Add stats for visits during marking

python · markshannon · Nov 18, 2024 · Nov 4, 2024 · Nov 4, 2024 · Nov 4, 2024
commit 8bd7606d90c8b2a1e610a510b3a0836942bfaac0
diff --git a/Include/cpython/pystats.h b/Include/cpython/pystats.h
@@ -98,6 +98,7 @@ typedef struct _object_stats {
 typedef struct _gc_stats {
     uint64_t collections;
     uint64_t object_visits;
+    uint64_t mark_visits;
     uint64_t objects_collected;
     uint64_t objects_marked;
 } GCStats;

diff --git a/Python/gc.c b/Python/gc.c
@@ -1360,15 +1360,23 @@ IS_IN_VISITED(PyGC_Head *gc, int visited_space)
 struct container_and_flag {
     PyGC_Head *container;
     int visited_space;
+    int mark;
     uintptr_t size;
 };
 
 /* A traversal callback for adding to container) */
 static int
 visit_add_to_container(PyObject *op, void *arg)
 {
-    OBJECT_STAT_INC(object_visits);
     struct container_and_flag *cf = (struct container_and_flag *)arg;
+#ifdef Py_STATS
+    if (cf->mark) {
+        GC_STAT_ADD(1, mark_visits, 1);
+    }
+    else {
+        OBJECT_STAT_INC(object_visits);
+    }
+#endif
     int visited = cf->visited_space;
     assert(visited == get_gc_state()->visited_space);
     if (!_Py_IsImmortal(op) && _PyObject_IS_GC(op)) {
@@ -1390,6 +1398,7 @@ expand_region_transitively_reachable(PyGC_Head *container, PyGC_Head *gc, GCStat
     struct container_and_flag arg = {
         .container = container,
         .visited_space = gcstate->visited_space,
+        .mark = 0,
         .size = 0
     };
     assert(GC_NEXT(gc) == container);
@@ -1493,6 +1502,7 @@ gc_mark(PyThreadState *tstate, struct gc_collection_stats *stats)
     struct container_and_flag arg = {
         .container = &reachable,
         .visited_space = gcstate->visited_space,
+        .mark = 1,
         .size = 0
     };
     while (!gc_list_is_empty(&reachable)) {

diff --git a/Python/specialize.c b/Python/specialize.c
@@ -229,6 +229,7 @@ print_gc_stats(FILE *out, GCStats *stats)
 {
     for (int i = 0; i < NUM_GENERATIONS; i++) {
         fprintf(out, "GC[%d] collections: %" PRIu64 "\n", i, stats[i].collections);
+        fprintf(out, "GC[%d] marking visits: %" PRIu64 "\n", i, stats[i].mark_visits);
         fprintf(out, "GC[%d] object visits: %" PRIu64 "\n", i, stats[i].object_visits);
         fprintf(out, "GC[%d] objects marked: %" PRIu64 "\n", i, stats[i].objects_marked);
         fprintf(out, "GC[%d] objects collected: %" PRIu64 "\n", i, stats[i].objects_collected);

diff --git a/Tools/scripts/summarize_stats.py b/Tools/scripts/summarize_stats.py
@@ -1118,6 +1118,7 @@ def calc_gc_stats(stats: Stats) -> Rows:
                 Count(gen["collections"]),
                 Count(gen["objects collected"]),
                 Count(gen["object visits"]),
+                Count(gen["marking visits"]),
                 Count(gen["objects marked"]),
             )
             for (i, gen) in enumerate(gc_stats)
@@ -1128,7 +1129,7 @@ def calc_gc_stats(stats: Stats) -> Rows:
         "GC collections and effectiveness",
         [
             Table(
-                ("Generation:", "Collections:", "Objects collected:", "Object visits:", "Objects marked:"),
+                ("Generation:", "Collections:", "Objects collected:", "Object visits:", "Marking visits:", "Objects marked:"),
                 calc_gc_stats,
             )
         ],