This patch implements automatic on-disk code caching that can be enabled
via an environment variable NODE_COMPILE_CACHE.
When set, whenever Node.js compiles a CommonJS or a ECMAScript Module,
it will use on-disk [V8 code cache][] persisted in the specified
directory to speed up the compilation. This may slow down the first
load of a module graph, but subsequent loads of the same module graph
may get a significant speedup if the contents of the modules do not
change. Locally, this speeds up loading of
test/fixtures/snapshot/typescript.js from ~130ms to ~80ms.
To clean up the generated code cache, simply remove the directory.
It will be recreated the next time the same directory is used for
`NODE_COMPILE_CACHE`.
Compilation cache generated by one version of Node.js may not be used
by a different version of Node.js. Cache generated by different versions
of Node.js will be stored separately if the same directory is used
to persist the cache, so they can co-exist.
Caveat: currently when using this with V8 JavaScript code coverage, the
coverage being collected by V8 may be less precise in functions that are
deserialized from the code cache. It's recommended to turn this off when
running tests to generate precise coverage.
Implementation details:
There is one cache file per module on disk. The directory layout
is:
- Compile cache directory (from NODE_COMPILE_CACHE)
- 8b23c8fe: CRC32 hash of CachedDataVersionTag + NODE_VERESION
- 2ea3424d:
- 10860e5a: CRC32 hash of filename + module type
- 431e9adc: ...
- ...
Inside the cache file, there is a header followed by the actual
cache content:
```
[uint32_t] code size
[uint32_t] code hash
[uint32_t] cache size
[uint32_t] cache hash
... compile cache content ...
```
When reading the cache file, we'll also check if the code size
and code hash match the code that the module loader is loading
and whether the cache size and cache hash match the file content
read. If they don't match, or if V8 rejects the cache passed,
we'll ignore the mismatch cache, and regenerate the cache after
compilation succeeds and rewrite it to disk.
PR-URL: https://github.com/nodejs/node/pull/52535
Refs: https://github.com/nodejs/node/issues/47472
Reviewed-By: Benjamin Gruenbaum <benjamingr@gmail.com>
Reviewed-By: Yagiz Nizipli <yagiz.nizipli@sentry.io>
Reviewed-By: Mohammed Keyvanzadeh <mohammadkeyvanzade94@gmail.com>
Currently we use external strings for internalized builtin source code.
However when a snapshot is taken, any external string whose resource
is not registered is flattened into a SeqString (see ref). The
result is that module source code stored in the snapshot does not
use external strings after deserialization. This patch registers an
external string resource for each internalized builtin's source. The
savings are substantial: ~1.9 MB of heap memory per isolate, or ~44%
of an otherwise empty isolate's heap usage:
```console
$ node --expose-gc -p 'gc(),process.memoryUsage().heapUsed'
4190968
$ ./node --expose-gc -p 'gc(),process.memoryUsage().heapUsed'
2327536
```
The savings can be even higher for user snapshots which may include
more internal modules.
The existing UnionBytes implementation was ill-suited, because it only
created an external string resource when ToStringChecked was called,
but we need to register the external string resources before the
isolate even exists. We change UnionBytes to no longer own the data,
and shift ownership of the data to a new external resource class
called StaticExternalByteResource. StaticExternalByteResource are
either statically allocated (for internalized builtin code) or owned
by the static `externalized_builtin_sources` map, so they will only be
destructed when static resources are destructed. We change JS2C to emit
statements to register a string resource for each internalized builtin.
Refs: https://github.com/v8/v8/blob/d2c8fbe9ccd1a6ce5591bb7dd319c3c00d6bf489/src/snapshot/serializer.cc#L633
PR-URL: https://github.com/nodejs/node/pull/47055
Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com>
Reviewed-By: Chengzhong Wu <legendecas@gmail.com>
The pattern of getting the address of the element at index 0 of a
container is generally used to materialize a pointer to the backing
data of a container, however `std::string` and `std::vector`
provide a `data()` accessor to retrieve the data pointer which
should be preferred.
This also ensures that in the case that the container is empty, the
data pointer access does not perform an errant memory access.
PR-URL: https://github.com/nodejs/node/pull/47750
Reviewed-By: Tobias Nießen <tniessen@tnie.de>
Reviewed-By: Daeyeon Jeong <daeyeon.dev@gmail.com>
Reviewed-By: Chengzhong Wu <legendecas@gmail.com>
Reviewed-By: Darshan Sen <raisinten@gmail.com>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Remove the dependency on ICU for this part, as well as the
hacky way of converting embedder main sources to UTF-8 via
V8 APIs. Allow `UnionBytes` to own the memory its pointing
to in order to simplify the code on the `BuiltinLoader` side.
