Struct pastix_data_s in CalculiX seems inconsistent with that in PaStiX4CalculiX

[uebian]

Struct pastix_data_s in CalculiX which is declared in the ccx_2.19/src/pastix.c is inconsistent with that in PaStiX4CalculiX, which is declared in pastixdata.h. As a result, there is a Segmentation fault on ccx_2.19/src/pastix.c:759, when calling pastix_data->iparm[IPARM_GPU_NBR]=0;. When debuging with GDB, pastix_data->iparm is always 0.

I'm using gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0 and following the instructions in CalculiX's README.INSTALL file when setting up the environment.

Is this a bug or a mistake from my side while setting up/compiling?

Any help is appreciated.

[Kabbone]

I honestly don't remember why I choose the same name, but I don't think this was on purpose because they are clearly not related to each other.
But this should not be a problem, because pastixdata.h is not included in pastix.c. The pastix_data you are refering to is of type pastix_data_t.
So the problem is located somewhere else.

[uebian]

In ccx, global variable pastix_data with type pastix_data_s is initilized on pastix.c:262 through calling PaStiX's API pastixInit. However, in PaStiX, this API's argument pastix_data is treated as type pastix_data_t which actually has different type from the caller.
As a result, when I set breakpoint on the end of pastixInit, the value pastix_data->iparm is not null. However, after it returns to CCX, on pastix.c:264, the pointer pastix_data is interpreted as a type that is different from where it is initilized and the value pastix_data->iparm is null because of that.
What's more, changing pastix_data_s in CCX to pastix_data_ss and copy the definition of pastix_data_s from pastixdata.h to pastix.c seems solve this problem, but I don't know if this is actually a bug in CalculiX or some mistakes when compiling this software.

[uebian]

I'm still getting Segmentation fault after re-compiling CalculiX and PaStiX4CalculiX, is there any follow up?