VYPR
← All advisories
Moderate severityNVD Advisory· Published Aug 12, 2021· Updated Aug 4, 2024

Use of unitialized value in TensorFlow Lite

CVE-2021-37682

Description

TensorFlow is an end-to-end open source platform for machine learning. In affected versions all TFLite operations that use quantization can be made to use unitialized values. For example. The issue stems from the fact that quantization.params is only valid if quantization.type is different that kTfLiteNoQuantization. However, these checks are missing in large parts of the code. We have patched the issue in GitHub commits 537bc7c723439b9194a358f64d871dd326c18887, 4a91f2069f7145aab6ba2d8cfe41be8a110c18a5 and 8933b8a21280696ab119b63263babdb54c298538. The fix will be included in TensorFlow 2.6.0. We will also cherrypick this commit on TensorFlow 2.5.1, TensorFlow 2.4.3, and TensorFlow 2.3.4, as these are also affected and still in supported range.

Affected packages

Versions sourced from the GitHub Security Advisory.

PackageAffected versionsPatched versions
tensorflowPyPI
< 2.3.42.3.4
tensorflowPyPI
>= 2.4.0, < 2.4.32.4.3
tensorflowPyPI
>= 2.5.0, < 2.5.12.5.1
tensorflow-cpuPyPI
< 2.3.42.3.4
tensorflow-cpuPyPI
>= 2.4.0, < 2.4.32.4.3
tensorflow-cpuPyPI
>= 2.5.0, < 2.5.12.5.1
tensorflow-gpuPyPI
< 2.3.42.3.4
tensorflow-gpuPyPI
>= 2.4.0, < 2.4.32.4.3
tensorflow-gpuPyPI
>= 2.5.0, < 2.5.12.5.1

Affected products

1

Patches

3
8933b8a21280

Fix a null pointer exception caused by branching on uninitialized data.

https://github.com/tensorflow/tensorflowMihai MaruseacJul 16, 2021via ghsa
commit
1 file changed · +3 −0
  • tensorflow/lite/kernels/depthwise_conv.cc+3 −0 modified
    @@ -176,6 +176,7 @@ TfLiteStatus Prepare(TfLiteContext* context, TfLiteNode* node) {
   if (data_type != kTfLiteFloat32) {
     TF_LITE_ENSURE_EQ(context, filter->quantization.type,
                       kTfLiteAffineQuantization);
+    TF_LITE_ENSURE(context, filter->quantization.type != kTfLiteNoQuantization);
     const auto* affine_quantization =
         reinterpret_cast<TfLiteAffineQuantization*>(
             filter->quantization.params);
@@ -195,6 +196,7 @@ TfLiteStatus Prepare(TfLiteContext* context, TfLiteNode* node) {
   }
 
   if (is_hybrid) {
+    TF_LITE_ENSURE(context, filter->quantization.type != kTfLiteNoQuantization);
     const auto* affine_quantization =
         reinterpret_cast<TfLiteAffineQuantization*>(
             filter->quantization.params);
@@ -495,6 +497,7 @@ TfLiteStatus EvalHybridPerChannel(TfLiteContext* context, TfLiteNode* node,
   op_params.weights_offset = 0;
   op_params.float_activation_min = output_activation_min;
   op_params.float_activation_max = output_activation_max;
+  TF_LITE_ENSURE(context, filter->quantization.type != kTfLiteNoQuantization);
   const auto* affine_quantization =
       reinterpret_cast<TfLiteAffineQuantization*>(filter->quantization.params);
   if (kernel_type == kReference) {
4a91f2069f71

Fix a null pointer exception caused by branching on uninitialized data.

https://github.com/tensorflow/tensorflowMihai MaruseacJul 16, 2021via ghsa
commit
1 file changed · +9 −0
  • tensorflow/lite/kernels/unidirectional_sequence_lstm.cc+9 −0 modified
    @@ -62,8 +62,12 @@ TfLiteStatus PopulateQuantizedLstmParams8x8_16(
       context,
       GetOutputSafe(context, node, lstm::full::kOutputTensor, &output_tensor));
 
