fix

Signed-off-by: Cody Yu <[email protected]>
vllm-project · Nov 4, 2024 · 87f5195 · 87f5195
1 parent 62354d7
commit 87f5195
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 3 deletions.
diff --git a/vllm/v1/core/kv_cache_manager.py b/vllm/v1/core/kv_cache_manager.py
@@ -68,10 +68,10 @@ def __init__(self, blocks: List[KVCacheBlock]) -> None:
         # Initialize the doubly linked list of free blocks.
         self.free_list_head = blocks[0]
         self.free_list_tail = blocks[-1]
-        for i in range(self.size):
+        for i in range(self.num_free_blocks):
             if i > 0:
                 blocks[i].prev_free_block = blocks[i - 1]
-            if i < self.size - 1:
+            if i < self.num_free_blocks - 1:
                 blocks[i].next_free_block = blocks[i + 1]
 
     def popleft(self) -> KVCacheBlock:

diff --git a/vllm/v1/core/scheduler.py b/vllm/v1/core/scheduler.py
@@ -210,7 +210,6 @@ def schedule(self) -> "SchedulerOutput":
         )
 
         self.finished_req_ids = set()
-        self.kv_cache_manager.async_remove_touched_blocks()
         return scheduler_output
 
     def _make_running_request_data(