fix: graceful fallback when attention backends fail to import

[sym-bot]

Problem

External attention backends (flash_attn, xformers, sageattention, etc.) may be installed but fail to import at runtime due to ABI mismatches. For example, when flash_attn is compiled against PyTorch 2.4 but used with PyTorch 2.8, the import fails with:

OSError: .../flash_attn_2_cuda.cpython-311-x86_64-linux-gnu.so: undefined symbol: _ZN3c104cuda9SetDeviceEab

The current code uses importlib.util.find_spec() to check if packages exist, but this only verifies the package is installed—not that it can actually be imported. When the import fails, diffusers crashes instead of falling back to native PyTorch attention.

Solution

Wrap all external attention backend imports in try-except blocks that catch ImportError and OSError. On failure:

1. Log a warning message explaining the issue
2. Set the corresponding _CAN_USE_* flag to False
3. Set the imported functions to None

This allows diffusers to gracefully degrade to PyTorch's native SDPA (scaled_dot_product_attention) instead of crashing.

Affected backends

• flash_attn (Flash Attention)
• flash_attn_3 (Flash Attention 3)
• aiter (AMD Instinct)
• sageattention (SageAttention)
• flex_attention (PyTorch Flex Attention)
• torch_npu (Huawei NPU)
• torch_xla (TPU/XLA)
• xformers (Meta xFormers)

Testing

Tested with PyTorch 2.8.0 and flash_attn 2.7.4.post1 (compiled for PyTorch 2.4).

• Before: crashes on from diffusers import ... with undefined symbol error
• After: logs warning and uses native attention successfully

Example warning output

WARNING:diffusers.models.attention_dispatch:flash_attn is installed but failed to import: .../flash_attn_2_cuda.cpython-311-x86_64-linux-gnu.so: undefined symbol: _ZN3c104cuda9SetDeviceEab. Falling back to native PyTorch attention.