From d72b0be33cdd561e557df1ce5350a14451b9af13 Mon Sep 17 00:00:00 2001
From: "Chendi.Xue" <chendi.xue@intel.com>
Date: Wed, 25 Feb 2026 09:31:07 -0600
Subject: [PATCH] [XPU]Fix for Qwen-OMNI crash (#35249)

Signed-off-by: Chendi Xue <chendi.xue@intel.com>
---
 vllm/_xpu_ops.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/vllm/_xpu_ops.py b/vllm/_xpu_ops.py
index e40b18f81..1f64aacd4 100644
--- a/vllm/_xpu_ops.py
+++ b/vllm/_xpu_ops.py
@@ -105,9 +105,10 @@ class xpu_ops:
             assert len(window_size) == 2
             real_window_size = (window_size[0], window_size[1])  # noqa: F841
 
-        # In encode attention, v maybe not contiguous and current
+        # In encode attention, k and v maybe not contiguous and current
         # kernel can't handle it
         if block_table is None:
+            k = k.contiguous()
             v = v.contiguous()
         return flash_attn_varlen_func(
             out=out,