Check for available icc file profiles #19975

jenshannoschwalm · 2025-12-22T09:44:18Z

While falling back to lab jzczhz blending we should check

if there is a profile defined
and if the corresponding input profile could be created.

If not, we can't do anything here and just leave out blending.

Should fix #19968

@MStraeten could you check?

MStraeten · 2025-12-22T14:41:04Z

unfortunately the check for the profile doesn't help.
running darktable in a debugger indcates, that profile is set an contains the filename of the now missing icc file.

jenshannoschwalm · 2025-12-22T18:51:40Z

@MStraeten could you test again (all squashed in one commit) ?

BTW - could you provide the profile file?

MStraeten · 2025-12-22T19:36:12Z

issue no longer exists with this pr
Its a quite old edit with an old icc file - if it's available then it's handled properly 20190321-5DmIV.icc.txt

jenshannoschwalm · 2025-12-23T04:00:35Z

Sorry to bother you but the latest force-pushed version tests for "icc file existing" at the proper place.

TurboGit

Thanks!

TurboGit · 2025-12-23T07:19:27Z

I'll wait for @MStraeten second testing round before merging. TIA.

MStraeten · 2025-12-23T11:30:20Z

the latest push breaks it again:

Process 91583 stopped
* thread #29, name = 'worker res 0', stop reason = EXC_BAD_ACCESS (code=1, address=0x408)
    frame #0: 0x0000000100865b5c libdarktable.dylib`_blendif_combine_channels(pixels=0x00000006e9800000, mask=0x00000006ed800000, stride=2176, blendif=256, parameters=0x000000017292c080, profile=0x000000017292c700) at blendif_rgb_jzczhz.c:245:115 [opt]
   242 	                    invert_mask, profile);
   243 	    else if(profile)
   244 	    {
-> 245 	      cmsHPROFILE *input = dt_colorspaces_get_profile(profile->type, profile->filename, DT_PROFILE_DIRECTION_IN)->profile;
   246 	      cmsHPROFILE *Lab = dt_colorspaces_get_profile(DT_COLORSPACE_LAB, "", DT_PROFILE_DIRECTION_ANY)->profile;
   247 	      cmsHTRANSFORM *xform = input ? cmsCreateTransform(input, TYPE_RGBA_FLT, Lab, TYPE_LabA_FLT, profile->intent, 0) : NULL;
   248
  thread #62, stop reason = EXC_BAD_ACCESS (code=1, address=0x408)
    frame #0: 0x0000000100865b5c libdarktable.dylib`_blendif_combine_channels(pixels=0x00000006e9b27800, mask=0x00000006ed8c9e00, stride=2176, blendif=256, parameters=0x000000017852a980, profile=0x000000017292c700) at blendif_rgb_jzczhz.c:245:115 [opt]
   242 	                    invert_mask, profile);
   243 	    else if(profile)
   244 	    {
-> 245 	      cmsHPROFILE *input = dt_colorspaces_get_profile(profile->type, profile->filename, DT_PROFILE_DIRECTION_IN)->profile;
   246 	      cmsHPROFILE *Lab = dt_colorspaces_get_profile(DT_COLORSPACE_LAB, "", DT_PROFILE_DIRECTION_ANY)->profile;
   247 	      cmsHTRANSFORM *xform = input ? cmsCreateTransform(input, TYPE_RGBA_FLT, Lab, TYPE_LabA_FLT, profile->intent, 0) : NULL;
   248
  thread #63, stop reason = EXC_BAD_ACCESS (code=1, address=0x408)
    frame #0: 0x0000000100865b5c libdarktable.dylib`_blendif_combine_channels(pixels=0x00000006e9e4f000, mask=0x00000006ed993c00, stride=2176, blendif=256, parameters=0x0000000178d32980, profile=0x000000017292c700) at blendif_rgb_jzczhz.c:245:115 [opt]
