From 0f72060d66d0ef453303800629c0f39ee6e60bac Mon Sep 17 00:00:00 2001 From: MOVZX Date: Fri, 9 Oct 2026 20:13:04 +0700 Subject: [PATCH] [Scheduler] Preserve per-group eviction order in deferred block free (port of upstream #48532) Repair: the port driver logged this patch as resolved but produced no commit; the -3 merge dropped the coordinator code hunk. Reapplied to HEAD. Co-Authored-By: Hinata Signed-off-by: MOVZX --- vllm/v1/core/kv_cache_coordinator.py | 13 +++++++++---- vllm/v1/core/kv_cache_manager.py | 6 ++++-- vllm/v1/core/sched/scheduler.py | 9 ++++++--- 3 files changed, 19 insertions(+), 9 deletions(-) diff --git a/vllm/v1/core/kv_cache_coordinator.py b/vllm/v1/core/kv_cache_coordinator.py index 9b91bb1920..f457bb9fdd 100644 --- a/vllm/v1/core/kv_cache_coordinator.py +++ b/vllm/v1/core/kv_cache_coordinator.py @@ -402,19 +402,24 @@ class KVCacheCoordinator(ABC): """Pop the request's bookkeeping from all single-type managers and return its blocks without returning them to the block pool. The caller must eventually pass the returned blocks to - `block_pool.free_blocks`, freeing them in reverse order (so that - tail blocks are evicted first). + `block_pool.free_blocks` as-is: they are already ordered so that + tail blocks are evicted first, within each KV cache group. Args: request_id: The request ID. Returns: - The request's blocks in allocation order. + The request's blocks in eviction order (each group's blocks + reversed, then concatenated group by group). """ blocks: list[KVCacheBlock] = [] for manager in self.single_type_managers: - blocks.extend(manager.pop_blocks_for_free(request_id)) + # Reverse within each group so the concatenation is already in + # eviction order, matching `free()`, which frees each group in its + # own reversed `free_blocks()` call. Reversing the flat list + # instead would emit whole groups in the wrong order. + blocks.extend(reversed(manager.pop_blocks_for_free(request_id))) return blocks def get_num_common_prefix_blocks(self, running_request_id: str) -> list[int]: diff --git a/vllm/v1/core/kv_cache_manager.py b/vllm/v1/core/kv_cache_manager.py index ad7bf18090..4e358da2ba 100644 --- a/vllm/v1/core/kv_cache_manager.py +++ b/vllm/v1/core/kv_cache_manager.py @@ -651,13 +651,15 @@ class KVCacheManager: def pop_blocks_for_free(self, request: Request) -> list[KVCacheBlock]: """Pop the request's bookkeeping and return its blocks without returning them to the block pool. The caller must eventually free - them in reverse order (so that tail blocks are evicted first). + them as-is: they are already ordered so that tail blocks are + evicted first, within each KV cache group. Args: request: The request to pop the blocks for. Returns: - The request's blocks in allocation order. + The request's blocks in eviction order (each group's blocks + reversed, then concatenated group by group). """ return self.coordinator.pop_blocks_for_free(request.request_id) diff --git a/vllm/v1/core/sched/scheduler.py b/vllm/v1/core/sched/scheduler.py index 5f17f1e017..e5eb8bc1c1 100644 --- a/vllm/v1/core/sched/scheduler.py +++ b/vllm/v1/core/sched/scheduler.py @@ -2831,7 +2831,9 @@ class Scheduler(SchedulerInterface): if not self.defer_block_free or fence_seq <= self.processed_step_seq: self.kv_cache_manager.block_pool.free_blocks(blocks) return - self.deferred_frees.append((fence_seq, blocks[::-1])) + # The immediate path above frees these as-is, so enqueue them as-is: + # the drain no longer reverses. + self.deferred_frees.append((fence_seq, blocks)) def _drain_deferred_frees(self): """Return deferred blocks whose fence step has completed. @@ -2845,8 +2847,9 @@ class Scheduler(SchedulerInterface): if fence > self.processed_step_seq: break _, blocks = self.deferred_frees.popleft() - # Free in reverse order so that the tail blocks are evicted first. - self.kv_cache_manager.block_pool.free_blocks(reversed(blocks)) + # Already in eviction order (reversed per KV cache group by + # `pop_blocks_for_free`), so free as-is. + self.kv_cache_manager.block_pool.free_blocks(blocks) def get_num_unfinished_requests(self) -> int: if self._pause_state == PauseState.PAUSED_ALL: