Building the error message from a NULL pointer raised ArgumentError instead of IO::TimeoutError.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Without pthread, mswin fell back to GETADDRINFO_IMPL 1, which ignores the timeout of Addrinfo.getaddrinfo and lets neither Timeout, Thread#kill nor Thread#raise stop the caller until getaddrinfo(3) returns. Start the resolver thread with _beginthreadex(3) so mswin takes GETADDRINFO_IMPL 2 as the pthread platforms do. MinGW stays on GETADDRINFO_IMPL 1.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
No caller reads the pthread_t, so raddrinfo_thread_create() no longer takes one, and a non-pthread implementation can sit behind it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An invalid open_timeout raised before the waiter was assigned, so the cleanup closed fd 0 left by memset and took the process's stdin with it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Winsock takes the length as int, so a String of 2GiB or more had its length truncated. The caller now sees a short write, as BasicSocket#send already reports one.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ADDRESS_FAMILIES[:family_name] used the Symbol :family_name as the key
instead of the local variable family_name, so the address family was
always nil. This is a follow-up to b896f80598 [Bug #21512].
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A getaddrinfo thread calls getaddrinfo(3) inside rb_thread_prevent_fork,
but on glibc its thread exit takes the resolver configuration lock again
after the thread has released the fork lock. A fork that lands then
leaves the child with that lock held, and every later name lookup in the
child blocks forever. getnameinfo(3) attaches the same resolver state,
even for a name in /etc/hosts, and its thread ran outside
rb_thread_prevent_fork altogether.
Make the lookup threads joinable and have fork join any that may still
be exiting. As a lookup thread finishes, still inside
rb_thread_prevent_fork, it swaps itself into a single slot with an
atomic exchange and joins the thread it displaced, which has usually
exited by then, and rb_thread_acquire_fork_lock joins whatever is left
in the slot. Once lookups settle, at most one finished thread is left
unjoined. The slot holds pthread_t as a pointer, with NULL for empty,
which assumes pthread_t is scalar and never 0.
getnameinfo threads now run within rb_thread_prevent_fork and register
as they exit, as getaddrinfo threads do, and where getnameinfo runs
without a helper thread it is called within rb_thread_prevent_fork too.
So fork also waits for a reverse lookup in progress, as it already did
for a forward one. raddrinfo_pthread_create now always creates joinable
threads, which drops its pthread_detach fallback for platforms without
pthread_attr_setdetachstate.
The slot is kept across fork, so every fork path that keeps running
Ruby code has to take the fork lock. On the BSDs, Process.daemon forks
with daemon(3) and did not take it, so the daemon would inherit the
registered thread, and its next fork would abort in pthread_join with
ESRCH. Take and release the fork lock around daemon(3) as rb_fork_ruby
does. daemon(3) can fail in the child as well as in the parent, so
compare pids to tell which process returned.
[Bug #22406]
hostname is added as a suffix to getaddrinfo error messages. for
example, a "name or service not known" error message becomes "name or
service not known for `invalidthost.com`.
The lengths passed to recv(2), send(2) and gethostname(2), and the lengths of sockaddr, are left as they are.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Debian GNU/kFreeBSD left the official releases after Debian 7 and was removed from debian-ports in 2023. The k*bsd*-gnu pattern also matched GNU/kNetBSD, which was abandoned long before that.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
EAI_ADDRFAMILY left the IPv6 result looking successful, so a later IPv4 resolution error was not recorded and TCPSocket.new raised Socket::ResolutionError with error_code 0. When the IPv6 notification failed, EAI_ADDRFAMILY was recorded instead and hid the IPv4 error.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The errno of the failed write(2) overwrote err, so the main thread took it for an EAI code, discarding addresses that had been resolved or raising with an unrelated message.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The loop only left on EAGAIN, so EOF or a read error other than EINTR spun forever while holding the GVL.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
When stdin is closed, a socket can get fd 0, and nfds stayed 0 if it was the only fd to wait for, so select never reported the connection.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The fd was closed but left in connection_attempt_fds, so it was passed to the next select and closed again in the cleanup, possibly after the number had been reused.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
remove_connection_attempt_fd shifts the remaining fds down, so advancing the index skipped the fd moved into the current slot. A connection that had already succeeded could go unnoticed until the next select, and a new attempt was started meanwhile.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The main thread polls has_syserr on every loop while the resolver thread may be setting it, so read it with the lock the resolver thread holds. Once it is seen set, err and ai were written before it and can be read as before.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The resolver thread checks notify and writes to it while holding the lock. Closing it outside the lock let the thread write a byte to an unrelated file that had reused the fd number.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
When a socket has buffered output (`sync == false`) and `#close_write` is
called, the write side was shut down via `shutdown(SHUT_WR)` without first
flushing the buffer. This had two bad consequences:
- the buffered bytes were silently dropped (never delivered to the peer);
- a subsequent `#close` tried to flush the still-populated write buffer
into the now `shutdown(SHUT_WR)` socket and raised `Errno::EPIPE`.
