[PR #10601/f7cac7e6 backport][3.12] Reduce WebSocket buffer slicing overhead #10640
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
![patchback[bot]](https://avatars.githubusercontent.com/in/21488?s=60&v=4){kind=small}
<!-- Thank you for your contribution! --> ## What do these changes do? Use a `const unsigned char *` for the buffer (Cython will automatically extract is using `__Pyx_PyBytes_AsUString`) as its a lot faster than copying around `PyBytes` objects. We do need to be careful that all slices are bounded and we bound check everything to make sure we do not do an out of bounds read since Cython does not bounds check C strings. I checked that all accesses to `buf_cstr` are proceeded by a bounds check but it would be good to get another set of eyes on that to verify in the `self._state == READ_PAYLOAD` block that we will never try to read out of bounds. <img width="376" alt="Screenshot 2025-03-19 at 10 21 54 AM" src="https://github.com/user-attachments/assets/a340ffa2-f09b-4aff-a4f7-c487dae186c8" /> ## Are there changes in behavior for the user? performance improvement ## Is it a substantial burden for the maintainers to support this? no There is a small risk that someone could remove a bounds check in the future and create a memory safety issue, however in this case its likely we would already be trying to read data that wasn't there if we are missing the bounds checking so the pure python version would throw if we are testing properly. --------- Co-authored-by: Sam Bull <[email protected]> (cherry picked from commit f7cac7e)
bdraco
approved these changes
Mar 30, 2025
CodSpeed Performance ReportMerging #10640 will improve performances by 16.71%Comparing Summary
Benchmarks breakdown
|
Codecov ReportAll modified and coverable lines are covered by tests ✅
✅ All tests successful. No failed tests found. Additional details and impacted files@@ Coverage Diff @@
## 3.12 #10640 +/- ##
=======================================
Coverage 98.09% 98.09%
=======================================
Files 126 126
Lines 37531 37532 +1
Branches 2120 2120
=======================================
+ Hits 36817 36818 +1
Misses 545 545
Partials 169 169
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
bdraco
deleted the
patchback/backports/3.12/f7cac7e63f18691e4261af353e84f9073b16624a/pr-10601
branch
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This is a backport of PR #10601 as merged into master (f7cac7e).
What do these changes do?
Use a
const unsigned char *for the buffer (Cython will automatically extract is using__Pyx_PyBytes_AsUString) as its a lot faster than copying aroundPyBytesobjects. We do need to be careful that all slices are bounded and we bound check everything to make sure we do not do an out of bounds read since Cython does not bounds check C strings.I checked that all accesses to
buf_cstrare proceeded by a bounds check but it would be good to get another set of eyes on that to verify in theself._state == READ_PAYLOADblock that we will never try to read out of bounds.Are there changes in behavior for the user?
performance improvement
Is it a substantial burden for the maintainers to support this?
no
There is a small risk that someone could remove a bounds check in the future and create a memory safety issue, however in this case its likely we would already be trying to read data that wasn't there if we are missing the bounds checking so the pure python version would throw if we are testing properly.