Does someone know why everything is extremely stuttery with Wayland in KDE Plasma 6? This includes the mouse cursor itself. Moving the mouse results in a lot of hitching and stuttering of the cursor itself. It’s as if the mouse cursor is getting updated at 20FPS or something like that.
Everything is perfectly smooth with X11.
I had similar log spam. I had to start chromium browsers with --disable-gpu-memory-buffer-video-frames to get rid of the spam while playing videos.
@RealNC That’s probably caused by the GSP Firmware. Try disabling it in the NVIDIA proprietary drivers.
Run the following command:
sudo nano /etc/modprobe.d/nvidia.conf
Then add this line:
options nvidia NVreg_EnableGpuFirmware=0
Still having this issue running 560.35.03-13
@edgarscirulis I did and rebooted. Verified with nvidia-smi -q | grep GSP and it says “N/A”.
Nothing changed. Mouse stutters like crazy. If I start an empty Weston sessions, it’s fine. But kwin_wayland is completely unusable.
@RealNC Is your taskbar floating? If yes, set it to non-floating. Is your taskbar transparent? If yes, set it to non-transparent.
Legend: Taskbar - Panel
So I tried that flag. It did get rid of the log spam, but it does not solve the fullscreen video stutter problem on Youtube, which is my main annoynace.
I have a laptop with an HDMI attached external monitor, the issue I’m having is games freezing very soon after starting if the game is on the external monitor.
Laptop: MSI GS66 Stealth 10UH
Memory: 32g
iGPU: Intel CometLake-H GT2
dGPU: GeForce RTX 3080 Mobile
Monitor (internal): Sharp LQ156M1JW23 (1920x1080, 300Hz)
Monitor (external): Sceptre F27 - HDMI - primary extended right (1920x1080, 75Hz)
OS(1): Fedora 37, Kernel 6.5.12-100.fc37.x86_64, Xfce 4.18.1, NVIDIA 545.29.06
OS(2): Fedora 40, Kernel 6.10.6-200.fc40.x86_64, Xfce 4.16.1, NVIDIA 560.35.03
I recently installed Fedora 40 as a 2nd bootable partition along side my existing Fedora 37 partition. On F40 I’m using driver version 560.35.03 and on F37 driver version 545.29.06.
On F37, I can start Steam games on both the internal or external monitor using Proton 8 and the games run fine.
On F40, again using Steam/Proton 8, the games will freeze if they’re started on the external monitor. It depends on the game, but usually doesn’t take very long. For instance Skyrim and Fallout 4 generally freeze during one of the loading screens, often even before loading a save file. Witcher 3 generally seems to load a save file fine and can even run for a while, but usually if I bring up the map, the game will freeze shortly after.
The problem only seems to occur with Windows games (via Proton). If I run a native linux game (via Steam) on the external monitor (eg Witcher 2, Portal 2, Half-Life) they seem to work fine. So maybe this is more Proton related, but then again, on F37 the games work fine and I’m using the same version of Proton and an older version of the NVIDIA drivers.
PS. I’ve also tried using Proton 9 on F40 and if the game is started on the external monitor, it freezes. With Proton 9 on F37 the games all play fine.
PPS. There’s a separate issue with Proton 9 that I assume is unrelated to the freezing issue, but perhaps they’re related, so here it is in the spirit of full disclosure. With Proton 8, the game will start on whichever monitor is configured as PRIMARY, regardless of the position relative to the internal monitor (ie, right, left, above, below). This is what I’d expect. But with Proton 9, the game will only start at position 0,0, whatever monitor is configured as PRIMARY seems to be ignored. So if the external monitor is configured as PRIMARY to the left or above, the game starts on the external monitor. But if the external monitor is configured as PRIMARY to the right or below, the game starts on the internal monitor.
I noticed these things in X11 as well when I upgraded from Plasma 5 to 6, so I had already disabled them.
There is still this bug where many Unity versions (eg. 2022.3.22f1) crashes when a project uses Vulkan API. is it something that will be fixed?
I confirm this bug when I launch applications depening upon wgpu, they run fine when I use DRI_PRIME=1 i.e. apps are not running on iGPU with 560.
I see errors like this,
Protocol error 0 on object wl_surface@19:
@abchauhan @amrits Can you please confirm whether this bug has been acknowledged ?
Operating System: RHEL 9.4 (Plow) GNOME 40.4.0 Intel® Core™ i9-14900KF × 32 NVIDIA GeForce RTX 4090/PCIe/SSE2. NVIDIA Driver Version: 560.35.03 NVML Version: 12.560.35.03 Wayland Display: Odyssey G70B Outputs 2 Current Mode: 3840x2160 @60.00
The Joined Displays (Single display option) has a flicking problem on 1 of the displays. It goes away when both displays are mirrored. It only happens in the full screen mode and disappears when the windows key is selected showing multiple open windows used.
I’ve reported this to KDE but to them it looks like a Nvidia bug. Sometimes, not always, but often, KRunner won’t open and it throws this error: krunner[8182]: qt.qpa.wayland: eglSwapBuffers failed with 0x300d, surface: 0x5a621732fb50
In any case this is important, the surface address stays the same.
