Граф коммитов

40 Коммитов

Автор SHA1 Сообщение Дата
Nobuyoshi Nakada bad61d34be
Qundef is not for ID 2021-10-13 12:07:57 +09:00
eileencodes b91b3bc771 Add a cache for class variables
Redo of 34a2acdac788602c14bf05fb616215187badd504 and
931138b00696419945dc03e10f033b1f53cd50f3 which were reverted.

GitHub PR #4340.

This change implements a cache for class variables. Previously there was
no cache for cvars. Cvar access is slow due to needing to travel all the
way up th ancestor tree before returning the cvar value. The deeper the
ancestor tree the slower cvar access will be.

The benefits of the cache are more visible with a higher number of
included modules due to the way Ruby looks up class variables. The
benchmark here includes 26 modules and shows with the cache, this branch
is 6.5x faster when accessing class variables.

```
compare-ruby: ruby 3.1.0dev (2021-03-15T06:22:34Z master 9e5105c) [x86_64-darwin19]
built-ruby: ruby 3.1.0dev (2021-03-15T12:12:44Z add-cache-for-clas.. c6be009) [x86_64-darwin19]

|         |compare-ruby|built-ruby|
|:--------|-----------:|---------:|
|vm_cvar  |      5.681M|   36.980M|
|         |           -|     6.51x|
```

Benchmark.ips calling `ActiveRecord::Base.logger` from within a Rails
application. ActiveRecord::Base.logger has 71 ancestors. The more
ancestors a tree has, the more clear the speed increase. IE if Base had
only one ancestor we'd see no improvement. This benchmark is run on a
vanilla Rails application.

Benchmark code:

```ruby
require "benchmark/ips"
require_relative "config/environment"

Benchmark.ips do |x|
  x.report "logger" do
    ActiveRecord::Base.logger
  end
end
```

Ruby 3.0 master / Rails 6.1:

```
Warming up --------------------------------------
              logger   155.251k i/100ms
Calculating -------------------------------------
```

Ruby 3.0 with cvar cache /  Rails 6.1:

```
Warming up --------------------------------------
              logger     1.546M i/100ms
Calculating -------------------------------------
              logger     14.857M (± 4.8%) i/s -     74.198M in   5.006202s
```

Lastly we ran a benchmark to demonstate the difference between master
and our cache when the number of modules increases. This benchmark
measures 1 ancestor, 30 ancestors, and 100 ancestors.

Ruby 3.0 master:

```
Warming up --------------------------------------
            1 module     1.231M i/100ms
          30 modules   432.020k i/100ms
         100 modules   145.399k i/100ms
Calculating -------------------------------------
            1 module     12.210M (± 2.1%) i/s -     61.553M in   5.043400s
          30 modules      4.354M (± 2.7%) i/s -     22.033M in   5.063839s
         100 modules      1.434M (± 2.9%) i/s -      7.270M in   5.072531s

Comparison:
            1 module: 12209958.3 i/s
          30 modules:  4354217.8 i/s - 2.80x  (± 0.00) slower
         100 modules:  1434447.3 i/s - 8.51x  (± 0.00) slower
```

Ruby 3.0 with cvar cache:

```
Warming up --------------------------------------
            1 module     1.641M i/100ms
          30 modules     1.655M i/100ms
         100 modules     1.620M i/100ms
Calculating -------------------------------------
            1 module     16.279M (± 3.8%) i/s -     82.038M in   5.046923s
          30 modules     15.891M (± 3.9%) i/s -     79.459M in   5.007958s
         100 modules     16.087M (± 3.6%) i/s -     81.005M in   5.041931s

Comparison:
            1 module: 16279458.0 i/s
         100 modules: 16087484.6 i/s - same-ish: difference falls within error
          30 modules: 15891406.2 i/s - same-ish: difference falls within error
```

Co-authored-by: Aaron Patterson <tenderlove@ruby-lang.org>
2021-06-18 10:02:44 -07:00
Aaron Patterson 07f055bb13
Revert "Filling cache values on cvar write"
This reverts commit 08de37f9fa.
This reverts commit e8ae922b62.
2021-05-11 13:31:00 -07:00
eileencodes e8ae922b62 Add a cache for class variables
This change implements a cache for class variables. Previously there was
no cache for cvars. Cvar access is slow due to needing to travel all the
way up th ancestor tree before returning the cvar value. The deeper the
ancestor tree the slower cvar access will be.

