sgl-project/sglang · error · ValueError

{name}_block_cnt and {name}_block_idx must both be provided

Error message

{name}_block_cnt and {name}_block_idx must both be provided or both be None

What it means

Block-sparsity metadata comes in pairs: {name}_block_cnt and {name}_block_idx. _check_and_expand_block enforces that both are provided together or both are None; supplying only one raises ValueError.

Source

Thrown at python/sglang/kernels/ops/attention/flash_attn/cute/block_sparsity.py:242

        hint_clause = f" Hint: {resolved_hint}" if resolved_hint else ""
        raise ValueError(
            f"{tensor_name}{context_clause} with shape {tensor.shape} cannot be expanded to expected shape {expected_shape}."
            f"{hint_clause}"
        )
    return tensor.expand(*expected_shape)


def _check_and_expand_block(
    name: str,
    cnt: torch.Tensor | None,
    idx: torch.Tensor | None,
    expected_count_shape: Tuple[int, ...],
    expected_index_shape: Tuple[int, ...],
    context: str | None,
    hint: str | Callable[[], str] | None,
) -> Tuple[torch.Tensor | None, torch.Tensor | None]:
    if (cnt is None) != (idx is None):
        raise ValueError(
            f"{name}_block_cnt and {name}_block_idx must both be provided or both be None"
        )
    if cnt is None or idx is None:
        return None, None
    if cnt.dtype != torch.int32 or idx.dtype != torch.int32:
        raise ValueError(f"{name}_block tensors must have dtype torch.int32")
    if cnt.device != idx.device:
        raise ValueError(
            f"{name}_block_cnt and {name}_block_idx must be on the same device"
        )
    if not cnt.is_cuda or not idx.is_cuda:
        raise ValueError(f"{name}_block tensors must live on CUDA")
    expanded_cnt = _expand_sparsity_tensor(
        cnt, expected_count_shape, f"{name}_block_cnt", context, hint
    )
    # [Note] Allow Compact block sparse indices
    # Allow the last dimension (n_blocks) of idx to be <= expected, since
    # FA4 only accesses indices 0..cnt-1 per query tile. This enables compact

View on GitHub (pinned to 0132848349)

Solutions

1. Provide both tensors or neither
2. Check call site: if cnt is not None: assert idx is not None

Example fix

# before
res = normalize_block_sparse_tensors(topk_cnt, None, ...)
# after
res = normalize_block_sparse_tensors(topk_cnt, topk_idx, ...)

assert (cnt is None) == (idx is None), 'cnt and idx must be provided together'

def valid_block_pair(cnt, idx) -> bool:\n    return (cnt is None) == (idx is None)

When it happens

Trigger: Calling normalize_block_sparse_tensors (or FA with block-sparse args) with e.g. topk_block_cnt set but topk_block_idx=None, or vice versa.

Common situations: Building only the count tensor and forgetting the index tensor; partial refactors of sparsity metadata plumbing; optional-arg defaults left None on one side.

Related errors

• Invalid attention metadata values.Sparsity should be in [0,
• {tensor_name}{context_clause} with shape {tensor.shape} cann
• {name}_block tensors must have dtype torch.int32
• {name}_block_cnt and {name}_block_idx must be on the same de
• {name}_block tensors must live on CUDA

AI-assisted analysis of sgl-project/sglang@0132848349 (2026-08-28). Data as JSON: /api/errors/b347f1aa7941016f. Report an issue: GitHub.