Use std::in_range for integer range checking by lakshayg · Pull Request #179400 · pytorch/pytorch

lakshayg · 2026-04-05T07:21:27Z

C++20 provides std::in_range which allows checking
if an integer value can fit in a given integer type. This PR
replaces such checks with in_range.

std::in_range is safer than direct comparisons
because it avoids implicit casts.

The call sites to update were obtained by grepping
for "numeric_limits.*min".

PyTorch supports C++20 now #176662

I have also proposed a clang-tidy checker to take
care of this automatically going forward llvm/llvm-project#190665

pytorch-bot · 2026-04-05T07:21:31Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/179400

📄 Preview Python docs built from this PR
📄 Preview C++ docs built from this PR
❓ Need help or want to give feedback on the CI? Visit the bot commands wiki

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures, 9 Unrelated Failures, 8 Unclassified Failures

As of commit 27907b0 with merge base b9a2479 ():

NEW FAILURES - The following jobs have failed:

trunk / linux-jammy-cuda13.0-py3.10-gcc11 / test-osdc (default, 3, 5, mt-l-x86aavx2-29-113-l4) (gh)
test/inductor/test_compiled_optimizers.py::CompiledOptimizerTests::test_bitwise_adam_capturable_foreach_cuda
windows-arm64-binary-wheel / wheel-py3_11-cpu-test (gh)
Process completed with exit code 1.

UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:

linux-binary-manywheel / manywheel-py3_10-xpu-build / build (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
linux-binary-manywheel / manywheel-py3_11-xpu-build / build (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
/__w/pytorch/pytorch/third_party/torch-xpu-ops/src/ATen/native/xpu/sycl/Indexing.cpp:978:30: error: ‘intel::reqd_sub_group_size’ scoped attribute directive ignored [-Werror=attributes]
linux-binary-manywheel / manywheel-py3_12-xpu-build / build (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
linux-binary-manywheel / manywheel-py3_13-xpu-build / build (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
linux-binary-manywheel / manywheel-py3_14-xpu-build / build (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
linux-binary-manywheel / manywheel-py3_14t-xpu-build / build (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
/__w/pytorch/pytorch/third_party/torch-xpu-ops/src/ATen/native/xpu/sycl/Indexing.cpp:978:30: error: ‘intel::reqd_sub_group_size’ scoped attribute directive ignored [-Werror=attributes]
linux-binary-manywheel / manywheel-py3_15-xpu-build / build (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
linux-binary-manywheel / manywheel-py3_15t-xpu-build / build (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)

FLAKY - The following job failed but was likely due to flakiness present on trunk:

trunk / linux-jammy-rocm-py3.10-mi355 / test (default, 4, 8, linux.rocm.gpu.gfx950.1) (gh) (disabled by #186868)
test_nn.py::TestNNDeviceTypeCUDA::test_LSTM_dropout_per_call_randomness_dropout_p_0_5_training_True_cuda

BROKEN TRUNK - The following jobs failed but was present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

trunk / win-vs2022-cpu-py3 / test (default, 1, 4, windows.4xlarge.nonephemeral) (gh) (trunk failure)
Process completed with exit code 2.
trunk / win-vs2022-cpu-py3 / test (default, 2, 4, windows.4xlarge.nonephemeral) (gh) (trunk failure)
Process completed with exit code 2.
trunk / win-vs2022-cpu-py3 / test (default, 3, 4, windows.4xlarge.nonephemeral) (gh) (trunk failure)
Process completed with exit code 2.
trunk / win-vs2022-cpu-py3 / test (default, 4, 4, windows.4xlarge.nonephemeral) (gh) (trunk failure)
Process completed with exit code 2.
trunk / win-vs2022-cpu-py3 / test (openreg, 1, 1, windows.4xlarge.nonephemeral) (gh) (trunk failure)
Process completed with exit code 2.

