You are not logged in.
I got segfault when I tried to run applications that require GPU, like `kitty`, `mpv`.
kitty
libEGL warning: egl: failed to create dri2 screen
[1] 2490 segmentation fault (core dumped) kittycoredump: http://0x0.st/HUYY.bin
mpv video.mp4
(+) Video --vid=1 (*) (h264 1280x720 29.970fps)
(+) Audio --aid=1 (*) (aac 2ch 48000Hz)
libEGL warning: egl: failed to create dri2 screen
[1] 2590 segmentation fault (core dumped) mpv video.mp4coredump: http://0x0.st/HUYg.bin
`eglinfo` also crashes with segfaults.
output: http://0x0.st/HUYx.txt
coredump: http://0x0.st/HUY3.bin
I have read some recent threads, which I think they seems related:
- [mesa problem / Pacman & Package Upgrade Issues / Arch Linux Forums](https://bbs.archlinux.org/viewtopic.php?id=291519)
- [X segfaulting today[Solved} / Pacman & Package Upgrade Issues / Arch Linux Forums](https://bbs.archlinux.org/viewtopic.php?pid=2140345)
After trying some suggestion, the problem still exists. I am not sure which part of my system cause this problem. I tried to enclose any things that I think it could be related below.
----------
Basic information
# Basic information
OS: Arch Linux x86_64
Kernel: 6.6.10-arch1-1
WM: sway
Terminal: foot
CPU: Intel i7-8700 (12) @ 4.600GHz
GPU: NVIDIA GeForce GTX 1050 Ti
GPU: Intel CoffeeLake-S GT2 [UHD Graphics 630] List of related packages and their version
nvidia-dkms (545.29.06-1)
nvidia-utils (545.29.06-1)
mesa (1:23.3.3-1)
mesa-utils (9.0.0-3)
sway-git (1.9.r7237.c5fd8c0-1 AUR package)
kitty (0.31.0-1)
mpv (1:0.37.0-1)Kernel parameters in `/etc/default/grub`. Ran `grub-mkconfig -o /boot/grub/grub.cfg`.
GRUB_CMDLINE_LINUX_DEFAULT="loglevel=3 quiet nvidia_drm.modeset=1 nvidia_drm.fbdev=1 initcall_blacklist=simpledrm_platform_driver_init"Early KMS in `/etc/mkinitcpio.conf`. Ran `mkinitcpio -p linux`.
MODULES=(nvidia nvidia_modeset nvidia_uvm nvidia_drm)No `xf86-video-*` installed.
pacman -Qs xf86-video-
# no outputScript to start sway:
# Resolve: invisible cursor
export WLR_NO_HARDWARE_CURSORS=1
# Resolve: random flickering
export WLR_RENDERER=vulkan
# Resolve: random flickering in xwayland
export XWAYLAND_NO_GLAMOR=1
# start firefox in wayland mode
export MOZ_ENABLE_WAYLAND=1
# fcitx5 integration
# export GTK_IM_MODULE=fcitx # unset it to use wayland module
export QT_IM_MODULE=fcitx
export XMODIFIERS=@im=fcitx
# start sway
exec sway --unsupported-gpu 2>&1 | systemd-cat -t swayJournal via `journalctl -b`
output: http://0x0.st/HUYv.txt
Last edited by kkoyung (2024-01-14 15:34:47)
Offline
https://wiki.archlinux.org/title/Core_d … _core_dump
EGL driver name: zinkYou end up using https://wiki.archlinux.org/title/OpenGL … kan_(Zink)
Jan 14 15:35:27 kkoyung-pc sway[700]: 00:00:00.251 [ERROR] [wlr] [render/vulkan/renderer.c:2221] Could not match drm and vulkan device
Jan 14 15:35:27 kkoyung-pc sway[700]: 00:00:00.251 [ERROR] [wlr] [render/wlr_renderer.c:204] Failed to create a Vulkan renderer
Jan 14 15:35:27 kkoyung-pc sway[700]: 00:00:00.251 [ERROR] [wlr] [render/wlr_renderer.c:284] Could not initialize renderer
Jan 14 15:35:27 kkoyung-pc sway[700]: 00:00:00.251 [ERROR] [wlr] [backend/drm/renderer.c:18] Failed to create renderer
Jan 14 15:35:27 kkoyung-pc sway[700]: 00:00:00.251 [ERROR] [wlr] [backend/drm/backend.c:260] Failed to initialize renderer
Jan 14 15:35:27 kkoyung-pc sway[700]: 00:00:00.252 [ERROR] [wlr] [backend/backend.c:220] Failed to create DRM backend
Jan 14 15:35:27 kkoyung-pc sway[700]: 00:00:00.252 [ERROR] [sway/server.c:144] !!! Proprietary Nvidia drivers are in use !!!
