http,http2: speed up header validation and write coalescing by anonrig · Pull Request #65332 · nodejs/node

anonrig · 2026-08-16T18:33:03Z

This speeds up the HTTP/1 and HTTP/2 hot paths that show up in
writeHead()/end(), small-body writes, and HTTP/2 header serialization.

HTTP/1

Use the token lookup table for names of length <= 10 (Connection,
Keep-Alive). Longer names stay on the regex path; a JS table over
those lengths is slower than V8's regex.
Cache OutgoingMessage lenient-validation and skip statusMessage
validation for the built-in reason phrases.
Reuse prebuilt HTTP/1.1 <code> <reason>\r\n status lines.
Avoid toLowerCase() for the common Title-Case / lowercase spellings
of Connection, Content-Length, Transfer-Encoding, and friends.
Coalesce headers with small Buffer bodies into one socket.write().
When res.end(chunk) is used with Transfer-Encoding: chunked and
headers have not been flushed, send headers + last chunk + terminator
in a single write (headers copied as latin1 so obs-text is preserved).
Keep incoming header name/value bytes in a packed Buffer. rawHeaders
/ headers materialize JS strings only when read. The default server
Host / Expect / body-header checks use flag bits and do not build
req.headers.
IncomingMessage._dump() / dumped bodies skip Buffer::Copy via
parser.setSkipBody().

HTTP/2

Skip toLowerCase() when header names are already lowercase.
Hoist buildNgHeaderString processing off a per-call closure.
Use a Set for sensitive / strict single-value header checks.
Reserve outgoing session storage while gathering
nghttp2_session_mem_send chunks.

These are incremental, behavior-preserving changes. They do not move
end-to-end benchmark/http/simple.js by 50%. Local wrk numbers vs
upstream/main (same machine, release build, 2x 5s) for the JS write /
validation work:

config	main	this PR	delta
buffer, len=4, chunked, c=50	84.6k req/s	91.2k	+7.9%
bytes, len=4, chunked, c=50	78.2k	78.3k	+0.1%
buffer, len=1024, chunked, c=50	73.4k	77.4k	+5.4%
bytes, len=1024, content-length, c=50	77.0k	81.4k	+5.8%

check_is_http_token for Connection is about +21%. Header-value
validation stays on regex (a JS Uint8Array scan was 30–50% slower).

The packed-header follow-up keeps bytes off the JS string heap until
something actually reads rawHeaders / headers. A per-request native
BaseObject was slower than creating the string array, so the bytes live
in one Buffer with flag bits for Host / Expect / body headers.

Tests

Extra checkInvalidHeaderChar cases in test/parallel/test-http-common.js.
test/parallel/test-http-chunked-end-coalesce.js checks the wire
format of coalesced chunked end() (string, buffer, UTF-8, trailers,
unusual Content-Length casing, latin1 header values).
test/parallel/test-http-native-headers.js checks lazy Host / Expect
lookups and materialization on req.headers access.

Local: 405/405 test/parallel/test-http-* (HTTP/1) passed against a
release build of the packed-headers tree. Earlier 776/776
test-http* / test-https* / test-http2* passed on the same branch
before the Buffer packing tweak.

nodejs-github-bot · 2026-08-16T18:33:08Z

Review requested:

@nodejs/http
@nodejs/http2
@nodejs/net

Replace regex-based header token/value checks with byte lookup tables, cache OutgoingMessage lenient-validation, and coalesce headers with small Buffer bodies into a single socket write. Scan IncomingMessage rawHeaders for Content-Length and Transfer-Encoding so optimizeEmptyRequests does not force req.headers construction. On the HTTP/2 path, skip toLowerCase for already-lowercase names, use a Set for sensitive/single-value header checks, and reserve outgoing session storage to avoid reallocs while gathering nghttp2 frames. Signed-off-by: Yagiz Nizipli <yagiz@nizipli.com>

Send headers, the last chunk, and the chunked terminator in a single write() when res.end() is used with Transfer-Encoding: chunked. Reuse prebuilt HTTP/1.1 status lines for default reason phrases, skip toLowerCase() on common outgoing header names, and hoist HTTP/2 header serialization off the per-call closure. Signed-off-by: Yagiz Nizipli <yagiz@nizipli.com>

