I feel like we should theoretically be able to implement this in terms of
0BRANCH, but since we don't have a mechanism for getting the "outer" value of
RSI inside the function, we can't just implement it as
: BRANCH 0 0BRANCH ;
In that case, RSI would point to EXIT when executing 0BRANCH.
-;; 0BRANCH is the fundamental mechanism for branching. If the top of the stack
-;; is zero, we jump by the given offset. 0BRANCH is given the offset as an
-;; integer after the word.
+;; BRANCH is the fundamental mechanism for branching. BRANCH reads the next word
+;; as a signed integer literal and jumps by that offset.
+BRANCH:
+ dq .start
+.start:
+ add rsi, [rsi] ; [RSI], which is the next word, contains the offset; we add this to the instruction pointer.
+ next ; Then, we can just continue execution as normal
+
+;; 0BRANCH is like BRANCH, but it jumps only if the top of the stack is zero.
ZBRANCH:
dq .start
.start:
ZBRANCH:
dq .start
.start:
cmp rax, 0
jnz .dont_branch
.do_branch:
cmp rax, 0
jnz .dont_branch
.do_branch:
- add rsi, [rsi] ; [RSI], which is the next word, contains the offset; we add this to the instruction pointer.
- next ; Then, we can just continue execution as normal
.dont_branch:
add rsi, 8 ; We need to skip over the next word, which contains the offset.
next
.dont_branch:
add rsi, 8 ; We need to skip over the next word, which contains the offset.
next
dq TELL
dq TELL
dq NEWLINE
dq TELL
dq TELL
dq NEWLINE