The `advanced` IPC serialization codec was implemented in JavaScript
(ChildProcessSerializer / ChildProcessDeserializer in
lib/internal/child_process/serialization.js). It allocated a wrapper
serializer/deserializer per message and crossed the JS/C++ boundary
several times for every message (writeHeader, writeValue, releaseBuffer,
readHeader, readValue and friends).
Move the codec into a native `ipc_serdes` binding that drives the V8
ValueSerializer/ValueDeserializer with a C++ delegate. The wire format
is preserved byte-for-byte: a big-endian uint32 length prefix followed
by the V8 payload, with ArrayBufferViews tagged as host objects so that
Node Buffers round-trip as Buffers rather than plain Uint8Arrays. The
JSON codec is left unchanged.
A cctest (test/cctest/test_node_ipc_serdes.cc) exercises the binding
directly, covering round-trips of primitives, objects, typed arrays and
Buffers (including the Buffer-vs-Uint8Array distinction) and asserting
the big-endian length-prefix framing.
Round-trip throughput
(benchmark/child_process/child-process-ipc-roundtrip):
payload before after change
64 B ~300k/s ~800k/s +166%
1 KiB ~272k/s ~616k/s +126%
16 KiB ~91k/s ~120k/s +32%
64 KiB ~30k/s ~35k/s +16%
The gain is largest for small messages, where per-message JavaScript
overhead dominated, and tapers for large messages, where the actual
serialization (already native) dominates.
Signed-off-by: Yagiz Nizipli <yagiz@nizipli.com>
PR-URL: https://github.com/nodejs/node/pull/63933
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: Filip Skokan <panva.ip@gmail.com>
04c04c8b5c