Zaloguj się
Deutsch English Español Polski
Średnia

Out of bounds access in tensorflow-lite

  • Ujawniono: 25.09.2020
  • Aktualizacja: 11.10.2026 13:19

Impact

In TensorFlow Lite, saved models in the flatbuffer format use a double indexing scheme: a model has a set of subgraphs, each subgraph has a set of operators and each operator has a set of input/output tensors. The flatbuffer format uses indices for the tensors, indexing into an array of tensors that is owned by the subgraph. This results in a pattern of double array indexing when trying to get the data of each tensor: https://github.com/tensorflow/tensorflow/blob/0e68f4d3295eb0281a517c3662f6698992b7b2cf/tensorflow/lite/kernels/kernel_util.cc#L36

However, some operators can have some tensors be optional. To handle this scenario, the flatbuffer model uses a negative -1 value as index for these tensors: https://github.com/tensorflow/tensorflow/blob/0e68f4d3295eb0281a517c3662f6698992b7b2cf/tensorflow/lite/c/common.h#L82

This results in special casing during validation at model loading time: https://github.com/tensorflow/tensorflow/blob/0e68f4d3295eb0281a517c3662f6698992b7b2cf/tensorflow/lite/core/subgraph.cc#L566-L580

Unfortunately, this means that the -1 index is a valid tensor index for any operator, including those that don't expect optional inputs and including for output tensors. Thus, this allows writing and reading from outside the bounds of heap allocated arrays, although only at a specific offset from the start of these arrays.

This results in both read and write gadgets, albeit very limited in scope.

Patches

We have patched the issue in several commits (46d5b0852, 00302787b7, e11f5558, cd31fd0ce, 1970c21, and fff2c83). We will release patch releases for all versions between 1.15 and 2.3.

We recommend users to upgrade to TensorFlow 1.15.4, 2.0.3, 2.1.2, 2.2.1, or 2.3.1.

Workarounds

A potential workaround would be to add a custom Verifier to the model loading code to ensure that only operators which accept optional inputs use the -1 special value and only for the tensors that they expect to be optional. Since this allow-list type approach is erro-prone, we advise upgrading to the patched code.

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.

Czego dotyczy

Kliknij technologię, aby zobaczyć wszystkie dotyczące jej podatności.

PlatformaPaczkaPodatne wersjePoprawka w wersji
Python (PyPI) tensorflow < 1.15.4, >= 2.0.0, < 2.0.3, >= 2.1.0, < 2.1.2, >= 2.2.0, < 2.2.1, >= 2.3.0, < 2.3.1 1.15.4, 2.0.3, 2.1.2, 2.2.1, 2.3.1
Python (PyPI) tensorflow-cpu < 1.15.4, >= 2.0.0, < 2.0.3, >= 2.1.0, < 2.1.2, >= 2.2.0, < 2.2.1, >= 2.3.0, < 2.3.1 1.15.4, 2.0.3, 2.1.2, 2.2.1, 2.3.1
Python (PyPI) tensorflow-gpu < 1.15.4, >= 2.0.0, < 2.0.3, >= 2.1.0, < 2.1.2, >= 2.2.0, < 2.2.1, >= 2.3.0, < 2.3.1 1.15.4, 2.0.3, 2.1.2, 2.2.1, 2.3.1

Odnośniki

Ocena CVSS 4.8/10
Wektor CVSS
CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:L/A:N
Słabość (CWE)
CWE-125, CWE-787
Platforma
Python (PyPI)
Paczka
tensorflow, tensorflow-cpu, tensorflow-gpu
Ujawniono
25.09.2020
Aktualizacja
11.10.2026 13:19

Źródła i licencje

OSV / GitHub Advisory Database - PYSEC-2020-134 - PyPA Advisory Database; CC BY 4.0

OSV / GitHub Advisory Database - PYSEC-2020-291 - PyPA Advisory Database; CC BY 4.0

OSV / GitHub Advisory Database - PYSEC-2020-326 - PyPA Advisory Database; CC BY 4.0

OSV / GitHub Advisory Database - GHSA-cvpc-8phh-8f45 - CC BY 4.0

Historia podatności: tensorflow-gpu

Wszystkie opisane podatności tej paczki (Python (PyPI)), od najnowszej. Widać, czy to jednorazowa wpadka, czy paczka, której trzeba pilnować.