   242 	                    invert_mask, profile);
   243 	    else if(profile)
   244 	    {
-> 245 	      cmsHPROFILE *input = dt_colorspaces_get_profile(profile->type, profile->filename, DT_PROFILE_DIRECTION_IN)->profile;
   246 	      cmsHPROFILE *Lab = dt_colorspaces_get_profile(DT_COLORSPACE_LAB, "", DT_PROFILE_DIRECTION_ANY)->profile;
   247 	      cmsHTRANSFORM *xform = input ? cmsCreateTransform(input, TYPE_RGBA_FLT, Lab, TYPE_LabA_FLT, profile->intent, 0) : NULL;
   248
  thread #64, stop reason = EXC_BAD_ACCESS (code=1, address=0x408)
    frame #0: 0x0000000100865b5c libdarktable.dylib`_blendif_combine_channels(pixels=0x00000006ea176800, mask=0x00000006eda5da00, stride=2176, blendif=256, parameters=0x000000017953a980, profile=0x000000017292c700) at blendif_rgb_jzczhz.c:245:115 [opt]
   242 	                    invert_mask, profile);
   243 	    else if(profile)
   244 	    {
-> 245 	      cmsHPROFILE *input = dt_colorspaces_get_profile(profile->type, profile->filename, DT_PROFILE_DIRECTION_IN)->profile;
   246 	      cmsHPROFILE *Lab = dt_colorspaces_get_profile(DT_COLORSPACE_LAB, "", DT_PROFILE_DIRECTION_ANY)->profile;
   247 	      cmsHTRANSFORM *xform = input ? cmsCreateTransform(input, TYPE_RGBA_FLT, Lab, TYPE_LabA_FLT, profile->intent, 0) : NULL;
   248
  thread #65, stop reason = EXC_BAD_ACCESS (code=1, address=0x408)
    frame #0: 0x0000000100865b5c libdarktable.dylib`_blendif_combine_channels(pixels=0x00000006ea49e000, mask=0x00000006edb27800, stride=2176, blendif=256, parameters=0x0000000179d42980, profile=0x000000017292c700) at blendif_rgb_jzczhz.c:245:115 [opt]
   242 	                    invert_mask, profile);
   243 	    else if(profile)
   244 	    {
-> 245 	      cmsHPROFILE *input = dt_colorspaces_get_profile(profile->type, profile->filename, DT_PROFILE_DIRECTION_IN)->profile;
   246 	      cmsHPROFILE *Lab = dt_colorspaces_get_profile(DT_COLORSPACE_LAB, "", DT_PROFILE_DIRECTION_ANY)->profile;
   247 	      cmsHTRANSFORM *xform = input ? cmsCreateTransform(input, TYPE_RGBA_FLT, Lab, TYPE_LabA_FLT, profile->intent, 0) : NULL;
   248
  thread #66, stop reason = EXC_BAD_ACCESS (code=1, address=0x408)
    frame #0: 0x0000000100865b5c libdarktable.dylib`_blendif_combine_channels(pixels=0x00000006ea7c5800, mask=0x00000006edbf1600, stride=2176, blendif=256, parameters=0x000000017a54a980, profile=0x000000017292c700) at blendif_rgb_jzczhz.c:245:115 [opt]
   242 	                    invert_mask, profile);
   243 	    else if(profile)
   244 	    {
-> 245 	      cmsHPROFILE *input = dt_colorspaces_get_profile(profile->type, profile->filename, DT_PROFILE_DIRECTION_IN)->profile;
   246 	      cmsHPROFILE *Lab = dt_colorspaces_get_profile(DT_COLORSPACE_LAB, "", DT_PROFILE_DIRECTION_ANY)->profile;
   247 	      cmsHTRANSFORM *xform = input ? cmsCreateTransform(input, TYPE_RGBA_FLT, Lab, TYPE_LabA_FLT, profile->intent, 0) : NULL;
   248
  thread #67, stop reason = EXC_BAD_ACCESS (code=1, address=0x408)
