From 0cd5353644d3d045ab33c7e8e19c182bfd7db911 Mon Sep 17 00:00:00 2001
From: "Li, Jiang" <jiang1.li@intel.com>
Date: Wed, 17 Dec 2025 15:25:12 +0800
Subject: [PATCH] [Bugfix][CPU] Fix CPU backend ROPE dispatch for VL models
 (#30829)

Signed-off-by: jiang1.li <jiang1.li@intel.com>
Signed-off-by: Li, Jiang <bigpyj64@gmail.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
---
 vllm/model_executor/layers/rotary_embedding/common.py | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/vllm/model_executor/layers/rotary_embedding/common.py b/vllm/model_executor/layers/rotary_embedding/common.py
index 3e6584dbc3da0..50660c6ecc223 100644
--- a/vllm/model_executor/layers/rotary_embedding/common.py
+++ b/vllm/model_executor/layers/rotary_embedding/common.py
@@ -264,6 +264,15 @@ class ApplyRotaryEmb(CustomOp):
 
         return output
 
+    def forward_cpu(
+        self,
+        x: torch.Tensor,
+        cos: torch.Tensor,
+        sin: torch.Tensor,
+    ) -> torch.Tensor:
+        # TODO (bigPYJ1151): need to enable fused CPU ROPE here
+        return self.forward_native(x, cos, sin)
+
     def extra_repr(self) -> str:
         s = f"is_neox_style={self.is_neox_style}"
         s += f"enable_fp32_compute={self.enable_fp32_compute}"