[BugFix][Fusion] Fix graph fusion failure problem (#5253)

Currently, the vllm pull request
(https://github.com/vllm-project/vllm/pull/24252) is causing operator
fusion to fail. This issue was previously fixed by patching the backend.
The root cause has been identified, and the problem can be resolved with
this pull request.

- vLLM version: release/v0.13.0
- vLLM main:
ad32e3e19c

---------

Signed-off-by: wxsIcey <1790571317@qq.com>

vllm_ascend/compilation/passes/norm_quant_fusion_pass.py

@@ -20,6 +20,7 @@ import torch._inductor.pattern_matcher as pm
 from torch._inductor.pattern_matcher import PatternMatcherPass
 from vllm.compilation.vllm_inductor_pass import VllmInductorPass
 from vllm.config import VllmConfig
+from vllm.config.compilation import Range
 from vllm.logger import logger
 
 
@@ -308,7 +309,7 @@ class AddRMSNormQuantFusionPass(VllmInductorPass):
         logger.debug("Replaced %s patterns", self.matched_count)
         self.end_and_log()
 
-    def is_applicable(self, runtime_shape: int | None = None) -> bool:
+    def is_applicable_for_range(self, compile_range: Range) -> bool:
         """
         Check if the pass is applicable for the current configuration.
         """

vllm_ascend/compilation/passes/qknorm_rope_fusion_pass.py

@@ -22,6 +22,7 @@ from torch._inductor.pattern_matcher import (PatternMatcherPass,
 from vllm.attention.layer import Attention
 from vllm.compilation.vllm_inductor_pass import VllmInductorPass
 from vllm.config import VllmConfig, get_layers_from_vllm_config
+from vllm.config.compilation import Range
 from vllm.logger import logger
 
 
@@ -283,7 +284,7 @@ class QKNormRopeFusionPass(VllmInductorPass):
                 pattern_idx += 1
         self.end_and_log()
 
-    def is_applicable(self, runtime_shape):
+    def is_applicable_for_range(self, compile_range: Range) -> bool:
         """
         Check if the pass is applicable for the current configuration.
         """