The binding data must be weak so that it won't keep the realm reachable
from strong GC roots indefinitely. The wrapper object of binding data
should be referenced from JavaScript, thus the binding data should be
reachable throughout the lifetime of the realm.
PR-URL: https://github.com/nodejs/node/pull/47688
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com>
Instead of storing the function template of BaseObject for checking
if an object is BaseObject by calling HasInstance, simply checks
the first internal field of the object, which is always set in the
BaseObject constructor. This is simpler and faster
(there is now no need to iterate over the inheritance for the check).
PR-URL: https://github.com/nodejs/node/pull/47217
Reviewed-By: Chengzhong Wu <legendecas@gmail.com>
Locally the hashing of the binding names sometimes has significant
presence in the profile of bindings, because there can be collisions,
which makes the cost of adding a new binding data non-trivial,
but it's wasteful to spend time on hashing them or dealing with
collisions at all, since we can just use the EmbedderObjectType
enum as the key, as the string names are not actually used beyond
debugging purposes and can be easily matched with a macro.
And since we can just use the enum as the key, we do not even
need the map and can just use an array with the enum as indices
for the lookup.
PR-URL: https://github.com/nodejs/node/pull/46620
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Chengzhong Wu <legendecas@gmail.com>
This allows us to mark weak/detached references in the heap snapshot.
Also mark weak/detached BaseObject with Detachedness::kDetached
so that the state of the reference can be displayed by frontend
consuming the heap snapshot.
PR-URL: https://github.com/nodejs/node/pull/44803
Reviewed-By: Chengzhong Wu <legendecas@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
- Wrap the initialization of the kSlot and kEmbedderType fields
into a BaseObject::SetInternalFields() method.
- Move the tagging of kEmbedderType field into
BaseObject::TagNodeObject()
- Add a variant of BaseObject::MakeLazilyInitializedJSTemplate()
that only needs IsolateData.
This makes it easier to create BaseObject subclasses.
PR-URL: https://github.com/nodejs/node/pull/44796
Reviewed-By: Chengzhong Wu <legendecas@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
This patch updates the layout of the BaseObjects to make sure
that the first embedder field of them is a "type" pointer, the
first 16 bits of which are the Node.js embedder ID, so that
cppgc will always skip over them. In addition we now use this
field to determine if the native object should be interpreted
as a Node.js embedder object in the serialization and deserialization
callbacks for the startup snapshot to improve the reliability.
Co-authored-by: Joyee Cheung <joyeec9h3@gmail.com>
PR-URL: https://github.com/nodejs/node/pull/43521
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Darshan Sen <raisinten@gmail.com>
When a process exits cleanly, i.e. because the event loop ends up
without things to wait for, the Node.js objects that are left on
the heap should be:
1. weak, i.e. ready for garbage collection once no longer
referenced, or
2. detached, i.e. scheduled for destruction once no longer
referenced, or
3. an unrefed libuv handle, i.e. does not keep the event loop
alive, or
4. an inactive libuv handle (essentially the same here)
There are a few exceptions to this rule, but generally,
if there are C++-backed Node.js objects on the heap
that do not fall into the above categories, we may be looking
at a potential memory leak. Most likely, the cause is a missing
`MakeWeak()` call on the corresponding object.
In order to avoid this kind of problem, we check the list
of BaseObjects for these criteria. In this commit, we only do so
when explicitly instructed to or when in debug mode
(where --verify-base-objects is always-on).
In particular, this avoids the kinds of memory leak issues
that were fixed in the PRs referenced below.
Refs: https://github.com/nodejs/node/pull/35488
Refs: https://github.com/nodejs/node/pull/35487
Refs: https://github.com/nodejs/node/pull/35481
PR-URL: https://github.com/nodejs/node/pull/35490
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>
Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com>
Reviewed-By: Benjamin Gruenbaum <benjamingr@gmail.com>
Enable JS wrapper objects to be used as transferable or cloneable
objects in `postMessage()` calls, by having them extend a C++-backed
class.
This requires a few internal changes:
- This commit adds the possibility for transferred objects to
read/write JS values at the end of the serialization/deserialization
phases.
- This commit adds the possibility for transferred objects to list
sub-transferables, e.g. typically the public JS wrapper class
would list its C++ handle in there.
- This commit adds usage of `BaseObject` in a few more places, because
now during deserialization weakly held objects can also be involved,
in addition to `MessagePort`s.
PR-URL: https://github.com/nodejs/node/pull/33772
Reviewed-By: Benjamin Gruenbaum <benjamingr@gmail.com>
Extend support for transferring objects à la `MessagePort` to other
types of `BaseObject` subclasses, as well as implement cloning
support for cases in which destructive transferring is not needed
or optional.
