Thread: Segmentation Fault of death

Code:

components.ni = ∋
   components.xbpe = &xbpe;

Setting pointers to point to local variables is generally a Bad Idea. I have no idea why you designed these as pointers (are they supposed to eventually turn into arrays?) but I think you'll be happier if you just malloc some space for these guys.

Well, the first issue is that you are not declaring any components type variable like:

Code:

struct components var_components;

From what I gather components is a DATA TYPE not a VARIABLE. Correct?

tabstop:
I guess what I'm trying to do is pass the address of the components in pass1 to breakup so that breakup can tokenize the sourceline and assign the correct component parts into the components in pass1, does that make sense? The structs are provided and cannot be changed.

claudiu:
I'm creating components in pass1.c

Code:

struct toklist
{
   char *lbl;
   char *opc;
   char *opr1;
   char *opr2;
   int *ni;
   int *xbpe;
} components;

I take it ni is then supposed to stand for "number of items"? And passed by pointer so that the function can change it. But you still have to have somewhere for components.ni to point to, and I'm not sure a local variable will do. (That address isn't legal while the function isn't running, so during the call to breakup that variable is invalid, and it will be invalid again when the function ends and returns to whence it came.)

ni actually represents 2 bits in SIC/XE assembly object code. opcode + nixbpe

I'll give that a shot but if I remember correctly the professor mentioned that we shouldn't have to allocate memory. Should I be allocating memory in both functions or just the calling function?

You should allocate memory before you use components. As I said yesterday you can't dereference a pointer that doesn't point anywhere (i.e. you can't just declare int *p; and then expect to be able to put a value in *p.) When you allocate memory using malloc() the pointer will point to a VALID memory address allocated on the heap.

I thought that was why I was using the following code:

Code:

char label[81], opcode[81], opr1[81], opr2[81];
int ni, xbpe;
components.lbl = label;
components.opc = opcode;
components.opr1 = opr1;
components.opr2 = opr2;
components.ni = ∋
components.xbpe = &xbpe;

Does that not allocate memory for the pointer to point to? I guess I'm not fully grasping all of this

when allocating the memory shouldn't I be doing the following:

Code:

components.lbl = malloc(sizeof(char) * 100); /*allocate a string of 100 chars; lbl points to it */

Code:

Program received signal SIGSEGV, Segmentation fault.
0x00000034c3470db0 in strcpy () from /lib64/tls/libc.so.6
(gdb) bt
#0  0x00000034c3470db0 in strcpy () from /lib64/tls/libc.so.6
#1  0x00000000004029f3 in breakup (
    sourceline=0x504ee0 ".CHECK   getsource removal of initial blank lines", 
    components=
      {lbl = 0x7fbffff790 ".CHECK", opc = 0x7fbffff750 "getsource", opr1 = 0xab8aa1e <Address 0xab8aa1e out of bounds>, opr2 = 0x40070d "strcspn", ni = 0x0, xbpe = 0x34c3207d8e}) at breakup.c:128
#2  0x0000000000401ee3 in pass1 () at pass1.c:63
#3  0x0000000000400ed1 in main ()

Wow, I just realized that the line it's trying to breakup shouldn't be getting passed at all from pass1 as it is a comment. Let me try to fix that first and then see if the out of bounds issues are resolved

Can you tell me what's wrong with this statement?

Code:

 while(strcmp(sourcelist[Scnt].substr(0, 1), ".") == 0)
      Scnt++;

I'm trying to increment the Scnt counter for as long as there's a "." in the first place of the string. But I get the following error:

Code:

pass1.c: In function `pass1':
pass1.c:60: error: request for member `substr' in something not a structure or union

hmmmm other than the fact that it's C++ lol is there a substr function in C?

Yes, check strstr().