From: Mark <markjordanthom@gmail.com>
Date: Sun, 22 Oct 2023 03:37:27 +0000 (-0600)
Subject: use Deutsch-Schorr-Waite to implement acyclic_term/1 (#2128)
X-Git-Tag: remove~14
X-Git-Url: https://git.sagredo.dev/?a=commitdiff_plain;h=6a913bc4ccc76a662392b7b1febb16e6ae4bf3da;p=scryer-prolog.git

use Deutsch-Schorr-Waite to implement acyclic_term/1 (#2128)
---

diff --git a/Cargo.lock b/Cargo.lock
index 133b6524..e054b9db 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -2047,7 +2047,6 @@ dependencies = [
  "to-syn-value",
  "to-syn-value_derive",
  "tokio",
- "topo_sort",
  "walkdir",
  "warp",
  "wasm-bindgen",
@@ -2577,12 +2576,6 @@ dependencies = [
  "tracing",
 ]
 
-[[package]]
-name = "topo_sort"
-version = "0.4.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "156552d3c80df430aaac98c605a4e0eb7da8d06029cce2d40b4a6b095a34b37e"
-
 [[package]]
 name = "tower-service"
 version = "0.3.2"
diff --git a/Cargo.toml b/Cargo.toml
index 6ed48fc0..5fb0db69 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -70,7 +70,6 @@ bytes = "1"
 dashu = "0.4.0"
 num-order = { version = "1.2.0" }
 rand = "0.8.5"
-topo_sort = { version = "0.4.0" }
 
 [target.'cfg(not(target_arch = "wasm32"))'.dependencies]
 libffi = { version = "3.2.0", optional = true }
diff --git a/src/heap_iter.rs b/src/heap_iter.rs
index 3c47c934..2e1bf6eb 100644
--- a/src/heap_iter.rs
+++ b/src/heap_iter.rs
@@ -1,6 +1,8 @@
-pub(crate) use crate::machine::gc::{CycleDetectorUMP, StacklessPreOrderHeapIter};
+#[cfg(test)]
+pub(crate) use crate::machine::gc::StacklessPreOrderHeapIter;
 
 use crate::atom_table::*;
+use crate::machine::cycle_detection::CycleDetectingIter;
 use crate::machine::heap::*;
 use crate::machine::stack::*;
 use crate::types::*;
@@ -504,11 +506,11 @@ impl<'a, ElideLists: ListElisionPolicy> Iterator for StackfulPreOrderHeapIter<'a
 }
 
 #[inline(always)]
-pub(crate) fn cycle_detecting_stackless_preorder_iter(
-    heap: &mut Vec<HeapCellValue>,
+pub(crate) fn cycle_detecting_stackless_preorder_iter<'a>(
+    heap: &'a mut [HeapCellValue],
     start: usize,
-) -> StacklessPreOrderHeapIter<CycleDetectorUMP> {
-    StacklessPreOrderHeapIter::<CycleDetectorUMP>::new(heap, start)
+) -> CycleDetectingIter<'a> {
+    CycleDetectingIter::new(heap, start)
 }
 
 #[inline(always)]
diff --git a/src/machine/cycle_detection.rs b/src/machine/cycle_detection.rs
new file mode 100644
index 00000000..8de3e8e6
--- /dev/null
+++ b/src/machine/cycle_detection.rs
@@ -0,0 +1,424 @@
+use crate::atom_table::*;
+use crate::types::*;
+
+/* Use the pointer reversal technique of the Deutsch-Schorr-Waite
+ * algorithm to detect cycles in Prolog terms.
+ *
+ * Much of the structure and nomenclature of the GC marking algorithm
+ * is adapted here but there are a few significant changes:
+ *
+ * - Forwarded cells now form a trail of bread crumbs leading back to self.start
+ * - Cells are only marked during the backward phase
+ * - Visiting subterms of a visited compound does not immediately shift to the backward phase
+ * - The heads of LIS structures are both marked and forwarded rather
+ *   than just forwarded to distinguish them from tails;
+ *   continue_forwarding() checks for this before entering the forward
+ *   phase
+ *
+ * Commonalities with the GC marking algorithm:
+ * - The contents of forwarded cells are modified only when they are unforwarded
+ * - Marked (but unforwarded!) cells immediately shift to the backward phase
+ */
+
+#[derive(Debug)]
+pub(crate) struct CycleDetectingIter<'a> {
+    pub(crate) heap: &'a mut [HeapCellValue],
+    start: usize,
+    current: usize,
+    next: u64,
+    cycle_found: bool,
+    mark_phase: bool,
+}
+
+impl<'a> CycleDetectingIter<'a> {
+    pub(crate) fn new(heap: &'a mut [HeapCellValue], start: usize) -> Self {
+        heap[start].set_forwarding_bit(true);
+        let next = heap[start].get_value();
+
+        Self {
+            heap,
+            start,
+            current: start,
+            next,
+            cycle_found: false,
+            mark_phase: true,
+        }
+    }
+
+    #[inline]
+    pub(crate) fn cycle_found(&self) -> bool {
+        self.cycle_found
+    }
+
+    #[inline]
+    fn cycle_detection_active(&self) -> bool {
+        self.mark_phase && !self.cycle_found
+    }
+
+    fn backward_and_return(&mut self) -> HeapCellValue {
+        let mut current = self.heap[self.current];
+        current.set_value(self.next);
+
+        if self.backward() {
+            // set the f and m bits on the heap cell at start
+            // so we invoke backward() and return None next call.
+
+            self.heap[self.current].set_forwarding_bit(false);
+            self.heap[self.current].set_mark_bit(self.mark_phase);
+        }
+
+        current
+    }
+
+    fn traverse_subterm(&mut self, h: usize, arity: usize) -> Option<usize> {
+        let mut last_cell_loc = h + arity - 1;
+
+        for idx in (h .. h + arity).rev() {
+            if self.heap[idx].get_forwarding_bit() {
+                if self.cycle_detection_active() {
+                    self.cycle_found = true;
+                    return None;
+                }
+
+                last_cell_loc -= 1;
+            } else if self.heap[idx].get_mark_bit() == self.mark_phase {
+                last_cell_loc -= 1;
+            } else {
+                break;
+            }
+        }
+
+        Some(last_cell_loc)
+    }
+
+    #[inline]
+    fn continue_forwarding(&self) -> bool {
+        self.heap[self.current].get_mark_bit() != self.mark_phase ||
+            self.heap[self.current].get_forwarding_bit()
+    }
+
+    fn forward(&mut self) -> Option<HeapCellValue> {
+        loop {
+            if self.continue_forwarding() {
+                match self.heap[self.current].get_tag() {
+                    tag @ HeapCellValueTag::AttrVar | tag @ HeapCellValueTag::Var => {
+                        let next = self.next as usize;
+
+                        if self.heap[next].get_forwarding_bit() {
+                            if self.current != next {
+                                return if self.cycle_detection_active() {
+                                    self.cycle_found = true;
+                                    None
+                                } else {
+                                    Some(self.backward_and_return())
+                                };
+                            } else if self.backward() {
+                                return None;
+                            }
+
+                            continue;
+                        } else if self.heap[next].get_mark_bit() == self.mark_phase {
+                            return Some(self.backward_and_return());
+                        }
+
+                        self.heap[next].set_forwarding_bit(true);
+
+                        let temp = self.heap[next].get_value();
+
+                        self.heap[next].set_value(self.current as u64);
+                        self.current = next;
+                        self.next = temp;
+
+                        if self.next < self.heap.len() as u64 {
+                            if self.heap[self.next as usize].get_mark_bit() == self.mark_phase {
+                                return Some(HeapCellValue::build_with(tag, next as u64));
+                            }
+                        }
+                    }
+                    HeapCellValueTag::Str => {
+                        let h = self.next as usize;
+                        let cell = self.heap[h];
+                        let arity = cell_as_atom_cell!(self.heap[h]).get_arity();