PR-URL: https://github.com/nodejs/node/pull/33772
Reviewed-By: Benjamin Gruenbaum <benjamingr@gmail.com>
Since f59ec2abee, `BaseObject` instances were tracked in heap snapshots
through their associated `CleanupHookCallback`s which were stored on
the `Environment`; however, this is inaccurate, because:
- Edges in heap dumps imply a keeps-alive relationship, but cleanup
hooks do not keep the `BaseObject`s that they point to alive.
- It loses information about whether `BaseObject` instances are
GC roots: Even weak `BaseObject`s are now, practically speaking,
showing up as hanging off a GC root when that isn’t actually the case
(e.g. in the description of nodejs/node#33468).
Thus, this is a partial revert of f59ec2abee.
Refs: https://github.com/nodejs/node/issues/33468
Refs: https://github.com/nodejs/node/pull/27018
PR-URL: https://github.com/nodejs/node/pull/33809
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Enable the state associated with the individual bindings, e.g. fs or
http2, to be moved out of the Environment class, in order for these
to be more modular and for Environment to be come less of a collection
of random data fields.
Do this by using a BaseObject as the data for callbacks, which can hold
the per-binding state. By default, no per-binding state is available,
although that can be configured when setting up the binding.
PR-URL: https://github.com/nodejs/node/pull/32538
Reviewed-By: James M Snell <jasnell@gmail.com>
Heap dumps can be taken either through the inspector or the public API
for it during an async_hooks init() hook, but at that point the
AsyncWrap in question is not done initializing yet and virtual methods
cannot be called on it.
Address this issue (somewhat hackily) by excluding `AsyncWrap`
instances which have not yet executed their `init()` hook fully
from heap dumps.
Fixes: https://github.com/nodejs/node/issues/28786
PR-URL: https://github.com/nodejs/node/pull/28789
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Rich Trott <rtrott@gmail.com>
Have clearer ownership relations between the `Http2Ping`,
`Http2Settings` and `Http2Session` objects.
Ping and Settings objects are now owned by the `Http2Session`
instance, and deleted along with it, so neither type of object
refers to the session after it is gone.
In the case of `Http2Ping`s, that deletion is slightly delayed,
so we explicitly reset its `session_` property.
Fixes: https://github.com/nodejs/node/issues/28088
PR-URL: https://github.com/nodejs/node/pull/28150
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Ujjwal Sharma <usharma1998@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>
`v8::Global` is essentially a nicer variant of `node::Persistent` that,
in addition to reset-on-destroy, also implements move semantics.
This commit makes the necessary replacements, removes
`node::Persistent` and (now-)unnecessary inclusions of the
`node_persistent.h` header, and makes some of the functions that
take Persistents as arguments more generic so that they work with all
`v8::PersistentBase` flavours.
PR-URL: https://github.com/nodejs/node/pull/27287
Reviewed-By: Gus Caplan <me@gus.host>
Reviewed-By: Eugene Ostroukhov <eostroukhov@google.com>
Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
This allows us to track the essentially-global objects in
Environment in the heap snapshot. Note that this patch only
tracks the fields that can be tracked correctly. There are
still several types of fields that cannot be tracked:
- v8::Data including v8::Private, v8::ObjectTemplate etc.
- Internal types that do not implement MemoryRetainer yet
- STL containers with MemoryRetainer* inside
- STL containers with numeric types inside that should not have their
nodes elided e.g. numeric keys in maps.
The `BaseObject`s are now no longer globals. They are tracked
as arguments in CleanupHookCallbacks referenced by the Environment
node. This model is closer to how their lifetime is managed
internally.
To track the per-environment strong persistent properties, this patch
divides them into those that are also `v8::Value` and those that
are just `v8::Data`. The values can be tracked by the current
memory tracker while the data cannot.
This patch also implements the `MemoryRetainer` interface in several
internal classes so that they can be tracked in the heap snapshot.
PR-URL: https://github.com/nodejs/node/pull/27018
Refs: https://github.com/nodejs/node/issues/26776
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: James M Snell <jasnell@gmail.com>
This patch:
- Refactors the `MemoryRetainer` API so that the impementer no longer
calls `TrackThis()` that sets the size of node on the top of the
stack, which may be hard to understand. Instead now they implements
`SelfSize()` to provide their self sizes. Also documents
the API in the header.