The chunked end() fast path concatenated headers with the body and wrote the result using the body encoding. That re-encoded obs-text header values as UTF-8 and reduced corked res.end() to a single socket.write(), which broke test-http-server-non-utf8-header and test-http-response-cork. Copy headers as latin1 into the combined buffer, and accept a single write after uncork. Signed-off-by: Yagiz Nizipli <yagiz@nizipli.com>

JS lookup tables were slower than V8's regex for header values and for token names longer than ~10 bytes. Restore the regex path for those cases, and use the table for names of length <= 10 so Connection / Keep-Alive stay on the faster path. Signed-off-by: Yagiz Nizipli <yagiz@nizipli.com>

anonrig · 2026-08-16T20:17:12Z

Local results from a release build of this branch vs upstream/main on the same machine (4 cores).

Correctness

779/779 test-http* / test-https* / test-http2* (parallel + sequential + pummel + async-hooks) passed.
The earlier CI failures (test-http-response-cork, test-http-server-non-utf8-header) were from writing coalesced chunked headers with the body encoding. Fixed by copying headers as latin1 and relaxing the cork test to mustCallAtLeast.

benchmark/http/simple.js (wrk, duration=5, c=50, 2 runs)

type	len	encoding	main req/s	PR req/s	delta
buffer	4	chunked	84,567	91,239	+7.9%
bytes	4	chunked	78,179	78,271	+0.1%
buffer	1024	chunked	73,366	77,353	+5.4%
bytes	1024	content-length	76,980	81,406	+5.8%

Micros (3 runs, ops/s)

A JS lookup table for all token lengths and for header-value checks was a regression (V8 regex is faster past ~10 bytes). Latest commit keeps regex for values and for tokens longer than 10, and uses the table for Connection / Keep-Alive:

bench	main	PR	delta
`check_is_http_token` `Connection`	44.2M	53.7M	+21%
`check_is_http_token` `Content-Length`	37.8M	36.6M	~noise
`check_invalid_header_char` `keep-alive`	51.3M	49.0M	~noise
`set_header` `Connection`	12.5M	12.8M	+2%

HTTP/2 headers.js was within noise (−7% to +0%).

So the measurable end-to-end win is the small-Buffer / chunked end() coalescing, on the order of 5–8%, not 50%. I am not claiming a 50% HTTP/HTTP2 speedup from this PR.

Copilot

Pull request overview

This PR optimizes Node.js HTTP/1.1 and HTTP/2 hot paths around header validation/serialization and write coalescing to reduce per-request overhead in writeHead()/end() and small-body responses.

Changes:

HTTP/1: faster header-name token validation, cached lenient-validation, reused prebuilt status lines, avoided common-case toLowerCase(), and added header/body write coalescing (including a combined chunked end() write).
HTTP/1 server: avoid forcing req.headers construction when checking for Content-Length / Transfer-Encoding by scanning rawHeaders.
HTTP/2: reduce lowercasing overhead in header processing and reserve outgoing send storage to reduce reallocations.

Reviewed changes

Copilot reviewed 10 out of 10 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
test/parallel/test-http-response-cork.js	Updates expectations to allow fewer `socket.write()` calls due to new coalescing behavior.
test/parallel/test-http-common.js	Adds more `checkInvalidHeaderChar()` test cases for strict vs lenient validation.
test/parallel/test-http-chunked-end-coalesce.js	New test validating wire format when chunked `res.end()` is coalesced into a single write.
src/node_http2.cc	Reserves outgoing HTTP/2 storage in send path to reduce reallocations.
lib/internal/http2/util.js	Refactors nghttp2 header string building to reduce lowercasing/closure overhead and uses `Set` for strict single-value checks.
lib/internal/http2/core.js	Avoids unconditional `toLowerCase()` in response header array scanning.
lib/_http_server.js	Prebuilds HTTP/1.1 status lines and avoids unnecessary `statusMessage` validation for defaults; uses `req._hasBodyHeaders()` for optimizeEmptyRequests.
lib/_http_outgoing.js	Coalesces headers with small bodies; adds combined chunked `end()` fast path; caches lenient-header-validation result; reduces common-case `toLowerCase()`.
lib/_http_incoming.js	Adds `IncomingMessage.prototype._hasBodyHeaders()` scanning `rawHeaders` to avoid constructing `req.headers`.
lib/_http_common.js	Speeds up `checkIsHttpToken()` by using a lookup table for short names and regex for longer ones.

