Segmentation fault (core dumped) in paddle._pir_ops.fft_c2c

[qiqicliff]

The following script lead to a segmentation fault.

import paddle
input_data = paddle.to_tensor([36028797018963968, 36028797018963968], dtype
    ='complex64')
result = paddle._pir_ops.fft_c2c(input_data, n=None, axis=-1, norm=
    'backward', name=None)

After running:

--------------------------------------
C++ Traceback (most recent call last):
--------------------------------------
0   paddle::pybind::eager_api_fft_c2c(_object*, _object*, _object*)
1   paddle::pybind::CastPyArg2Longs(_object*, std::string const&, long)

----------------------
Error Message Summary:
----------------------
FatalError: `Segmentation fault` is detected by the operating system.
  [TimeInfo: *** Aborted at 1739755405 (unix time) try "date -d @1739755405" if you are using GNU date ***]
  [SignalInfo: *** SIGSEGV (@0x8) received by PID 3405992 (TID 0x7fee4696c740) from PID 8 ***]

Segmentation fault (core dumped)

Versions

PaddlePaddle version: 2.6.1

[zhangyuqin1998]

Try to use result = paddle.fft.fft_c2c(input_data, n=None, axis=-1, norm='backward', forward=None, name=None)

Please note that the _pir_ops APIs are designed for static graph mode and should not be used in dynamic graph mode.