Darktable 5.6 crash on Arch Linux, due to config file?

[SuperBFG7]

Is there an existing issue for this?

• I checked and did not find my issue in the already reported ones

Describe the bug

After upgrade to darktable 5.6 it started up normally, showed the welcome screen and migrated DB and library. But the next time I tried to start it, it crashed after the splash screen. The main screen was never shown.

Steps to reproduce

1. Copy ~/.config/darktable/darktablerc from 5.4 backup
2. Start darktable -> shows info screen that open CL per device settings have changed and then works fine
3. Stop it (Ctrl-q)
4. Start darktable again -> crashes after splash screen

Expected behavior

start up

Logfile | Screenshot | Screencast

darktable_bt_BUBPS3.txt
darktablerc_5.4.txt
darktablerc_5.6.txt

Commit

No response

Where did you obtain darktable from?

distro packaging

darktable version

5.6.0

What OS are you using?

Linux

What is the version of your OS?

Arch Linux

Describe your system

HP ProDesk 4 Mini G1i
Core 7 Ultra 265T, 32 GB
Wayland, Hyprland, Intel OpenCL

Are you using OpenCL GPU in darktable?

Yes

If yes, what is the GPU card and driver?

Intel(R) OpenCL Graphics, Intel(R) Corporation, ID=32902

Please provide additional context if applicable. You can attach files too, but might need to rename to .txt or .zip

No response

[SuperBFG7]

Same behaviour when I delete the darktablerc file.

1. delete ~/.config/darktable/darktablerc
2. start darktable -> starts normally and creates a new config file
3. close darktable (ctrl-q) without doing anything (not changing any setting etc.)
4. start again -> segfault

[jenshannoschwalm]

To get into this a -d pipe -d opencl logfile would be helping, there have been more than one report about faulty intel OpenCL stuff on arch linux. At least make sure opencl does not crash with clinfo

[jlcala]

I've got the same issue on CachyOS 7.1.3-2. I'm on an AMD Ryzen 9 7940HS and Radeon 7700S using the amggpu driver.

[dt starting] as : darktable -d pipe -d opencl
[dt_get_sysresource_level] switched to `default'
total mem: 62082MB
available mem: 31041MB
singlebuff: 485MB
[opencl_init] opencl library 'libOpenCL' found on your system and loaded, preference 'default path'
Assertion 'close_nointr(fd) != -EBADF' failed at src/basic/fd-util.c:68, function safe_close(). Aborting.
fish: Job 1, 'darktable -d pipe -d opencl' terminated by signal SIGABRT (Abort)

[SuperBFG7]

Thanks for helping to narrow this down. clinfo runs fine and to rule out an opencl issue, I started darktable with --disable-opencl and get the same behaviour. Very strange.

I've attached outputs for:
darktable -d pipe -d opencl &> dt_pipe_opencl.txt
darktable -d pipe -d opencl --disable-opencl &> dt_pipe_noopencl.txt
darktable -d all --disable-opencl &> dt_all.txt

dt_pipe_opencl.txt
dt_pipe_noopencl.txt
dt_all.txt

[jlcala]

The latest commit has solved the issue for me, thanks!

[jenshannoschwalm]

@SuperBFG7 can you confirm too its fixed with nightly appimage?

[SuperBFG7]

Hi @jenshannoschwalm

thank you for your help getting to the bottom of this. Unfortunately the latest release does not resolve it for me. First startup was fine (asking me for some defaults).
Second startup it segfaulted again and I've attached the output of -d pipe -d opencl for the third startup:
dt-5.7.0+68.g4d32ac411e.txt

Finally I tried it with --disable-opencl again: worked once, but the second startup with OpenCL disabled fails.

I'm starting to suspect some old config or DB entry I inherited over the years. Maybe I should just start over with a fresh installation and reconfigure and reimport everything...

Here's an attempt to narrow it down further:

When I delete ~/.config/darktable it works repeatedly.
When I copy over my old darktablerc to that virgin config folder, it still works repeatedly.
When I copy over my old library.db and data.db it worked once, failed the second time (with the below output), worked the third time, failed the fourth time, worked a couple of times, failed again, ... - how can this be?

$ ./Darktable-5.7.0+68.g4d32ac411e-x86_64.AppImage 
/tmp/.mount_DarktaBbgDOK/usr/bin/../lib/libcrypto.so.3: version `OPENSSL_3.3.0' not found (required by /usr/lib/libssl.so.3)
Failed to load module: /tmp/.mount_DarktaBbgDOK/usr/lib/gio/modules/libgiolibproxy.so
Fontconfig warning: using without calling FcInit()

(org.darktable.darktable:48361): Pango-CRITICAL **: 21:07:49.683: pango_layout_get_line_count: assertion 'layout != NULL' failed

(org.darktable.darktable:48361): Pango-CRITICAL **: 21:07:49.683: pango_layout_get_pixel_extents: assertion 'PANGO_IS_LAYOUT (layout)' failed

(org.darktable.darktable:48361): Pango-CRITICAL **: 21:07:49.683: pango_layout_line_unref: assertion 'private->ref_count > 0' failed
Segmentation fault         (core dumped) ./Darktable-5.7.0+68.g4d32ac411e-x86_64.AppImage
Unable to attach: program terminated with signal SIGSEGV, Segmentation fault.
/home/daniel/Downloads/48361: No such file or directory.
warning: Currently logging to /tmp/darktable_bt_T65SS3.txt.  Turn the logging off and on to make the new setting effective.
/tmp/.mount_DarktaBbgDOK/usr/share/darktable/gdb_commands:2: Error in sourced command file:
No stack.

The first part I got with the previous, successful startups as well, only the pango part seems to be different:

$ ./Darktable-5.7.0+68.g4d32ac411e-x86_64.AppImage 
/tmp/.mount_DarktanDCFJl/usr/bin/../lib/libcrypto.so.3: version `OPENSSL_3.3.0' not found (required by /usr/lib/libssl.so.3)
Failed to load module: /tmp/.mount_DarktanDCFJl/usr/lib/gio/modules/libgiolibproxy.so
Fontconfig warning: using without calling FcInit()

[jenshannoschwalm]

@victoryforce i think you might know if the appimage might miss some.