Bug on multiprocessing.shared_memory.ShareableList

[JunLeecus]

Hello, I meet some bug when I use multiprocessing.shared_memory.ShareableList.
I want to save image bytes in a ShareableList, but some bytes are dropped. My codes are simple,

import numpy as np
fake_data = np.empty([720, 1280], dtype=np.uint8)
shm_list =  ShareableList([fake_data.tobytes()])
...
def process(image_bytes: bytes):
        data: np.ndarray[np.uint8] = np.frombuffer(image_bytes, np.uint8)
        image = data.reshape((720, 1280))
        shm_list[0] = image_bytes
        print(np.frombuffer(list(shm_list)[0], np.uint8).shape)
        print(np.frombuffer(image_bytes, np.uint8).shape)
...
shm_list.shm.close()
shm_list.shm.unlink()

this with the output

it seems some bytes are dropped.

My env

Description:    Ubuntu 20.04.4 LTS
python version : Python 3.8.10

This bug can be reproduce on our 2 Thinkpad X1 PC wirh arch x86_64, but normal on aarch64 device.

Thanks for help!

[GarvitMehta44]

shm_link.shm.print()

[JunLeecus]

ShareableList.shm is a SharedMemory instance without print method in python3.8.10

[JunLeecus]

I findout some of the reason. Actually, my image is an inflect of c++ Mat by pybind without copy to python. When I copy the image in python by np.array(image), things will be right. But I still want to know if there is any RLE in SharedMemory?