The compat layer always responded with waitForTrailers set, so every
response paid for a wantTrailers C++ -> JS callback, an empty
sendTrailers() submission scheduled through setImmediate(), and an
extra empty DATA frame on the wire, even though the vast majority of
responses never register any trailers.
When the headers are flushed as part of response.end() and no
trailers have been registered, there is no further opportunity to
add trailers, so waitForTrailers can be skipped altogether. Headers
flushed early (writeHead, write, flushHeaders) keep the previous
behavior so trailers can still be added while streaming.
Trailers added after response.end() are now silently dropped,
matching the HTTP/1 response.addTrailers() semantics.
Also reuse a shared options object for Http2ServerRequest instances
created without explicit options.
h2load, 1 KiB response payload, -c 4 -m 100, mean of 6 alternating
runs: compat API 43.1k -> 49.9k req/s (+15.7% cumulative vs main).
Signed-off-by: Matteo Collina <hello@matteocollina.com>
PR-URL: https://github.com/nodejs/node/pull/64265
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Gürgün Dayıoğlu <hey@gurgun.day>
Reviewed-By: Tim Perry <pimterry@gmail.com>
This brings back several improvements that were reverted by mistake when
landing https://redirect.github.com/nodejs/node/pull/64154
- Update the documentation about how the removal of side effects of
source collection
- Add non-enumerable `requireStack` and `topLevelAwaitLocations`
properties to `ERR_REQUIRE_ASYNC_MODULE`, the latter is only
populated when --experimental-print-required-tla is enabled
- Add "Required module: <url>" to the error message to identify the
required ESM entry point regardless of whether the flag is enabled
- Fix TLA caret column from 0-based to 1-based
- Store module source via a private symbol instead of a public property
- Use `hasAsyncGraph` (post-instantiation) in `throwIfAsyncGraph`
instead of walking the graph before instantiation
- Merge the require stack checking into the
`common.expectRequiredTLAError` helper.
- Removed tests that are made redundant by the snapshot tests
Signed-off-by: Joyee Cheung <joyeec9h3@gmail.com>
PR-URL: https://github.com/nodejs/node/pull/64260
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com>
Previously in order to collect the locations of the TLA, we wait
until right before evalutation to ensure instantiation is
completed so that we can use
v8::Module::GetStalledTopLevelAwaitMessages(). Now we try to
add an additioanl shortcut to the source code in the module wraps
instead during compilation for modules that contain TLAs and
use acron to locate the TLAs when we need to throw
ERR_REQUIRE_AYNSC_MODULE, so we can do this as early as before
instantiation and do not need to run the module again to collect
the locations.
In addition, we now collect the require stack for
ERR_REQUIRE_ASYNC_MODULE too for better metadata in the errors.
Signed-off-by: Joyee Cheung <joyeec9h3@gmail.com>
PR-URL: https://github.com/nodejs/node/pull/64154
Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com>
Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>
The text said functions with more than 7 arguments fall back to the
generic call path, contradicting the preceding sentence that Fast FFI
calls support up to 8 arguments.
The real limit is architecture- and type-dependent: the hard cap is 8
public arguments (src/ffi/fast.cc), but register pressure lowers the
effective count to 7 integer/pointer arguments on AArch64 and 6 on
x86-64, while floating-point arguments can use up to 8 on both. Describe
this instead of the previous, inaccurate single threshold.
Signed-off-by: Daijiro Wachi <daijiro.wachi@gmail.com>
PR-URL: https://github.com/nodejs/node/pull/63960
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Paolo Insogna <paolo@cowtech.it>
When `options.path` passed to `http.request()` contains an absolute
URL, `http.request` has been sending it directly as the request target
in the HTTP 1.1 message. If the receiving server is a proxy,
the proxy server typically forwards the request to the destination
specified in the request target and ignores the Host header.
This means eventually the request can be forwarded to a destination
that is not consistent with `options.host`, depending on how the
receiving server behaves. Mimatched Host header and request target
also violates RFC 9112 Section 3.2, which we have been entirely
leaving to the users to verify.
This patch documents this behavior and warns that the user needs
to ensure the `path`, `option` and `headers` conform to the RFC.
If the receiving server is known to be a proxy server because
the request is routed by Node.js' built-in HTTP proxy support,
we now do a best-effort check to verify that the authority in
`options.path` (if absolute), Host headers and `options.host`
agree at request construction time. Node.js will give up on the
require target rewriting and throw an error when they don't match
at request construction.
Signed-off-by: Joyee Cheung <joyeec9h3@gmail.com>
PR-URL: https://github.com/nodejs/node/pull/64108
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: Tim Perry <pimterry@gmail.com>
Reviewed-By: Ulises Gascón <ulisesgascongonzalez@gmail.com>
When OpenSSL reports no peer temporary key object for TLS key
agreement, fall back to the negotiated TLS Supported Group name.
