bpo-37257: obmalloc: stop simple arena thrashing (#14039)

GH-14039:  allow (no more than) one wholly empty arena on the usable_arenas list.

This prevents thrashing in some easily-provoked simple cases that could end up creating and destroying an arena on each loop iteration in client code.   Intuitively, if the only arena on the list becomes empty, it makes scant sense to give it back to the system unless we know we'll never need another free pool again before another arena frees a pool.  If the latter obtains, then - yes - this will "waste" an arena.

1 files changed, 7 insertions, 2 deletions

diff --git a/Objects/obmalloc.c b/Objects/obmalloc.c
index fc7bef6..622da3a 100644
--- a/Objects/obmalloc.c
+++ b/Objects/obmalloc.c
@@ -1758,7 +1758,12 @@ pymalloc_free(void *ctx, void *p)
     /* All the rest is arena management.  We just freed
      * a pool, and there are 4 cases for arena mgmt:
      * 1. If all the pools are free, return the arena to
-     *    the system free().
+     *    the system free().  Except if this is the last
+     *    arena in the list, keep it to avoid thrashing:
+     *    keeping one wholly free arena in the list avoids
+     *    pathological cases where a simple loop would
+     *    otherwise provoke needing to allocate and free an
+     *    arena on every iteration.  See bpo-37257.
      * 2. If this is the only free pool in the arena,
      *    add the arena back to the `usable_arenas` list.
      * 3. If the "next" arena has a smaller count of free
@@ -1767,7 +1772,7 @@ pymalloc_free(void *ctx, void *p)
      *    nfreepools.
      * 4. Else there's nothing more to do.
      */
-    if (nf == ao->ntotalpools) {
+    if (nf == ao->ntotalpools && ao->nextarena != NULL) {
         /* Case 1.  First unlink ao from usable_arenas.
          */
         assert(ao->prevarena == NULL ||