Illegal Instruction Debugging (RISC-V core)

Question

I am testing a program on a RISC-V core in baremetal. I am using no FPGA but running it in QuestaSim (RTL simulation).

The program consists of the main launching a function.

The problem is that when the function is supposed to return to main, an illegal instruction rises:

    Error: Illegal instruction 0x00010413 at PC 0x00000004!

I have inspected:

1. Core traces - Here I found this beauty when the execution is failing. From my understanding, the jalr instruction near the end is supposed to return the main, but x1 holds a zero...

            78923501           78909 1c0023e6 00008067 jalr             x0, x1, 0            x1:1c000a8e
            78926501           78912 1c000a8e 00000493 addi             x9, x0, 0            x9=00000000
            78927501           78913 1c000a92 0060006f jal              x0, 6              
            78929501           78915 1c000a98 01012d03 lw               x26, 16(x2)         x26=00000000  x2:100041b0  PA:100041c0
            78930501           78916 1c000a9a 01412c83 lw               x25, 20(x2)         x25=00000000  x2:100041b0  PA:100041c4
            78931501           78917 1c000a9c 01812c03 lw               x24, 24(x2)         x24=00000000  x2:100041b0  PA:100041c8
            78932501           78918 1c000a9e 01c12b83 lw               x23, 28(x2)         x23=00000000  x2:100041b0  PA:100041cc
            78950501           78936 1c000aa0 02012b03 lw               x22, 32(x2)         x22=00000000  x2:100041b0  PA:100041d0
            78951501           78937 1c000aa2 02412a83 lw               x21, 36(x2)         x21=00000000  x2:100041b0  PA:100041d4
            78952501           78938 1c000aa4 02812a03 lw               x20, 40(x2)         x20=00000000  x2:100041b0  PA:100041d8
            78953501           78939 1c000aa6 02c12983 lw               x19, 44(x2)         x19=00000000  x2:100041b0  PA:100041dc
            78954501           78940 1c000aa8 00900533 add              x10, x0, x9         x10=00000000  x9:00000000
            78955501           78941 1c000aac 03012903 lw               x18, 48(x2)         x18=00000000  x2:100041b0  PA:100041e0
            78956501           78942 1c000aae 03412483 lw               x9, 52(x2)           x9=00000000  x2:100041b0  PA:100041e4
            78974501           78960 1c000ab0 03812403 lw               x8, 56(x2)           x8=00000000  x2:100041b0  PA:100041e8
            78975501           78961 1c000ab2 03c12083 lw               x1, 60(x2)           x1=00000000  x2:100041b0  PA:100041ec
            78976501           78962 1c000ab4 04010113 addi             x2, x2, 64           x2=100041f0  x2:100041b0
            78978501           78964 1c000ab6 00008067 jalr             x0, x1, 0            x1:00000000
            79000501           78986 00000000 0fb12e27 sw               x59, 252(x2)        f27:00000000  x2:100041f0  PA:100042ec

2. Disassembled code - It confirms that jalr is returning to main (its the last instruction before the called function to terminate in the disassembled).

1c000ab6:   8082                    ret 

Possible issues:

1. Function pointers + struct? I have implemented custom structs with parameters and function pointers for class-based behaviour. This might be the issue as I am not sure It is implemented following best practices 100%. Still, the code executes fine. An example:

void api_init(my_struct *_entity){
  // -- init my function pointers
  _entity->init_streams            = (void *)my_init;
  _entity->compute                 = (void *)my_compute;

The application instantiates an _entity, then calls these functions in the following way:

(*_entity)->my_compute();

2. Stack? I am allocating lot of pointers to memory buffers, functions, etc. However, I tried to augment it in the linker script, but seems not to be real issue.

MY QUESTIONS

1. Does anyone have any experience with this kind of issues?
2. Can you suggest any method to debug this?

Thanks in advance to anyone will answer.

Have a nice day!