+
+                        let last_cell_loc = match self.traverse_subterm(h + 1, arity) {
+                            Some(last_cell_loc) => last_cell_loc,
+                            None => return None,
+                        };
+
+                        if last_cell_loc == h {
+                            if self.backward() {
+                                return None;
+                            }
+
+                            continue;
+                        }
+
+                        if self.cycle_detection_active() {
+                            for idx in (h + 1 .. last_cell_loc).rev() {
+                                if self.heap[idx].get_forwarding_bit() {
+                                    self.cycle_found = true;
+                                    return None;
+                                }
+                            }
+                        }
+
+                        self.heap[last_cell_loc].set_forwarding_bit(true);
+
+                        self.next = self.heap[last_cell_loc].get_value();
+                        self.heap[last_cell_loc].set_value(self.current as u64);
+                        self.current = last_cell_loc;
+
+                        return Some(cell);
+                    }
+                    HeapCellValueTag::Lis => {
+                        let mut cell = self.heap[self.current];
+                        cell.set_value(self.next);
+
+                        let last_cell_loc = match self.traverse_subterm(self.next as usize, 2) {
+                            Some(last_cell_loc) => last_cell_loc,
+                            None => return None,
+                        };
+
+                        if (last_cell_loc + 1) as u64 == self.next {
+                            if self.backward() {
+                                return None;
+                            }
+
+                            continue;
+                        } else if last_cell_loc as u64 == self.next {
+                            // car cells of lists are both marked and forwarded.
+                            self.heap[last_cell_loc].set_mark_bit(self.mark_phase);
+                        }
+
+                        self.heap[last_cell_loc].set_forwarding_bit(true);
+
+                        self.next = self.heap[last_cell_loc].get_value();
+                        self.heap[last_cell_loc].set_value(self.current as u64);
+                        self.current = last_cell_loc;
+
+                        return Some(cell);
+                    }
+                    HeapCellValueTag::PStrLoc => {
+                        let h = self.next as usize;
+                        let cell = self.heap[h];
+                        let last_cell_loc = h + 1;
+
+                        if self.heap[last_cell_loc].get_forwarding_bit() {
+                            if self.cycle_detection_active() {
+                                self.cycle_found = true;
+                                return None;
+                            } else if self.backward() {
+                                return None;
+                            }
+
+                            continue;
+                        }
+
+                        self.heap[last_cell_loc].set_forwarding_bit(true);
+
+                        self.next = self.heap[last_cell_loc].get_value();
+                        self.heap[last_cell_loc].set_value(self.current as u64);
+                        self.current = last_cell_loc;
+
+                        return Some(cell);
+                    }
+                    HeapCellValueTag::PStrOffset => {
+                        let h = self.next as usize;
+                        let cell = self.heap[h];
+                        let last_cell_loc = h + 1;
+
+                        if self.heap[h].get_tag() == HeapCellValueTag::PStr {
+                            if self.heap[last_cell_loc].get_forwarding_bit() {
+                                if self.cycle_detection_active() {
+                                    self.cycle_found = true;
+                                    return None;
+                                } else if self.backward() {
+                                    return None;
+                                }
+
+                                continue;
+                            }
+
+                            self.heap[last_cell_loc].set_forwarding_bit(true);
+
+                            self.next = self.heap[last_cell_loc].get_value();
+                            self.heap[last_cell_loc].set_value(self.current as u64);
+                            self.current = last_cell_loc;
+                        } else {
+                            debug_assert!(self.heap[h].get_tag() == HeapCellValueTag::CStr);
+
+                            self.next = self.heap[h].get_value();
+                            self.heap[h].set_value(self.current as u64);
+                            self.current = h;
+                        }
+
+                        return Some(cell);
+                    }
+                    tag @ HeapCellValueTag::Atom => {
+                        let cell = HeapCellValue::build_with(tag, self.next);
+                        let arity = AtomCell::from_bytes(cell.into_bytes()).get_arity();
+
+                        if arity == 0 {
