Function forward_linear_attention 

pub fn forward_linear_attention<Q: QuantWeights>(
    gpu: &dyn GpuBackend,
    cfg: &Qwen35ForwardConfig,
    k: &Qwen35Kernels,
    layer: &LinearAttentionLayer<'_, Q>,
    state: &LinearAttentionState,
    scratch: &LinearAttentionScratch,
    x_in: DevicePtr,
    x_buf: DevicePtr,
    stream: u64,
    intra_dump: Option<&dyn Fn(&str, DevicePtr, u32) -> Result<()>>,
) -> Result<DevicePtr>

2026-09-25: Runs one GDN decoder layer for one token and returns x_buf, into which the layer’s output residual stream (scratch.x_final) is copied. intra_dump, when given, receives named intermediate buffers after a stream synchronize.