The following code block is to swap values stored at two memory addresses $s1 and $s3:

lw $s0, 0($s1)

lw $s2, 0($s3)

sw $s0, 0($s3)

sw $s2, 0($s1)

 

However, in a multiprocessing environment, this wouldn't work since between the first line and the last line, if some other thread changes the content of the memory at addresses $s1 and/or $s3, it will create a mess (the swap would not be implemented correctly). In such an environment, you need LL and SC to ensure that the swap is atomic: when the last line is run, the content at addresses $s1 and $s3 has not been changed since after the first line. Modify the above code (which is properly commented) to implement the atomic swap.

Answer rating: 100% (QA)

The issue here is that between the code lines the current thread could be interrupted and another thread could alter the values stored in memory leadi... View the full answer