http: implement slab allocation for HTTP header parsing #61375

mertcanaltin · 2026-01-13T20:01:58Z

fix: // TODO(bnoordhuis) Use slab allocation, O(n) allocs is bad.

before:

http/bench-parser-fragmented.js n=100000 len=8 frags=2: 951,629.0928924936
http/bench-parser-fragmented.js n=100000 len=16 frags=2: 677,295.1203426437
http/bench-parser-fragmented.js n=100000 len=8 frags=4: 759,881.7857984888
http/bench-parser-fragmented.js n=100000 len=16 frags=4: 509,278.74034996366
http/bench-parser-fragmented.js n=100000 len=8 frags=8: 680,706.0381048893
http/bench-parser-fragmented.js n=100000 len=16 frags=8: 466,898.1840452692

after:
52% faster HTTP header parsing

http/bench-parser-fragmented.js n=100000 len=8 frags=2: 1,231,534.676937666
http/bench-parser-fragmented.js n=100000 len=16 frags=2: 890,580.6675901335
http/bench-parser-fragmented.js n=100000 len=8 frags=4: 1,117,723.8010822271
http/bench-parser-fragmented.js n=100000 len=16 frags=4: 789,398.1180369954
http/bench-parser-fragmented.js n=100000 len=8 frags=8: 869,642.0937523921
http/bench-parser-fragmented.js n=100000 len=16 frags=8: 711,293.8345939534

nodejs-github-bot · 2026-01-13T20:02:04Z

Review requested:

@nodejs/http
@nodejs/net

Copilot

Pull request overview

This pull request implements slab allocation for HTTP header parsing to optimize performance by reducing heap allocations. The implementation addresses a TODO comment about O(n) allocations being inefficient and achieves a reported 52% performance improvement in HTTP header parsing benchmarks.

Changes:

Added a 128-byte fixed-size buffer (slab) to each StringPtr instance for caching small strings
Modified Save() and Update() methods to use slab allocation before falling back to heap allocation
Added using_slab_ flag to track slab usage state alongside existing on_heap_ flag

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

codecov · 2026-01-13T21:19:02Z

Codecov Report

❌ Patch coverage is 87.71930% with 7 lines in your changes missing coverage. Please review.
✅ Project coverage is 88.51%. Comparing base (daeafc0) to head (56393d3).
⚠️ Report is 12 commits behind head on main.

Files with missing lines	Patch %	Lines
src/node_http_parser.cc	87.71%	1 Missing and 6 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #61375      +/-   ##
==========================================
- Coverage   88.51%   88.51%   -0.01%     
==========================================
  Files         704      704              
  Lines      208776   208834      +58     
  Branches    40301    40313      +12     
==========================================
+ Hits       184803   184847      +44     
- Misses      15966    15969       +3     
- Partials     8007     8018      +11

Files with missing lines	Coverage Δ
src/node_http_parser.cc	`84.87% <87.71%> (+0.07%)`	⬆️

... and 29 files with indirect coverage changes

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

Copilot

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 1 comment.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-01-13T21:22:37Z

src/node_http_parser.cc

-    Reset();
-  }
-
+  // Memory impact: ~8KB per parser (66 StringPtr × 128 bytes).


The memory calculation is slightly inaccurate. 66 × 128 = 8,448 bytes, which is approximately 8.25KB, not 8KB. Consider updating to '~8.4KB' or '~8.5KB' for better accuracy.

Suggested change

// Memory impact: ~8KB per parser (66 StringPtr × 128 bytes).

// Memory impact: ~8.4KB per parser (66 StringPtr × 128 bytes).

Thank you overly pedantic AI code review. The ~8KB is accurate enough here.

Yes, I agree.

addaleax · 2026-01-15T16:42:18Z

src/node_http_parser.cc

+  bool on_heap_ = false;
+  bool using_slab_ = false;
+  size_t size_ = 0;
+  char slab_[kSlabSize];


This overall seems like we're re-inventing MaybeStackBuffer here – I think we could just re-use that?