Restarting KRunner fixes it until it eventually breaks again.
Yes, it is acknowledged: 560 release feedback & discussion - #371 by abchauhan with NVBug #4840658
There’s lot of stutters in Gnome 46, Fedora 40 with 560 on 4090 mobile with optimus; a lot more when compared to driver version 555. Disabling GSP firmware is of no use. I wonder if there’s any other variable which fixes stutter for those who had disabled GSP firmware.
Not sure if anyone had this problem but with the 560 driver and X11, KDE lockscreen “hangs” here, the clock is correct on the screen but I get no login dialog after resuming from a screenlock (a workaround is to go to console and login as root and kill the lockscreen process). Does not happen with 550.
I see nothing special in the logs either.
@shelter : Yes, can confirm I’ve noticed the same thing multiple times in my 560 testing. I went back to 555 because proton/32-bit apps don’t seem to work using the graphics-driver PPA installation, but before that I had seen enough of those kscreenlocker hangs to consider writing an alias to simplify killing it.
I have this stacktrace appearing frequently when running GravityMark Vulkan RT test, though application itself continues to run normally:
WARNING: CPU: 6 PID: 1761 at include/linux/rwsem.h:80 follow_pte+0x1f0/0x220
[ 70.672846] Modules linked in: rfcomm snd_seq_dummy snd_hrtimer nf_nat_tftp nf_nat_ftp nft_masq nvidia_drm(PO) nvidia_modeset(PO) nf_conntrack_sane nf_conntrack_netbios_ns nf_conntrack_broadcast nf_conntrack_tftp nf_conntrack_ftp nft_fib_inet nft_fib_ipv4 nft_fib_ipv6 nft_fib nvidia_uvm(PO) vboxnetadp(O) vboxnetflt(O) nft_reject_inet nf_reject_ipv4 nf_reject_ipv6 nft_reject nft_ct vboxdrv(O) nvidia(PO) nft_chain_nat nf_tables ebtable_nat ebtable_broute ip6table_nat ip6table_mangle ip6table_raw ip6table_security iptable_nat nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 iptable_mangle iptable_raw iptable_security ip_set ebtable_filter ebtables ip6table_filter iptable_filter qrtr uhid bnep snd_ctl_led snd_soc_skl_hda_dsp snd_soc_hdac_hdmi snd_sof_probes snd_soc_intel_hda_dsp_common binfmt_misc vfat fat snd_hda_codec_realtek snd_hda_codec_generic snd_hda_scodec_component snd_soc_dmic snd_sof_pci_intel_cnl snd_sof_intel_hda_generic soundwire_intel soundwire_cadence snd_sof_intel_hda_common snd_sof_intel_hda_mlink
[ 70.672866] snd_sof_intel_hda snd_sof_pci snd_sof_xtensa_dsp snd_sof iwlmvm snd_sof_utils soundwire_generic_allocation soundwire_bus snd_soc_avs snd_soc_hda_codec snd_soc_skl snd_soc_hdac_hda mac80211 snd_hda_ext_core snd_soc_sst_ipc snd_soc_sst_dsp snd_soc_acpi_intel_match snd_soc_acpi intel_uncore_frequency snd_soc_core intel_uncore_frequency_common intel_tcc_cooling x86_pkg_temp_thermal intel_powerclamp coretemp snd_hda_codec_hdmi uvcvideo snd_compress libarc4 kvm_intel ac97_bus snd_pcm_dmaengine snd_hda_intel uvc videobuf2_vmalloc snd_intel_dspcfg videobuf2_memops snd_intel_sdw_acpi videobuf2_v4l2 kvm btusb videobuf2_common snd_hda_codec iwlwifi videodev btrtl btintel snd_hda_core dell_laptop iTCO_wdt spi_nor processor_thermal_device_pci_legacy intel_pmc_bxt btbcm processor_thermal_device btmtk rapl bluetooth intel_rapl_msr mei_hdcp mei_pxp mtd ee1004 iTCO_vendor_support mc dell_wmi snd_hwdep intel_cstate snd_seq processor_thermal_wt_hint intel_uncore alienware_wmi snd_seq_device processor_thermal_rfim