    frame #0: 0x0000000100865b5c libdarktable.dylib`_blendif_combine_channels(pixels=0x00000006eaaed000, mask=0x00000006edcbb400, stride=2176, blendif=256, parameters=0x000000017ad52980, profile=0x000000017292c700) at blendif_rgb_jzczhz.c:245:115 [opt]
   242 	                    invert_mask, profile);
   243 	    else if(profile)
   244 	    {
-> 245 	      cmsHPROFILE *input = dt_colorspaces_get_profile(profile->type, profile->filename, DT_PROFILE_DIRECTION_IN)->profile;
   246 	      cmsHPROFILE *Lab = dt_colorspaces_get_profile(DT_COLORSPACE_LAB, "", DT_PROFILE_DIRECTION_ANY)->profile;
   247 	      cmsHTRANSFORM *xform = input ? cmsCreateTransform(input, TYPE_RGBA_FLT, Lab, TYPE_LabA_FLT, profile->intent, 0) : NULL;
   248
  thread #68, stop reason = EXC_BAD_ACCESS (code=1, address=0x408)
    frame #0: 0x0000000100865b5c libdarktable.dylib`_blendif_combine_channels(pixels=0x00000006eae0c000, mask=0x00000006edd83000, stride=2176, blendif=256, parameters=0x000000017b55a980, profile=0x000000017292c700) at blendif_rgb_jzczhz.c:245:115 [opt]
   242 	                    invert_mask, profile);
   243 	    else if(profile)
   244 	    {
-> 245 	      cmsHPROFILE *input = dt_colorspaces_get_profile(profile->type, profile->filename, DT_PROFILE_DIRECTION_IN)->profile;
   246 	      cmsHPROFILE *Lab = dt_colorspaces_get_profile(DT_COLORSPACE_LAB, "", DT_PROFILE_DIRECTION_ANY)->profile;
   247 	      cmsHTRANSFORM *xform = input ? cmsCreateTransform(input, TYPE_RGBA_FLT, Lab, TYPE_LabA_FLT, profile->intent, 0) : NULL;
   248
  thread #69, stop reason = EXC_BAD_ACCESS (code=1, address=0x408)
    frame #0: 0x0000000100865b5c libdarktable.dylib`_blendif_combine_channels(pixels=0x00000006eb12b000, mask=0x00000006ede4ac00, stride=2176, blendif=256, parameters=0x000000017bd62980, profile=0x000000017292c700) at blendif_rgb_jzczhz.c:245:115 [opt]
   242 	                    invert_mask, profile);
   243 	    else if(profile)
   244 	    {
-> 245 	      cmsHPROFILE *input = dt_colorspaces_get_profile(profile->type, profile->filename, DT_PROFILE_DIRECTION_IN)->profile;
   246 	      cmsHPROFILE *Lab = dt_colorspaces_get_profile(DT_COLORSPACE_LAB, "", DT_PROFILE_DIRECTION_ANY)->profile;
   247 	      cmsHTRANSFORM *xform = input ? cmsCreateTransform(input, TYPE_RGBA_FLT, Lab, TYPE_LabA_FLT, profile->intent, 0) : NULL;
   248
  thread #70, stop reason = EXC_BAD_ACCESS (code=1, address=0x408)
    frame #0: 0x0000000100865b5c libdarktable.dylib`_blendif_combine_channels(pixels=0x00000006eb44a000, mask=0x00000006edf12800, stride=2176, blendif=256, parameters=0x000000017c56a980, profile=0x000000017292c700) at blendif_rgb_jzczhz.c:245:115 [opt]
   242 	                    invert_mask, profile);
   243 	    else if(profile)
   244 	    {
-> 245 	      cmsHPROFILE *input = dt_colorspaces_get_profile(profile->type, profile->filename, DT_PROFILE_DIRECTION_IN)->profile;
   246 	      cmsHPROFILE *Lab = dt_colorspaces_get_profile(DT_COLORSPACE_LAB, "", DT_PROFILE_DIRECTION_ANY)->profile;
   247 	      cmsHTRANSFORM *xform = input ? cmsCreateTransform(input, TYPE_RGBA_FLT, Lab, TYPE_LabA_FLT, profile->intent, 0) : NULL;
   248
Target 0: (darktable) stopped.