Flush the buffer before shutting down the write side, matching the
write-only branch (which flushes via `rb_io_close`) and the non-socket
`IO#close_write` path. `IO#close_write` for a plain `IO` wrapping a socket
fd is fixed in the same way.
Assisted-By: devx/2c064143-8655-4631-a7cd-6395bb241bdc
Only test_mode_settings itself was checked to be a Hash, and its delay and error values were passed to rb_hash_aref as they were.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The delay in `test_mode_settings` was stored as a raw VALUE, so a delay of n milliseconds slept for 2n+1.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Winsock reports a failed non-blocking connect(2) only through the
exceptfds of select(2), which IO#wait_writable never asks for. So
Addrinfo#connect_internal kept waiting on a socket whose connection had
already been refused, and burned the whole connect_timeout before
raising Errno::ETIMEDOUT instead of the real error.
Ask for the priority event on Windows alone. The M:N thread scheduler
refuses any wait carrying another event, so requesting it where it
cannot fire would take the wait off that scheduler for nothing.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Darwin 27 answers the retry connect(2) on a refused nonblocking socket
with EISCONN, so the retry idiom in Addrinfo#connect_internal returned
an unconnected socket. SO_ERROR still holds the real error, so consult
it after wait_writable, as wait_connectable() in init.c already does.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Replace `tool/update-deps` with a Ruby source scanner so dependency
updates no longer require GNU make, a configured build, or compiler
preprocessor output.
Keep compact source mappings and scanner declarations in dependency
files, and expand complete rules into the build directory. Avoid the
temporary fixture headers required by the compiler-driven prototype.
Use the generated rules from GNU make, BSD make, and NMake builds.
Support out-of-tree and read-only source trees, and retain expanded
dependencies shipped in release archives when baseruby is unavailable.
IO#wait_writable is defined in core since Ruby 3.2, so this guard
never requires io/wait when running on the ruby built from this tree.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Set RUBY_TYPED_THREAD_SAFE_FREE on TypedData types whose dfree function
is trivially safe, and only frees its own memory
Types not yet marked as THREAD_SAFE_FREE:
- id2ref_tbl_type: sets the process-global id2ref_tbl to NULL.
- mutex_data_type: unlinks itself from another thread's keeping_mutexes list.
- autoload_data_type: deletes nodes from a shared intrusive list.
- autoload_const_type: deletes itself from a shared list.
- rb_cont_data_type: mutates the shared fiber pool and the global first_jit_cont list.
- rb_fiber_data_type: delegates to cont_free.
- FiberPoolDataType: mutates shared fiber pool free-lists.
- ractor_data_type: reconfigures global VM event-hook flags and call caches.
- exported_object_registry: frees a global table under the VM lock.
- rb_box_data_type (Box::Entry): unlinks classext from other live class and module objects.
- box_ext_cleanup_type: dereferences another String object and calls unlink.
- monitor_data_type: uses the default free, so the flag has no effect.
Co-authored-by: Luke Gruber <luke.gruber@shopify.com>
`File.join` is a hotspot for common libraries such as Zeitwerk
and Bootsnap. It has a fairly flexible signature, but 99% of
the time it's called with just two (or a small number of) UTF-8 strings.
