Since libuv 1.21.0, pipes on Windows support `writev` on the
libuv side.
This allows for some simplification, and makes the `StreamBase`
API more uniform (multi-buffer `Write()` is always supported now,
including when used by other non-JS consumers like HTTP/2).
PR-URL: https://github.com/nodejs/node/pull/21527
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Don't set `writable` to true when a socket connects if the socket is
already in an ending state.
In the existing code, afterConnect always set `writable` to true. This
has been the case for a long time, but previous to commit
9b7a6914a7, the socket would still be
destroyed by `destroySoon` and emit a `'close'` event. Since that
commit removed this masking behavior, we have relied on maybeDestroy to
destroy the socket when the readble state is ended, and that won't
happen if `writable` is set to true.
If the socket has `allowHalfOpen` set to true, then `destroy` will still
not be called and `'close'` will not be emitted.
PR-URL: https://github.com/nodejs/node/pull/21290
Fixes: https://github.com/nodejs/node/issues/21268
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Refactor writable part (the _write and _writev functions) in net.Socket
and http2.Http2Stream classes.
Also involves adding a generic "WriteGeneric" method to the Http2Stream
class based on net.Socket._writeGeneric, but behind a symbol.
PR-URL: https://github.com/nodejs/node/pull/20643
Reviewed-By: Anatoli Papirovski <apapirovski@mac.com>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
`DuplexBase` was added to prevent the "no-half-open enforcer" from
being inherited by `net.Socket`. The main reason to use it instead
of `Duplex` was that it allowed to not copy the options object but
since commit 5e3f516 the options object is copyed anyway so it is
no longer useful.
PR-URL: https://github.com/nodejs/node/pull/19779
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Daniel Bevenius <daniel.bevenius@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Move tracking of `socket.bytesWritten` to C++ land.
This makes it easier to provide this functionality for all
`StreamBase` instances, and in particular should keep working
when they have been 'consumed' in C++ in some way (e.g. for
the network sockets that are underlying to TLS or HTTP2 streams).
Also, this parallels `socket.bytesRead` a lot more now.
PR-URL: https://github.com/nodejs/node/pull/19551
Reviewed-By: James M Snell <jasnell@gmail.com>
... in addition to the event names they currently use.
Currently, various internal streams have different events that
indicate that the underlying resource has successfully been
established. This commit adds ready event for fs and net
sockets to standardize on emitting ready for all of these streams.
PR-URL: https://github.com/nodejs/node/pull/19408
Fixes: https://github.com/nodejs/node/issues/19304
Reviewed-By: Anna Henningsen <anna@addaleax.net>
This adds a rule that forbids the use of native Error constructors in
the `lib` directory. This is to encourage use of the `internal/errors`
mechanism. The rule is disabled for errors that are not created with
the `internal/errors` module but are still assigned an error code.
PR-URL: https://github.com/nodejs/node/pull/19373
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: James M Snell <jasnell@gmail.com>
Currently the writable side of the socket is closed as soon as `UV_EOF`
is read regardless of the state of the socket. This allows the handle
to be closed before `'end'` is emitted and thus `'close'` can be
emitted before `'end'` if the socket is paused.
This commit prevents the handle from being closed until `'end'` is
emitted ensuring the correct order of events.
PR-URL: https://github.com/nodejs/node/pull/19241
Fixes: https://github.com/nodejs/node/issues/19166
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Benjamin Gruenbaum <benjamingr@gmail.com>
This improves error handling for streams in a few ways.
1. It ensures that no user defined methods (_read, _write, ...) are run
after .destroy has been called.
2. It introduces an explicit error to tell the user if they are write to
write, etc to the stream after it has been destroyed.
3. It makes streams always emit close as the last thing after they have
been destroyed
4. Changes the default _destroy to not gracefully end streams.
It also updates net, http2, zlib and fs to the new error handling.
PR-URL: https://github.com/nodejs/node/pull/18438
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Instead of exposing internals of async_hooks & async_wrap throughout
the code base, create necessary helper methods within the internal
async_hooks that allows easy usage by Node.js internals. This stops
every single internal user of async_hooks from importing a ton of
functions, constants and internal Aliased Buffers from C++ async_wrap.
Adds functions initHooksExist, afterHooksExist, and destroyHooksExist
to determine whether the related emit methods need to be triggered.
Adds clearDefaultTriggerAsyncId and clearAsyncIdStack on the JS side
as an alternative to always calling C++.
Moves async_id_symbol and trigger_async_id_symbol to internal
async_hooks as they are never used in C++.
