wlroots
wlroots copied to clipboard
SEGFAULT on launch of sway: SEGV (/usr/lib/libgbm.so.1+0x3962)
When I currently start sway now, no matter what version, it always segfaults right away.
I believe this bug will go away when I restart my machine. I'll hold off for now on doing that, but will have to reboot the machine in 24h from now.
- sway version: sway version 1.1-rc1-34-g5b1a8d62 (Jun 9 2019, branch 'master')
- wlroots version (aur package wlroots-git): 0.6.0.r34.gd201fc35-1
- coredump of sway with bt-full: http://sprunge.us/WpYSSQ
- sway.log of
sway -d -c /etc/sway/config
: http://sprunge.us/cIRVRx - /etc/sway/config: http://sprunge.us/DeGrCO
wlroots has migrated to gitlab.freedesktop.org. This issue has been moved to:
https://gitlab.freedesktop.org/wlroots/wlroots/-/issues/1718
coredump of sway with bt-full
Red herring. When an executable is compiled with ASan (meson build -Db_sanitize=address
) it won't produce coredumps (stack traces will be included in the log instead). So this stack trace is from a previous crash.
So the interesting part is sway.log. libinput fails to initialize because it doesn't find any input devices, but then destroying the DRM backend segfaults:
2019-06-09 21:18:15 - [backend/libinput/backend.c:55] Initializing libinput
2019-06-09 21:18:16 - [backend/libinput/backend.c:84] libinput initialization failed, no input devices
2019-06-09 21:18:16 - [backend/libinput/backend.c:85] Set WLR_LIBINPUT_NO_DEVICES=1 to suppress this check
2019-06-09 21:18:16 - [backend/multi/backend.c:32] Failed to initialize backend.
2019-06-09 21:18:16 - [sway/server.c:192] Failed to start backend
2019-06-09 21:18:16 - [sway/tree/output.c:258] Disabling output 'eDP-1'
2019-06-09 21:18:16 - [sway/tree/workspace.c:161] Destroying workspace '1'
2019-06-09 21:18:16 - [sway/desktop/transaction.c:412] Transaction 0x604000046890 committing with 2 instructions
2019-06-09 21:18:16 - [sway/desktop/transaction.c:280] Applying transaction 0x604000046890
2019-06-09 21:18:16 - [sway/tree/output.c:278] Destroying output 'eDP-1'
2019-06-09 21:18:16 - [sway/desktop/transaction.c:412] Transaction 0x6040000468d0 committing with 1 instructions
2019-06-09 21:18:16 - [sway/desktop/transaction.c:280] Applying transaction 0x6040000468d0
2019-06-09 21:18:16 - [backend/drm/drm.c:1591] Emitting destruction signal for 'eDP-1'
2019-06-09 21:18:16 - [backend/drm/drm.c:1007] De-allocating CRTC 0 for output 'eDP-1'
2019-06-09 21:18:16 - [backend/drm/atomic.c:38] Atomic test failed: Permission denied
2019-06-09 21:18:16 - [backend/drm/atomic.c:55] eDP-1: Atomic commit failed (modeset): Permission denied
AddressSanitizer:DEADLYSIGNAL
=================================================================
==19060==ERROR: AddressSanitizer: SEGV on unknown address 0x61603f000217 (pc 0x7f5ab5793963 bp 0x606000020720 sp 0x7ffe32b118c8 T0)
==19060==The signal is caused by a READ memory access.
#0 0x7f5ab5793962 (/usr/lib/libgbm.so.1+0x3962)
#1 0x7f5aabd01d82 (/usr/lib/dri/i965_dri.so+0x1a0d82)
#2 0x7f5aabd01dec (/usr/lib/dri/i965_dri.so+0x1a0dec)
#3 0x7f5aad032a3e (/usr/lib/../lib/libEGL_mesa.so.0+0x1ca3e)
#4 0x7f5aad021488 in eglMakeCurrent (/usr/lib/../lib/libEGL_mesa.so.0+0xb488)
#5 0x7f5ab57ce6c0 (/usr/lib/libEGL.so.1+0x36c0)
#6 0x7f5ab664d1f0 in wlr_egl_make_current ../render/egl.c:333
#7 0x7f5ab664e3f0 in gles2_destroy ../render/gles2/renderer.c:367
#8 0x7f5ab663e1d1 in finish_drm_renderer ../backend/drm/renderer.c:57
#9 0x7f5ab6639b1d in backend_destroy ../backend/drm/backend.c:51
#10 0x7f5ab6643a7f in multi_backend_destroy ../backend/multi/backend.c:54
#11 0x56165e806a7c in server_start ../sway/server.c:193
#12 0x56165e805275 in main ../sway/main.c:384
#13 0x7f5ab63eace2 in __libc_start_main (/usr/lib/libc.so.6+0x23ce2)
#14 0x56165e7e612d in _start (/data/appData/archlinux/sway-git/src/sway/build/sway/sway+0x3912d)
AddressSanitizer can not provide additional info.
SUMMARY: AddressSanitizer: SEGV (/usr/lib/libgbm.so.1+0x3962)
==19060==ABORTING
2019-06-09 21:18:16 - [swaybg/main.c:521] Missing a required Wayland interface
Red herring. When an executable is compiled with ASan (meson build -Db_sanitize=address) it won't produce coredumps
OK, thanks! :))
FYI: After going to standby (systemctl suspend) and coming back from it, the problem was gone.