| Summary: | kwin_wayland 6.2.0 Consumes All CPU Resources | ||
|---|---|---|---|
| Product: | [Plasma] kwin | Reporter: | erdrick016 |
| Component: | wayland-generic | Assignee: | KWin default assignee <kwin-bugs-null> |
| Status: | RESOLVED DUPLICATE | ||
| Severity: | major | CC: | frmdstryr, kdedev, nate, xaver.hugl |
| Priority: | NOR | Keywords: | regression |
| Version: | 6.2.0 | ||
| Target Milestone: | --- | ||
| Platform: | Neon | ||
| OS: | Linux | ||
| Latest Commit: | Version Fixed In: | ||
| Sentry Crash Report: | |||
| Attachments: |
htop At System Idle
Hotspot - Bottom Up View |
||
Can you run perf to record some perf data? sudo perf record --call-graph dwarf -p $(pidof kwin_wayland) and then open the dump file in hotspot At quick glance, it looks like kwin uses the software renderer kwin_wayland is taking up to 200% CPU on a VM, before and after a dist upgrade to 24.04. It started after Plasma updated to 6.2 Here's the output of perf llvmpipe-0 1405 1418.543079: 1 cycles:P: ffffffff88ac4b06 [unknown] ([kernel.kallsyms]) ffffffff88a1a3d5 [unknown] ([kernel.kallsyms]) ffffffff88a0e026 [unknown] ([kernel.kallsyms]) ffffffff88d9658e [unknown] ([kernel.kallsyms]) ffffffff88d9f610 [unknown] ([kernel.kallsyms]) ffffffff88b50dbf [unknown] ([kernel.kallsyms]) ffffffff89c3ca64 [unknown] ([kernel.kallsyms]) ffffffff89c3ced3 [unknown] ([kernel.kallsyms]) ffffffff89c2d025 [unknown] ([kernel.kallsyms]) ffffffff89c2d0b3 [unknown] ([kernel.kallsyms]) ffffffff89c2b83b [unknown] ([kernel.kallsyms]) ffffffff89e00f4b [unknown] ([kernel.kallsyms]) 7cb3cbe80f4d [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.543130: 1600 cycles:P: ffffffff89ba178b [unknown] ([kernel.kallsyms]) ffffffff88c0594d [unknown] ([kernel.kallsyms]) ffffffff88c06362 [unknown] ([kernel.kallsyms]) ffffffff88c072b6 [unknown] ([kernel.kallsyms]) ffffffff88aa96ee [unknown] ([kernel.kallsyms]) ffffffff89c2b87d [unknown] ([kernel.kallsyms]) ffffffff89e00f4b [unknown] ([kernel.kallsyms]) ffffffff88b50c59 [unknown] ([kernel.kallsyms]) ffffffff89c3ca64 [unknown] ([kernel.kallsyms]) ffffffff89c3ced3 [unknown] ([kernel.kallsyms]) ffffffff89c2d025 [unknown] ([kernel.kallsyms]) ffffffff89c2d0b3 [unknown] ([kernel.kallsyms]) ffffffff89c2b83b [unknown] ([kernel.kallsyms]) ffffffff89e00f4b [unknown] ([kernel.kallsyms]) 7cb3cbe80f4d [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.543384: 1 cycles:P: ffffffff88ac4b06 [unknown] ([kernel.kallsyms]) ffffffff88a1a3d5 [unknown] ([kernel.kallsyms]) ffffffff88a0e026 [unknown] ([kernel.kallsyms]) ffffffff88d9658e [unknown] ([kernel.kallsyms]) ffffffff88d9f610 [unknown] ([kernel.kallsyms]) ffffffff88b50dbf [unknown] ([kernel.kallsyms]) ffffffff89c3ca64 [unknown] ([kernel.kallsyms]) ffffffff89c3ced3 [unknown] ([kernel.kallsyms]) ffffffff89c2d025 [unknown] ([kernel.kallsyms]) ffffffff89c2d0b3 [unknown] ([kernel.kallsyms]) ffffffff89c2b579 [unknown] ([kernel.kallsyms]) ffffffff89e00f8b [unknown] ([kernel.kallsyms]) 7cb3cbe7ff20 [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.543944: 58021 cycles:P: 7cb3cbe7fdfd [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.543981: 55083 cycles:P: 7cb3cbe7fdeb [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.544024: 108871 cycles:P: 7cb3cbe7f621 [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.544156: 88794 cycles:P: 7cb3cbe7ce0a [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.544209: 81305 cycles:P: 7cb3cbe7d7a4 [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.544250: 134360 cycles:P: 7cb3cbe80334 [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.544316: 253553 cycles:P: 7cb3cbe801dd [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.545041: 352165 cycles:P: 7cb3cbe7c862 [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.545107: 201841 cycles:P: 7cb3cbe7d537 [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.545183: 328382 cycles:P: 7cb3cbe7fe3a [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.545197: 184382 cycles:P: 7cb3cbe7e9f5 [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.545280: 246407 cycles:P: 7cb3cbe7d7a4 [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.545317: 383286 cycles:P: 7cb3cbe7fd23 [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.545384: 339701 cycles:P: 7cb3cbe7c9e6 [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.545471: 454590 cycles:P: 7cb3cbe7f697 [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.546048: 416213 cycles:P: 7cb3cbe7e5a1 [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.546086: 510665 