search

google-coral / pycoral

Python API for ML inferencing and transfer-learning on Coral devices
https://coral.ai
Apache License 2.0
360 stars 149 forks source link

Failed to load delegate from libedgetpu.so.1 & Inappropriate ioctl for device with M.2 Accelerator A+E key #152

Open PinxuanHuang opened 1 month ago

PinxuanHuang commented 1 month ago

Description

I'm using the M.2 Accelerator A + E to do some PoC.

Currently I open the Kernel Config to build the necessary kernel module (coral tpu driver) and follow the official example to verify. I can see /dev/apex_0 after I insmod gaket & apex, but when I run the example python3 examples/classify_image.py --model test_data/mobilenet_v2_1.0_224_inat_bird_quant_edgetpu.tflite --labels test_data/inat_bird_labels.txt --input test_data/parrot.jpg in the container (debian:10) it shows this error and also Inappropriate ioctl for device error.

image

image

image

The error also appear in Frigate NVR. It's weird that it shows no TPU device was detected, but there's /dev/apex_0 and also libedgetpu.so.1.0 under /usr/lib/aarch-linux-gnu directory.

image

image

image

What do I need to do for fixing it?

image

image

CONFIG_STAGING=y
CONFIG_STAGING_GASKET_FRAMEWORK=m
CONFIG_STAGING_APEX_DRIVER=m

Thanks.

reference [1] https://coral.ai/docs/m2/get-started/#requirements

[2] https://coral.ai/docs/notes/build-coral/#required-components

[3] https://coral.googlesource.com/linux-imx/+/refs/heads/dkms/drivers/staging/gasket/

[4] https://github.com/google-coral/libedgetpu#building

Click to expand! ### Issue Type Bug ### Operating System Linux ### Coral Device M.2 Accelerator A+E ### Other Devices _No response_ ### Programming Language Python 3.7 ### Relevant Log Output ```shell python3 examples/classify_image.py --model test_data/mobilenet_v2_1.0_224_inat_bird_quant_edgetpu.tflite --labels test_data/inat_bird_labels.txt --input test_data/parrot.jpg I tflite/edgetpu_manager_direct.cc:453] No matching device is already opened for shared ownership. I driver/usb/local_usb_device.cc:944] EnumerateDevices: vendor:0x1a6e, product:0x89a I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[2] port[0] I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[1] port[2] I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[1] port[0] I driver/usb/local_usb_device.cc:944] EnumerateDevices: vendor:0x18d1, product:0x9302 I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[2] port[0] I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[1] port[2] I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[1] port[0] I tflite/edgetpu_context_direct.cc:106] USB always DFU: False (default) I tflite/edgetpu_context_direct.cc:128] USB bulk-in queue capacity: default I tflite/edgetpu_context_direct.cc:67] Performance expectation: Max (default) I ./driver/mmio/host_queue.h:266] Starting in normal mode I driver/kernel/kernel_registers.cc:83] Opening /dev/apex_0. read_only=0 I driver/kernel/kernel_registers.cc:97] mmap_offset=0x0000000000040000, mmap_size=4096 I tflite/edgetpu_context_direct.cc:401] Failed to open device [Apex (PCIe)] at [/dev/apex_0]: Internal: Could not mmap: Operation not permitted Traceback (most recent call last): File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 160, in load_delegate delegate = Delegate(library, options) File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 119, in __init__ raise ValueError(capture.message) ValueError During handling of the above exception, another exception occurred: Traceback (most recent call last): File "examples/classify_image.py", line 123, in main() File "examples/classify_image.py", line 73, in main interpreter = make_interpreter(*args.model.split('@')) File "/usr/lib/python3/dist-packages/pycoral/utils/edgetpu.py", line 87, in make_interpreter delegates = [load_edgetpu_delegate({'device': device} if device else {})] File "/usr/lib/python3/dist-packages/pycoral/utils/edgetpu.py", line 52, in load_edgetpu_delegate return tflite.load_delegate(_EDGETPU_SHARED_LIB, options or {}) File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 163, in load_delegate library, str(e))) ValueError: Failed to load delegate from libedgetpu.so.1 ```
mdvthu commented 4 weeks ago

So close to returning my Accelerator because of this bug: Raspberry Pi 5 with both Debian (Raspbian) and Ubuntu 24.04 LTS. Has anyone made any progress?