CVE-2021-29591— Stack overflow due to looping TFLite subgraph
Quick assessment
Affected
tensorflow tensorflow
Exploitation
No confirmed in-the-wild exploitation; assess based on exposure
Recommended action
Check the vendor advisory and references for a fixed version. If immediate upgrade is impossible, restrict exposure and increase monitoring.
Google TensorFlow是美国谷歌（Google）公司的一套用于机器学习的端到端开源平台。 TensorFlow 2.4.2,2.3.3,2.2.3，2.1.4存在安全漏洞，该漏洞源于TFlite图的节点之间不能有循环。
Possible ATT&CK Techniques 1 AI
T1068 · Exploitation for Privilege Escalation
Get alerts for future matching vulnerabilities
Log in to subscribe
I. Basic Information for CVE-2021-29591
Vulnerability Information
Have questions about the vulnerability? See if Shenlong's analysis helps!
View Shenlong Deep Dive ↗
Although we use advanced large model technology, its output may still contain inaccurate or outdated information.Shenlong tries to ensure data accuracy, but please verify and judge based on the actual situation.
Vulnerability Title
Stack overflow due to looping TFLite subgraph
Source: CVE Program / CVE List V5
Vulnerability Description
TensorFlow is an end-to-end open source platform for machine learning. TFlite graphs must not have loops between nodes. However, this condition was not checked and an attacker could craft models that would result in infinite loop during evaluation. In certain cases, the infinite loop would be replaced by stack overflow due to too many recursive calls. For example, the `While` implementation(https://github.com/tensorflow/tensorflow/blob/106d8f4fb89335a2c52d7c895b7a7485465ca8d9/tensorflow/lite/kernels/while.cc) could be tricked into a scneario where both the body and the loop subgraphs are the same. Evaluating one of the subgraphs means calling the `Eval` function for the other and this quickly exhaust all stack space. 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. Please consult our security guide(https://github.com/tensorflow/tensorflow/blob/master/SECURITY.md) for more information regarding the security model and how to contact us with issues and questions.
Source: CVE Program / CVE List V5
CVSS Information
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:L/A:H
Source: CVE Program / CVE List V5
Vulnerability Type
不可达退出条件的循环(无限循环)
Source: CVE Program / CVE List V5
Vulnerability Title
Google TensorFlow 安全漏洞
Source: CNNVD (China National Vulnerability Database)
Vulnerability Description
Google TensorFlow是美国谷歌（Google）公司的一套用于机器学习的端到端开源平台。 TensorFlow 2.4.2,2.3.3,2.2.3，2.1.4存在安全漏洞，该漏洞源于TFlite图的节点之间不能有循环。
Source: CNNVD (China National Vulnerability Database)
CVSS Information
N/A
Source: CNNVD (China National Vulnerability Database)
Vulnerability Type
N/A
Source: CNNVD (China National Vulnerability Database)
Affected Products
| Vendor | Product | Affected Versions | CPE | Subscribe |
|---|---|---|---|---|
| tensorflow | tensorflow | < 2.1.4 | - |
|
II. Public POCs for CVE-2021-29591
| # | POC Description | Source Link | Shenlong Link |
|---|
AI-Generated POC
Premium
No public POC found.
Login to generate AI POCIII. Intelligence Information for CVE-2021-29591
请登录查看更多情报信息。
Patches & Fixes for CVE-2021-29591 (2)
Vendor Advisories for CVE-2021-29591 (1)
-
https://github.com/tensorflow/tensorflow/security/advisories/GHSA-cwv3-863g-39vx x_refsource_CONFIRM
Same Patch Batch · tensorflow · 2021-05-14 · 108 CVEs total
| CVE-2021-29605 | 7.1 HIGH | Integer overflow in TFLite memory allocation |
| CVE-2021-29606 | 7.1 HIGH | Heap OOB read in TFLite |
| CVE-2021-29614 | 7.1 HIGH | Interpreter crash from `tf.io.decode_raw` |
| CVE-2021-29601 | 6.3 MEDIUM | Integer overflow in TFLite concatentation |
| CVE-2021-29613 | 6.3 MEDIUM | Incomplete validation in `tf.raw_ops.CTCLoss` |
| CVE-2021-29607 | 5.3 MEDIUM | Incomplete validation in `SparseSparseMinimum` |
| CVE-2021-29608 | 5.3 MEDIUM | Heap OOB and null pointer dereference in `RaggedTensorToTensor` |
| CVE-2021-29609 | 5.3 MEDIUM | Incomplete validation in `SparseAdd` |
| CVE-2021-29571 | 4.5 MEDIUM | Memory corruption in `DrawBoundingBoxesV2` |
| CVE-2021-29592 | 4.4 MEDIUM | Null pointer dereference in TFLite's `Reshape` operator |
| CVE-2021-29610 | 3.6 LOW | Invalid validation in `QuantizeAndDequantizeV2` |
| CVE-2021-29611 | 3.6 LOW | Incomplete validation in `SparseReshape` |
| CVE-2021-29612 | 3.6 LOW | Heap buffer overflow in `BandedTriangularSolve` |
| CVE-2021-29526 | 2.5 LOW | Division by 0 in `Conv2D` |
| CVE-2021-29527 | 2.5 LOW | Division by 0 in `QuantizedConv2D` |
| CVE-2021-29525 | 2.5 LOW | Division by 0 in `Conv2DBackpropInput` |
| CVE-2021-29584 | 2.5 LOW | CHECK-fail due to integer overflow |
| CVE-2021-29583 | 2.5 LOW | Heap buffer overflow and undefined behavior in `FusedBatchNorm` |
| CVE-2021-29574 | 2.5 LOW | Undefined behavior in `MaxPool3DGradGrad` |
| CVE-2021-29573 | 2.5 LOW | Division by 0 in `MaxPoolGradWithArgmax` |
Showing top 20 of 108 CVEs. View all on vendor page → →
IV. Related Vulnerabilities
Same product: tensorflow
Same vendor: tensorflow
-
CVE-2026-2492
TensorFlow HDF5 Library Uncontrolled Search Path Element Loc -
CVE-2023-33976
TensorFlow segfault in array_ops.upper_bound -
CVE-2024-3660
Arbitrary code injection vulnerability in Keras framework < -
CVE-2023-25661
Denial of Service in TensorFlow -
CVE-2023-25660
TensorFlow vulnerable to seg fault in `tf.raw_ops.Print`
Same weakness: CWE-835
-
CVE-2026-87082
Net::IDN::Punycode versions before 2.590 for Perl hang, cras -
CVE-2026-82560
Pod::Text versions before 6.1.1 for Perl allow CPU and memor -
CVE-2026-61633
NanoMQ: Infinite Loop in UNSUBSCRIBE Decoder Leading to Remo -
CVE-2026-84446
libheif: Sequence decode timing-table initialization allows -
CVE-2026-93690
uri-js through 4.4.1 Denial of Service via removeDotSegments
V. Comments for CVE-2021-29591
No comments yet