+  TF_LITE_ENSURE(context,
+                 cell_state->quantization.type != kTfLiteNoQuantization);
   auto* cell_state_params =
       static_cast<TfLiteAffineQuantization*>(cell_state->quantization.params);
+  TF_LITE_ENSURE(context,
+                 output_tensor->quantization.type != kTfLiteNoQuantization);
   auto* proj_params = static_cast<TfLiteAffineQuantization*>(
       output_tensor->quantization.params);
   if (cell_clip > 0.0) {
@@ -160,6 +164,8 @@ TfLiteStatus PopulateQuantizedLstmParams8x8_16(
       TfLiteTensor* intermediate;
       TF_LITE_ENSURE_OK(context,
                         GetIntermediatesSafe(context, node, i, &intermediate));
+      TF_LITE_ENSURE(context,
+                     intermediate->quantization.type != kTfLiteNoQuantization);
       auto* params = static_cast<TfLiteAffineQuantization*>(
           intermediate->quantization.params);
       intermediate_scale.push_back(params->scale->data[0]);
@@ -174,6 +180,7 @@ TfLiteStatus PopulateQuantizedLstmParams8x8_16(
   // is ignored.
   TfLiteTensor* hidden;
   TF_LITE_ENSURE_OK(context, GetIntermediatesSafe(context, node, 4, &hidden));
+  TF_LITE_ENSURE(context, hidden->quantization.type != kTfLiteNoQuantization);
   auto* hidden_params =
       static_cast<TfLiteAffineQuantization*>(hidden->quantization.params);
   intermediate_scale.push_back(hidden_params->scale->data[0]);
@@ -760,6 +767,8 @@ TfLiteStatus PopulatePrecomputedZPTimesWeightsWithBias(TfLiteContext* context,
 
   const TfLiteTensor* intermediate =
       &context->tensors[node->intermediates->data[4]];
+  TF_LITE_ENSURE(context,
+                 intermediate->quantization.type != kTfLiteNoQuantization);
   const auto* params =
       static_cast<TfLiteAffineQuantization*>(intermediate->quantization.params);
   const int32_t hidden_zp = params->zero_point->data[0];
537bc7c72343

Fix a null pointer exception caused by branching on uninitialized data.

https://github.com/tensorflow/tensorflowMihai MaruseacJul 16, 2021via ghsa
commit
1 file changed · +7 −0
  • tensorflow/lite/kernels/svdf.cc+7 −0 modified
    @@ -256,14 +256,21 @@ TfLiteStatus Prepare(TfLiteContext* context, TfLiteNode* node) {
                                                      output_temp_size_array));
 
     // Calculate effective scales.
+    TF_LITE_ENSURE(context, input->quantization.type != kTfLiteNoQuantization);
     auto* input_params =
         reinterpret_cast<TfLiteAffineQuantization*>(input->quantization.params);
+    TF_LITE_ENSURE(context,
+                   weights_feature->quantization.type != kTfLiteNoQuantization);
     auto* weights_feature_params = reinterpret_cast<TfLiteAffineQuantization*>(
         weights_feature->quantization.params);
+    TF_LITE_ENSURE(context, state->quantization.type != kTfLiteNoQuantization);
     auto* state_params =
         reinterpret_cast<TfLiteAffineQuantization*>(state->quantization.params);
+    TF_LITE_ENSURE(context,
+                   weights_time->quantization.type != kTfLiteNoQuantization);
     auto* weight_time_params = reinterpret_cast<TfLiteAffineQuantization*>(
         weights_time->quantization.params);
+    TF_LITE_ENSURE(context, output->quantization.type != kTfLiteNoQuantization);
     auto* output_params = reinterpret_cast<TfLiteAffineQuantization*>(
         output->quantization.params);
     const double effective_scale_1 = input_params->scale->data[0] *

Vulnerability mechanics

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

References

10

News mentions

0

No linked articles in our index yet.