RK3588板上推理模型遇到段错误

[复制链接] · 发表于 2025-4-6 13:47:57

项目中一个魔改的YOLO Classify模型，在PC端上可以正常将ONNX转换成RKNN模型并通过模拟器推理出正确结果，但是将该RKNN模型放到RK3588的板子上，使用Rknntoolkit-lite2进行推理，出现"段错误"并终止。图1这是在PC虚拟机上转换模型并模拟器推理的输出，图2是RK3588板上的推理终端信息，我看二者的Feature Tensor Information Table和Const Tensor Information Table都是完全一致的，大佬们在部署模型时是否遇到过类似问题？以及能否提供些解决思路？相关代码模型以及数据都已经打包，如果需要的话直接解压就可以运行，谢谢！通过网盘分享的文件：yolo.zip
链接: https://pan.baidu.com/s/1kEnZYqCwQFE0hFneIxGmhA?pwd=fe9h 提取码: fe9h
--来自百度网盘超级会员v6的分享

只看该作者 · 发表于 2025-4-7 08:56:17

先确认驱动版本和rknn版本对的上

只看该作者 · 发表于 2025-4-7 15:53:25

jefferyzhang 发表于 2025-4-7 08:56
先确认驱动版本和rknn版本对的上

感谢您回复，我的驱动版本和RKNN版本应该都是最新的，下面是相应的信息
“RKNN Runtime Information, librknnrt version: 2.3.0 (c949ad889d@2024-11-07T11:35:33)
RKNN Driver Information, version: 0.9.8
RKNN Model Information, version: 6, toolkit version: 2.3.0(compiler version: 2.3.0 (c949ad889d@2024-11-07T11:39:30)), target: RKNPU v2, target platform: rk3588”,
为了获得更详细的报错信息，我进行了连板推理，下面是一些我截的其中关键的报错信息，您能否看一下这里面有无能够帮助定位解决问题的信息？“E RKNNAPI: Catch runtime Crash! Stack backtrace:
rknn_server() [0x40c584]
rknn_server(_Z14signal_handleriP9siginfo_tPv+0xbc) [0x41967c]
linux-vdso.so.1(__kernel_rt_sigreturn+0) [0x7fb988d7d0]
/lib/librknnrt.so(+0x4fbc3c) [0x7fafdfdc3c]
/lib/librknnrt.so(+0x2cdd80) [0x7fafbcfd80]
/lib/librknnrt.so(+0x2d7054) [0x7fafbd9054]
/lib/librknnrt.so(+0x2d8f4c) [0x7fafbdaf4c]
/lib/librknnrt.so(rknn_run+0x2d0) [0x7faf9bebe8]
rknn_server() [0x40e640]
rknn_server() [0x40d6e4]
rknn_server() [0x40dd24]
rknn_server() [0x406e48]
/lib/aarch64-linux-gnu/libc.so.6(__libc_start_main+0xe8) [0x7fb9400dd8]
rknn_server() [0x407378]

E RKNNAPI: rknn_run failed, runtime abort!
E inference: Traceback (most recent call last):
  File "rknn/api/rknn_log.py", line 344, in rknn.api.rknn_log.error_catch_decorator.error_catch_wrapper
  File "rknn/api/rknn_base.py", line 2682, in rknn.api.rknn_base.RKNNBase.inference
  File "rknn/api/rknn_runtime.py", line 482, in rknn.api.rknn_runtime.RKNNRuntime.run
Exception: rknn run failed. error code: RKNN_ERR_RUNTIME_ABORT

W inference: ===================== WARN(69) =====================
E rknn-toolkit2 version: 2.3.0
Traceback (most recent call last):
  File "rknn/api/rknn_log.py", line 344, in rknn.api.rknn_log.error_catch_decorator.error_catch_wrapper
  File "rknn/api/rknn_base.py", line 2682, in rknn.api.rknn_base.RKNNBase.inference
  File "rknn/api/rknn_runtime.py", line 482, in rknn.api.rknn_runtime.RKNNRuntime.run
Exception: rknn run failed. error code: RKNN_ERR_RUNTIME_ABORT