Wszystkie (427) w archiwum
Wszystkich wpisów
427
Pierwsza → ostatnia
04.2019 07.2024
Bez poprawki
0
Wykorzystywana w atakach
0

Rozkład istotności

  • Krytyczna 6
  • Wysoka 109
  • Średnia 207
  • Niska 105

Ujawnienia w latach

Istotność:
  1. Out of bounds access in tensorflow-lite ten wpis
    Średnia · 4.8 CVE-2020-15211 < 1.15.4, >= 2.0.0, < 2.0.3, >= 2.1.0, < 2.1.2, >= 2.2.0, < 2.2.1, >= 2.3.0, < 2.3.1 1.15.4, 2.0.3, 2.1.2, 2.2.1, 2.3.1
  2. TensorFlow has segfault in array_ops.upper_bound
    Wysoka · 7.5 CVE-2023-33976 < 2.12.1 2.12.1
  3. TensorFlow vulnerable to Out-of-Bounds Read in DynamicStitch
    Wysoka · 7.5 CVE-2023-25659 < 2.11.1 2.11.1
  4. TensorFlow vulnerable to seg fault in `tf.raw_ops.Print`
    Wysoka · 7.5 CVE-2023-25660 < 2.11.1 2.11.1
  5. TensorFlow vulnerable to integer overflow in EditDistance
    Wysoka · 7.5 CVE-2023-25662 < 2.11.1 2.11.1
  6. TensorFlow has Null Pointer Error in TensorArrayConcatV2
    Wysoka · 7.5 CVE-2023-25663 < 2.11.1 2.11.1
  7. TensorFlow has Null Pointer Error in SparseSparseMaximum
    Wysoka · 7.5 CVE-2023-25665 < 2.11.1 2.11.1
  8. TensorFlow has Floating Point Exception in AudioSpectrogram
    Wysoka · 7.5 CVE-2023-25666 < 2.11.1 2.11.1
  9. TensorFlow has a heap out-of-buffer read vulnerability in the QuantizeAndDequantize operation
    Krytyczna · 9.8 CVE-2023-25668 < 2.11.1 2.11.1
  10. TensorFlow has Floating Point Exception in AvgPoolGrad with XLA
    Wysoka · 7.5 CVE-2023-25669 < 2.11.1 2.11.1
  11. TensorFlow has Null Pointer Error in QuantizedMatMulWithBiasAndDequantize
    Wysoka · 7.5 CVE-2023-25670 < 2.11.1 2.11.1
  12. TensorFlow has Null Pointer Error in LookupTableImportV2
    Wysoka · 7.5 CVE-2023-25672 < 2.11.1 2.11.1
  13. TensorFlow has Floating Point Exception in TensorListSplit with XLA
    Wysoka · 7.5 CVE-2023-25673 < 2.11.1 2.11.1
  14. TensorFlow has Null Pointer Error in RandomShuffle with XLA enable
    Wysoka · 7.5 CVE-2023-25674 < 2.11.1 2.11.1
  15. TensorFlow has Segfault in Bincount with XLA
    Wysoka · 7.5 CVE-2023-25675 < 2.11.1 2.11.1
  16. TensorFlow has null dereference on ParallelConcat with XLA
    Wysoka · 7.5 CVE-2023-25676 < 2.11.1 2.11.1
  17. TensorFlow has double free in Fractional(Max/Avg)Pool
    Wysoka · 8.0 CVE-2023-25801 < 2.11.1 2.11.1
  18. TensorFlow has Floating Point Exception in TFLite in conv kernel
    Wysoka · 7.5 CVE-2023-27579 < 2.11.1 2.11.1
  19. TensorFlow vulnerable to Out-of-Bounds Read in GRUBlockCellGrad
    Wysoka · 7.5 CVE-2023-25658 < 2.11.1 2.11.1
  20. Out of bounds write in grappler in Tensorflow
    Wysoka · 7.1 CVE-2022-41902 < 2.8.4, >= 2.9.0, < 2.9.3, >= 2.10.0, < 2.10.1 2.8.4, 2.9.3, 2.10.1
  21. FractionalMaxPool and FractionalAVGPool heap out-of-bounds acess
    Wysoka · 7.1 CVE-2022-41900 < 2.8.4, >= 2.9.0, < 2.9.3, >= 2.10.0, < 2.10.1 2.8.4, 2.9.3, 2.10.1
  22. TensorFlow vulnerable to OOB write in `scatter_nd` in TF Lite
    Wysoka · 7.0 CVE-2022-35939 < 2.7.2, >= 2.8.0, < 2.8.1, >= 2.9.0, < 2.9.1 2.7.2, 2.8.1, 2.9.1