Renames newUid to newAsyncId for added clarity of its purpose.
Adjusts usage throughout the codebase, as well as in a couple of tests.
PR-URL: https://github.com/nodejs/node/pull/18720
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Ali Ijaz Sheikh <ofrobots@google.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Encapsulate stream requests more:
- `WriteWrap` and `ShutdownWrap` classes are now tailored to the
streams on which they are used. In particular, for most streams
these are now plain `AsyncWrap`s and do not carry the overhead
of unused libuv request data.
- Provide generic `Write()` and `Shutdown()` methods that wrap
around the actual implementations, and make *usage* of streams
easier, rather than implementing; for example, wrap objects
don’t need to be provided by callers anymore.
- Use `EmitAfterWrite()` and `EmitAfterShutdown()` handlers to
call the corresponding JS handlers, rather than always trying
to call them. This makes usage of streams by other C++ code
easier and leaner.
Also fix up some tests that were previously not actually testing
asynchronicity when the comments indicated that they would.
PR-URL: https://github.com/nodejs/node/pull/18676
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Anatoli Papirovski <apapirovski@mac.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
This commit moves error creation helpers scattered around
under lib/ into lib/internal/errors.js in the hope of being clearer
about the differences of errors that we throw into the user land.
- Move util._errnoException and util._exceptionWithHostPort
into internal/errors.js and simplify their logic so it's
clearer what the properties these helpers create.
- Move the errnoException helper in dns.js to internal/errors.js
into internal/errors.js and rename it to dnsException. Simplify
it's logic so it no longer calls errnoException and skips
the unnecessary argument checks.
PR-URL: https://github.com/nodejs/node/pull/18546
Reviewed-By: James M Snell <jasnell@gmail.com>
Removes a few lines of C++ code while making `isIPv4()` about 3x faster.
`isIPv6()` and `isIP()` for the IPv6 case stay about the same.
I removed the homegrown `isIPv4()` in lib/dns.js that utilized a lookup
table. It is in fact a little faster than the new `isIPv4()` function
but:
1. The difference is only measurable at around 10M iterations, and
2. The function is a "probably IPv4" heuristic, not a proper validator.
PR-URL: https://github.com/nodejs/node/pull/18398
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Jan Krems <jan.krems@gmail.com>
Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Hidden via a symbol because I'm unsure exactly what the API should look
like in the end.
Removes the need to use _unrefActive for efficiently refreshing
timeouts.
It still uses it under the hood but that could be replaced with
insert() directly if it were in the same file.
PR-URL: https://github.com/nodejs/node/pull/18065
Reviewed-By: Anatoli Papirovski <apapirovski@mac.com>
This avoids routing writes through the full LibuvStreamWrap
write machinery. In particular, it enables the next commit,
because otherwise the callback passed to `_write()`
would not be called synchronously for pipes on Windows
(because the latter does not support `uv_try_write()`,
even for blocking I/O).
PR-URL: https://github.com/nodejs/node/pull/18019
Reviewed-By: Anatoli Papirovski <apapirovski@mac.com>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
The existing version of defaultTriggerAsyncIdScope creates an Array
for the callback's arguments which is highly inefficient. Instead,
use rest syntax and allow V8 to do that work for us. This yields
roughly 2x performance for this particular function.
PR-URL: https://github.com/nodejs/node/pull/18004
Reviewed-By: Andreas Madsen <amwebdk@gmail.com>
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
This makes `net.Sockets` use actual Timeout objects in a `[kTimeout]`
symbol property, rather than making the socket itself a timer and
appending properties to it directly.
This should make the code generally easier to understand, and might
also prevent some deopts from properties being changes on the socket
itself.
Also moves the Timeout constructor into an internal module.
PR-URL: https://github.com/nodejs/node/pull/17704
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Anatoli Papirovski <apapirovski@mac.com>
Previously the getter would mutate the kDefaultTriggerAsncId value. This
refactor changes the setter to bind the current kDefaultTriggerAsncId to
a scope, such that the getter doesn't have to mutate its own value.
PR-URL: https://github.com/nodejs/node/pull/17273
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Currently, writeQueueSize is never used in C++ and barely used
within JS. Instead of constantly updating the value on the JS
object, create a getter that will retrieve the most up-to-date
value from C++.
For the vast majority of cases though, create a new prop on
Socket.prototype[kLastWriteQueueSize] using a Symbol. Use this
to track the current write size, entirely in JS land.
PR-URL: https://github.com/nodejs/node/pull/17650
Reviewed-By: Anna Henningsen <anna@addaleax.net>
killagu