Advisory

GHSA-q263-fvxm-m5mw

Also known as: BIT-tensorflow-2020-26271, CVE-2020-26271, PYSEC-2020-257, PYSEC-2020-302, PYSEC-2020-337

Summary

Heap out of bounds access in MakeEdge in TensorFlow

Details

Impact
Under certain cases, loading a saved model can result in accessing uninitialized memory while building the computation graph. The `MakeEdge` function creates an edge between one output tensor of the src node (given by output_index) and the input slot of the dst node (given by input_index). This is only possible if the types of the tensors on both sides coincide, so the function begins by obtaining the corresponding DataType values and comparing these for equality:

  DataType src_out = src->output_type(output_index);
  DataType dst_in = dst->input_type(input_index);
  //...

However, there is no check that the indices point to inside of the arrays they index into. Thus, this can result in accessing data out of bounds of the corresponding heap allocated arrays.

In most scenarios, this can manifest as unitialized data access, but if the index points far away from the boundaries of the arrays this can be used to leak addresses from the library.

Patches
We have patched the issue in GitHub commit 0cc38aaa4064fd9e79101994ce9872c6d91f816b and will release TensorFlow 2.4.0 containing the patch. TensorFlow nightly packages after this commit will also have the issue resolved.

Since this issue also impacts TF versions before 2.4, we will patch all releases between 1.15 and 2.3 inclusive.

For more information
Please consult our security guide for more information regarding the security model and how to contact us with issues and questions.

Severity

Medium Basis: The source's own band: MODERATE

Dates

Published
2020-12-10 19:07 UTC
Modified
2026-07-08 06:00 UTC

Affected packages and fixes

15 affected packages; all PyPI · All systems
PackageAffectedFix
tensorflowEarlier versionspip install --upgrade tensorflow==1.15.5
tensorflowintroduced 2.0.0, fixed 2.0.4pip install --upgrade tensorflow==2.0.4
tensorflowintroduced 2.1.0, fixed 2.1.3pip install --upgrade tensorflow==2.1.3
tensorflowintroduced 2.2.0, fixed 2.2.2pip install --upgrade tensorflow==2.2.2
tensorflowintroduced 2.3.0, fixed 2.3.2pip install --upgrade tensorflow==2.3.2
tensorflow-cpuEarlier versionspip install --upgrade tensorflow-cpu==1.15.5
tensorflow-cpuintroduced 2.0.0, fixed 2.0.4pip install --upgrade tensorflow-cpu==2.0.4
tensorflow-cpuintroduced 2.1.0, fixed 2.1.3pip install --upgrade tensorflow-cpu==2.1.3
tensorflow-cpuintroduced 2.2.0, fixed 2.2.2pip install --upgrade tensorflow-cpu==2.2.2
tensorflow-cpuintroduced 2.3.0, fixed 2.3.2pip install --upgrade tensorflow-cpu==2.3.2
tensorflow-gpuEarlier versionspip install --upgrade tensorflow-gpu==1.15.5
tensorflow-gpuintroduced 2.0.0, fixed 2.0.4pip install --upgrade tensorflow-gpu==2.0.4
tensorflow-gpuintroduced 2.1.0, fixed 2.1.3pip install --upgrade tensorflow-gpu==2.1.3
tensorflow-gpuintroduced 2.2.0, fixed 2.2.2pip install --upgrade tensorflow-gpu==2.2.2
tensorflow-gpuintroduced 2.3.0, fixed 2.3.2pip install --upgrade tensorflow-gpu==2.3.2

Fix commands are the export's own generic commands. Ojuju never runs them. Check them against your setup first.

Check your own computer: guard packages check

View this record on OSV.dev

Source

Source: GitHub Advisory Database (reviewed advisories) · CC BY 4.0

Sources & licences