Question

I’m working on a homework problem out of Stalling’s Operating Systems: Internals and Design Principals. As such, I am not requesting an answer but any help understanding this would be helpful as I am having a tough time wrapping my head around this one. There are two processes executing:

Both processes are executing the following code:

shared int x;
x = 10;
while (1) 
{
    x = x - 1;
    x = x + 1;
    if (x != 10)
        printf(“x is %d”,x)
}

The first part of this question asked how “x is 10” could be printed, which I had no trouble tracing the logic for.

However, the second part asks how can “x is 8” be printed. Additionally it gives the hint:

“You should remember that the increment/decrements at the source language level are not done atomically, that is, the assembly language code:”

LD r0,X
INCR r0
STO r0,X

“implements the single C instruction x=x+1”

Despite the hint, I am having trouble seeing how a process can ever reach the printf section without the value being incremented back up to at least 9. What am I missing?

Answer 1

Like this:

Thread 1              Thread 2

Load x  (10)
Decr x   (9)
Store x  (9)
                      Load x  (9)
Load x   (9)
Incr x  (10)
Store x (10)
                      Decr x  (8)
                      Store x (8)
Load x   (8)
Print x  (8)
                      // ...

The lesson here is that if an operation is not atomic, then writes can be lost if they get overwritten.