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/jerry/gra/2025.4.7/yolo_convert.py", line 110, in <module>
outputs = rknn.inference(inputs=[img, img_T],
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jerry/anaconda3/envs/rknn_23/lib/python3.12/site-packages/rknn/api/rknn.py", line 308, in inference
return self.rknn_base.inference(inputs=inputs, data_format=data_format,
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "rknn/api/rknn_log.py", line 349, in rknn.api.rknn_log.error_catch_decorator.error_catch_wrapper
  File "rknn/api/rknn_log.py", line 95, in rknn.api.rknn_log.RKNNLog.e
ValueError: Traceback (most recent call last):
  File "rknn/api/rknn_log.py", line 344, in rknn.api.rknn_log.error_catch_decorator.error_catch_wrapper
  File "rknn/api/rknn_base.py", line 2682, in rknn.api.rknn_base.RKNNBase.inference
  File "rknn/api/rknn_runtime.py", line 482, in rknn.api.rknn_runtime.RKNNRuntime.run
Exception: rknn run failed. error code: RKNN_ERR_RUNTIME_ABORT
”

只看该作者 · 发表于 2025-4-7 15:59:20

我使用连板调试以获得更多的报错信息定位问题，下面是ADB连板下的报错信息，不知道有无帮助
“E RKNNAPI: Catch runtime Crash! Stack backtrace:
rknn_server() [0x40c584]
rknn_server(_Z14signal_handleriP9siginfo_tPv+0xbc) [0x41967c]
linux-vdso.so.1(__kernel_rt_sigreturn+0) [0x7fb988d7d0]
/lib/librknnrt.so(+0x4fbc3c) [0x7fafdfdc3c]
/lib/librknnrt.so(+0x2cdd80) [0x7fafbcfd80]
/lib/librknnrt.so(+0x2d7054) [0x7fafbd9054]
/lib/librknnrt.so(+0x2d8f4c) [0x7fafbdaf4c]
/lib/librknnrt.so(rknn_run+0x2d0) [0x7faf9bebe8]
rknn_server() [0x40e640]
rknn_server() [0x40d6e4]
rknn_server() [0x40dd24]
rknn_server() [0x406e48]
/lib/aarch64-linux-gnu/libc.so.6(__libc_start_main+0xe8) [0x7fb9400dd8]
rknn_server() [0x407378]

E RKNNAPI: rknn_run failed, runtime abort!
E inference: Traceback (most recent call last):
  File "rknn/api/rknn_log.py", line 344, in rknn.api.rknn_log.error_catch_decorator.error_catch_wrapper
  File "rknn/api/rknn_base.py", line 2682, in rknn.api.rknn_base.RKNNBase.inference
  File "rknn/api/rknn_runtime.py", line 482, in rknn.api.rknn_runtime.RKNNRuntime.run
Exception: rknn run failed. error code: RKNN_ERR_RUNTIME_ABORT

W inference: ===================== WARN(69) =====================
E rknn-toolkit2 version: 2.3.0
Traceback (most recent call last):
  File "rknn/api/rknn_log.py", line 344, in rknn.api.rknn_log.error_catch_decorator.error_catch_wrapper
  File "rknn/api/rknn_base.py", line 2682, in rknn.api.rknn_base.RKNNBase.inference
  File "rknn/api/rknn_runtime.py", line 482, in rknn.api.rknn_runtime.RKNNRuntime.run
Exception: rknn run failed. error code: RKNN_ERR_RUNTIME_ABORT

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/jerry/gra/2025.4.7/yolo_convert.py", line 110, in <module>
outputs = rknn.inference(inputs=[img, img_T],
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jerry/anaconda3/envs/rknn_23/lib/python3.12/site-packages/rknn/api/rknn.py", line 308, in inference
return self.rknn_base.inference(inputs=inputs, data_format=data_format,
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "rknn/api/rknn_log.py", line 349, in rknn.api.rknn_log.error_catch_decorator.error_catch_wrapper
  File "rknn/api/rknn_log.py", line 95, in rknn.api.rknn_log.RKNNLog.e
ValueError: Traceback (most recent call last):
  File "rknn/api/rknn_log.py", line 344, in rknn.api.rknn_log.error_catch_decorator.error_catch_wrapper
  File "rknn/api/rknn_base.py", line 2682, in rknn.api.rknn_base.RKNNBase.inference
  File "rknn/api/rknn_runtime.py", line 482, in rknn.api.rknn_runtime.RKNNRuntime.run
Exception: rknn run failed. error code: RKNN_ERR_RUNTIME_ABORT
”

RK3588板上推理模型遇到段错误

本帖子中包含更多资源