about summary refs log tree commit homepage
DateCommit message (Collapse)
2013-01-18rename fastopen => kgio_fastopen in Kgio::Socket
In the unlikely case the Ruby Socket class implements its own "fastopen" method, we will avoid conflicting.
2012-12-30accept: do not set blocking if non-blocking is set
This is prone to race conditions in multiprocess situations where one process is relying on non-blocking operation while another (likely newer process) relies on blocking operation. Since the blocking process can always fall back to calling rb_io_wait_readable(), use that instead and give up some scalability for higher reliability. Those interested in avoiding thundering herds will have to stop/start their processes using blocking sockets (and tolerate some downtime).
2012-12-27fastopen: fix argument order in RDoc example
Oops :x
2012-12-27read_write: remove unused variable
2012-12-27implement TCP Fast Open support (client + server)
Server support just requires exposing one constant for setsockopt: Kgio::TCP_FASTOPEN Client support implements a new Kgio::Socket#fastopen method. This new method wraps the the sendto() syscall. With TCP Fast Open, the sendto() syscall is overloaded for stream sockets to implement the functionality of both connect() + write() Since it only makes sense to use _blocking_ I/O for sendto(), TFO clients are only supported in Ruby implementations with native threads.
2012-12-13Kgio::Socket.new retains compatibility with Socket.new
This allows us to create an unconnected socket, just like the normal Socket class it inherits from.
2012-12-12connect: factor out tcp_getaddr() function
This will be reused for TCP fast open support.
2012-12-12connect: split out my_socket() function
This makes the retry logic for mismatched libc headers/kernel versions easier to understand and follow.
2012-11-21tryopen: include errno.h header just in case
errno.h is not guaranteed to be included in existing headers, so we need to #include it to ensure errno and friends are usable. Thanks to stuart on the kgio mailing list for noticing ref: <062571308.133355.1353536890665.JavaMail.sas1@172.29.251.236>
2012-07-11test workaround for platforms with unreliable signals
Ruby may not respond well to signals on all platforms, especially not after fork()-ing in the face of a running pthread (timer thread on 1.9.2). SIGKILL bypasses Ruby (and all userspace) signal handling on Debian GNU/kFreeBSD.
2012-07-11test/lib_read_write: wait for readability before tryread
On FreeBSD, writing to a loopback TCP socket does not guarantee immediate readability on the other end. Tested on Debian GNU/kFreeBSD 6.0
2012-07-11test_poll: skip signal torture on Debian GNU/kfreebsd
This cascades test failures on a platform with questionable signal/fork handling. Tested on: Debian GNU/kFreeBSD 6.0
2012-06-01add `#kgio_writev` and `#kgio_trywritev`
Add methods for using writev(2) syscall for sending array of string in a single syscall. This is more efficient than concatenating strings on Ruby side or sending them one by one. `#kgio_trywritev` returns array of strings which are not sent to the socket. If there were objects other than string, they could be converted using `#to_s` method, but this is not strictly applied, cause `#kgio_*writev` tries to write at most `sysconf(_SC_IOV_MAX)` items at once (for Linux its value is 1024). First string of returned array could be part of string from array, so that you should assume it is not in consistent state. `#kgio_writev` semantic differs a bit from `#kgio_write` in term of buffers mutability: currently `#kgio_write` tries to follow string changes made concurrently, but `#kgio_writev` works with array's lightweight copy. Signed-off-by: Eric Wong <normalperson@yhbt.net>
2012-05-30tryopen: avoid ambiguous name for subst function
Define rb_thread_blocking_region as a macro for MRI 1.8 to prevent confusing output in tools such as valgrind/gdb.
2012-05-30use rb_str_subseq for tail string on write
Use rb_str_subseq for taking string's tail. rb_str_subseq do not allocate additional memory in this case. And although it prevents from collecting original string, it seems that tests wins both in performance and in memory usage. Use fallback to rb_str_substr on ruby1.8 Signed-off-by: Eric Wong <normalperson@yhbt.net>
2012-05-30Fix UnixClientReadServerWrite test class name
Signed-off-by: Eric Wong <normalperson@yhbt.net>
2012-03-24kgio 2.7.4 - small fixes and cleanups v2.7.4
Fix build for platforms lacking both TCP_CORK _and_ TCP_NOPUSH There are many test case fixes and cleanups, too.
2012-03-24test: more workaround for FreeBSD 9.0
Followup-to: e26358413c9d87e1ce8f6cda5cf0b8dd53979ed2
2012-03-23test/lib_read_write: test workarounds for TCP in FreeBSD 9.0
Under load, TCP sockets may not register as readable right away after the writer finishes. This can be expected for implementations where loopback TCP is a closer simulation of non-local TCP traffic. These test failures were noticed under FreeBSD 9.0.
2012-03-23test_tryopen: fix horribly-named test for EACCES
We can't actually test for EPERM without changing permissions/ownership, and we can't do that without root...
2012-03-22test/lib_read_write: increase test reliability
IO#readpartial may not drain the socket buffers enough for kgio_write to succeed on some platforms. So use IO#read for read-in-full behavior.
2012-03-19test/*: remove assert_nothing_raised
It makes test failures hard to track down, tests will already fail if exceptions are thrown and we'll get nice backtraces.