UNSTABLE - The following jobs are marked as unstable, possibly due to flakiness on trunk:

vllm-test / vllm-x-pytorch-test / test (vllm_basic_correctness_test, 1, 1, linux.g6.4xlarge.experimental.nvidia.gpu, unstable) (gh) (#178678)
RuntimeError: 1 pytest runs failed: ['pytest --reruns 2 --reruns-delay 10 -v -s basic_correctness/test_cumem.py']
vllm-test / vllm-x-pytorch-test / test (vllm_basic_models_test, 1, 1, linux.g6.4xlarge.experimental.nvidia.gpu, unstable) (gh) (#178678)
RuntimeError: 1 pytest runs failed: ['HF_DATASETS_OFFLINE=0 TRANSFORMERS_OFFLINE=0 pytest --reruns 2 --reruns-delay 10 -v -s models/test_initialization.py']
vllm-test / vllm-x-pytorch-test / test (vllm_entrypoints_test, 1, 1, linux.g6.4xlarge.experimental.nvidia.gpu, unstable) (gh) (#178678)
RuntimeError: 2 pytest runs failed: ['pytest --reruns 2 --reruns-delay 10 -v -s entrypoints/llm --ignore=entrypoints/llm/test_generate.py --ignore=entrypoints/llm/test_collective_rpc.py', 'pytest --reruns 2 --reruns-delay 10 -v -s entrypoints/offline_mode']

This comment was automatically generated by Dr. CI and updates every 15 minutes.

github-actions · 2026-04-05T07:25:33Z

Attention! PyTorch one of the C-stable API file was changed

You MUST NOT change existing function declarations in this, as this header defines a stable C ABI. If you need to change the signature for a function, introduce a new v2 version of the function and modify code generation to target the new version of the function.

Resolved

Skylion007

So the reason this is useful is removing that possibility of accidental implicit casts when reading the doc and it's therefore safer. You should update the PR description to reflect that.

Skylion007

Not sure about the stable API stuff

lakshayg · 2026-04-05T15:56:18Z

Yeah, I am not sure either. I believe @janeyx99 will be able to check if this makes sense or should be reverted.

lakshayg · 2026-04-28T03:40:31Z

@pytorchmergebot merge

pytorch-bot · 2026-04-28T03:40:35Z

This PR has pending changes requested. Please address the comments and update the PR before merging.

lakshayg · 2026-04-28T03:41:48Z

@Skylion007 can you please re-review? I have reverted the changes related to stable API

PR author addressed by reverting

janeyx99 · 2026-04-29T15:49:12Z

@pytorchbot merge

pytorchmergebot · 2026-04-29T15:52:05Z

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging

Check the merge workflow status
here

pytorchmergebot · 2026-05-18T07:51:43Z

@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here

pytorchmergebot · 2026-05-18T07:51:44Z

Rebase failed due to Command git -C /home/runner/work/pytorch/pytorch rebase refs/remotes/origin/viable/strict pull/179400/head returned non-zero exit code 1

Rebasing (1/1)
Auto-merging aten/src/ATen/native/Pool.h
CONFLICT (content): Merge conflict in aten/src/ATen/native/Pool.h
Auto-merging torch/csrc/stable/tensor_struct.h
Auto-merging torch/nativert/graph/Serialization.cpp
error: could not apply cf1a505a031... Use std::in_range for integer range checking
hint: Resolve all conflicts manually, mark them as resolved with
hint: "git add/rm <conflicted_files>", then run "git rebase --continue".
hint: You can instead skip this commit: run "git rebase --skip".
hint: To abort and get back to the state before "git rebase", run "git rebase --abort".
hint: Disable this message with "git config set advice.mergeConflict false"
Could not apply cf1a505a031... # Use std::in_range for integer range checking

Raised by https://github.com/pytorch/pytorch/actions/runs/26020550472

cyyever · 2026-05-20T01:36:04Z

@pytorchbot merge -i

pytorchmergebot · 2026-05-20T01:38:53Z

Merge started

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging

Check the merge workflow status
here

C++20 provides [`std::in_range`](https://en.cppreference.com/w/cpp/utility/in_range.html) which allows checking if an integer value can fit in a given integer type. This PR replaces such checks with in_range. std::in_range is safer than direct comparisons because it avoids implicit casts. The call sites to update were obtained by grepping for `"numeric_limits.*min"`. PyTorch supports C++20 now pytorch#176662 I have also proposed a clang-tidy checker to take care of this automatically going forward llvm/llvm-project#190665 Pull Request resolved: pytorch#179400 Approved by: https://github.com/janeyx99

atalman · 2026-05-21T13:01:32Z

@claude review if this PR caused following failures: https://github.com/pytorch/pytorch/actions/runs/26198707681/job/77106554779