The benefits of the cache are more visible with a higher number of
included modules due to the way Ruby looks up class variables. The
benchmark here includes 26 modules and shows with the cache, this branch
is 6.5x faster when accessing class variables.

```
compare-ruby: ruby 3.1.0dev (2021-03-15T06:22:34Z master 9e5105ca45) [x86_64-darwin19]
built-ruby: ruby 3.1.0dev (2021-03-15T12:12:44Z add-cache-for-clas.. c6be0093ae) [x86_64-darwin19]

|         |compare-ruby|built-ruby|
|:--------|-----------:|---------:|
|vm_cvar  |      5.681M|   36.980M|
|         |           -|     6.51x|
```

Benchmark.ips calling `ActiveRecord::Base.logger` from within a Rails
application. ActiveRecord::Base.logger has 71 ancestors. The more
ancestors a tree has, the more clear the speed increase. IE if Base had
only one ancestor we'd see no improvement. This benchmark is run on a
vanilla Rails application.

Benchmark code:

```ruby
require "benchmark/ips"
require_relative "config/environment"

Benchmark.ips do |x|
  x.report "logger" do
    ActiveRecord::Base.logger
  end
end
```

Ruby 3.0 master / Rails 6.1:

```
Warming up --------------------------------------
              logger   155.251k i/100ms
Calculating -------------------------------------
```

Ruby 3.0 with cvar cache /  Rails 6.1:

```
Warming up --------------------------------------
              logger     1.546M i/100ms
Calculating -------------------------------------
              logger     14.857M (± 4.8%) i/s -     74.198M in   5.006202s
```

Lastly we ran a benchmark to demonstate the difference between master
and our cache when the number of modules increases. This benchmark
measures 1 ancestor, 30 ancestors, and 100 ancestors.

Ruby 3.0 master:

```
Warming up --------------------------------------
            1 module     1.231M i/100ms
          30 modules   432.020k i/100ms
         100 modules   145.399k i/100ms
Calculating -------------------------------------
            1 module     12.210M (± 2.1%) i/s -     61.553M in   5.043400s
          30 modules      4.354M (± 2.7%) i/s -     22.033M in   5.063839s
         100 modules      1.434M (± 2.9%) i/s -      7.270M in   5.072531s

Comparison:
            1 module: 12209958.3 i/s
          30 modules:  4354217.8 i/s - 2.80x  (± 0.00) slower
         100 modules:  1434447.3 i/s - 8.51x  (± 0.00) slower
```

Ruby 3.0 with cvar cache:

```
Warming up --------------------------------------
            1 module     1.641M i/100ms
          30 modules     1.655M i/100ms
         100 modules     1.620M i/100ms
Calculating -------------------------------------
            1 module     16.279M (± 3.8%) i/s -     82.038M in   5.046923s
          30 modules     15.891M (± 3.9%) i/s -     79.459M in   5.007958s
         100 modules     16.087M (± 3.6%) i/s -     81.005M in   5.041931s

Comparison:
            1 module: 16279458.0 i/s
         100 modules: 16087484.6 i/s - same-ish: difference falls within error
          30 modules: 15891406.2 i/s - same-ish: difference falls within error
```

Co-authored-by: Aaron Patterson <tenderlove@ruby-lang.org>
2021-05-11 12:04:27 -07:00
Nobuyoshi Nakada 187c164d71
Suppress unused-variable warning
`key` is not used outside this assertion.
2020-08-10 17:49:08 +09:00
卜部昌平 32623150cd reroute redefinition of NDEBUG
NDEBUG can be defined via a command-line argument.  Should take care of
such situations.
2020-04-21 13:07:19 +09:00
Koichi Sasada b9007b6c54 Introduce disposable call-cache.
This patch contains several ideas:

(1) Disposable inline method cache (IMC) for race-free inline method cache
    * Making call-cache (CC) as a RVALUE (GC target object) and allocate new
      CC on cache miss.
    * This technique allows race-free access from parallel processing
      elements like RCU.
(2) Introduce per-Class method cache (pCMC)
    * Instead of fixed-size global method cache (GMC), pCMC allows flexible
      cache size.
    * Caching CCs reduces CC allocation and allow sharing CC's fast-path
      between same call-info (CI) call-sites.
(3) Invalidate an inline method cache by invalidating corresponding method
    entries (MEs)
    * Instead of using class serials, we set "invalidated" flag for method
      entry itself to represent cache invalidation.
    * Compare with using class serials, the impact of method modification
      (add/overwrite/delete) is small.
    * Updating class serials invalidate all method caches of the class and
      sub-classes.
    * Proposed approach only invalidate the method cache of only one ME.

See [Feature #16614] for more details.
2020-02-22 09:58:59 +09:00
卜部昌平 eb92159d72 Revert https://github.com/ruby/ruby/pull/2486
This reverts commits: 10d6a3aca7 8ba48c1b85 fba8627dc1 dd883de5ba
6c6a25feca 167e6b48f1 7cb96d41a5 3207979278 595b3c4fdd 1521f7cf89
c11c5e69ac cf33608203 3632a812c0 f56506be0d 86427a3219 .

The reason for the revert is that we observe ABA problem around
inline method cache.  When a cache misshits, we search for a
method entry.  And if the entry is identical to what was cached
before, we reuse the cache.  But the commits we are reverting here
introduced situations where a method entry is freed, then the
identical memory region is used for another method entry.  An
inline method cache cannot detect that ABA.

Here is a code that reproduce such situation:

```ruby
require 'prime'

class << Integer
  alias org_sqrt sqrt
  def sqrt(n)
    raise
  end

  GC.stress = true
  Prime.each(7*37){} rescue nil # <- Here we populate CC
  class << Object.new; end

  # These adjacent remove-then-alias maneuver
  # frees a method entry, then immediately
  # reuses it for another.
  remove_method :sqrt
  alias sqrt org_sqrt
end

Prime.each(7*37).to_a # <- SEGV
```
2019-10-03 12:45:24 +09:00
卜部昌平 3632a812c0 refactor add rb_id_table_foreach_with_replace_with_key
This is a pure refactoring to reduce copy & paste.  Also the new
function is made visible from other parts of the interpreter, to
be used later.
2019-09-30 10:26:38 +09:00
tenderlove 91793b8967 Add `GC.compact` again.
🙏

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@67620 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2019-04-20 01:19:47 +00:00
tenderlove 744e5df715 Reverting compaction for now
For some reason symbols (or classes) are being overridden in trunk

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@67598 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2019-04-17 09:41:41 +00:00
tenderlove 3c55b643ae Adding `GC.compact` and compacting GC support.
This commit adds the new method `GC.compact` and compacting GC support.
Please see this issue for caveats:

  https://bugs.ruby-lang.org/issues/15626

[Feature #15626]

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@67576 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2019-04-17 03:17:25 +00:00
kazu 25c1fd3b90 Reverting all commits from r67479 to r67496 because of CI failures
Because hard to specify commits related to r67479 only.
So please commit again.


git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@67499 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2019-04-10 09:15:21 +00:00
nobu b8e3a4bd2c id_table.c: use NULL as ID* instead of Qundef
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@67495 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2019-04-10 06:44:43 +00:00
nobu 5ae753dfa3 Adjusted styles
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@67493 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2019-04-10 06:44:41 +00:00
tenderlove 3ef4db15e9 Adding `GC.compact` and compacting GC support.
This commit adds the new method `GC.compact` and compacting GC support.
Please see this issue for caveats:

  https://bugs.ruby-lang.org/issues/15626

[Feature #15626]

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@67479 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2019-04-09 20:32:04 +00:00
nobu 6924066a65 adjust styles [ci skip]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@58646 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2017-05-10 00:39:26 +00:00
ko1 4c2d014e92 swithc id_table data structure.
* id_table.c: swtich to "simple open addressing with quadratic probing"
  by Yura Sokolov.  For more detail measurements, see [Feature #12180]
* id_table.c: remove other algorithms to simplify the source code.


git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57416 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2017-01-25 03:03:52 +00:00
nobu 00ed779cb0 id_table.c: fix typo
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57369 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2017-01-19 05:24:34 +00:00
nobu dfe1d4fda6 id_table.c: extend, don't shrink
* id_table.c (hash_table_extend): should not shrink the table than
  the previous capacity.  [ruby-core:76534] [Bug #12614]

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@55896 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2016-08-14 13:35:42 +00:00
nobu a62dcd947c id_table.h: dummy sentinel
* id_table.h (rb_id_table_iterator_result): add dummy sentinel
  member because C standard prohibits a trailing comma.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@55821 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2016-08-06 00:35:33 +00:00
ngoto 02cafdf491 capa should be even number on 64-bit SPARC for 8-byte word alignment
* id_table.c (list_id_table_init): When unaligned word access is
  prohibited and sizeof(VALUE) is 8 (64-bit machines),
  capa should always be even number for 8-byte word alignment
  of the values of a table. This code assumes that sizeof(ID) is 4,
  sizeof(VALUE) is 8, and xmalloc() returns 8-byte aligned memory.
  This fixes bus error on 64-bit SPARC Solaris 10.
  [Bug #12406][ruby-dev:49631]


git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@55086 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2016-05-20 14:55:23 +00:00
nobu 439224a590 RUBY_ASSERT
* error.c (rb_assert_failure): assertion with stack dump.
* ruby_assert.h (RUBY_ASSERT): new header for the assertion.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@53615 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2016-01-22 08:33:55 +00:00
normal b147b5a3b8 id_table: const correctness for _size and _memsize
This allows us to swap in rb_id_table_memsize for st_memsize
(which takes a "const st_table *") more easily.

It also makes sense to do the same for rb_id_table_size,
too; as the table cannot be altered when accessing size.

* id_table.h (rb_id_table_size): const arg
  (rb_id_table_memsize): ditto
* id_table.c (st_id_table_size): ditto
  (st_id_table_memsize): ditto
  (list_id_table_size): ditto
  (list_id_table_memsize): ditto
  (hash_id_table_size): ditto
  (hash_id_table_memsize): ditto
  (mix_id_table_size): ditto
  (mix_id_table_memsize): ditto

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52428 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-11-02 22:18:32 +00:00
ko1 840e6b6307 * id_table.c (mix_id_table_insert): do not touch list during
list->hash transition because GC can run during transition.




git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52421 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-11-01 08:17:25 +00:00
nobu 7f13f878cf id_table.c: fix prototype names
* id_table.c: fix prototype names, missing underscore prefixes.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51950 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-09-27 00:23:18 +00:00
nobu 33ed298379 id_table.c: fix prototype names
* id_table.c: fix prototype names, missing underscore prefixes.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-09-27 00:17:53 +00:00
nobu 6b64ffd225 id_table.c: suppress warnings
* id_table.c (UNUSED): mark implementation functions maybe-unused
  to suppress warnings by old gcc.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51948 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-09-27 00:10:47 +00:00
hsbt 1bbe442f80 * id_table.c: fix typo. [ci skip][fix GH-1031] Patch @davydovanton
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51932 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-09-25 00:47:17 +00:00
nobu 80e386e92c id_table.c: fix types
* id_table.c (insert_into_chain, insert_into_main): fix argument
  types in prototype declarations.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51884 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-09-17 01:01:39 +00:00
nobu a3b9b9870e id_table.c: aliases
* id_table.c (IMPL_TYPE, IMPL_VOID): make aliases if supported on
  the platform.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51683 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-08-26 01:20:31 +00:00
nobu 38cfb957dd id_table.h: callback function types
* id_table.h (rb_id_table_foreach_func_t): define callback
  function type for rb_id_table_foreach().

* id_table.h (rb_id_table_foreach_values_func_t): ditto for
  rb_id_table_foreach_values().

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51682 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-08-26 00:47:10 +00:00
nobu 86fd28d11e id_table.c: adjust indent
* id_table.c (list_table_extend, fix_empty): adjust indent.
  (hash_id_table_foreach_values): ditto.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51681 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-08-26 00:46:58 +00:00
nobu f92ed1943c id_table.c: constify
* id_table.c (find_empty): constify static data.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51680 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-08-26 00:46:34 +00:00
nobu 341c84ed22 id_table.c: fix for C89 compilers
* id_table.c (list_table_extend, hash_table_extend): remove C99
  features.  [ruby-dev:49239] [Bug #11487]

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51678 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-08-26 00:26:02 +00:00
nobu 235e7738d6 id_table.c: TOKEN_PASTE
* id_table.c (IMPL1): use TOKEN_PASTE, and prevent `op' from
  expansion.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51562 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-08-13 04:41:54 +00:00
nobu 2c3b813bd3 id_table.c: prefix first
* id_table.c (IMPL): prepend id_table to the argument before its
  expansion.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51550 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-08-12 13:20:21 +00:00
ko1 1761312577 * id_table.c: IMPL() macro accept op as _opname instead of opname
because jemalloc seems to replace the word `free' to `je_free'.



git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51549 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-08-12 12:59:29 +00:00
ko1 e514914919 * id_table.c (mix_id_table_insert): fix memory leak.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51548 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-08-12 12:51:56 +00:00
ko1 c35ff11ae5 * id_table.h: introduce ID key table.
[Feature #11420]
  This table only manage ID->VALUE table to reduce overhead of st.
  Some functions prefixed rb_id_table_* are provided.
* id_table.c: implement rb_id_table_*.
  There are several algorithms to implement it.
  Now, there are roughly 4 types:
    * st
    * array
    * hash (implemented by  Yura Sokolov)
    * mix of array and hash
  The macro ID_TABLE_IMPL can choose implementation.
  You can see detailes about them at the head of id_table.c.
  At the default, I choose 34 (mix of list and hash).
  This is not final decision.
  Please report your suitable parameters or
  your data structure.
  * symbol.c: introduce rb_id_serial_t and rb_id_to_serial()
    to represent ID by serial number.
  * internal.h: use id_table for method tables.
  * class.c, gc.c, marshal.c, vm.c, vm_method.c: ditto.



git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51541 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-08-12 08:43:55 +00:00