jenshannoschwalm · 2025-12-23T12:59:18Z

Very interesting. It seems the list is not correct. Will further investigate ...

We maintain a global list of all profiles while committing params via _ioppr_generate_profile_info(). In case of icc file profiles we didn't test for the file being available and missed to return that as an error condition. We now test & log such error conditions. The colorin & colorout modules handled this "gracefully" already but since introducing the fallback to lab blending this resulted in a crash while traversing the list. While being here added info about missing matrix and nonlinearlut in the log.

If we set input,output or work profile to a icc file profile we should report if the file was not available and report that. This is only done while calculating preview pipe to avoid repeating messages.

jenshannoschwalm · 2025-12-23T20:17:38Z

@TurboGit and @MStraeten

Found the issue leading to crashes while traversing the list. It was basically a missing test for existance of icc files resulting in a bad list.

1st commit would be for 5.4.1 and master
2nd commit reporting via control log for master

MStraeten · 2025-12-23T22:40:53Z

thats it, it properly reports the missing icc file
thanks for the fix

jenshannoschwalm added the bugfix pull request fixing a bug label Dec 22, 2025

jenshannoschwalm modified the milestones: 5.6, 5.4.1 Dec 22, 2025

jenshannoschwalm force-pushed the missing_lab_profile branch from a3d4b54 to 06f53c8 Compare December 22, 2025 18:42

jenshannoschwalm force-pushed the missing_lab_profile branch from 06f53c8 to 28ee35f Compare December 23, 2025 03:58

TurboGit approved these changes Dec 23, 2025

View reviewed changes

jenshannoschwalm added 2 commits December 23, 2025 21:11

control_log missing icc profiles

9f5cbc2

If we set input,output or work profile to a icc file profile we should report if the file was not available and report that. This is only done while calculating preview pipe to avoid repeating messages.

jenshannoschwalm force-pushed the missing_lab_profile branch from 28ee35f to 9f5cbc2 Compare December 23, 2025 20:12

jenshannoschwalm changed the title ~~Check for available lab profile~~ Check for available icc file profiles Dec 23, 2025

jenshannoschwalm added the priority: high core features are broken and not usable at all, software crashes label Dec 23, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Check for available icc file profiles #19975

Check for available icc file profiles #19975

jenshannoschwalm commented Dec 22, 2025

Uh oh!

MStraeten commented Dec 22, 2025

Uh oh!

jenshannoschwalm commented Dec 22, 2025

Uh oh!

MStraeten commented Dec 22, 2025

Uh oh!

jenshannoschwalm commented Dec 23, 2025

Uh oh!

TurboGit left a comment

Uh oh!

TurboGit commented Dec 23, 2025

Uh oh!

MStraeten commented Dec 23, 2025

Uh oh!

jenshannoschwalm commented Dec 23, 2025

Uh oh!

jenshannoschwalm commented Dec 23, 2025

Uh oh!

MStraeten commented Dec 23, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Check for available icc file profiles #19975

Are you sure you want to change the base?

Check for available icc file profiles #19975

Conversation

jenshannoschwalm commented Dec 22, 2025

Uh oh!

MStraeten commented Dec 22, 2025

Uh oh!

jenshannoschwalm commented Dec 22, 2025

Uh oh!

MStraeten commented Dec 22, 2025

Uh oh!

jenshannoschwalm commented Dec 23, 2025

Uh oh!

TurboGit left a comment

Choose a reason for hiding this comment

Uh oh!

TurboGit commented Dec 23, 2025

Uh oh!

MStraeten commented Dec 23, 2025

Uh oh!

jenshannoschwalm commented Dec 23, 2025

Uh oh!

jenshannoschwalm commented Dec 23, 2025

Uh oh!

MStraeten commented Dec 23, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants