Greetings I'm running a Debian testing (11) system using Nouveau as a driver for 2 graphics cards: 1. Nvidia 1050 Ti (GP107) and a Nvidia 570 (GF110) driving 5 monitors 1 - 3840x2160 and 4 - 1920x1080s. The 5th monitor was added some about 8 weeks ago and since life got interesting. Previously I would use an uptime that would last anywhere from 4 to 6 months but after adding the 5th monitor - - - well the best has been some few days and the worst - - - a few hours. I starting digging to try and find possible issues. First thing I found was the idea of adding firmware which was an interesting exercise but was successful and now somewhere between 400 and 700 seconds after reboot I'm seeing this: [ 534.790587] nouveau 0000:02:00.0: firmware: direct-loading firmware nouveau/nvc8_fuc084 I was cheering when I got this far - - - - except this flaw was hiding another one that I hope you might be able to help with. The error I see (using dmesg) looks something like this: [25375.252874] perf: interrupt took too long (3168 > 3150), lowering kernel.perf_event_max_sample_rate to 63000 [35577.509444] perf: interrupt took too long (3963 > 3960), lowering kernel.perf_event_max_sample_rate to 50250 [54648.710595] perf: interrupt took too long (4991 > 4953), lowering kernel.perf_event_max_sample_rate to 40000 [77975.516742] nouveau 0000:01:00.0: fifo: FB_FLUSH_TIMEOUT [85039.583604] nouveau 0000:01:00.0: DRM: core notifier timeout [85041.583597] nouveau 0000:01:00.0: DRM: base-0: timeout [85041.585023] nouveau 0000:01:00.0: bus: MMIO read of 00000000 FAULT at 690400 [ IBUS ] [85041.585259] nouveau 0000:01:00.0: bus: MMIO read of 00000000 FAULT at 616d48 [ IBUS ] [85043.587213] nouveau 0000:01:00.0: DRM: core notifier timeout [85045.587202] nouveau 0000:01:00.0: DRM: base-1: timeout [85047.587302] nouveau 0000:01:00.0: DRM: core notifier timeout [85049.587289] nouveau 0000:01:00.0: DRM: base-2: timeout [85051.628464] nouveau 0000:01:00.0: DRM: core notifier timeout [85053.628538] nouveau 0000:01:00.0: DRM: core notifier timeout [85085.181271] nouveau 0000:01:00.0: DRM: core notifier timeout [85087.181829] nouveau 0000:01:00.0: DRM: core notifier timeout [85089.181909] nouveau 0000:01:00.0: DRM: core notifier timeout That's the error but here's another data point from early in the boot cycle: [ 1.989397] nouveau 0000:01:00.0: firmware: direct-loading firmware nvidia/gp107/acr/ucode_unload.bin [ 1.989401] nouveau 0000:01:00.0: pmu: firmware unavailable [ 1.989535] nouveau 0000:01:00.0: firmware: direct-loading firmware nvidia/gp107/gr/fecs_bl.bin In digging for what I might be able to do I found this final communication from late February with a subject line of: Re: [PATCH v2] drm/nouveau/pmu: fix timeout on GP108 Is what is discussed in this thread the same issue that I'm having on a GP107? If it is the same issue do I need to apply both of the patches? (If the answer to the second question is yes how do I apply the patch? I've found this technique at 'stackoverflow': 1) git clone <path_to_kernel_sources> 2) git checkout 13fac179aa50556ba3c60790a9beb6ca9d0b1b8b 3) git apply <patch_file> is this the right way to patch the kernel? (Never done this and really don't need my main computer down for even hours nevermind weeks.) Hopefully my explanation is clear enough - - - - if more information from dmesg files are needed - - - well I have 5 or 6 to choose from (called from the second m/c on the network that I reboot the main m/c from). TIA _______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/dri-devel
