[Bugfix][Kernel] fix potential cuda graph broken for merge_attn_states kernel (#16693)

Signed-off-by: DefTruth <qiustudent_r@163.com>

csrc/attention/merge_attn_states.cu

@@ -107,7 +107,8 @@ __global__ void merge_attn_states_kernel(
 
 #define LAUNCH_MERGE_ATTN_STATES(scalar_t, NUM_THREADS)                     \
   {                                                                         \
-    vllm::merge_attn_states_kernel<scalar_t, NUM_THREADS><<<grid, block>>>( \
+    vllm::merge_attn_states_kernel<scalar_t, NUM_THREADS>                   \
+        <<<grid, block, 0, stream>>>(                                       \
             reinterpret_cast<scalar_t*>(output.data_ptr()), output_lse_ptr, \
             reinterpret_cast<scalar_t*>(prefix_output.data_ptr()),          \
             reinterpret_cast<float*>(prefix_lse.data_ptr()),                \
@@ -122,10 +123,10 @@ __global__ void merge_attn_states_kernel(
  * @param output [n,h,d] The output tensor to store the merged attention states.
  * @param output_lse [h,d] Optional tensor to store the log-sum-exp values.
  * @param prefix_output [n,h,d] The prefix attention states.
- * @param prefix_lse [h,d] The log-sum-exp values for the prefix attention
+ * @param prefix_lse [h,n] The log-sum-exp values for the prefix attention
  * states.
  * @param suffix_output [n,h,d] The suffix attention states.
- * @param suffix_lse [h,d] The log-sum-exp values for the suffix attention
+ * @param suffix_lse [h,n] The log-sum-exp values for the suffix attention
  * states.
  */
 template <typename scalar_t>
@@ -146,13 +147,17 @@ void merge_attn_states_launcher(torch::Tensor& output,
   if (output_lse.has_value()) {
     output_lse_ptr = output_lse.value().data_ptr<float>();
   }
-  // process one pack elements per thread. float -> 4, half/bf16 -> 8
+  // Process one pack elements per thread. for float, the
+  // pack_size is 4 for half/bf16, the pack_size is 8.
   const uint threads_per_head = head_size / pack_size;
   const uint total_threads = num_tokens * num_heads * threads_per_head;
 
   dim3 block(NUM_THREADS);
   dim3 grid((total_threads + NUM_THREADS - 1) / NUM_THREADS);
 
+  const c10::cuda::OptionalCUDAGuard device_guard(prefix_output.device());
+  auto stream = at::cuda::getCurrentCUDAStream();
+
   LAUNCH_MERGE_ATTN_STATES(scalar_t, NUM_THREADS);
 }