If we optimistically optimize for that use case we can cut down a large
number of type and encoding checks, significantly speeding up the method.
The one remaining expensive check we could try to optimize is `str_null_check`.
Given it's common to use the same base string for joining, we could memoize it.
Also we could precompute it for literal strings.
```
compare-ruby: ruby 4.1.0dev (2026-01-17T14:40:03Z master 00a3b71eaf) +PRISM [arm64-darwin25]
built-ruby: ruby 4.1.0dev (2026-01-18T12:10:38Z spedup-file-join 069bab58d4) +PRISM [arm64-darwin25]
warming up....
| |compare-ruby|built-ruby|
|:-------------|-----------:|---------:|
|two_strings | 2.475M| 9.444M|
| | -| 3.82x|
|many_strings | 551.975k| 2.346M|
| | -| 4.25x|
|array | 514.946k| 522.034k|
| | -| 1.01x|
|mixed | 621.236k| 633.189k|
| | -| 1.02x|
```
`fast fallback` cannot be used with explicitly specified local port,
because concurrent binds to the same `local_host:local_port`
can raise `Errno::EADDRINUSE`.
This issue is more likely to occur on hosts with `IPV6_V6ONLY` disabled,
because IPv6 binds can also occupy IPv4-mapped IPv6 address space.
This change fixes a bug in which specifying 0 for timeout-related options (such as the `timeout` option of `Addrinfo.getaddrinfo`) incorrectly results in an infinite wait.
(This change overwrites https://github.com/ruby/ruby/pull/15626 .)
This change adds host information to the error messages shown when a timeout occurs while passing timeout options to `TCPSocket.new` or `Socket.tcp`, for improved usability.
(When the `fast_fallback option` is enabled, there may be multiple possible destinations, so the host name is shown instead of an IP address.)
As part of this change, the error messages in `Addrinfo.getaddrinfo` and `Addrinfo#connect_internal`, both of which are used by `Socket.tcp`, have also been improved in the same way.
* `Socket.tcp` and `TCPSocket.new` raises `IO::TiemoutError` with user specified timeout
In https://github.com/ruby/ruby/pull/11880, `rsock_connect()` was changed to raise `IO::TimeoutError` when a user-specified timeout occurs.
However, when `TCPSocket.new` attempts to connect to multiple destinations, it does not use `rsock_connect()`, and instead raises `Errno::ETIMEDOUT` on timeout.
As a result, the exception class raised on timeout could differ depending on whether there were multiple destinations or not.
To align this behavior with the implementation of `rsock_connect()`, this change makes `TCPSocket.new` raise `IO::TimeoutError` when a user-specified timeout occurs.
Similarly, `Socket.tcp` is updated to raise `IO::TimeoutError` when a timeout occurs within the method.
(Note that the existing behavior of `Addrinfo#connect_internal`, which Socket.tcp depends on internally and which raises `Errno::ETIMEDOUT` on timeout, is not changed.)
* [ruby/net-http] Raise `Net::OpenTimeout` when `TCPSocket.open` raises `IO::TimeoutError`.
With the changes in https://github.com/ruby/ruby/pull/15602, `TCPSocket.open` now raises `IO::TimeoutError` when a user-specified timeout occurs.
This change updates #connect to handle this case accordingly.
https://github.com/ruby/net-http/commit/f64109e1cf
This change updates the behavior so that, when there is only a single destination and `open_timeout` is specified, the remaining `open_timeout` duration is used as the connection timeout.
This change fixes a bug where, with `Socket.tcp`’s `fast_fallback option` disabled, specifying `open_timeout` could unintentionally pass a negative value to `Addrinfo#connect_internal`, `causing an ArgumentError`.
