test: fix Task 1 CPU unit coverage

test/registered/unit/managers/test_prefill_adder.py

@@ -1,7 +1,12 @@
+import sys
 import unittest
 from types import SimpleNamespace
 from unittest.mock import MagicMock
 
+for _mod in ("sgl_kernel", "sgl_kernel.kvcacheio"):
+    if _mod not in sys.modules:
+        sys.modules[_mod] = MagicMock()
+
 from sglang.srt.managers.schedule_batch import Req
 from sglang.srt.managers.schedule_policy import AddReqResult, PrefillAdder
 from sglang.srt.mem_cache.base_prefix_cache import (
@@ -453,7 +458,6 @@ class TestPrefillAdder(CustomTestCase):
             running_batch,
             page_size=64,
             rem_input_tokens=4096,
-            rem_total_tokens=4096,
         )
         req = self.create_mock_req("req", priority=0, max_new_tokens=16)
         req.extend_input_len = 256
@@ -474,7 +478,7 @@ class TestPrefillAdder(CustomTestCase):
 
         self.assertNotEqual(result, AddReqResult.NO_TOKEN)
         params = self.mock_tree_cache.init_load_back.call_args.args[0]
-        self.assertEqual(params.mem_quota, 256)
+        self.assertEqual(params.mem_quota, 320)
 
 
 if __name__ == "__main__":

test/registered/unit/mem_cache/test_radix_cache_unit.py

@@ -17,6 +17,13 @@ Usage:
     python -m pytest test_radix_cache_unit.py::TestRadixCache::test_insert_basic
 """
 
+import sys
+import unittest.mock
+
+for _mod in ("sgl_kernel", "sgl_kernel.kvcacheio"):
+    if _mod not in sys.modules:
+        sys.modules[_mod] = unittest.mock.MagicMock()
+
 from sglang.srt.mem_cache.common import available_and_evictable_str
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 
@@ -27,7 +34,6 @@ register_amd_ci(est_time=5, suite="stage-b-test-1-gpu-small-amd")
 import random
 import time
 import unittest
-import unittest.mock
 
 import torch
 
@@ -743,6 +749,7 @@ class TestRadixCache(unittest.TestCase):
             # Should have 1 page (split at page_size=2)
             self.assertEqual(len(node.hash_value), 1)
 
+    @unittest.skipIf(not torch.cuda.is_available(), "CUDA is required")
     def test_memory_allocated(self):
         keys, values = [], []