Jan 14 15:35:27 kkoyung-pc sway[700]: 2024-01-14 15:35:27 - [swaybg-1.2.0/main.c:582] wl_display_roundtrip failed
…
Jan 14 15:35:28 kkoyung-pc sway[700]: Failed to initialize glamor, falling back to swSkip
export WLR_RENDERER=vulkanSee https://wiki.archlinux.org/title/Sway#F … VIDIA_GPUs in doubt.
---
Edit: Also https://wiki.archlinux.org/title/Sway#U … s_renderer - do you have https://archlinux.org/packages/?name=vu … ion-layers ?
----
Also maybe remove "nvidia_drm.fbdev=1" again, it causes
Jan 14 15:35:18 kkoyung-pc kernel: [drm:nv_drm_atomic_commit [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00000100] Flip event timeout on head 0
Jan 14 15:35:18 kkoyung-pc kernel: [drm:nv_drm_atomic_commit [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00000100] Flip event timeout on head 1and is reported to be still buggy. You don't need that since nvidia_drm.modeset=1 will disable the simpledrm device.
Jan 14 15:35:18 kkoyung-pc kernel: i915 0000:00:02.0: [drm] Cannot find any crtc or sizes
Jan 14 15:35:18 kkoyung-pc kernel: i915 0000:00:02.0: [drm] Cannot find any crtc or sizes
Jan 14 15:35:18 kkoyung-pc kernel: i915 0000:00:02.0: [drm] Cannot find any crtc or sizesYou've all outputs attached to the nvidia GPU? Can you disable the IGP? (Might save your some battery)
Last edited by seth (2024-01-14 09:14:46)
Offline
Skiping `export WLR_RENDERER=vulkan` doesn't help. kitty and mpv still crash with segfault. However, "journalctl -b" showed different errors.
Jan 14 19:27:53 kkoyung-pc sway[701]: 00:00:00.093 [ERROR] [sway/server.c:144] !!! Proprietary Nvidia drivers are in use !!!
Jan 14 19:27:53 kkoyung-pc sway[701]: 00:00:00.156 [ERROR] [wlr] [EGL] command: eglQueryDmaBufModifiersEXT, error: EGL_BAD_PARAMETER (0x300c), message: "EGL_BAD_PARAMETER error: In eglQueryDmaBufModifiersEXT: Invalid format
Jan 14 19:27:53 kkoyung-pc sway[701]: "
Jan 14 19:27:53 kkoyung-pc sway[701]: 00:00:00.156 [ERROR] [wlr] [render/egl.c:856] Failed to query dmabuf number of modifiers
Jan 14 19:27:53 kkoyung-pc sway[701]: 00:00:00.156 [ERROR] [wlr] [EGL] command: eglQueryDmaBufModifiersEXT, error: EGL_BAD_PARAMETER (0x300c), message: "EGL_BAD_PARAMETER error: In eglQueryDmaBufModifiersEXT: Invalid format
Jan 14 19:27:53 kkoyung-pc sway[701]: "
Jan 14 19:27:53 kkoyung-pc sway[701]: 00:00:00.156 [ERROR] [wlr] [render/egl.c:856] Failed to query dmabuf number of modifiers
Jan 14 19:27:54 kkoyung-pc sway[701]: 00:00:00.763 [ERROR] [wlr] [types/wlr_layer_shell_v1.c:291] A configure is sent to an uninitialized wlr_layer_surface_v1 0x561e56e96780
Jan 14 19:27:54 kkoyung-pc sway[701]: 00:00:00.763 [ERROR] [wlr] [types/wlr_layer_shell_v1.c:291] A configure is sent to an uninitialized wlr_layer_surface_v1 0x561e56dbc980
Jan 14 19:27:54 kkoyung-pc sway[701]: Failed to initialize glamor, falling back to swFull journal: http://0x0.st/HUgd.txt
I didn't have "vulkan-validation-layers", so I tried again after installing it. Same errors appear.