 File "/pytorch/.ci/pytorch/smoke_test/check_binary_symbols.py", line 494, in <module>
    main()
  File "/pytorch/.ci/pytorch/smoke_test/check_binary_symbols.py", line 487, in main
    check_stable_api_symbols(install_root)
  File "/pytorch/.ci/pytorch/smoke_test/check_binary_symbols.py", line 293, in check_stable_api_symbols
    symbols_stable = _compile_and_extract_symbols(
  File "/pytorch/.ci/pytorch/smoke_test/check_binary_symbols.py", line 140, in _compile_and_extract_symbols
    raise RuntimeError(f"Compilation failed: {result.stderr}")
RuntimeError: Compilation failed: In file included from /opt/python/cp310-cp310/bin/../lib/python3.10/site-packages/torch/include/torch/headeronly/util/Exception.h:4,
                 from /opt/python/cp310-cp310/bin/../lib/python3.10/site-packages/torch/include/torch/csrc/inductor/aoti_torch/c/shim.h:6,
                 from /opt/python/cp310-cp310/bin/../lib/python3.10/site-packages/torch/include/torch/csrc/stable/accelerator.h:3,
                 from /tmp/tmptdpe3lcy/test.cpp:2:
/opt/python/cp310-cp310/bin/../lib/python3.10/site-packages/torch/include/torch/csrc/stable/tensor_struct.h: In member function ‘int8_t torch::stable::Tensor::get_device() const’:
/opt/python/cp310-cp310/bin/../lib/python3.10/site-packages/torch/include/torch/csrc/stable/tensor_struct.h:309:14: error: ‘in_range’ is not a member of ‘std’
  309 |         std::in_range<int8_t>(device_index),
      |              ^~~~~~~~
/opt/python/cp310-cp310/bin/../lib/python3.10/site-packages/torch/include/torch/headeronly/macros/Macros.h:202:64: note: in definition of macro ‘C10_UNLIKELY’
  202 | #define C10_UNLIKELY(expr) (__builtin_expect(static_cast<bool>(expr), 0))
      |                                                                ^~~~

claude · 2026-05-21T13:02:03Z

Claude finished @atalman's task in 49s —— View job

CI Failure Diagnosis for PR #179400

Read the failing file (torch/csrc/stable/tensor_struct.h)
Diagnose the root cause
Provide fix guidance

Root Cause

This is the same issue that caused the original revert. The fix was not applied before re-merging.

At torch/csrc/stable/tensor_struct.h:309:

std::in_range<int8_t>(device_index),

std::in_range is a C++20 feature (<utility>). The stable API headers in torch/csrc/stable/ are consumed by external projects (like vllm) that compile with C++17 (-std=gnu++17). The check_stable_api_symbols smoke test validates this by compiling stable headers with C++17, and it correctly catches the failure:

error: 'in_range' is not a member of 'std'

Fix

Revert line 309 in torch/csrc/stable/tensor_struct.h back to the C++17-compatible bounds check:

STD_TORCH_CHECK(
    device_index >= std::numeric_limits<int8_t>::min() &&
        device_index <= std::numeric_limits<int8_t>::max(),
    "Device index is out of range of return type int8_t, please use get_device_index() instead.");

The rest of the std::in_range changes (in non-stable headers/source files) are fine since PyTorch itself compiles with C++20. Only torch/csrc/stable/ must remain C++17-compatible because external consumers compile these headers with older standards.

TL;DR: The torch/csrc/stable/ directory must not use C++20 features. This file needs to be excluded from the std::in_range refactoring.

atalman · 2026-05-21T13:42:29Z

@pytorchmergebot revert -c nosignal -m "Broke nightly builds. Please test with ciflow/binaries when remerging"

pytorchmergebot · 2026-05-21T13:44:54Z

@pytorchbot successfully started a revert job. Check the current status here.
Questions? Feedback? Please reach out to the PyTorch DevX Team

pytorchmergebot · 2026-05-21T13:45:00Z

@lakshayg your PR has been successfully reverted.

This reverts commit d7e3761. Reverted #179400 on behalf of https://github.com/atalman due to Broke nightly builds. Please test with ciflow/binaries when remerging ([comment](#179400 (comment)))

lakshayg · 2026-05-22T16:32:38Z

@pytorchmergebot rebase

pytorchmergebot · 2026-05-22T16:34:40Z

@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here

pytorchmergebot · 2026-05-22T16:34:41Z

Rebase failed due to

Aborting rebase because rebasing the branch resulted in the same sha as the target branch.
This usually happens because the PR has already been merged.  Please rebase locally and push.

