What does sys_break do?

Question

I was reading the list of Linux system calls and found sys_break, whose description is as follows.

Syntax: int sys_break()

Source: kernel/sys.c

Action: return -ENOSYS

Details: call exists only for compatibility

Does anyone know what sys_break do? Or does it do nothing?

Answer 1

I might be wrong, but I assume it is replaced by brk(2) system call which is used to control the amount of memory allocated to the data segment of the process. The original break call was deprecated probably because break is a keyword in C programming language. And I found following comment in Unix V6 source code (written either in or before 1976):

/* break system call.
* -- bad planning: "break" is a dirty word in C.
*/
sbreak()
{
register a, n, d;
int i;
/* set n to new data size
* set d to new-old
* set n to new total size
*/
...
}

So, before C programming language was invented, Unix was written in assembler which didn't define break as reserved word.

sys_break itself as syscall number 17 was introduced in Unix V1 (this is PDP-11 assembler):

# V1/u2.s - 1971-11-03
sysbreak: / set the program break
    mov u.break,r1 / move users break point to r1
    cmp r1,$core / is it the same or lower than core?
    blos    1f / yes, 1f
    cmp r1,sp / is it the same or higher than the stack?
    bhis    1f / yes, 1f
    bit $1,r1 / is it an odd address
    beq 2f / no, its even
    clrb    (r1)+ / yes, make it even
2: / clear area between the break point and the stack
    cmp r1,sp / is it higher or same than the stack
    bhis    1f / yes, quit
    clr (r1)+ / clear word
    br  2b / go back
1:
    jsr r0,arg; u.break / put the "address" in u.break (set new 
                / break point)
    br  sysret4 / br sysret

Now if you compare the V6 and V1, you can see that the meaning of the syscall has changed over time. Originally it was used to set the breakpoint for process, in V6 it is basically the brk(2) syscall.