2012-03-19fix for non-Linux, non-TCP_NOPUSH platforms
We don't need to care for TCP_NOPUSH in read_write.c, it's entirely in autopush.c and no-op on platforms without TCP_CORK/TCP_NOPUSH. TCP_CORK/TCP_NOPUSH are non-POSIX, so it's entirely possible some Free systems will lack them. Reported-by: Edho Arief <edho@myconan.net>
2012-03-15kgio 2.7.3 - compatibility fixes v2.7.3
Fixed build and autopush support under Debian GNU/kFreeBSD. Test case fixes for timing-sensitive tests.
2012-03-07HACKING: add instructions for running tests
Unlike most Rubyists, I prefer GNU make to Rake.
2012-03-07test_poll: workaround for timing-sensitive test on slow hosts
poll(2) may return successfully before it gets interrupted by the signal. Found and fix confirmed by 375gnu on the kgio mailing list. ref: <CAAB-Kcm=_CRa4UoSQt+C4cHk6z2Rpfsv6_KXPHV3R34Gt6sLiQ@mail.gmail.com>
2012-03-05test: increase delta range for timing-sensitive test
This appears to be needed for Debian GNU/kFreeBSD under KVM.
2012-03-05autopush: fix/enable under Debian GNU/kFreeBSD
It seems autopush support in our autopush code has always been broken outside of Linux-based systems, as we never marked the socket as having pending data.
2012-03-05accept4: require SOCK_NONBLOCK/SOCK_CLOEXEC macros
The check for the accept4() function actually succeeds on a stock installation of Debian GNU/kFreeBSD 6.0, but the eglibc headers fail to define the necessary flags.
2012-01-08kgio 2.7.2 - for older, and older Rubies v2.7.2
Fix a missing #include for Ruby 1.8.5 users. No need to upgrade to this (nor 2.7.1) if you're on a modern version of Ruby.
2012-01-08tryopen: remember to include ancient_ruby.h for 1.8.5
Apparently the old Ruby 1.8.6 installation lying around isn't old enough.
2012-01-08kgio 2.7.1 - compatibility with older Rubies v2.7.1
This release fixes some compatibility issues with people stuck on older versions of Ruby/RubyGems. * define RARRAY_PTR/RARRAY_LEN macros for Ruby 1.8.6 * test/test_autopush: skip strace tests if not available * gemspec: disable development dependencies for old systems
2012-01-08gemspec: disable development dependencies for old systems
"Enterprise" users are sometimes stuck on older Rubies/RubyGems and this is still required for them.
2012-01-08test/test_autopush: skip strace tests if not available
No need to completely fail on a test.
2012-01-07define RARRAY_PTR/RARRAY_LEN macros for Ruby 1.8.6
Apparently Ruby 1.8.6 is still in use...
2011-12-13kgio 2.7.0 - minor updates v2.7.0
When running under Ruby trunk/2.0.0dev, all IO objects created by kgio will be close-on-exec by default to match the (future) 2.0.0 behavior. accept()ed sockets in kgio have always been close-on-exec by default.. Singleton Kgio.accept_* methods are deprecated as the kgio_accept/kgio_tryaccept methods all take an additional flags argument. There are various, test, documentation, and error message improvements.
2011-11-17doc: update documentation regarding accept method flags
There's no reason for SOCK_NONBLOCK with Ruby, and SOCK_CLOEXEC has always been on-by-default with accept() wrappers.
2011-11-17connect,tryopen: set close-on-exec flag for new fds on Ruby 2.0+
All IO objects created by Kgio will have FD_CLOEXEC descriptor flag set on it when run under Ruby 2.0.0dev. This matches the upcoming behavior of Ruby 2.0.0dev for IO objects in the core and standard library. This change does not affect users on Ruby 1.9.3 and earlier. accept()-ed sockets in kgio have _always_ had FD_CLOEXEC set by default.
2011-11-15tests: remove tests for IO#nonblock? after accept
There's no point in testing a Ruby implementation detail and these tests fail under OpenBSD where the accept()-ed socket inherits the O_NONBLOCK flag from the parent.
2011-11-15accept: deprecate singleton Kgio.accept_* methods
The kgio_accept and kgio_tryaccept methods now take an additional flags argument, so there's no reason to set global flags anywhere.
2011-11-15accept: always set O_NONBLOCK for accept()-ed sockets in 1.8
This is mostly an implementation detail, but it's already true on OpenBSD (and maybe other BSDs), and also requires no additional syscalls on newer Linux systems.
2011-08-29test_poll: test for closing a polled IO in sighandler
This needs to work similarly to IO.select.
2011-08-24.wrongdoc.yml: add public/private email addresses
We want feedback!
2011-08-24wait.c: fix misspelling in rdoc
eye kan spel!
2011-08-09LICENSE: s/GNU C Library/kgio/
This is not glibc and I'm not Ulrich Drepper.
2011-08-09TODO: update SSL/TLS support status :)
Monkeys!
2011-07-21use rb_update_max_fd() under MRI 1.9.3+
This helps exec() and similar methods close descriptors on shutdown.
2011-07-15connect: more descriptive error for TCP port
We want more descriptive error messages and don't want crazy stuff like floats.
2011-07-14kgio 2.6.0 - minor feature update and cleanups v2.6.0
We now export SOCK_NONBLOCK, SOCK_CLOEXEC constants in the Kgio namespace to make kgio_tryaccept/kgio_accept easier-to-use. There are also some minor internal cleanups.
2011-07-14add prototype for rb_thread_blocking_io_region()
It's no in the public headers, but Ruby 1.9.3 will have it and it's still superior to rb_thread_blocking_region() even though it's not ideal.