cycles:P: 7cb3cbe7f060 [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.546213: 390340 cycles:P: 7cb3cbe7f663 [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.546286: 481423 cycles:P: 7cb3cbe7e423 [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.546370: 439702 cycles:P: 7cb3cbe8018c [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.546475: 521880 cycles:P: 7cb3cbe7dfa7 [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.546544: 501851 cycles:P: 7cb3cbe7e8a5 [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.546679: 571722 cycles:P: 7cb3cbe7fe8c [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.547105: 555383 cycles:P: 7cb3cbe80a8e [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.547147: 621911 cycles:P: 7cb3cbe80244 [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.547324: 527185 cycles:P: 7cb3cbe7e47c [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.547387: 599467 cycles:P: 7cb3cbe80a94 [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.547535: 561513 cycles:P: 7cb3cbe7e47c [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.547619: 628699 cycles:P: 7cb3cbe7e40b [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.547756: 602575 cycles:P: 7cb3cbe7fd05 [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.547862: 663275 cycles:P: 7cb3cbe80a2d [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.548141: 650061 cycles:P: 7cb3cbe7fa9f [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.548171: 708140 cycles:P: 7cb3cbe7fdeb [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.548394: 639110 cycles:P: 7cb3cbe7fcee [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.548445: 715168 cycles:P: 7cb3cbe7d8de [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.548644: 664642 cycles:P: 7cb3cbe801dd [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.548721: 734632 cycles:P: 7cb3cbe7e295 [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.548907: 692295 cycles:P: 7cb3cbe7c8ec [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.549021: 766722 cycles:P: 7cb3cbe7ee22 [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.549198: 732591 cycles:P: 7cb3cbe7e444 [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.549316: 777309 cycles:P: 7cb3cbe7ee2e [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.549486: 745955 cycles:P: 7cb3cbe7ebaa [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.549615: 789995 cycles:P: 7cb3cbe7e3dd [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.549772: 760740 cycles:P: 7cb3cbe8019c [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.549927: 807776 cycles:P: 7cb3cbe7e7e5 [unknown] (/tmp/perf-1354.map) llvmpipe-1 1406 1418.550085: 783640 cycles:P: 7cb3cbe7dc61 [unknown] (/tmp/perf-1354.map) llvmpipe-0 1405 1418.550239: 814474 cycles:P: Created attachment 174653 [details]
Hotspot - Bottom Up View
Here's my perf data as requested. LMK if it would be helpful for me to install debug symbols or something, or maybe you need a different view in Hotspot?
Small data point: I upgraded from Neon 22.04 (Kernel: 6.8.0-40-generic) -> Neon 24.04 (Kernel: 6.8.0-45-generic) and things became slightly better. On 22.04, my 4x CPUs were pegged at 100%, all the time. Now, on 24.04, when nothing is changing on screen at all, they settle down to something akin to background noise. But even just simply typing this comment out makes my CPU usage jump all over the place. Perhaps this is due to https://bugs.kde.org/493940 ? Looking at the commit for that, shouldn't the new timer also be stopped if a repaint is pending (eg after the if at line 251)? *** This bug has been marked as a duplicate of bug 493295 *** |
Created attachment 174636 [details] htop At System Idle SUMMARY After upgrading Plasma from 6.1.5 -> 6.2, it became unusable as kwin_wayland is consuming all my CPU. I'm running a VM (under QEMU) with 4x CPUs allocated to it and it was fine before the upgrade, now Wayland is unusable but X11 is fine. STEPS TO REPRODUCE 1. Login to a Wayland session. OBSERVED RESULT See attachment but kwin_wayland is consuming all my CPU resources, yielding the system nigh unusable. EXPECTED RESULT kwin_wayland would consume some small percentage of CPU. SOFTWARE/OS VERSIONS Linux/KDE Plasma: KDE neon 6.2 KDE Plasma Version: 6.2.0 KDE Frameworks Version: 6.6.0 Qt Version: 6.7.2 ADDITIONAL INFORMATION Processors: 4 × 12th Gen Intel® Core™ i9-12900HK Memory: 7.6 GiB of RAM Graphics Processor: llvmpipe Manufacturer: QEMU Product Name: Standard PC (Q35 + ICH9, 2009) System Version: pc-q35-7.1