[ 70.672886] dell_smm_hwmon dell_smbios cfg80211 processor_thermal_rapl snd_pcm thunderbolt pcspkr intel_rapl_common dell_wmi_sysman dcdbas r8169 firmware_attributes_class snd_timer processor_thermal_wt_req intel_wmi_thunderbolt dell_wmi_descriptor wmi_bmof mei_me processor_thermal_power_floor snd processor_thermal_mbox i2c_i801 spi_intel_pci mei spi_intel i2c_smbus soundcore idma64 rfkill realtek intel_soc_dts_iosf intel_pch_thermal int3403_thermal intel_pmc_core int340x_thermal_zone intel_vsec pmt_telemetry int3400_thermal intel_hid pmt_class acpi_thermal_rel sparse_keymap joydev acpi_tad acpi_pad loop dm_multipath nfnetlink i915 i2c_algo_bit drm_buddy ttm crct10dif_pclmul nvme crc32_pclmul crc32c_intel polyval_clmulni polyval_generic drm_display_helper nvme_core hid_multitouch ucsi_acpi ghash_clmulni_intel sha512_ssse3 sha256_ssse3 typec_ucsi sha1_ssse3 cec nvme_auth video typec i2c_hid_acpi i2c_hid wmi pinctrl_cannonlake serio_raw scsi_dh_rdac scsi_dh_emc scsi_dh_alua ip6_tables ip_tables fuse i2c_dev
[ 70.672909] CPU: 6 PID: 1761 Comm: nv_queue Tainted: P W O 6.10.9-200.fc40.x86_64 #1
[ 70.672910] Hardware name: Dell Inc. Dell G15 5510/0H0F1D, BIOS 1.24.0 04/02/2024
[ 70.672911] RIP: 0010:follow_pte+0x1f0/0x220
[ 70.672912] Code: cc cc cc 48 81 e2 00 00 00 c0 48 09 c2 48 f7 d2 48 85 fa 75 20 e8 a0 f4 ff ff 48 8b 35 99 37 86 01 48 81 e6 00 00 00 c0 eb 89 <0f> 0b 48 3b 1f 0f 83 42 fe ff ff bd ea ff ff ff eb b2 49 8b 3c 24
[ 70.672913] RSP: 0018:ffffb1d6412c7b68 EFLAGS: 00010246
[ 70.672914] RAX: 0000000000000000 RBX: 00007fa1a3419000 RCX: ffffb1d6412c7bb0
[ 70.672914] RDX: ffffb1d6412c7ba8 RSI: 00007fa1a3419000 RDI: ffff9b99ad490f20
[ 70.672915] RBP: ffffb1d6412c7bf0 R08: ffffb1d6412c7d48 R09: 0000000000000000
[ 70.672916] R10: 0000000000000200 R11: ffff9b99d2b34008 R12: ffffb1d6412c7bb0
[ 70.672916] R13: ffffb1d6412c7ba8 R14: ffff9b9981dea100 R15: 0000000000000000
[ 70.672917] FS: 0000000000000000(0000) GS:ffff9b9cff300000(0000) knlGS:0000000000000000
[ 70.672918] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 70.672919] CR2: 00007f5c995f267c CR3: 00000002ec428004 CR4: 00000000007706f0
[ 70.672919] PKRU: 55555554
[ 70.672920] Call Trace:
[ 70.672920] <TASK>
[ 70.672921] ? follow_pte+0x1f0/0x220
[ 70.672922] ? __warn.cold+0x8e/0xe8
[ 70.672923] ? follow_pte+0x1f0/0x220
[ 70.672925] ? report_bug+0xff/0x140
[ 70.672926] ? handle_bug+0x3c/0x80
[ 70.672928] ? exc_invalid_op+0x17/0x70
[ 70.672929] ? asm_exc_invalid_op+0x1a/0x20
[ 70.672931] ? follow_pte+0x1f0/0x220
[ 70.672932] ? unmap_page_range+0x17b0/0x18a0
[ 70.672934] follow_phys+0x49/0x110
[ 70.672936] untrack_pfn+0x55/0x120
[ 70.672938] unmap_single_vma+0xa6/0xe0
[ 70.672939] zap_page_range_single+0x122/0x1d0
[ 70.672941] unmap_mapping_range+0x116/0x140
[ 70.672943] nv_revoke_gpu_mappings+0x67/0xb0 [nvidia]
[ 70.673112] _nv000723rm+0x35/0xf6 [nvidia]
[ 70.673335] rm_execute_work_item+0x110/0x170 [nvidia]
[ 70.673611] os_execute_work_item+0x5e/0x80 [nvidia]
[ 70.673804] _main_loop+0x8f/0x150 [nvidia]
[ 70.673979] ? __pfx__main_loop+0x10/0x10 [nvidia]
[ 70.674153] kthread+0xcf/0x100
[ 70.674155] ? __pfx_kthread+0x10/0x10
[ 70.674157] ret_from_fork+0x31/0x50
[ 70.674158] ? __pfx_kthread+0x10/0x10
[ 70.674159] ret_from_fork_asm+0x1a/0x30
[ 70.674161] </TASK>
[ 70.674161] ---[ end trace 0000000000000000 ]---
nvidia-bug-report.log.gz (1.5 MB)
The stutter issues might be due to the DE (Gnome), got no stutters on Plasma 6 and Cosmic Desktop. I also noticed that removing NVreg_EnableGpuFirmware=0 actually improved the frame rate in Gnome against what many others have mentioned here.
I have been able to reproduce this now.
First, you wanna load into another game, for example, Baldur’s Gate 3. Then, load into World of Warcraft while keeping BG3 running. I get this freeze almost instantly every time when my character loads in with the “Pageflip timed out! This is a kernel bug” log messages.
This does not effect the other game, once the screen unfreezes, WoW is in a frozen state and needs to be killed, but the other game is fine and continues to run with no problem.