+                            return Some(self.backward_and_return());
+                        } else if self.backward() {
+                            return None;
+                        }
+                    }
+                    HeapCellValueTag::PStr => {
+                        if self.backward() {
+                            return None;
+                        }
+                    }
+                    _ => {
+                        return Some(self.backward_and_return());
+                    }
+                }
+            } else if self.backward() {
+                return None;
+            }
+        }
+    }
+
+    fn pivot_subterm(&mut self) {
+        self.current -= 1;
+
+        let temp = self.heap[self.current + 1].get_value();
+
+        self.heap[self.current + 1].set_value(self.next);
+        self.next = self.heap[self.current].get_value();
+        self.heap[self.current].set_value(temp);
+
+        self.heap[self.current].set_forwarding_bit(true);
+    }
+
+    fn continue_backward(&mut self) -> bool {
+        self.heap[self.current].set_forwarding_bit(false);
+
+        if self.current == self.start {
+            return false;
+        }
+
+        let temp = self.heap[self.current].get_value();
+
+        match self.heap[temp as usize].get_tag() {
+            HeapCellValueTag::Str => {
+                let mut new_str_back_link = self.current;
+
+                for idx in (0 .. self.current).rev() {
+                    if self.heap[idx].get_tag() == HeapCellValueTag::Atom {
+                        if cell_as_atom_cell!(self.heap[idx]).get_arity() > 0 {
+                            new_str_back_link = idx;
+                            break;
+                        }
+                    }
+
+                    if self.heap[idx].get_mark_bit() != self.mark_phase {
+                        if !self.heap[idx].get_forwarding_bit() {
+                            new_str_back_link = idx;
+                            break;
+                        }
+                    }
+                }
+
+                self.heap[self.current].set_mark_bit(self.mark_phase);
+                self.heap[self.current].set_value(self.next);
+
+                let back_link_cell = self.heap[new_str_back_link];
+
+                self.next = back_link_cell.get_value();
+                self.heap[new_str_back_link].set_value(temp);
+                self.current = new_str_back_link;
+
+                read_heap_cell!(back_link_cell,
+                    (HeapCellValueTag::Atom, (_name, arity)) => {
+                        if arity > 0 {
+                            self.heap[self.current].set_mark_bit(self.mark_phase);
+                            return true;
+                        }
+                    }
+                    _ => {}
+                );
+
+                self.heap[self.current].set_forwarding_bit(true);
+                false
+            }
+            HeapCellValueTag::Lis => {
+                if self.heap[self.current].get_mark_bit() == self.mark_phase {
+                    true
+                } else {
+                    self.heap[self.current - 1].set_mark_bit(self.mark_phase);
+                    self.heap[self.current].set_mark_bit(self.mark_phase);
+
+                    if self.heap[self.current - 1].get_forwarding_bit() {
+                        self.next = self.current as u64 - 1;
+                        self.heap[self.current].set_value(self.next);
+                        self.current = temp as usize;
+
+                        true
+                    } else {
+                        self.pivot_subterm();
+                        false
+                    }
+                }
+            }
+            _ => {
+                self.heap[self.current].set_mark_bit(self.mark_phase);
+                true
+            }
+        }
+    }
+
+    fn backward(&mut self) -> bool {
+        while self.continue_backward() {
+            let temp = self.heap[self.current].get_value();
+
+            self.heap[self.current].set_value(self.next);
+            self.next = self.current as u64;
+            self.current = temp as usize;
+        }
+
+        if self.current == self.start {
+            return true;
+        }
+
+        false
+    }
+
+    fn invert_marker(&mut self) {
+        self.cycle_found = false;
+
+        if self.heap[self.start].get_forwarding_bit() {
+            while !self.backward() {}
+        }
+
+        self.mark_phase = false;
+        self.heap[self.start].set_forwarding_bit(true);
+
+        self.next = self.heap[self.start].get_value();
+        self.current = self.start;
+
+        while let Some(_) = self.forward() {}
+    }
+}
+
+impl<'a> Iterator for CycleDetectingIter<'a> {
+    type Item = HeapCellValue;
+
+    #[inline]
+    fn next(&mut self) -> Option<Self::Item> {
+        self.forward()
+    }
+}
+
+
+impl<'a> Drop for CycleDetectingIter<'a> {
+    fn drop(&mut self) {
+        self.invert_marker();
+
+        if self.current == self.start {
+            return;
+        }
+
+        while !self.backward() {}
+    }
+}
diff --git a/src/machine/gc.rs b/src/machine/gc.rs
index b7f257a5..715adde7 100644
--- a/src/machine/gc.rs
+++ b/src/machine/gc.rs
@@ -55,39 +55,6 @@ impl UnmarkPolicy for IteratorUMP {
     }
 }
 