I took a look to the coredump of kitty. It seems that it still uses zink. Here are some part of the "coredump info":
PID: 2683 (kitty)
UID: 1000 (kkoyung)
GID: 1000 (kkoyung)
Signal: 11 (SEGV)
Timestamp: Sun 2024-01-14 19:33:36 HKT (18s ago)
Command Line: kitty
Executable: /usr/bin/kitty
Control Group: /user.slice/user-1000.slice/session-1.scope
Unit: session-1.scope
Slice: user-1000.slice
Session: 1
Owner UID: 1000 (kkoyung)
Boot ID: 3a2c52df806b4b6f88e6a85baa631ab5
Machine ID: 190dd6ad8d8f4ac9908973c2ad68b159
Hostname: kkoyung-pc
Storage: /var/lib/systemd/coredump/core.kitty.1000.3a2c52df806b4b6f88e6a85baa631ab5.2683.1705232016000000.zst (present)
Size on Disk: 5.6M
Message: Process 2683 (kitty) of user 1000 dumped core.
Stack trace of thread 2683:
#0 0x00007fda98c8837d n/a (zink_dri.so + 0xa8837d)
#1 0x00007fda98c9454d n/a (zink_dri.so + 0xa9454d)
#2 0x00007fda982c257d n/a (zink_dri.so + 0xc257d)
#3 0x00007fda982c53e7 n/a (zink_dri.so + 0xc53e7)
#4 0x00007fda9f18d41f n/a (libEGL_mesa.so.0 + 0x1d41f)
#5 0x00007fda9f18017d n/a (libEGL_mesa.so.0 + 0x1017d)
#6 0x00007fdaa0697f5d glfwCreateWindow (glfw-wayland.so + 0x17f5d)
#7 0x00007fdaa163c61b n/a (fast_data_types.so + 0x3c61b)
#8 0x00007fdaa27fb2e1 n/a (libpython3.11.so.1.0 + 0x1fb2e1)
#9 0x00007fdaa27da8ec _PyObject_MakeTpCall (libpython3.11.so.1.0 + 0x1da8ec)
#10 0x00007fdaa27e4c23 _PyEval_EvalFrameDefault (libpython3.11.so.1.0 + 0x1e4c23)
#11 0x00007fdaa280bb90 _PyFunction_Vectorcall (libpython3.11.so.1.0 + 0x20bb90)
#12 0x00007fdaa27de1d7 _PyObject_FastCallDictTstate (libpython3.11.so.1.0 + 0x1de1d7)
#13 0x00007fdaa281429d _PyObject_Call_Prepend (libpython3.11.so.1.0 + 0x21429d)
#14 0x00007fdaa28d7c02 n/a (libpython3.11.so.1.0 + 0x2d7c02)
#15 0x00007fdaa27da8ec _PyObject_MakeTpCall (libpython3.11.so.1.0 + 0x1da8ec)
#16 0x00007fdaa27e4c23 _PyEval_EvalFrameDefault (libpython3.11.so.1.0 + 0x1e4c23)
#17 0x00007fdaa289c484 n/a (libpython3.11.so.1.0 + 0x29c484)
#18 0x00007fdaa289be6c PyEval_EvalCode (libpython3.11.so.1.0 + 0x29be6c)
#19 0x00007fdaa28b2c42 n/a (libpython3.11.so.1.0 + 0x2b2c42)
#20 0x00007fdaa27f383a n/a (libpython3.11.so.1.0 + 0x1f383a)
#21 0x00007fdaa27f2987 PyObject_Vectorcall (libpython3.11.so.1.0 + 0x1f2987)
#22 0x00007fdaa27e4c23 _PyEval_EvalFrameDefault (libpython3.11.so.1.0 + 0x1e4c23)
#23 0x00007fdaa280bb90 _PyFunction_Vectorcall (libpython3.11.so.1.0 + 0x20bb90)
#24 0x00007fdaa28c5757 n/a (libpython3.11.so.1.0 + 0x2c5757)
#25 0x00007fdaa272faa7 n/a (libpython3.11.so.1.0 + 0x12faa7)
#26 0x0000564c15d6760a main (kitty + 0x260a)
#27 0x00007fdaa2445cd0 n/a (libc.so.6 + 0x27cd0)
#28 0x00007fdaa2445d8a __libc_start_main (libc.so.6 + 0x27d8a)
#29 0x0000564c15d67715 _start (kitty + 0x2715)---
I removed "nvidia_drm.fbdev=1". I also removed "initcall_blacklist=simpledrm_platform_driver_init", which should be implied by "nvidia_drm.modeset=1".
