PyTorch Memory Corruption Vulnerability in RNN Utility Functions

A critical memory corruption vulnerability has been identified in PyTorch version 2.6.0. The issue arises in the function 'torch.nn.utils.rnn.unpack_sequence', where calling the function on an empty 'PackedSequence' leads to a segmentation fault, indicating an invalid memory access. This vulnerability requires local exploitation.

Impact

Exploitation of this vulnerability causes a segmentation fault, leading to a hard crash of the program.

Reproduction

The vulnerability can be reproduced by importing the 'torch' library and the 'PackedSequence' and 'unpack_sequence' functions from 'torch.nn.utils.rnn'. After creating an empty tensor and an empty 'PackedSequence' with no batch sizes, 'unpack_sequence' is called on the empty 'PackedSequence', which results in a segmentation fault.