Use __cpp_aligned_new instead of hand-rolling the implementation#11293
Use __cpp_aligned_new instead of hand-rolling the implementation#11293facebook-github-bot merged 1 commit intopytorch:mainfrom tamird:export-D75806635
__cpp_aligned_new instead of hand-rolling the implementation#11293Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/11293
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit f7fbffe with merge base bca2cf5 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
@pytorchbot label "release notes: none" |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
11 similar comments
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
posix_memalign on all non-CRT platforms__cpp_aligned_new instead of hand-rolling the implementation
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
swolchok
left a comment
There was a problem hiding this comment.
Do we know whether this actually works on Apple platforms? #10660 (the PR that added the previous implementation) cited some concerns about aligned_alloc.
I'm inclined to accept this on the grounds that we should presume that standard library functions that exist on a platform actually work until proven otherwise, but I'd like to let @lucylq chime in.
We have tests on apple platforms, so that should be sufficient to prove that it works. Yes? |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
1 similar comment
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
1 similar comment
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
2 similar comments
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
|
@swolchok this keeps failing to land with:
Yet this PR shows no merge conflicts. This has persisted through several rebases. Do you know what's going on? |
|
The problem seems to be: which is an undocumented field. Why is that used to determine whether it is mergeable or not? |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
1 similar comment
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
how are you updating the PR? if you sent it out internally, you need to update the internal diff and export that diff to github; I see force pushes which looks like maybe you are updating the PR through some separate mechanism like working with github directly. |
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
No, those force pushes are done by phabricator. It seems that PRs that have both internal and external changes are broken because the |
…11293) Summary: Pull Request resolved: #11293 The current feature detection seems to be incorrect, resulting in fallback to the manual implementation, which in turn returns untagged pointers under hwasan. Use new expressions instead to delegate the work to the C++ compiler. Reviewed By: lucylq Differential Revision: D75806635
|
This pull request was exported from Phabricator. Differential Revision: D75806635 |
The current feature detection seems to be incorrect, resulting in fallback to
the manual implementation, which in turn returns untagged pointers under
hwasan.
Use new expressions instead to delegate the work to the C++ compiler.
Differential Revision: D75806635