Both outputs are attached to the nvidia GPU. I disabled the iGPU from BIOS. Nice recommendation.
---
Edit:
Side note: When I used nvidia driver 535 before, I had to set "WLR_RENDERER=vulkan" and "XWAYLAND_NO_GLAMOR=1" to get rid of random flickering. After upgrading to 545.29.06, a few random flickering came back. After removing "WLR_RENDERER=vulkan" with nvidia driver 545.29.06, those random flickering are gone.
Last edited by kkoyung (2024-01-14 12:04:44)
Offline
You still run into libEGL_mesa and zink
printenv to see whether you've the environment to trigger that
Otherwise https://wiki.archlinux.org/title/Vulkan … initialize - but the libEGL_mesa.so.0 there doesn't really help.
Offline
I cannot find any weird things in "printenv".
In "journalctl -b", my system run into libEGL_mesa and zink while sway is starting. I guess the problem comes from sway, so I tried to switch from the AUR package "sway-git" back to "sway" (1:1.8.1-5) in Arch Linux repository. After switching, the problem is gone.
Not sure the actual reason causing this issue. I may spend some time investigating it later. But my system works fine now. I call it a day.
Thanks, seth.
---------
For anyone who has similar issue, here is my current configuration for your reference.
List of related packages and their version
nvidia-dkms (545.29.06-1)
nvidia-utils (545.29.06-1)
mesa (1:23.3.3-1)
mesa-utils (9.0.0-3)
sway (1:1.8.1-5)
kitty (0.31.0-1)
mpv (1:0.37.0-1)Kernel parameters in `/etc/default/grub`. Run `grub-mkconfig -o /boot/grub/grub.cfg` after changing.
GRUB_CMDLINE_LINUX_DEFAULT="loglevel=3 quiet nvidia_drm.modeset=1 nvidia_drm.fbdev=1 initcall_blacklist=simpledrm_platform_driver_init"Early KMS in `/etc/mkinitcpio.conf`. Run `mkinitcpio -p linux` after changing.
MODULES=(nvidia nvidia_modeset nvidia_uvm nvidia_drm)No `xf86-video-*` installed.
pacman -Qs xf86-video-
# no outputScript to start sway:
# Resolve: invisible cursor
export WLR_NO_HARDWARE_CURSORS=1
# Resolve: random flickering
# export WLR_RENDERER=vulkan # Skip this
# Resolve: random flickering in xwayland
export XWAYLAND_NO_GLAMOR=1
# start firefox in wayland mode
export MOZ_ENABLE_WAYLAND=1
# fcitx5 integration
# export GTK_IM_MODULE=fcitx # unset it to use wayland module
export QT_IM_MODULE=fcitx
export XMODIFIERS=@im=fcitx
# start sway
exec sway --unsupported-gpu 2>&1 | systemd-cat -t swayEdit: missing word
Last edited by kkoyung (2024-01-14 15:34:04)
Offline