tensorflow / tensorflow/tensorflow

Multiple segmentation faults and aborted in some modules

Open
#96,209 9 comments 0 reactions 1 assignee View on GitHub

@Kayyuri is already working on this.

Since Aug 13, 2026.

awaiting PR merge comp:ops TF 2.19 type:bug
Dominant language
C++
Stars
200k
Forks
76.9k
Avg merge
2d 3h
Merged PRs (30d)
433

Description

Issue type

Bug

Have you reproduced the bug with TensorFlow Nightly?

Yes

Source

binary

TensorFlow version

v1.12.1-127710-gd3d06fc4b99 2.20.0-dev20250623

Custom code

Yes

OS platform and distribution

Linux Fedora 42

Mobile device

Linux Fedora 42

Python version

3.10.16

Bazel version

No response

GCC/compiler version

No response

CUDA/cuDNN version

CPU

GPU model and memory

No response

Current behavior?

Hi,

For a research paper, we carried out a large-scale benchmark of Pynguin, an Automatic Unit Test Generation Tool for Python, to test its new feature that can find Python interpreter crashes. In this benchmark, we found more than 20 potential bugs in tensorflow, and we are making this issue to report them.

(Sorry for making only one issue, there were so many different causes, and I didn't want to spam by splitting it into 20 smaller issues.)

Standalone code to reproduce the issue

The different tests that crash are available here: https://github.com/BergLucas/tensorflow-replication/tree/pynguin-subprocess-paper

Impacted functions
  • DispatchServer (tensorflow.python.data.experimental.service.server_lib.DispatchServer)
    • Potential duplicate of:
      • #76720
      • #93391
  • PyOperation (tensorflow.python.client._pywrap_tf_session.PyOperation)
  • PyTensor (tensorflow.python.client._pywrap_tf_session.PyTensor)
  • __init__ (tensorflow.python.eager.monitoring.__init__)
  • _initialize_logical_devices (tensorflow.python.eager.context._initialize_logical_devices)
  • add_c_function (tensorflow.python.eager.context.add_c_function)
  • check_alive (tensorflow.python.eager.context.check_alive)
  • clear_error (tensorflow.python.eager.executor.clear_error)
  • delete_config_key_value (tensorflow.python.eager.context.delete_config_key_value)
  • ensure_initialized (tensorflow.python.eager.context.ensure_initialized)
  • get_api_def (tensorflow.python.framework.c_api_util.get_api_def)
  • get_memory_info (tensorflow.python.eager.context.get_memory_info)
  • get_registered_kernels_for_op (tensorflow.python.framework.kernels.get_registered_kernels_for_op)
  • is_async (tensorflow.python.eager.executor.is_async)
  • join (tensorflow.python.data.experimental.service.server_lib.join)
  • monitor (tensorflow.python.eager.profiler_client.monitor)
  • quick_execute (tensorflow.python.eager.execute.quick_execute)
  • reset_memory_stats (tensorflow.python.eager.context.reset_memory_stats)
  • set (tensorflow.python.eager.monitoring.set)
  • start_tracing (tensorflow.python.eager.profiler_client.start_tracing)
  • start (tensorflow.python.profiler.profiler_v2.start)
  • value (tensorflow.python.eager.monitoring.value)
  • variable_accessed (tensorflow.python.eager.tape.variable_accessed)
  • watched_variables (tensorflow.python.eager.record.watched_variables)
  • watched_variables (tensorflow.python.eager.tape.watched_variables)
  • xla_replica_id (tensorflow.compiler.tf2xla.ops.gen_xla_ops.xla_replica_id)

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.