  23. TensorFlow vulnerable to OOB read in `Gather_nd` in TF Lite
    Wysoka · 7.0 CVE-2022-35937 < 2.7.2, >= 2.8.0, < 2.8.1, >= 2.9.0, < 2.9.1 2.7.2, 2.8.1, 2.9.1
  24. Code injection in `saved_model_cli` in TensorFlow
    Wysoka · 7.8 CVE-2022-29216 < 2.6.4, >= 2.7.0, < 2.7.2, >= 2.8.0, < 2.8.1 2.6.4, 2.7.2, 2.8.1
  25. Segfault and OOB write due to incomplete validation in `EditDistance` in TensorFlow
    Wysoka · 7.1 CVE-2022-29208 < 2.6.4, >= 2.7.0, < 2.7.2, >= 2.8.0, < 2.8.1 2.6.4, 2.7.2, 2.8.1
  26. NULL Pointer Dereference and Access of Uninitialized Pointer in TensorFlow
    Krytyczna · 9.3 ghsa-h6gw-r52c-724r < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  27. Improper Validation of Integrity Check Value in TensorFlow
    Wysoka · 7.0 ghsa-43q8-3fv7-pr5x < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  28. Out of bounds read in Tensorflow
    Wysoka · 8.8 CVE-2022-23594 >= 2.7.0, < 2.7.1 2.7.1
  29. Read and Write outside of bounds in TensorFlow
    Wysoka · 8.8 CVE-2022-23560 < 2.5.3, >= 2.6.0, < 2.6.3 2.5.3, 2.6.3
  30. Insecure temporary file in Tensorflow
    Wysoka · 7.1 CVE-2022-23563 < 2.5.3, >= 2.6.0, < 2.6.3 2.5.3, 2.6.3
  31. Out of bounds read in Tensorflow
    Wysoka · 8.1 CVE-2022-23592 >= 2.7.0, < 2.8.0 2.8.0
  32. Stack overflow in TensorFlow
    Wysoka · 7.5 CVE-2022-23591 < 2.5.3, >= 2.6.0, < 2.6.3 2.5.3, 2.6.3
  33. Out of bounds read and write in Tensorflow
    Wysoka · 8.8 CVE-2022-23574 < 2.5.3, >= 2.6.0, < 2.6.3 2.5.3, 2.6.3
  34. Uninitialized variable access in Tensorflow
    Wysoka · 7.6 CVE-2022-23573 < 2.5.3, >= 2.6.0, < 2.6.3 2.5.3, 2.6.3
  35. Integer overflow in Tensorflow
    Wysoka · 7.6 CVE-2022-23562 < 2.5.3, >= 2.6.0, < 2.6.3 2.5.3, 2.6.3
  36. Out of bounds write in TFLite
    Wysoka · 8.8 CVE-2022-23561 < 2.5.3, >= 2.6.0, < 2.6.3 2.5.3, 2.6.3
  37. Integer overflow in TFLite
    Wysoka · 8.8 CVE-2022-23559 < 2.5.3, >= 2.6.0, < 2.6.3 2.5.3, 2.6.3
  38. Out of bounds write in Tensorflow
    Wysoka · 8.8 CVE-2022-23566 < 2.5.3, >= 2.6.0, < 2.6.3 2.5.3, 2.6.3
  39. Use after free in `DecodePng` kernel
    Wysoka · 7.6 CVE-2022-23584 < 2.5.3, >= 2.6.0, < 2.6.3 2.5.3, 2.6.3
  40. Integer overflow in TFLite array creation
    Wysoka · 7.6 CVE-2022-23558 < 2.5.3, >= 2.6.0, < 2.6.3 2.5.3, 2.6.3
  41. Integer overflow in TensorFlow
    Wysoka · 8.8 CVE-2022-23587 < 2.5.3, >= 2.6.0, < 2.6.3 2.5.3, 2.6.3
  42. Heap overflow in Tensorflow
    Wysoka · 7.6 CVE-2022-21740 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  43. Division by zero in TFLite
    Wysoka · 6.5 CVE-2022-21741 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  44. Null pointer dereference in TensorFlow
    Wysoka · 6.5 CVE-2022-21739 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  45. Assertion failure based denial of service in Tensorflow
    Wysoka · 6.5 CVE-2022-21737 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  46. Integer overflow leading to crash in Tensorflow
