time the following appears in kernel log: kernel: i915 0000:00:02.0: [drm] *ERROR* CPU pipe A FIFO underrun Happens on both X11 and Wayland on both KDE plasma and Mate (though less). I first encountered the issue with Linux 6.8.0-11 and it still happens with Ubunutu's 6.8.0-28 and Mainline 6.8.7. Kernels 6.6, 6.5, 6.1 and older did not have that issue. Hardware is: Dell XPS 9350, Vendor Intel, Driver i915, Skylake GT2 [HD Graphics 520] Software: Ubuntu 23.10, KDE Plasma 5.27.10 / Mate 1.26 Attaching kernel log for 6.8.0-28 To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2062951/+subscriptions -- Mailing list: https://launchpad.net/~kernel-packages Post to : kernel-packages@lists.launchpad.net Unsubscribe : https://launchpad.net/~kernel-packages More help : https://help.launchpad.net/ListHelp