-pub(crate) struct CycleDetectorUMP {
-    mark_phase: bool,
-    focus: usize,
-}
-
-impl UnmarkPolicy for CycleDetectorUMP {
-    #[inline(always)]
-    fn forward_attr_var(iter: &mut StacklessPreOrderHeapIter<Self>) -> Option<HeapCellValue> {
-        iter.forward_var()
-    }
-
-    #[inline]
-    fn invert_marker(iter: &mut StacklessPreOrderHeapIter<Self>) {
-        iter.iter_state.mark_phase = false;
-        invert_marker(iter);
-    }
-
-    #[inline]
-    fn mark_phase(&self) -> bool {
-        self.mark_phase
-    }
-
-    #[inline(always)]
-    fn report_var_link(_iter: &StacklessPreOrderHeapIter<Self>) -> bool {
-        true
-    }
-
-    #[inline(always)]
-    fn record_focus(iter: &mut StacklessPreOrderHeapIter<Self>) {
-        iter.iter_state.focus = iter.current;
-    }
-}
-
 struct MarkerUMP {}
 
 impl UnmarkPolicy for MarkerUMP {
@@ -177,31 +144,6 @@ impl<'a> StacklessPreOrderHeapIter<'a, IteratorUMP> {
     }
 }
 
-impl<'a> StacklessPreOrderHeapIter<'a, CycleDetectorUMP> {
-    pub(crate) fn new(heap: &'a mut [HeapCellValue], start: usize) -> Self {
-        heap[start].set_forwarding_bit(true);
-        let next = heap[start].get_value();
-
-        Self {
-            heap,
-            start,
-            current: start,
-            next,
-            iter_state: CycleDetectorUMP { mark_phase: true, focus: 0 },
-        }
-    }
-
-    #[inline]
-    pub(crate) fn focus(&self) -> usize {
-        self.iter_state.focus
-    }
-
-    #[inline(always)]
-    pub(crate) fn current(&self) -> usize {
-        self.current
-    }
-}
-
 impl<'a, UMP: UnmarkPolicy> StacklessPreOrderHeapIter<'a, UMP> {
     fn backward_and_return(&mut self) -> HeapCellValue {
         let mut current = self.heap[self.current];
diff --git a/src/machine/machine_state_impl.rs b/src/machine/machine_state_impl.rs
index 6a7cb95d..5007e85a 100644
--- a/src/machine/machine_state_impl.rs
+++ b/src/machine/machine_state_impl.rs
@@ -1131,66 +1131,23 @@ impl MachineState {
 
     #[inline]
     pub fn is_cyclic_term(&mut self, value: HeapCellValue) -> bool {
-        use topo_sort::TopoSort;
-
         let value = self.store(self.deref(value));
 
-        if value.is_constant() || value.is_stack_var() {
+        if value.is_stack_var() || value.is_constant() {
             return false;
         }
 
         let h = self.heap.len();
         self.heap.push(value);
 
-        let found_cycle = (|| {
-            let mut topo_graph = TopoSort::new();
+        let cycle_found = {
             let mut iter = cycle_detecting_stackless_preorder_iter(&mut self.heap, h);
-
-            while let Some(cell) = iter.next() {
-                let focus = iter.focus();
-
-                read_heap_cell!(cell,
-                    (HeapCellValueTag::Atom, (_name, arity)) => {
-                        if arity > 0 {
-                            // focus is actually the location of Str(s) here.
-                            let s = iter.current() - arity;
-
-                            topo_graph.insert(focus, vec![s]);
-                            topo_graph.insert(s, (s + 1 .. s + arity + 1).collect::<Vec<_>>());
-                        }
-                    }
-                    (HeapCellValueTag::Str, s) => {
-                        topo_graph.insert(focus, vec![s]);
-                    }
-                    (HeapCellValueTag::Lis | HeapCellValueTag::PStrLoc | HeapCellValueTag::PStrOffset, l) => {
-                        if l <= focus && focus < l + 2 {
-                            // TopoSort doesn't consider focus ->
-                            // focus to induce a cycle so in this case
-                            // it must be checked manually.
-                            return true;
-                        }
-
-                        topo_graph.insert(focus, (l .. l + 2).collect::<Vec<_>>());
-                    }
-                    (HeapCellValueTag::AttrVar | HeapCellValueTag::Var, h) => {
-                        if focus != h {
-                            topo_graph.insert(focus, vec![h]);
-                        }
-                    }
-                    _ => {
-                    }
-                );
-
-                if topo_graph.cycle_detected() {
-                    break;
-                }
-            }
-
-            topo_graph.cycle_detected()
-        })();
+            while let Some(_) = iter.next() {}
+            iter.cycle_found()
+        };
 
         self.heap.pop();
-        found_cycle
+        cycle_found
     }
 
     // arg(+N, +Term, ?Arg)
diff --git a/src/machine/mod.rs b/src/machine/mod.rs
index 665366c3..c44df762 100644
--- a/src/machine/mod.rs
+++ b/src/machine/mod.rs
@@ -6,6 +6,7 @@ pub mod code_walker;
 pub mod loader;
 pub mod compile;
 pub mod copier;
+pub mod cycle_detection;
 pub mod disjuncts;
 pub mod dispatch;
 pub mod gc;