This lets getEphemeralKeyInfo() identify PQ, hybrid, and other
group-based key agreement as TLSGroup without synthesizing a size.
Keep existing DH and ECDH results unchanged when OpenSSL provides
a recognizable temporary key.
Fixes: https://github.com/nodejs/node/issues/59452
Signed-off-by: Filip Skokan <panva.ip@gmail.com>
PR-URL: https://github.com/nodejs/node/pull/64119
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com>
Every unconnected send(), and the implicit bind on first send,
resolved the destination through dns.lookup() even when it was
already a literal IP. The address resolves to itself, so the call is
redundant, and tools that instrument dns.lookup() record a lookup for
every datagram sent to an IP.
Skip the resolver for a literal IP of the socket's family and report
it on the next tick, keeping dns.lookup()'s asynchronous contract. A
custom lookup function is still consulted for every address.
Refs: https://github.com/DataDog/dd-trace-js/issues/2984
Signed-off-by: Ruben Bridgewater <ruben@bridgewater.de>
PR-URL: https://github.com/nodejs/node/pull/64133
Reviewed-By: Bryan English <bryan@bryanenglish.com>
Reviewed-By: Gürgün Dayıoğlu <hey@gurgun.day>
Reviewed-By: James M Snell <jasnell@gmail.com>
Add a samplePerIteration option to monitorEventLoopDelay that records
event loop delay from libuv event loop iterations instead of the timer
interval sampler. The default remains interval-based; existing uses of
monitorEventLoopDelay() keep behaving the same unless the
samplePerIteration option is passed through.
Signed-off-by: Pablo Erhard <pablo.erhardhernandez@datadoghq.com>
PR-URL: https://github.com/nodejs/node/pull/62935
Reviewed-By: Bryan English <bryan@bryanenglish.com>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: James M Snell <jasnell@gmail.com>
V8's Minor Mark-Sweep collector, enabled with the --minor-ms flag
(available since Node.js 22), reports garbage collection performance
entries with kind kGCTypeMinorMarkSweep (value 2). perf_hooks exposed
constants for every other GC kind (major, minor, incremental, weakcb)
but not this one, so consumers inspecting performanceEntry.detail.kind
had no constant to compare against and saw an unmapped value.
Expose it as perf_hooks.constants.NODE_PERFORMANCE_GC_MINOR_MARK_SWEEP,
mirroring the existing v8::GCType mappings, and document it alongside
the other GC kinds.
Signed-off-by: Attila Szegedi <attila.szegedi@datadoghq.com>
PR-URL: https://github.com/nodejs/node/pull/63877
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com>
Reviewed-By: Ilyas Shabi <ilyasshabi94@gmail.com>
Reviewed-By: Chengzhong Wu <legendecas@gmail.com>
PR #62782 updated the keepAliveTimeout default from 5s to 65s and
correctly updated server.keepAliveTimeout (line 2006), but the
http.createServer() options table at line 3775 was not updated.
Fix the remaining reference to reflect the current default of 65000ms.
Signed-off-by: Jahanzaib iqbal <jahanzaib@gsoftconsulting.com>
PR-URL: https://github.com/nodejs/node/pull/63974
Reviewed-By: Tim Perry <pimterry@gmail.com>
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
Port the domain module from createHook (async_hooks) to
AsyncLocalStorage using the AsyncContextFrame-based implementation.
Key changes:
- Use AsyncLocalStorage for domain context propagation instead of
async_hooks.createHook()
- Lazy initialization that triggers AsyncContextFrame prototype swap
on first domain use
- Use enterWith instead of ALS.run() so domain context is NOT
automatically restored on exception - this matches the original
domain.run() behavior where exit() only runs on success
- Add ERR_ASYNC_RESOURCE_DOMAIN_REMOVED error for AsyncResource.domain
- Update DEP0097 to End-of-Life status
- Remove tests that relied on the removed MakeCallback domain property
The domain module now uses the AsyncContextFrame version of
AsyncLocalStorage directly for proper context propagation across
async boundaries.
Signed-off-by: Matteo Collina <hello@matteocollina.com>
PR-URL: https://github.com/nodejs/node/pull/61095
Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com>
Reviewed-By: Rafael Gonzaga <rafael.nunu@hotmail.com>
Reviewed-By: Robert Nagy <ronagy@icloud.com>
Reviewed-By: Paolo Insogna <paolo@cowtech.it>
Reviewed-By: Marco Ippolito <marcoippolito54@gmail.com>
Reviewed-By: Gürgün Dayıoğlu <hey@gurgun.day>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Moshe Atlow <moshe@atlow.co.il>
Add shared bit-length helpers for WebCrypto operations that accept bit
sequences whose length is not byte-aligned.
