This both save time for when it will be eventually needed,
and avoid mutating heap pages after a potential fork.
Instrumenting some large Rails app, I've witnessed up to
58% of String instances having their coderange still unknown.
[Feature #18885]
For now, the optimizations performed are:
- Run a major GC
- Compact the heap
- Promote all surviving objects to oldgen
Other optimizations may follow.
* Add rb_io_path and rb_io_open_descriptor.
* Use rb_io_open_descriptor to create PTY objects
* Rename FMODE_PREP -> FMODE_EXTERNAL and expose it
FMODE_PREP I believe refers to the concept of a "pre-prepared" file, but
FMODE_EXTERNAL is clearer about what the file descriptor represents and
aligns with language in the IO::Buffer module.
* Ensure that rb_io_open_descriptor closes the FD if it fails
If FMODE_EXTERNAL is not set, then it's guaranteed that Ruby will be
responsible for closing your file, eventually, if you pass it to
rb_io_open_descriptor, even if it raises an exception.
* Rename IS_EXTERNAL_FD -> RUBY_IO_EXTERNAL_P
* Expose `rb_io_closed_p`.
* Add `rb_io_mode` to get IO mode.
---------
Co-authored-by: KJ Tsanaktsidis <ktsanaktsidis@zendesk.com>
[Feature #18885]
For now, the optimizations performed are:
- Run a major GC
- Compact the heap
- Promote all surviving objects to oldgen
Other optimizations may follow.
[Feature #19443]
Until recently most libc would cache `getpid()` so this was a cheap check to make.
However as of glibc version 2.25 the PID cache is removed and calls to getpid() always
invoke the actual system call which significantly degrades the performance of existing applications.
The reason glibc removed the cache is that some libraries were bypassing fork(2)
by issuing system calls themselves, causing stale cache issues.
That isn't a concern for Ruby as bypassing MRI's primitive for forking would
render the VM unusable, so we can safely cache the PID.
rb_pid_t is 32 bits on some platforms, which will cause a warning on GCC
due to POSFIXABLE always returning true.
include/ruby/internal/arithmetic/fixnum.h:43:31: warning: comparison is always true due to limited range of data type [-Wtype-limits]
[Feature #19443]
It's not uncommon for database client and similar network libraries
to protect themselves from Process.fork by regularly checking Process.pid
Until recently most libc would cache `getpid()` so this was a cheap
check to make.
However as of glibc version 2.25 the PID cache is removed and calls to
`getpid()` always invoke the actual system call which significantly degrades
the performance of existing applications.
The reason glibc removed the cache is that some libraries were bypassing
`fork(2)` by issuing system calls themselves, causing stale cache issues.
That isn't a concern for Ruby as bypassing MRI's primitive for forking
would render the VM unusable, so we can safely cache the PID.
* Revert "Remove special handling of `SIGCHLD`. (#7482)"
This reverts commit 44a0711eab.
* Revert "Remove prototypes for functions that are no longer used. (#7497)"
This reverts commit 4dce12bead.
* Revert "Remove SIGCHLD `waidpid`. (#7476)"
This reverts commit 1658e7d966.
* Fix change to rjit variable name.
The last part of the sentence was accidentally put in enumeration, It
made an impression that it's one of the rules, not the result of
applying the rules.
Replacing `...` with `*pids` seems to clarify the expected variadic arguments.
Note that the expected arguments are two or more with a signal and pids.
That is, the method must have at least one pid, which cannot be omitted:
```console
% ruby -e 'Process.kill(0)'
-e:1:in `kill': wrong number of arguments (given 1, expected 2+) (ArgumentError)
from -e:1:in `<main>'
```
The forked child process is a grandchild process from the viewpoint of
the process which invoked the caller process. That means the child is
detached at that point, and it does not need to fork twice.
First, rb_mjit_fork should call rb_thread_atfork to stop threads after
fork in the child process. Unfortunately, we cannot use rb_fork_ruby to
prevent this kind of mistakes because MJIT needs special handling of
waiting_pid and mjit_pause/resume.
Second, mjit_waitpid_finished should be checked regardless of
trap_interrupt. It doesn't seem like the flag is not set when SIGCHLD is
handled for an MJIT child process.
The argument of `rb_syswait` is now `rb_pid_t` which may differ from
`int`. Also it is an undefined behavior to take the result of casted
void function (in `rb_protect`).
As discussed in [Bug #18911], I'm adding some documentation to
`Process._fork` to clarify that it is not expected to cover
calls to `Process.daemon`.
[Bug #18911]: https://bugs.ruby-lang.org/issues/18911
Co-authored-by: Yusuke Endoh <mame@ruby-lang.org>
Burdette Lamar
Nobuyoshi Nakada
Jean Boussier
Samuel Williams
NARUSE, Yui
Daniel DeLorme
Aaron Patterson
Peter Zhu
Takashi Kokubun
zverok
Maciek Rząsa
S-H-GAMELINKS
Hiroshi SHIBATA
David Carlier
Koichi ITO
Ivo Anjo