```
❯ ruby -rsocket -e 'p Socket.tcp("localhost", 9292, open_timeout: 1, fast_fallback: false)'
/Users/misaki-shioi/src/install/lib/ruby/4.0.0+0/socket.rb:64:in 'IO#wait_writable': time interval must not be negative (ArgumentError)
sock.wait_writable(timeout) or
^^^^^^^
from /Users/misaki-shioi/src/install/lib/ruby/4.0.0+0/socket.rb:64:in 'Addrinfo#connect_internal'
from /Users/misaki-shioi/src/install/lib/ruby/4.0.0+0/socket.rb:141:in 'Addrinfo#connect'
from /Users/misaki-shioi/src/install/lib/ruby/4.0.0+0/socket.rb:964:in 'block in Socket.tcp_without_fast_fallback'
from /Users/misaki-shioi/src/install/lib/ruby/4.0.0+0/socket.rb:231:in 'Array#each'
from /Users/misaki-shioi/src/install/lib/ruby/4.0.0+0/socket.rb:231:in 'Addrinfo.foreach'
from /Users/misaki-shioi/src/install/lib/ruby/4.0.0+0/socket.rb:945:in 'Socket.tcp_without_fast_fallback'
from /Users/misaki-shioi/src/install/lib/ruby/4.0.0+0/socket.rb:671:in 'Socket.tcp'
from -e:1:in '<main>'
```
Socket.tcp launches ruby threads to resolve hostnames, and those threads
communicate through a queue implemented with `IO.pipe`. When the thread
that called `Socket.tcp` is killed, the resolver threads still try to
communicate through the pipe even though it may be closed. The method
`Socket.tcp_with_fast_fallback` tries to deal with this by killing the
threads in an ensure block, and then closing the pipe. However, calling
`Thread#kill` is not a blocking operation, it only sets a flag on the
thread telling it to raise during the next interrupt. The thread needs
to be joined to ensure it is terminated. The following script
demonstrates the issue:
```ruby
require "socket"
ts = []
5.times do
ts << Thread.new do
loop do
1_000.times do |i|
puts "#{i}"
t = Thread.new do
Socket.tcp("ruby-lang.org", 80)
end
sleep 0.001
t.kill
end
end
end
end
ts.each(&:join)
```
output:
```
/Users/luke/workspace/ruby-dev/ruby-build-debug/.ext/common/socket.rb:1019:in 'IO#write': closed stream (IOError)
from /Users/luke/workspace/ruby-dev/ruby-build-debug/.ext/common/socket.rb:1019:in 'IO#putc'
from /Users/luke/workspace/ruby-dev/ruby-build-debug/.ext/common/socket.rb:1019:in 'block in Socket::HostnameResolutionResult#add'
from /Users/luke/workspace/ruby-dev/ruby-build-debug/.ext/common/socket.rb:1017:in 'Thread::Mutex#synchronize'
from /Users/luke/workspace/ruby-dev/ruby-build-debug/.ext/common/socket.rb:1017:in 'Socket::HostnameResolutionResult#add'
from /Users/luke/workspace/ruby-dev/ruby-build-debug/.ext/common/socket.rb:980:in 'Socket.resolve_hostname'
from /Users/luke/workspace/ruby-dev/ruby-build-debug/.ext/common/socket.rb:719:in 'block (2 levels) in Socket.tcp_with_fast_fallback'
/Users/luke/workspace/ruby-dev/ruby-build-debug/.ext/common/socket.rb:1019:in 'IO#write': closed stream (IOError)
from /Users/luke/workspace/ruby-dev/ruby-build-debug/.ext/common/socket.rb:1019:in 'IO#putc'
from /Users/luke/workspace/ruby-dev/ruby-build-debug/.ext/common/socket.rb:1019:in 'block in Socket::HostnameResolutionResult#add'
from /Users/luke/workspace/ruby-dev/ruby-build-debug/.ext/common/socket.rb:1017:in 'Thread::Mutex#synchronize'
from /Users/luke/workspace/ruby-dev/ruby-build-debug/.ext/common/socket.rb:1017:in 'Socket::HostnameResolutionResult#add'
from /Users/luke/workspace/ruby-dev/ruby-build-debug/.ext/common/socket.rb:978:in 'Socket.resolve_hostname'
from /Users/luke/workspace/ruby-dev/ruby-build-debug/.ext/common/socket.rb:719:in 'block (2 levels) in Socket.tcp_with_fast_fallback'
```
