diff --git a/tensorrt_llm/_torch/auto_deploy/transform/library/kvcache.py b/tensorrt_llm/_torch/auto_deploy/transform/library/kvcache.py
index bb12d726b0d7..e4aadeab3d6e 100644
--- a/tensorrt_llm/_torch/auto_deploy/transform/library/kvcache.py
+++ b/tensorrt_llm/_torch/auto_deploy/transform/library/kvcache.py
@@ -432,6 +432,15 @@ def _apply(
                         # None sentinel: pass literal None positionally, no resource allocated.
                         cache_in_nodes.append(None)
                     else:
+                        # A window that can't slide within max_seq_len is functionally
+                        # full attention; normalize it to 0 so the layer shares the
+                        # full-attention pool instead of forking a redundant
+                        # single-window pool.
+                        if (
+                            isinstance(resource_handler, KVPagedResourceHandler)
+                            and resource_handler.sliding_window >= cm.info.max_seq_len
+                        ):
+                            resource_handler.sliding_window = 0
                         resource_name = cm.add_resource(k, resource_handler)
                         node = self._process_cache_node(gm, resource_name)
                         cache_in_nodes.append(node)
diff --git a/tests/integration/test_lists/waives.txt b/tests/integration/test_lists/waives.txt
index dafbb1db1c6d..ad2867423401 100644
--- a/tests/integration/test_lists/waives.txt
+++ b/tests/integration/test_lists/waives.txt
@@ -13,7 +13,6 @@ accuracy/test_llm_api.py::TestMistralNemo12B::test_fp8 SKIP (https://nvbugs/5413
 accuracy/test_llm_api_autodeploy.py::TestGemma4MoE::test_bf16 SKIP (https://nvbugs/6158397)
 accuracy/test_llm_api_autodeploy.py::TestGemmaE2B::test_gemma4_e2b_it SKIP (https://nvbugs/6194934)
 accuracy/test_llm_api_autodeploy.py::TestMiniMaxM2::test_finegrained_fp8 SKIP (https://nvbugs/6158397)
-accuracy/test_llm_api_autodeploy.py::TestModelRegistryAccuracy::test_autodeploy_from_registry[google_gemma-3-1b-it-False] SKIP (https://nvbugs/6248764)
 accuracy/test_llm_api_autodeploy.py::TestModelRegistryAccuracy::test_autodeploy_from_registry[mistralai_Ministral-8B-Instruct-2410-False] SKIP (https://nvbugs/6248769)
 accuracy/test_llm_api_autodeploy.py::TestModelRegistryAccuracy::test_autodeploy_from_registry[nvidia_Llama-3.1-8B-Instruct-NVFP4-True] SKIP (https://nvbugs/6245279)
 accuracy/test_llm_api_autodeploy.py::TestNemotronNanoV3::test_accuracy[bf16-4-trtllm] SKIP (https://nvbugs/6185150)