PR-URL: https://github.com/nodejs/node/pull/46119
Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Rafael Gonzaga <rafael.nunu@hotmail.com>
Reviewed-By: Darshan Sen <raisinten@gmail.com>
Bindings that need to be loaded in distinct contexts of node::Realm in
the same node::Environment instance needs to share the per-isolate
templates.
Add a new binding registration callback, which is called with
per-IsolateData. This allows bindings to define templates and
share them across realms eagerly, and avoid accidental modification on
the templates when the per-context callback is called multiple times.
This also tracks loaded bindings and built-in modules with node::Realm.
PR-URL: https://github.com/nodejs/node/pull/45547
Refs: https://github.com/nodejs/node/issues/42528
Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
- Move the file reading code in LoadBuiltinModuleSource into
util.h so that it can be reused by other C++ code, and
return an error code from it when there is a failure for
the caller to generate an error.
- Throw an error when reading local builtins fails in
LoadBulitinModuleSource.
PR-URL: https://github.com/nodejs/node/pull/38904
Reviewed-By: Chengzhong Wu <legendecas@gmail.com>
`GetProcessTitle()` otherwise runs an infinite loop when
`uv_setup_argv()` has not been called (yet). This is a problem
e.g. in assertions from static constructors, which run before
`main()` and thus before `argc` and `argv` become available.
To solve that, do not allocate more than 1MB of storage for the
title and bail out if we reach that point.
PR-URL: https://github.com/nodejs/node/pull/35492
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Gus Caplan <me@gus.host>
Reviewed-By: Tobias Nießen <tniessen@tnie.de>
Reviewed-By: Rich Trott <rtrott@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
The getter passed a stack-allocated, fixed-size buffer to
uv_get_process_title() but neglected to check the return value.
When the total length of the command line arguments exceeds the size of
the buffer, libuv returns UV_ENOBUFS and doesn't modify the contents of
the buffer. The getter then proceeded to return whatever garbage was on
the stack at the time of the call, quite possibly reading beyond the
end of the buffer.
Add a GetProcessTitle() helper that reads the process title into a
dynamically allocated buffer that is resized when necessary.
Fixes: https://github.com/nodejs/node/issues/31631
PR-URL: https://github.com/nodejs/node/pull/31633
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Gus Caplan <me@gus.host>
Reviewed-By: David Carlier <devnexen@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>
This patch moves the serialization of coverage profiles into
C++. With this we no longer need to patch `process.reallyExit`
and hook into the exit events, but instead hook into relevant
places in C++ which are safe from user manipulation. This also
makes the code easier to reuse for other types of profiles.
PR-URL: https://github.com/nodejs/node/pull/26874
Reviewed-By: Ben Coe <bencoe@gmail.com>
Where appropriate, use a helper that wraps around
`ArrayBufferView::Buffer()` or `ArrayBufferView::CopyContents()`
rather than `Buffer::Data()`, as that may help to avoid materializing
the underlying `ArrayBuffer` when reading small typed arrays from C++.
This allows keeping the performance benefits of the faster creation of
heap-allocated small typed arrays in many cases.
PR-URL: https://github.com/nodejs/node/pull/26301
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
So that it's easier to tell whether we are manipulating per-process
global states that may need to be treated with care to avoid races.
Also added comments about these variables and moved some of them
to a more suitable compilation unit:
- Move `v8_initialized` to `util.h` since it's only used in
`util.cc` and `node.cc`
- Rename `process_mutex` to `tty_mutex` and move it into
`node_errors.cc` since that's the only place it's used
to guard the tty.
- Move `per_process_opts_mutex` and `per_process_opts`
into `node_options.h` and rename them to
`per_process::cli_options[_mutex]`
- Rename `node_isolate[_mutex]` to `per_process::main_isolate[_mutex]`
PR-URL: https://github.com/nodejs/node/pull/25302
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Franziska Hinkelmann <franziska.hinkelmann@gmail.com>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
gcc 8+ recognizes that space has not been left for the pid and that the
return value of snprintf() isn't checked. Leave a little space for the
pid to prevent `-Wformat-truncation`.
PR-URL: https://github.com/nodejs/node/pull/24810
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
The agent code is supposed to manage multiple writers/clients.
Adding the default writer into the mix breaks that encapsulation.
Instead, manage default options through a separate "virtual"
default client handle, and keep the file writer management
all to the actual users.
PR-URL: https://github.com/nodejs/node/pull/21867
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Eugene Ostroukhov <eostroukhov@google.com>
Reviewed-By: Ali Ijaz Sheikh <ofrobots@google.com>
Joyee Cheung