[Bugfix][Core] Fix CPU memory leak from Request reference cycle in prefix caching (#34183)

Signed-off-by: Roger Wang <hey@rogerw.io>
2026-02-09 21:03:32 -08:00
parent 4cde2e0159
commit 8a5e0e2b2b
3 changed files with 14 additions and 12 deletions
--- a/tests/v1/core/test_async_scheduler.py
+++ b/tests/v1/core/test_async_scheduler.py
@@ -236,7 +236,7 @@ def test_prefix_caching_for_multi_turn():
        req._all_token_ids = req.prompt_token_ids.copy()
        req.all_token_ids = ConstantList(req._all_token_ids)
        req.block_hashes = []
-        req.block_hashes = req.get_hash_new_full_blocks()
+        req.update_block_hashes()

    # Schedule the next-turn requests.
    for req in next_turn_requests: