When --jit-wait is specified, `unit_queue.length` is always 0 and it's
not a good metric.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64097 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
When all compilation finishes or the number of JIT-ed code reaches
--jit-max-cache, this compacts all generated code to a single .so file
and re-loads all methods from it.
In the future, it may trigger compaction more frequently and/or limit
the maximum times of compaction to prevent unlimited memory usage.
So the current behavior is experimental, but at least the performance
improvement in this commit won't be removed.
=== Benchmark ===
In this benchmark, I'll compare following four conditions:
* trunk: r64082
* trunk JIT: r64082 w/ --jit
* single-so JIT: This commit w/ --jit
* objfcn JIT: This branch https://github.com/k0kubun/ruby/tree/objfcn w/ --jit,
which is shinh's objfcn https://github.com/shinh/ruby/tree/objfcn rebased from this commit
```
$ uname -a
Linux bionic 4.15.0-29-generic #31-Ubuntu SMP Tue Jul 17 15:39:52 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
```
* Micro benchmark
Using this script https://gist.github.com/k0kubun/10e6d3387c9ab1b134622b2c9d76ef51,
calls some amount of different methods that just return `nil`. The following tables
are its average duration seconds of 3 measurements.
Smaller is better.
** 1 method (seconds)
| | trunk | trunk JIT | single-so JIT | objfcn JIT |
|:------|:------------------|:------------------|:------------------|:------------------|
| Time | 5.576067774333296 | 5.915551971666446 | 5.833641665666619 | 5.845915191666639 |
| Ratio | 1.00x | 1.06x | 1.05x | 1.05x |
** 50 methods (seconds)
| | trunk | trunk JIT | single-so JIT | objfcn JIT |
|:------|:------------------|:------------------|:------------------|:------------------|
| Time | 3.1661167996666677| 6.125825928333342 | 4.135432743666665 | 3.750358728333348 |
| Ratio | 1.00x | 1.93x | 1.31x | 1.18x |
** 1500 methods (seconds)
| | trunk | trunk JIT | single-so JIT | objfcn JIT |
|:------|:------------------|:------------------|:------------------|:------------------|
| Time | 5.971650823666664 | 19.579182102999994| 10.511108153999961| 10.854653588999932|
| Ratio | 1.00x | 3.28x | 1.76x | 1.82x |
* Discourse
Using the same benchmark strategy as https://bugs.ruby-lang.org/issues/14490 with
this branch https://github.com/k0kubun/discourse/commits/benchmark2 forked from discourse
v1.8.11 to support running trunk.
1. Run ruby script/bench.rb to warm up profiling database
2. Run RUBYOPT='--jit-verbose=1 --jit-max-cache=10000' RAILS_ENV=profile bin/puma -e production
3. WAIT 5-15 or so minutes for all jitting to stop so we have no cross talk
4. Run ab -n 100 http://localhost:9292/
5. Wait for all new jitting to finish
6. Run ab -n 100 http://localhost:9292/
** Response time (ms)
Here is the response time milliseconds for each percentile.
Skipping 99%ile because it's the same as 100%ile in 100 calls.
| | trunk| trunk|single|objfcn|
| | | JIT|so JIT| JIT|
|:----|:-----|:-----|:-----|:-----|
| 50% | 38 | 45 | 41 | 43 |
| 66% | 39 | 50 | 44 | 44 |
| 75% | 47 | 51 | 46 | 45 |
| 80% | 49 | 52 | 47 | 47 |
| 90% | 50 | 63 | 50 | 52 |
| 95% | 60 | 79 | 52 | 55 |
| 98% | 91 | 114 | 91 | 91 |
|100% | 97 | 133 | 96 | 99 |
** Ratio (smaller is better)
Here is the response time increase ratio against no-JIT trunk's one.
| | trunk| trunk|single|objfcn|
| | | JIT|so JIT| JIT|
|:----|:-----|:-----|:-----|:-----|
| 50% | 1.00x| 1.18x| 1.08x| 1.13x|
| 66% | 1.00x| 1.28x| 1.13x| 1.13x|
| 75% | 1.00x| 1.09x| 0.98x| 0.96x|
| 80% | 1.00x| 1.06x| 0.96x| 0.96x|
| 90% | 1.00x| 1.26x| 1.00x| 1.04x|
| 95% | 1.00x| 1.32x| 0.87x| 0.92x|
| 98% | 1.00x| 1.25x| 1.00x| 1.00x|
|100% | 1.00x| 1.37x| 0.99x| 1.02x|
While 50 and 60 %ile are still worse than no-JIT trunk, 75, 80, 90, 95,
98 and 100% are not slower than that.
So now it's a little harder to say "MJIT slows down Rails applications".
Probably I can close [Bug #14490] now. Let's start improving it.
Close https://github.com/ruby/ruby/pull/1921
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64094 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* lib/unicode_normalize/normalize.rb: Fix the range check for trailing
Hangul jamo characters in Unicode normalization. Different from
leading or vowel jamos, where LBASE and VBASE are actual characters,
a value equal to TBASE expresses the absence of a trailing jamo.
This fix is technically correct, but there was no bug because
the regular expressions in lib/unicode_normalize/tables.rb
eliminate jamos equal to TBASE from normalization processing.
* test/test_unicode_normalize.rb: Add preventive test
test_no_trailing_jamo based on
d134809cd3
just for the case we ever get a regression.
This closes issue #14934, thanks to MaLin (Lin Ma) for reporting.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64087 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* test/test_unicode_normalize.rb: extract tests that do not depend
on NormalizationTest.txt data file from conditionally constructed
part of TestUnicodeNormalize class, to always run them even if
the data file isn't found.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64086 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* test/test_unicode_normalize.rb: search ucd directory
first if it exists. This change follows r61415.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64085 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
and lazily delete them on termination.
This will be needed to create a large so file later.
The large number of .o files will be probably compacted before the large so
file is created.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64077 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* ruby.c (external_str_new_cstr): strings come from the external
should be tainted. [ruby-dev:50596] [Bug #14941]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64071 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
This adds a password_hash keyword argument to
WEBrick::HTTPAuth::Htpasswd#initialize. If set to :bcrypt, it
will create bcrypt hashes instead of crypt hashes, and will
raise an exception if the .htpasswd file uses crypt hashes.
If :bcrypt is used, then instead of calling
BasicAuth.make_passwd (which uses crypt),
WEBrick::HTTPAuth::Htpasswd#set_passwd will set the bcrypt
password directly. It isn't possible to change the
make_passwd API to accept the password hash format, as that
would break configurations who use Htpasswd#auth_type= to set
a custom auth_type.
This modifies WEBrick::HTTPAuth::BasicAuth to handle checking
both crypt and bcrypt hashes.
There are commented out requires for 'string/crypt', to handle
when String#crypt is deprecated and the undeprecated version is
moved to a gem.
There is also a commented out warning for the case when
the password_hash keyword is not specified and 'string/crypt'
cannot be required. I think the warning makes sense to nudge
users to using bcrypt.
I've updated the tests to test nil, :crypt, and :bcrypt values
for the password_hash keyword, skipping the bcrypt tests if the
bcrypt library cannot be required.
[ruby-core:88111] [Feature #14940]
From: Jeremy Evans <code@jeremyevans.net>
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64060 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
in a short term, and add retries to prevent random CI failures by it.
I remember this and will address it later for sure.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64032 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* test/ruby/test_io.rb (test_select_leak): increase timeout seconds
to pass this test on a high-load machine.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64019 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* object.c (rb_f_integer): prefer `base` optional argument over
keyword arguments. this issue should be resolved more generally
by separating keyword arguments from hashes in the future.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64015 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
with --jit.
`./ruby --jit -Itest/lib test/ruby/test_rubyoptions.rb -n "TestRubyOptions#test_verbose"`
has failed without -DMJIT_FORCE_ENABLE like:
1) Failure:
TestRubyOptions#test_verbose [/home/travis/build/k0kubun/mjit-test/test/ruby/test_rubyoptions.rb:118]:
<"ruby 2.6.0dev (2018-07-21 trunk 64010) +JIT [x86_64-linux]"> expected but was
<"ruby 2.6.0dev (2018-07-21 trunk 64010) [x86_64-linux]">.
but the test should pass even without -DMJIT_FORCE_ENABLE.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64011 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
When do_select is interrupted and raise happens from
RUBY_VM_CHECK_INTS_BLOCKING, the original FD sets we copied
do not get freed, leading to a memory leak. Wrap up all the
FD sets into a Ruby object to ensure the GC can release an
allocations made for rb_fdset_t.
