2.5

CVSS Score

3.1

Basic Information

CVE ID

CVE-2021-29590

GHSA ID

GHSA-24x6-8c7m-hv3f

EPSS Score

0.0259%

CWE

CWE-125

Published

5/21/2021

Updated

11/13/2024

KEV Status

Technology

Python

Concerned about an active attack path?

Talk to our security experts and see Miggo in action.

Miggo Vulnerability Database

→

CVE-2021-29590

CVE-2021-29590: Heap OOB read in TFLite's implementation of `Minimum` or `Maximum`

Impact

The implementations of the Minimum and Maximum TFLite operators can be used to read data outside of bounds of heap allocated objects, if any of the two input tensor arguments are empty.

This is because the broadcasting implementation indexes in both tensors with the same index but does not validate that the index is within bounds:

auto maxmin_func = [&](int indexes[N]) {
  output_data[SubscriptToIndex(output_desc, indexes)] =
    op(input1_data[SubscriptToIndex(desc1, indexes)],
        input2_data[SubscriptToIndex(desc2, indexes)]);
};

Patches

We have patched the issue in GitHub commit 953f28dca13c92839ba389c055587cfe6c723578.

The fix will be included in TensorFlow 2.5.0. We will also cherrypick this commit on TensorFlow 2.4.2, TensorFlow 2.3.3, TensorFlow 2.2.3 and TensorFlow 2.1.4, as these are also affected and still in supported range.

For more information

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

Attribution

This vulnerability has been reported by members of the Aivul Team from Qihoo 360.

(GitHub Advisory)

2.5

CVSS Score

3.1

Basic Information

CVE ID

CVE-2021-29590

GHSA ID

GHSA-24x6-8c7m-hv3f

EPSS Score

0.0259%

CWE

CWE-125

Published

5/21/2021

Updated

11/13/2024

KEV Status

Technology

Python

Technical Details

CVSS Vector

CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:N/I:N/A:L

Package Name	Ecosystem	Vulnerable Versions	First Patched Version
tensorflow	pip	< 2.1.4	2.1.4
tensorflow	pip	>= 2.2.0, < 2.2.3	2.2.3
tensorflow	pip	>= 2.3.0, < 2.3.3	2.3.3
tensorflow	pip	>= 2.4.0, < 2.4.2	2.4.2
tensorflow-cpu	pip	>= 0, < 2.1.4	2.1.4
tensorflow-cpu	pip	>= 2.2.0, < 2.2.3	2.2.3
tensorflow-cpu	pip	>= 2.3.0, < 2.3.3	2.3.3
tensorflow-cpu	pip	>= 2.4.0, < 2.4.2	2.4.2
tensorflow-gpu	pip	>= 0, < 2.1.4	2.1.4
tensorflow-gpu	pip	>= 2.2.0, < 2.2.3	2.2.3
tensorflow-gpu	pip	>= 2.3.0, < 2.3.3	2.3.3
tensorflow-gpu	pip	>= 2.4.0, < 2.4.2	2.4.2

Vulnerability Intelligence
Miggo AI

Root Cause Analysis

The vulnerability stems from the broadcasting implementation in MaximumMinimumBroadcastSlow, which handles element-wise operations for Minimum/Maximum operators. The lambda function directly accesses input tensors using potentially out-of-bounds indices when inputs are empty, as shown in the provided code snippet. The patch adds input validation at the operator entry point (Eval function in maximum_minimum.cc), but the core vulnerability exists in this reference implementation's indexing logic.

Vulnerable functions

Only Mi**o us*rs **n s** t*is s**tion

WAF Protection Rules

WAF Rule

### Imp**t T** impl*m*nt*tions o* t** `Minimum` *n* `M*ximum` T*Lit* op*r*tors **n ** us** to r*** **t* outsi** o* *oun*s o* ***p *llo**t** o*j**ts, i* *ny o* t** two input t*nsor *r*um*nts *r* *mpty. T*is is ****us* [t** *ro****stin* impl*m*nt*tion

Reasoning

T** vuln*r**ility st*ms *rom t** *ro****stin* impl*m*nt*tion in M*ximumMinimum*ro****stSlow, w*i** **n*l*s *l*m*nt-wis* op*r*tions *or Minimum/M*ximum op*r*tors. T** l*m*** `*un*tion` *ir**tly ****ss*s input t*nsors usin* pot*nti*lly out-o*-*oun*s in

2.5

Basic Information

Concerned about an active attack path?

CVE-2021-29590: Heap OOB read in TFLite's implementation of `Minimum` or `Maximum`

Impact

Patches

For more information

Attribution

2.5

Basic Information

Technical Details

Vulnerability IntelligenceMiggo AI

Root Cause Analysis

Vulnerable functions

WAF Protection Rules

WAF Rule

Reasoning

Vulnerability Intelligence
Miggo AI