- Refactors `MemoryTracker` so it calls `MemoryInfoName()` and
`SelfSize()` of `MemoryRetainer` to retrieve info about them, and
separate `node_names` and `edge_names` so the edges can be properly
named with reference names and the nodes can be named with class
names. (Previously the nodes are named with reference names while the
edges are all indexed and appear as array elements).
- Adds `SET_MEMORY_INFO_NAME()`, `SET_SELF_SIZE()` and
`SET_NO_MEMORY_INFO()` convenience macros
- Fixes a few `MemoryInfo` calls in some `MemoryRetainers` to track
their references properly.
- Refactors the heapdump tests to check both node names and edge names,
distinguishing between wrapped JS nodes (without prefixes)
and embedder wrappers (prefixed with `Node / `).
PR-URL: https://github.com/nodejs/node/pull/23072
Reviewed-By: Anna Henningsen <anna@addaleax.net>
- Use camel case names for memory retainers inherited from AsyncWrap
instead of their provider names (which are all in upper case)
- Assign class names to wraps so that they appear in the heap snapshot
as nodes with class names as node names. Previously some nodes are
named with reference names, which are supposed to be edge names
instead.
PR-URL: https://github.com/nodejs/node/pull/21939
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Gus Caplan <me@gus.host>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Tobias Nießen <tniessen@tnie.de>
This will enable more detailed heap snapshots based on
a newer V8 API.
This commit itself is not tied to that API and could
be backported.
PR-URL: https://github.com/nodejs/node/pull/21742
Reviewed-By: James M Snell <jasnell@gmail.com>
Clean up after `BaseObject` instances when the `Environment`
is being shut down. This takes care of closing non-libuv resources
like `zlib` instances, which do not require asynchronous shutdown.
Many thanks for Stephen Belanger, Timothy Gu and Alexey Orlenko for
reviewing the original version of this commit in the Ayo.js project.
Refs: https://github.com/ayojs/ayo/pull/88
PR-URL: https://github.com/nodejs/node/pull/19377
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: James M Snell <jasnell@gmail.com>
This commit renames the handle parameter for the BaseObject constructor
to object instead of handle.
The motivation for doing this is that when stepping through an
inheritance chain it can sometimes be a little confusing when
HandleWrap is in involved. HandleWrap has a handle parameter
but calls the object that is passed to AsyncWrap object, but
then when you end up in BaseObject it is named handle.
PR-URL: https://github.com/nodejs/node/pull/20570
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: James M Snell <jasnell@gmail.com>
- Instead of storing a pointer whose type refers to the specific
subclass of `BaseObject`, just store a `BaseObject*` directly.
This means in particular that one can cast to classes along
the way of the inheritance chain without issues, and that
`BaseObject*` no longer needs to be the first superclass
in the case of multiple inheritance.
In particular, this renders hack-y solutions to this problem (like
ddc19be6de) obsolete and addresses
a `TODO` comment of mine.
- Move wrapping/unwrapping methods to the `BaseObject` class.
We use these almost exclusively for `BaseObject`s, and I hope
that this gives a better idea of how (and for what) these are used
in our code.
- Perform initialization/deinitialization of the internal field
in the `BaseObject*` constructor/destructor. This makes the code
a bit more obviously correct, avoids explicit calls for this
in subclass constructors, and in particular allows us to avoid
crash situations when we previously called `ClearWrap()`
during GC.
This also means that we enforce that the object passed to the
`BaseObject` constructor needs to have an internal field.
This is the only reason for the test change.
- Change the signature of `MakeWeak()` to not require a pointer
argument. Previously, this would always have been the same
as `this`, and no other value made sense. Also, the parameter
was something that I personally found somewhat confusing
when becoming familiar with Node’s code.
- Add a `TODO` comment that motivates switching to real inheritance
for the JS types we expose from the native side. This patch
brings us a lot closer to being able to do that.
- Some less significant drive-by cleanup.
Since we *effectively* already store the `BaseObject*` pointer
anyway since ddc19be6de, I do not
think that this is going to have any impact on diagnostic tooling.
Fixes: https://github.com/nodejs/node/issues/18897
PR-URL: https://github.com/nodejs/node/pull/20455
Reviewed-By: Gus Caplan <me@gus.host>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Daniel Bevenius <daniel.bevenius@gmail.com>
Replace v8::Persistent with node::Persistent, a specialization that
resets the persistent handle on destruction. Prevents accidental
resource leaks when forgetting to call .Reset() manually.
I'm fairly confident this commit fixes a number of resource leaks that
have gone undiagnosed so far.
PR-URL: https://github.com/nodejs/node/pull/18656
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com>
Joyee Cheung