Skip to main content

Loop with Unreachable Exit Condition ('Infinite Loop')

CVE-2021-37686

Medium
5.5/10

Summary

TensorFlow is an end-to-end open source platform for machine learning. In affected versions the strided slice implementation in TFLite has a logic bug which can allow an attacker to trigger an infinite loop. This arises from newly introduced support for [ellipsis in axis definition](https://github.com/tensorflow/tensorflow/blob/149562d49faa709ea80df1d99fc41d005b81082a/tensorflow/lite/kernels/strided_slice.cc#L103-L122). An attacker can craft a model such that `ellipsis_end_idx` is smaller than `i` (e.g., always negative). In this case, the inner loop does not increase `i` and the `continue` statement causes execution to skip over the preincrement at the end of the outer loop. We have patched the issue in GitHub commit dfa22b348b70bb89d6d6ec0ff53973bacb4f4695. TensorFlow prior to 2.3.4, 2.4.x prior to 2.4.3, 2.5.x prior to 2.5.1, 2.6.x prior to 2.6.0-rc2 are affected with this vulnerability

  • LOW
  • LOCAL
  • NONE
  • UNCHANGED
  • NONE
  • LOW
  • NONE
  • HIGH

CWE-835 - Loop with Unreachable Exit Condition

Loops with multiple exits and flags detract from the quality of an application. They tend to make control structures difficult to understand, and introduce the risk of non-termination and other structural problems. The vulnerability “loop with unreachable exit condition” enables attackers to exploit this flaw, leading to denial of service.

References

Advisory Timeline

  • Published