    Wysoka · 6.5 CVE-2022-21738 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  47. `CHECK`-fails when building invalid tensor shapes in Tensorflow
    Wysoka · 6.5 CVE-2022-23569 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  48. Overflow and uncaught divide by zero in Tensorflow
    Wysoka · 6.5 CVE-2022-21729 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  49. `CHECK`-failures in Tensorflow
    Wysoka · 6.5 CVE-2022-21734 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  50. Division by zero in Tensorflow
    Wysoka · 6.5 CVE-2022-21735 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  51. Undefined behavior in `SparseTensorSliceDataset`
    Wysoka · 7.6 CVE-2022-21736 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  52. Type confusion leading to segfault in Tensorflow
    Wysoka · 6.5 CVE-2022-21731 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  53. Integer overflows in Tensorflow
    Wysoka · 6.5 CVE-2022-23567 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  54. Integer overflows in Tensorflow
    Wysoka · 6.5 CVE-2022-23568 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  55. Out of bounds read in Tensorflow
    Wysoka · 8.1 CVE-2022-21726 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  56. Out of bounds read in Tensorflow
    Wysoka · 8.1 CVE-2022-21728 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  57. Integer overflow in Tensorflow
    Wysoka · 7.6 CVE-2022-21727 < 2.5.3, >= 2.6.0, < 2.6.3 2.5.3, 2.6.3
  58. Out of bounds read in Tensorflow
    Wysoka · 8.1 CVE-2022-21730 < 2.5.3, >= 2.6.0, < 2.6.3, >= 2.7.0, < 2.7.1 2.5.3, 2.6.3, 2.7.1
  59. Access to invalid memory during shape inference in `Cudnn*` ops
    Wysoka · 7.8 CVE-2021-41221 >= 2.6.0, < 2.6.1, >= 2.5.0, < 2.5.2, < 2.4.4 2.6.1, 2.5.2, 2.4.4
  60. Use after free / memory leak in `CollectiveReduceV2`
    Wysoka · 7.8 CVE-2021-41220 >= 2.6.0, < 2.6.1 2.6.1
  61. Incomplete validation in boosted trees code
    Krytyczna · 9.3 CVE-2021-41208 >= 2.6.0, < 2.6.1, >= 2.5.0, < 2.5.2, < 2.4.4 2.6.1, 2.5.2, 2.4.4
Pokaż więcej (20) Widać 61 z 116

Dowiedz się, które podatności dotyczą Twojej aplikacji

Po podłączeniu integracji DockRay i włączeniu raportowania oprogramowania DockRay porównuje zgłoszone wersje komponentów ze znanymi podatnościami i powiadamia o potwierdzonych dopasowaniach. Zakres zależy od obsługiwanej technologii i dostępnych biuletynów.

Podłącz swoją aplikację
Newsletter DockRay

Bezpieczeństwo aplikacji w Twojej skrzynce

Najważniejsze podatności, konkretne zalecenia i praktyczne wskazówki dla Twoich aplikacji. Wybierz technologie i śledź to, co ma znaczenie dla Twojego projektu.

  • Codzienne zestawienie nowych podatności, także krytycznych.
  • Powiadomienia o ważnych aktualizacjach.
  • Praktyczne porady o bezpieczeństwie i monitoringu aplikacji.

Pozostaw puste, aby otrzymywać informacje o wszystkich technologiach.

Polityka prywatności

Zapis jest bezpłatny. Wypiszesz się jednym kliknięciem w wiadomości.