Use the helpers for cSHAKE output, ECDH-derived bits, HMAC/KMAC key
generation/import/derivation, and KMAC sign/verify output. Preserve the
requested bit length in CryptoKey algorithm metadata while storing and
exporting rounded-up byte material with unused low bits cleared.
Keep byte-multiple validation for algorithms whose specs require it.
Extend the lower-end of KMAC's key length support.
Enable cSHAKE customization and functionName parameters.
Signed-off-by: Filip Skokan <panva.ip@gmail.com>
PR-URL: https://github.com/nodejs/node/pull/63988
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com>
This changes enables compression within OpenSSL *without* enabling
record compression, so this only affects compression of certificates
delivered within the TLS handshake. This certificate compression remains
disabled by default for now, but becomes available via the new
certificateCompression option in TLS context APIs.
Enabling this shrinks handshakes significantly, and also reduces
fingerprintability of Node.js client handshakes, as these are enabled in
all modern browsers by default.
Signed-off-by: Tim Perry <pimterry@gmail.com>
PR-URL: https://github.com/nodejs/node/pull/62217
Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Moves support for passing CryptoKey instances to node:crypto APIs to
End-of-Life status (DEP0203).
Moves support for passing a non-extractable CryptoKey instances to
KeyObject.from() End-of-Life status (DEP0204).
Closes#55293
Signed-off-by: Filip Skokan <panva.ip@gmail.com>
PR-URL: https://github.com/nodejs/node/pull/63188
Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
In `Writer`, there are two queues that matter: the slot queue
and the pending queue. The sync `writeSync`/`writevSync` only
use the slot queue. If writes cannot be accepted directly into
slots, they return `false`. The sync methods *never* enqueue
into the pending queue. The async `write`/`writev` will first
attempt to add to the slot queue; if it is full, then it will
attempt to add to the pending queue; if that is also full, the
backpressure policy kicks in.
Signed-off-by: James M Snell <jasnell@gmail.com>
PR-URL: https://github.com/nodejs/node/pull/63697
Reviewed-By: Trivikram Kamat <trivikr.dev@gmail.com>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: Gürgün Dayıoğlu <hey@gurgun.day>
The CJS mock.timers.setTime example asserted the timer ran right after
setTime() with no tick() call, which contradicts the documented behavior
and the ESM example above it. The mock.module example called .fn() on an
export named foo. Both threw if run.
Signed-off-by: Emmanuel Yusufu Kimaswa <kimaswaemma36@gmail.com>
PR-URL: https://github.com/nodejs/node/pull/63656
Reviewed-By: Chemi Atlow <chemi@atlow.co.il>
Reviewed-By: Jacob Smith <jacob@frende.me>
Add fast API trampolines for AArch64 and x86_64.
IsJitMemorySupported() maps an RW page, writes a ret instruction
(0xD65F03C0 AArch64 / 0xC3 x86_64), and mprotects it to RX. A
successful RX transition is treated as the support signal. The page is
deliberately not executed: this check may run during normal operation
(when an FFI function is first created), and executing freshly written
code from a capability probe could SIGSEGV/SIGKILL the process on
systems that block executable memory. The real trampoline emitter
performs the same mprotect at creation time and falls back to libffi
when it is rejected.
The result is computed once via std::call_once and cached for the
lifetime of the process, so concurrent callers never observe a
provisional value.
Wired into CreateFastFFIMetadata() for early nullptr bail-out when
JIT memory is unavailable.
Windows stub returns false (no trampolines yet on this branch).
IsFastCallEligible() validates at parse time whether a signature can
use the fast-call path, covering:
- Return and argument type eligibility (numeric, pointer; no structs)
- Argument count cap (8, matching V8 fast-call limit)
- Per-ABI register pressure limits that mirror the trampoline emitters
(AArch64 and x86_64 SysV). Platforms without an emitter, including
Win64, are reported ineligible.
- Buffer and float args cannot coexist, and buffer args additionally
consume an extra GP register slot on both supported ABIs.
The arg/arg-name lengths are checked before the per-arg loop so a
malformed signature cannot index out of bounds.
Returns nullptr from CreateFastFFIMetadata() for ineligible signatures,
falling back to libffi.
Co-authored-by: Bryan English <bryan@bryanenglish.com>
Signed-off-by: Paolo Insogna <paolo@cowtech.it>
Signed-off-by: Bryan English <bryan@bryanenglish.com>
Assisted-By: OpenAI:GPT-5.5 <openai/gpt-5.5>
PR-URL: https://github.com/nodejs/node/pull/63068
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: Bryan English <bryan@bryanenglish.com>