Weird error after struct-to-struct assignment

Question

I declared the following types and function in my program.

typedef char TYPE_name_type;
typedef char TYPE_name_category;
typedef struct {
   TYPE_name_category category;
   TYPE_name_type id;
} TYPE_cod_name;
typedef struct {
   TYPE_name_type name;
   TYPE_cod_name image;
} TYPE_beta_name;
typedef struct {
   TYPE_cod_name lefty;
   TYPE_cod_name righty;
} TYPE_codomain_pair;
typedef char TYPE_thread_ref;
typedef struct {
   TYPE_beta_name items[2];
   unsigned int length;
} TYPE_beta_env;
typedef struct {
   TYPE_thread_ref pid;
   TYPE_beta_env env;
} TYPE_thread_env;
typedef struct {
   TYPE_thread_env items[3];
   unsigned int length;
} TYPE_beta_type;

TYPE_beta_type FUNCTION_post_Reset
(TYPE_beta_type  V34,
 TYPE_thread_ref V35) {
  TYPE_beta_type result;
  TYPE_beta_type V78 = (TYPE__EMPTY_CONST_beta_type);
  TYPE_thread_env V79;
  TYPE_beta_type V79_cont;
  unsigned int V79_index;
  V79_cont = V34;
  V79_index = 0;
  V79 = V79_cont.items[0];
  if (V79_cont.length > 0) {
    while (1) {
      if (V79.pid == V35) {
         V78 = TYPE__OP_thread_env__CONCAT_OP__beta_type(V78, V79);
      }
      if(V79_index == (V79_cont.length - 1)) break;
      V79_index ++;
      V79 = V79_cont.items[V79_index];
      assert (V79.pid == V79_cont.items[V79_index].pid);
    }
  }
  return V78;
}

I noticed that the program sometimes halts at the assert at the end of function right after the struct-to-struct assignment at the previous line. I tried several versions of gcc. It works fine with 4.3.6, 4.7.2 but always fails with 4.6.3.

I am working on a 32-bit Fedora 16 with kernel 3.6.11.

Could it be due to a problem on gcc 4.3.6 or I am overlooking something?

Also, what happens when you step through your code with a debugger? — SecurityMatt
– SecurityMatt, Commented Feb 15, 2013 at 9:30
The code is very convoluted, and hard to follow, but have you tried running in a debugger? It will stop when the assert fails and you can see the states of variables etc. — Some programmer dude
– Some programmer dude, Commented Feb 15, 2013 at 9:31
it's generated code and i use the TYPE_ prefix and ugly variable names to avoid name conflicts. — sevan
– sevan, Commented Feb 15, 2013 at 9:32
no i haven-t tried to debug it. i'll do that. thanks for the suggestion — sevan
– sevan, Commented Feb 15, 2013 at 9:33

cmaster - reinstate monica · Accepted Answer · 2013-06-05 19:58:18Z

1

That's the consequence of one of the worst mistakes one can do in programming: using fixed buffer sizes. You are almost guaranteed to run into trouble with it sooner or later, because the requirements will change, and your fixed buffer size will not be enough.

Here the problem is items being only of size 3, and I am certain that you will find V79_index to be 3 upon the failed assert.

Ps: I know that all the world uses fixed buffer sizes. But the whole world once used goto. And I consider both harmful.

answered Jun 5, 2013 at 19:58

cmaster - reinstate monica

41.1k9 gold badges69 silver badges110 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Jim Fell Over a year ago

Granted, dynamic buffers are nice, but there are applications where a fixed buffer size cannot be avoided, such as embedded, bare-metal applications that do not support dynamic memory allocation.

cmaster - reinstate monica Over a year ago

@JimFell Right, there is no rule without exception when it comes to programming. Including the rules about goto and fixed buffer sizes... Nevertheless, when you have a choice, avoid both.

Collectives™ on Stack Overflow

Weird error after struct-to-struct assignment

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related