This leak existed since Ruby 2.0.0 (r36430)
[Bug #14929]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64007 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Spurious wakeup is unavoidable with MJIT; and any real
code must be able to deal with spurious wakeup anyways.
[ruby-core:87882] https://bugs.ruby-lang.org/issues/14901
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64005 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* doc/*: Move entry to default gems category from standard library.
* lib/shell/*: Added `Shell::VERSION` and re-used it with @RELEASE_VERSION.
* test/shell/test_command_processor.rb: added missing require for test library.
* tool/sync_default_gems.rb: Support shell library. We need to ignore shellwords.rb when
syncing shell* files.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64004 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
by promoting jit to feature flag.
mjit.h: update comment about mjit_opts.on
test_rubyoptions.rb: add test for switching JIT enablement
"--jit" flag usage may be deprecated later, but not discussed yet.
[Feature #14878]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63995 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* ruby.c (process_options): as DidYouMean requires Rubygems, disable
the former when the latter is disabled too.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63977 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Prevents response splitting and HTML injection attacks in
poorly-written applications which blindly pass along user input
in redirects.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63964 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
I suspect CI test machine was overloaded at the time,
or swapping at the time due to parallel build with mjit.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63962 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* ruby-runner.c (ruby_libm_func): removed as the following test
actually doesn't need the path of libm.
* test/fiddle/test_pointer.rb (Fiddle::TestPointer#test_free=):
removed never used constants and instance variables.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63957 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Don't leave runaway threads as it could affect other tests.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63952 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Threads should not be active during these tests, but
maybe they are...
ref: https://bugs.ruby-lang.org/issues/14906
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63950 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Objects loaded during iseq deserialization using arrays need to be added
to the compile time mark array so that they stay alive until iseqs
finish loading.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63920 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* object.c (mutable_obj_clone): `Kernel#clone` should not copy
FL_USER* flags because they are copied unexpectedly.
Unexpected copy will break internal data consistency.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63912 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* dir.c (glob_helper): fix directory glob which resulted in lacking
the first byte. adjust the length of basename to be appended as
well as removing the heading path, not the length of the joined
path. [ruby-dev:50588] [Bug #14899]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63909 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Any single object allocation can reduce object count, and
object counts are global, so multi-threading leads to unpredictable
results.
See also similar commits: r60699 and r62608
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63885 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Thanks to Greg for the fix. [ruby-core:87860] [Bug #14867]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63880 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
We need to preserve "IGNORE" behavior from Ruby 2.5 and earlier.
We can't rely on SA_NOCLDWAIT any more, since we always need
system() and MJIT to work; so we fake that behavior using
dedicated reaper (currently in timer-thread).
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63879 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
I theorize there can be UDP packet loss even over loopback if
the kernel is under memory pressure. Retry sending periodically
until recvmsg succeeds.
i[ruby-core:87842] [Bug #14898]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63872 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* variable.c (rb_const_search): call #const_missing method on
private constants, as well as uninitialized constants.
[Feature #14328]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63871 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* compile.c (iseq_peephole_optimize): remove unreachable jump
instruction only. if it is labeled and referred from other
instructions, it is reachable and must not be removed.
[ruby-core:87830] [Bug #14897]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63870 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
r63655 was tightly coupled to handle_frames and some assumptions seems
to have been broken by r63763.
To partially resolve Bug#14892, this reverts the optimization for now. I
want to make MJIT CI happy first and then I'll probably retry r63655 by
partially reverting r63763 for sp changes.
The skipped test is not fixed yet.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63863 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* Both timegmw and gmtimew ignores leap second if the timezone doesn't
have leap seconds on the first call of init_leap_second_info()
* Add Bug::Time.reset_leap_second_info for testing
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63857 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