Thanks! , I've refactored this to use MaybeStackBuffer as the backing store for a allocator.

new benchmark result :

asis (main):

./out/Release/node benchmark/http/bench-parser.js

http/bench-parser.js n=100000 len=4: 1,728,678.1850683796 http/bench-parser.js n=100000 len=8: 1,395,311.5272280464 http/bench-parser.js n=100000 len=16: 981,492.7277523204 http/bench-parser.js n=100000 len=32: 641,907.2347759664

./out/Release/node benchmark/http/bench-parser-fragmented.js

http/bench-parser-fragmented.js n=100000 frags=2 len=8: 1,092,175.0170709684 http/bench-parser-fragmented.js n=100000 frags=4 len=8: 883,046.0674095292 http/bench-parser-fragmented.js n=100000 frags=8 len=8: 772,164.5013385202 http/bench-parser-fragmented.js n=100000 frags=2 len=16: 766,964.5365185371 http/bench-parser-fragmented.js n=100000 frags=4 len=16: 640,379.4448008832 http/bench-parser-fragmented.js n=100000 frags=8 len=16: 520,572.0305757982

new:

./out/Release/node benchmark/http/bench-parser.js

http/bench-parser.js n=100000 len=4: 1,764,556.6527588428 http/bench-parser.js n=100000 len=8: 1,500,760.6980788389 http/bench-parser.js n=100000 len=16: 1,095,013.3922327897 http/bench-parser.js n=100000 len=32: 646,285.1635436564

./out/Release/node benchmark/http/bench-parser-fragmented.js

http/bench-parser-fragmented.js n=100000 frags=2 len=8: 1,361,774.7618279774 http/bench-parser-fragmented.js n=100000 frags=4 len=8: 1,263,148.5873261988 http/bench-parser-fragmented.js n=100000 frags=8 len=8: 1,056,199.0301452405 http/bench-parser-fragmented.js n=100000 frags=2 len=16: 1,021,757.915898309 http/bench-parser-fragmented.js n=100000 frags=4 len=16: 923,235.2934387976 http/bench-parser-fragmented.js n=100000 frags=8 len=16: 826,609.3826364499

addaleax · 2026-01-15T16:44:28Z

src/node_http_parser.cc

      str_ = str;
-    } else if (on_heap_ || str_ + size_ != str) {
-      // Non-consecutive input, make a copy on the heap.
-      // TODO(bnoordhuis) Use slab allocation, O(n) allocs is bad.


I'm wondering if this was inteded to refer to something broader than what this PR does; I assume genuinely efficient allocation would involve allocating slab memory for a single HTTP message instead of for each StringPtr

right! I tried both ideas, using MaybeStackBuffer and a single shared 8KB allocator for all strings: 56393d3

benchmark results show +30-46% improvement on fragmented input, with no regressions on normal cases.

mertcanaltin · 2026-01-15T18:09:11Z

uh sorry, I'm forget send benchmark file

http: implement slab allocation for HTTP header parsing

20bde47

nodejs-github-bot added c++ Issues and PRs that require attention from people who are familiar with C++. http_parser Issues and PRs related to the HTTP Parser dependency or the http_parser binding. needs-ci PRs that need a full CI run. labels Jan 13, 2026

lint

9115556

mcollina requested a review from Copilot January 13, 2026 21:10

Copilot started reviewing on behalf of mcollina January 13, 2026 21:11 View session

Copilot AI reviewed Jan 13, 2026

View reviewed changes

mcollina requested review from addaleax, Copilot, jasnell and joyeecheung January 13, 2026 21:19

Copilot started reviewing on behalf of mcollina January 13, 2026 21:20 View session

Copilot AI reviewed Jan 13, 2026

View reviewed changes

addaleax reviewed Jan 15, 2026

View reviewed changes

mertcanaltin added 2 commits January 15, 2026 21:13

benchmark: add fragmented HTTP header parsing benchmark

5def150

implement shared slab allocation

56393d3

mertcanaltin requested a review from addaleax January 15, 2026 19:10

	// Memory impact: ~8KB per parser (66 StringPtr × 128 bytes).
	// Memory impact: ~8.4KB per parser (66 StringPtr × 128 bytes).

Uh oh!

http: implement slab allocation for HTTP header parsing #61375

Are you sure you want to change the base?

http: implement slab allocation for HTTP header parsing #61375

Conversation

mertcanaltin commented Jan 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

nodejs-github-bot commented Jan 13, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

codecov bot commented Jan 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

jasnell Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

mertcanaltin Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

addaleax Jan 15, 2026

Choose a reason for hiding this comment

Uh oh!

mertcanaltin Jan 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mertcanaltin Jan 15, 2026

Choose a reason for hiding this comment

Uh oh!

addaleax Jan 15, 2026

Choose a reason for hiding this comment

Uh oh!

mertcanaltin Jan 15, 2026

Choose a reason for hiding this comment

Uh oh!

mertcanaltin commented Jan 15, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

mertcanaltin commented Jan 13, 2026 •

edited

Loading

codecov bot commented Jan 13, 2026 •

edited

Loading

mertcanaltin Jan 15, 2026 •

edited

Loading