[Bugfix] Fix tests/kernels/test_mamba_ssm_ssd.py (#16623)

Signed-off-by: Tyler Michael Smith <[email protected]>

Author: tlrmchlsmth

tests/kernels/test_mamba_ssm_ssd.py

@@ -5,6 +5,8 @@
 import torch.nn.functional as F
 from einops import rearrange, repeat
 
+from vllm.model_executor.layers.mamba.mamba2_metadata import (
+    _seq_idx_to_chunk_indices_offsets)
 from vllm.model_executor.layers.mamba.ops.ssd_combined import (
     mamba_chunk_scan_combined)
 from vllm.platforms import current_platform
@@ -160,14 +162,14 @@ def end_boundary(n: int):
 
         # get the metadata
         cu_seqlens = torch.tensor((0, ) + spec, device=device).cumsum(dim=0)
-        sed_idx = torch.zeros(cu_seqlens[-1],
+        seq_idx = torch.zeros(cu_seqlens[-1],
                               dtype=torch.int32,
                               device=cu_seqlens.device)
         for i, (srt, end) in enumerate(zip(
                 cu_seqlens,
                 cu_seqlens[1:],
         )):
-            sed_idx[srt:end] = i
+            seq_idx[srt:end] = i
 
         # for cont batch
         if IND_E is None:
@@ -177,7 +179,7 @@ def end_boundary(n: int):
         IND_E = [end_boundary(x + y) for x, y in zip(IND_S, spec)]
 
         yield ([Y_min[s, IND_S[s]:IND_E[s]] for s in range(num_examples)],
-               cu_seqlens, sed_idx.unsqueeze(0), (A, dt2, X2, B2, C2))
+               cu_seqlens, seq_idx.unsqueeze(0), (A, dt2, X2, B2, C2))
 
 
 @pytest.mark.parametrize("itype",
@@ -266,12 +268,15 @@ def test_mamba_chunk_scan_cont_batch(d_head, n_heads, seq_len_chunk_size_cases,
     exhausted: dict = {}  # map: eg -> boolean indicating example is exhausted
 
     states = None
-    for Y_min, cu_seqlens, sed_idx, (A, dt, X, B,
+    for Y_min, cu_seqlens, seq_idx, (A, dt, X, B,
                                      C) in generate_continous_batched_examples(
                                          cases, num_examples, seqlen,
                                          last_taken, exhausted, n_heads,
                                          d_head, itype):
 
+        chunk_indices, chunk_offsets = _seq_idx_to_chunk_indices_offsets(
+            seq_idx, chunk_size)
+
         Y, new_states = mamba_chunk_scan_combined(
             X,
             dt,
@@ -281,7 +286,9 @@ def test_mamba_chunk_scan_cont_batch(d_head, n_heads, seq_len_chunk_size_cases,
             chunk_size,
             D=None,
             cu_seqlens=cu_seqlens,
-            seq_idx=sed_idx,
+            seq_idx=seq_idx,
+            chunk_indices=chunk_indices,
+            chunk_offsets=chunk_offsets,
             return_varlen_states=True,
             initial_states=states,
         )