It looks like a bug. The unhappy path contains both `node++` and `node=node->next`. Note that this is in the code following "Let’s go back to the code we showed for value speculation in C:", which is actually different from the preceding code it's supposed to be a copy of. I guess it's a typo.
Comments
That is the trick.
In the happy path you are not assigning(`node=next`).
It is taken care of by `node++`, which removes the loop dependency and the processor can use the full instruction level parallelism.
It looks like a bug. The unhappy path contains both `node++` and `node=node->next`. Note that this is in the code following "Let’s go back to the code we showed for value speculation in C:", which is actually different from the preceding code it's supposed to be a copy of. I guess it's a typo.
The author has fixed this discrepancy now.
It looks as if the difference is in delaying the check for next == NULL.