cs_preprocessor_data.c:206: Erreur fatale.

[zeph67]

Hi !

I'm using Code_Saturne 3.0.0. When I run a calculation, the preprocessor step seems to be okay. But finally I get this error :

cs_preprocessor_data.c:206: Erreur fatale.

Aucun fichier ou répertoire "mesh_input" trouvé.


Pile d'appels :
1: 0x7f409cf2a60d <cs_preprocessor_data_read_headers+0x158d> (libsaturne.so.0)
2: 0x7f409ceaecd6 <cs_run+0xd6> (libsaturne.so.0)
3: 0x7f409ceaeb1a <main+0x14a> (libsaturne.so.0)
4: 0x7f409c4fb76d <__libc_start_main+0xed> (libc.so.6)
5: 0x402aa9 <> (cs_solver)
Fin de la pile

... which is surprising, since the mesh is recognized by the preprocessor.
Any idea ?

Thanks in advance.

[Yvan Fournier]

Hello,

As per the forum recommendations, could you post a bit more info ? Your xml file possibly cs_user_script.py files, as well as the preprocess.log file would be a start, and passing the output of "ls" in the execution directory would provide more info.

Regards,

Yvan

[zeph67]

Sorry, I first didn't post them, because I was almost sure that the error encountered, was somehow classical...

No xml file, since I don't use the GUI.

cs_user_scripts.py and preprocessor.log are joint hereafter.

And ls gives the following :

compile.log cs_user_scripts.py error preprocessor.log summary
cs_solver cs_user_scripts.py.txt listing src_saturne

The presence of listing shows that the preproc step was achieved, and the calculation was about to run.


Thanks in advance.

[Yvan Fournier]

Hello,

Did you read the preprocessor.log ? It seems to end prematurely (no "normal end" or error message), so I do not know what happened, but either the end of your posted file is missing, or there was an undetected preprocessing error.

Did you test the same mesh with a previous version of Code_Saturne ?

As your mesh seems to be is very small, could you post it here ?

Regards,

Yvan

[zeph67]

Yvan, your questions led me to the answer.

Yes, I read the preprocessor.log, and the fact that a "fin normale du préprocesseur" was missing surprised me, but I thought it was a specificity of this new version...

I first used a .msh mesh, that I also converted to .unv format. This latter works (preproc OK, calculation OK).

Thanks a lot for your help.

[Yvan Fournier]

Hello,

The .msh mesh should still work, or lead to an error message, so as it seems there may be a bug, we are still interested in your .msh file (either on the forum or by private message or sent to the saturne-support at edf dot fr if it is confidential), so as to check it and debug the code if necessary.

Regards,

Yvan

[zeph67]

Here it is...

Maybe a bad compilation (I did several months ago)...

Thanks !

[Yvan Fournier]

Hello,

It is a bug: if you replace in preprocessor/pre-post/ecs_pre_gmsh.c, line 502:

ECS_MALLOC(elt_pos_som_ent[ient], nbr_elt, ecs_size_t);

with:

ECS_MALLOC(elt_pos_som_ent[ient], nbr_elt + 1, ecs_size_t);

and recompile, the bug is fixed.

I'll push the fix under version control tomorrow, so the bug will be fixed in bug-fix release 3.0.1 (which we'll probably release within a few days; I'm just waiting for a few minor GUI-related bug fixes.

Thanks for the feedback.

Regards,

Yvan