VYPR
← All advisories
High severityNVD Advisory· Published Feb 3, 2022· Updated May 5, 2025

Integer overflows in Tensorflow

CVE-2022-23567

Description

Integer overflows in TensorFlow's Sparse*Cwise* ops can cause denial of service via large allocations or CHECK-fails.

AI Insight

LLM-synthesized narrative grounded in this CVE's description and references.

Integer overflows in TensorFlow's Sparse*Cwise* ops can cause denial of service via large allocations or CHECK-fails.

Vulnerability

The implementations of Sparse*Cwise* ops in TensorFlow are vulnerable to integer overflows due to missing validation on input tensor shapes. This allows constructing a large TensorShape with user-provided dimensions, leading to large allocations or CHECK-fails. Affected versions: TensorFlow 2.5.0 to 2.5.2, 2.6.0 to 2.6.2, 2.7.0 to 2.7.0 (since 2.7.1 is fixed), and 2.8.0-rc0 to 2.8.0-rc1 (since 2.8.0 is fixed). [1][3][4]

Exploitation

An attacker can provide crafted input tensors to the Sparse*Cwise* ops, triggering integer overflow during shape computation. No special privileges are required; the attacker only needs to supply malicious input to a TensorFlow model or operation. The overflow leads to either an out-of-memory condition (OOM) or a CHECK failure, causing denial of service. [1]

Impact

Successful exploitation results in denial of service: either a large memory allocation exhausting resources (OOM) or an assertion failure that crashes the process. No code execution or data disclosure is indicated. The impact is limited to availability. [1]

Mitigation

The fix is included in TensorFlow 2.8.0, and cherry-picked to versions 2.7.1, 2.6.3, and 2.5.3. Users should upgrade to these patched versions. No workaround is mentioned. The vulnerability is not listed in CISA's KEV. [1][3][4]

References
  1. NVD - CVE-2022-23567
  2. advisory-database/vulns/tensorflow-cpu/PYSEC-2022-76.yaml at main · pypa/advisory-database
  3. advisory-database/vulns/tensorflow-gpu/PYSEC-2022-131.yaml at main · pypa/advisory-database

AI Insight generated on May 21, 2026. Synthesized from this CVE's description and the cited reference URLs; citations are validated against the source bundle.

Affected packages

Versions sourced from the GitHub Security Advisory.

PackageAffected versionsPatched versions
tensorflowPyPI
< 2.5.32.5.3
tensorflowPyPI
>= 2.6.0, < 2.6.32.6.3
tensorflowPyPI
>= 2.7.0, < 2.7.12.7.1
tensorflow-cpuPyPI
< 2.5.32.5.3
tensorflow-cpuPyPI
>= 2.6.0, < 2.6.32.6.3
tensorflow-cpuPyPI
>= 2.7.0, < 2.7.12.7.1
tensorflow-gpuPyPI
< 2.5.32.5.3
tensorflow-gpuPyPI
>= 2.6.0, < 2.6.32.6.3
tensorflow-gpuPyPI
>= 2.7.0, < 2.7.12.7.1

Affected products

5

Patches

2
e952a89b7026

Prevent overflow in sparse dense cwise ops.

https://github.com/tensorflow/tensorflowMihai MaruseacDec 10, 2021via ghsa
commit
1 file changed · +3 −1
  • tensorflow/core/kernels/sparse_dense_binary_op_shared.cc+3 −1 modified
    @@ -99,7 +99,9 @@ class SparseDenseBinaryOpShared : public OpKernel {
 
     const auto indices_mat = indices_t->matrix<int64_t>();
     const auto shape_vec = shape_t->vec<int64_t>();
-    const auto lhs_dims = BCast::FromShape(TensorShape(shape_vec));
+    TensorShape lhs_shape;
+    OP_REQUIRES_OK(ctx, TensorShape::BuildTensorShape(shape_vec, &lhs_shape));
+    const auto lhs_dims = BCast::FromShape(lhs_shape);
     const auto rhs_dims = BCast::FromShape(dense_t->shape());
     BCast b(lhs_dims, rhs_dims, false);  // false for keeping the same num dims.
1b54cadd1939

Add missing validation to sparse dense cwise ops.

https://github.com/tensorflow/tensorflowMihai MaruseacDec 10, 2021via ghsa
commit
1 file changed · +13 −0
  • tensorflow/core/kernels/sparse_dense_binary_op_shared.cc+13 −0 modified
    @@ -78,11 +78,24 @@ class SparseDenseBinaryOpShared : public OpKernel {
                     "but received shapes: ",
                     values_t->shape().DebugString(), " and ",
                     shape_t->shape().DebugString()));
+    OP_REQUIRES(
+        ctx, TensorShapeUtils::IsVector(shape_t->shape()),
+        errors::InvalidArgument("Input sp_shape must be a vector. Got: ",
+                                shape_t->shape().DebugString()));
     OP_REQUIRES(
         ctx, values_t->dim_size(0) == indices_t->dim_size(0),
         errors::InvalidArgument(
             "The first dimension of values and indices should match. (",
             values_t->dim_size(0), " vs. ", indices_t->dim_size(0), ")"));
+    OP_REQUIRES(
+        ctx, shape_t->shape().dim_size(0) == indices_t->shape().dim_size(1),
+        errors::InvalidArgument(
+            "Number of dimensions must match second dimension of indices. ",
+            "Got ", shape_t->shape().dim_size(0),
+            " dimensions, indices shape: ", indices_t->shape().DebugString()));
+    OP_REQUIRES(ctx, shape_t->NumElements() > 0,
+                errors::InvalidArgument(
+                    "The shape argument requires at least one element."));
 
     const auto indices_mat = indices_t->matrix<int64_t>();
     const auto shape_vec = shape_t->vec<int64_t>();

Vulnerability mechanics

Generated on May 9, 2026. Inputs: CWE entries + fix-commit diffs from this CVE's patches. Citations validated against bundle.

References

9

News mentions

0

No linked articles in our index yet.