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

  if (chunk) {
    if (this.finished) {
      onError(this,
              new ERR_STREAM_WRITE_AFTER_END(),
              typeof callback !== 'function' ? nop : callback);
      return this;
    }

    if (this[kSocket]) {
      this[kSocket].cork();
    }

-    write_(this, chunk, encoding, null, true);
+    write_(this, chunk, encoding, onFinish.bind(undefined, this), true);


Pass NativeHttpHeaders from the parser instead of a JS string array. IncomingMessage materializes rawHeaders/headers only when read. Host/Expect/body-header checks use C++ has/get. Skip Buffer::Copy for dumped bodies via parser.setSkipBody(). Signed-off-by: Yagiz Nizipli <yagiz@nizipli.com>

Store header name/value bytes in a packed Buffer instead of a per-request native BaseObject or JS string array. IncomingMessage materializes rawHeaders only when read. Host/Expect/body-header checks use flag bits so the default server path never creates header strings. Dumped bodies skip Buffer::Copy via setSkipBody(). Signed-off-by: Yagiz Nizipli <yagiz@nizipli.com>

anonrig · 2026-08-16T20:55:58Z

Follow-up: incoming header bytes now stay off the JS string heap until something actually reads them.

kOnHeadersComplete passes a packed Buffer (magic + count + flags + name/value bytes) instead of a JS string array or a per-request native BaseObject. IncomingMessage.rawHeaders / headers materialize strings only on first read. The default HTTP/1.1 server Host / Expect / Content-Length / Transfer-Encoding checks use flag bits written while packing, so simple.js-style handlers that never touch req.headers do not create header strings. _dump() sets parser.setSkipBody(true) so dumped bodies skip Buffer::Copy.

A first attempt used a NativeHttpHeaders BaseObject per request (Function::NewInstance + MakeWeak). That was slower than creating the JS string array — especially with many headers — so it was replaced with one Buffer.

test/parallel/test-http-native-headers.js covers the lazy Host check (including 400 without Host) and materialization. 405/405 test/parallel/test-http-* passed on the packed-buffer tree.

I am not updating the simple.js wrk table from this environment; run-to-run noise here is large enough to hide a real single-digit delta. The earlier +5–8% numbers are for the write/validation commits, not a claim that packed headers moved end-to-end simple.js by that much (wrk GETs have no body, and the handler never reads req.headers).

nodejs-github-bot added lib / src Issues and PRs related to general changes in the lib or src directory. needs-ci PRs that need a full CI run. labels Aug 16, 2026

anonrig added 2 commits August 16, 2026 18:59

anonrig force-pushed the cursor/http-http2-perf-f24c branch from 76c8e14 to 62327c4 Compare August 16, 2026 18:59

anonrig added 2 commits August 16, 2026 19:37

ronag approved these changes Aug 16, 2026

View reviewed changes

ronag requested a lite review from Copilot August 16, 2026 20:20

Copilot started reviewing on behalf of ronag August 16, 2026 20:20 View session

Copilot AI reviewed Aug 16, 2026

View reviewed changes

anonrig added 2 commits August 16, 2026 20:32

Navigation Menu

Uh oh!

Uh oh!

http,http2: speed up header validation and write coalescing - #65332

http,http2: speed up header validation and write coalescing#65332
anonrig wants to merge 6 commits into
nodejs:mainfrom
anonrig:cursor/http-http2-perf-f24c

anonrig commented Aug 16, 2026 •

edited

Loading

Uh oh!

nodejs-github-bot commented Aug 16, 2026

Uh oh!

anonrig commented Aug 16, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

anonrig commented Aug 16, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Uh oh!

Uh oh!

Conversation

anonrig commented Aug 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

HTTP/1

HTTP/2

Tests

Uh oh!

nodejs-github-bot commented Aug 16, 2026

Uh oh!

anonrig commented Aug 16, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

anonrig commented Aug 16, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

anonrig commented Aug 16, 2026 •

edited

Loading