Raised by https://github.com/pytorch/pytorch/actions/runs/26299810141

lakshayg self-assigned this Apr 5, 2026

lakshayg added the ciflow/trunk Trigger trunk jobs on your pull request label Apr 5, 2026

lakshayg requested a review from janeyx99 as a code owner April 5, 2026 07:21

lakshayg added the topic: not user facing topic category label Apr 5, 2026

lakshayg requested a review from mikaylagawarecki as a code owner April 5, 2026 07:21

pytorchbot added the open source label Apr 5, 2026

Skylion007 reviewed Apr 5, 2026

View reviewed changes

Comment thread torch/csrc/stable/c/shim_function_versions.txt

Skylion007 approved these changes Apr 5, 2026

View reviewed changes

Skylion007 previously requested changes Apr 5, 2026

View reviewed changes

Skylion007 reviewed Apr 5, 2026

View reviewed changes

Comment thread torch/csrc/shim_common.cpp Outdated

lakshayg force-pushed the cpp20-in-range branch 2 times, most recently from 96c22f6 to 81764b7 Compare April 7, 2026 07:52

liangel-02 added the triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module label Apr 7, 2026

lakshayg force-pushed the cpp20-in-range branch 3 times, most recently from 1b511df to 76b2297 Compare April 13, 2026 22:10

lakshayg force-pushed the cpp20-in-range branch from 76b2297 to cf1a505 Compare April 27, 2026 19:57

janeyx99 approved these changes Apr 29, 2026

View reviewed changes

janeyx99 requested a review from Skylion007 April 29, 2026 15:48

pytorchmergebot added the merging label Apr 29, 2026

lakshayg force-pushed the cpp20-in-range branch from 5866245 to c516309 Compare May 19, 2026 17:37

pytorchmergebot added the merging label May 20, 2026

pytorchmergebot closed this in d7e3761 May 20, 2026

pytorchmergebot removed the merging label May 20, 2026

lakshayg mentioned this pull request May 20, 2026

C++20 Rollout Tracking #176662

Open

atalman added the ciflow/binaries Trigger all binary build and upload jobs on the PR label May 21, 2026

pytorchmergebot reopened this May 21, 2026

atalman mentioned this pull request May 21, 2026

Linux py 3.15 wheel builds #184600

Closed

1 task

lakshayg force-pushed the cpp20-in-range branch 2 times, most recently from 5a92e3d to dba62d8 Compare May 26, 2026 17:47

lakshayg force-pushed the cpp20-in-range branch from dba62d8 to cc4c3ab Compare June 9, 2026 00:02

lakshayg added 2 commits June 9, 2026 15:25

Use std::in_range for integer range checking

030178c

Revert changes to public headers

27907b0

lakshayg force-pushed the cpp20-in-range branch from cc4c3ab to 27907b0 Compare June 9, 2026 22:25

Conversation

lakshayg commented Apr 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pytorch-bot Bot commented Apr 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/179400

❌ 2 New Failures, 9 Unrelated Failures, 8 Unclassified Failures

Uh oh!

github-actions Bot commented Apr 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Attention! PyTorch one of the C-stable API file was changed

Uh oh!

Uh oh!

Skylion007 left a comment

Choose a reason for hiding this comment

Uh oh!

Skylion007 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

lakshayg commented Apr 5, 2026

Uh oh!

lakshayg commented Apr 28, 2026

Uh oh!

pytorch-bot Bot commented Apr 28, 2026

Uh oh!

lakshayg commented Apr 28, 2026

Uh oh!

janeyx99 commented Apr 29, 2026

Uh oh!

pytorchmergebot commented Apr 29, 2026

Merge started

Uh oh!

pytorchmergebot commented May 18, 2026

Uh oh!

pytorchmergebot commented May 18, 2026

Uh oh!

cyyever commented May 20, 2026

Uh oh!

pytorchmergebot commented May 20, 2026

Merge started

Uh oh!

atalman commented May 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

claude Bot commented May 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CI Failure Diagnosis for PR #179400

Root Cause

Fix

Uh oh!

atalman commented May 21, 2026

Uh oh!

pytorchmergebot commented May 21, 2026

Uh oh!

pytorchmergebot commented May 21, 2026

Uh oh!

lakshayg commented May 22, 2026

Uh oh!

pytorchmergebot commented May 22, 2026

Uh oh!

pytorchmergebot commented May 22, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

8 participants

lakshayg commented Apr 5, 2026 •

edited

Loading

pytorch-bot Bot commented Apr 5, 2026 •

edited

Loading

github-actions Bot commented Apr 5, 2026 •

edited

Loading

atalman commented May 21, 2026 •

edited

Loading

claude Bot commented May 21, 2026 •

edited

Loading