github
/
ruby
зеркало из https://github.com/github/ruby.git
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность
80 256 коммитов 219 Ветки 1 255 Теги 390 MiB
Граф коммитов

136 Коммитов

Автор SHA1 Сообщение Дата
Jean Boussier 5cc44f48c5 Refactor rb_shape_transition_shape_capa to not accept capacity 
This way the groth factor is encapsulated, which allows
rb_shape_transition_shape_capa to be smarter about ideal sizes.
 2023-10-10 14:47:54 +02:00
Alan Wu 85aa28e8a6 RJIT: Remove Type::CArray and limit use of Type::CString 
See previous similar YJIT commit.
 2023-08-28 17:14:33 -04:00
Takashi Kokubun cd8d20cd1f
YJIT: Compile exception handlers (#8171) 
Co-authored-by: Maxime Chevalier-Boisvert <maximechevalierb@gmail.com>
 2023-08-08 16:06:22 -07:00
Alan Wu f302e725e1
Remove __bp__ and speed-up bmethod calls (#8060) 
Remove rb_control_frame_t::__bp__ and optimize bmethod calls

This commit removes the __bp__ field from rb_control_frame_t. It was
introduced to help MJIT, but since MJIT was replaced by RJIT, we can use
vm_base_ptr() to compute it from the SP of the previous control frame
instead. Removing the field avoids needing to set it up when pushing new
frames.

Simply removing __bp__ would cause crashes since RJIT and YJIT used a
slightly different stack layout for bmethod calls than the interpreter.
At the moment of the call, the two layouts looked as follows:

                   ┌────────────┐    ┌────────────┐
                   │ frame_base │    │ frame_base │
                   ├────────────┤    ├────────────┤
                   │    ...     │    │    ...     │
                   ├────────────┤    ├────────────┤
                   │    args    │    │    args    │
                   ├────────────┤    └────────────┘<─prev_frame_sp
                   │  receiver  │
    prev_frame_sp─>└────────────┘
                     RJIT & YJIT      interpreter

Essentially, vm_base_ptr() needs to compute the address to frame_base
given prev_frame_sp in the diagrams. The presence of the receiver
created an off-by-one situation.

Make the interpreter use the layout the JITs use for iseq-to-iseq
bmethod calls. Doing so removes unnecessary argument shifting and
vm_exec_core() re-entry from the interpreter, yielding a speed
improvement visible through `benchmark/vm_defined_method.yml`:

     patched:   7578743.1 i/s
      master:   4796596.3 i/s - 1.58x  slower

C-to-iseq bmethod calls now store one more VALUE than before, but that
should have negligible impact on overall performance.

Note that re-entering vm_exec_core() used to be necessary for firing
TracePoint events, but that's no longer the case since
9121e57a5f.

Closes ruby/ruby#6428
 2023-07-17 13:57:58 -04:00
Peter Zhu 3223181284 Remove RARRAY_CONST_PTR_TRANSIENT 
RARRAY_CONST_PTR now does the same things as RARRAY_CONST_PTR_TRANSIENT.
 2023-07-13 14:48:14 -04:00
Alan Wu 296782ab60
YJIT: Fix autosplat miscomp for blocks with optionals (#8006) 
* YJIT: Fix autosplat miscomp for blocks with optionals

When passing an array as the sole argument to `yield`, and the yieldee
takes more than 1 optional parameter, the array is expanded similar
to `*array` splat calls. This is called "autosplat" in
`setup_parameters_complex()`.

Previously, YJIT did not detect this autosplat condition. It passed the
array without expanding it, deviating from interpreter behavior.
Detect this conditon and refuse to compile it.

Fixes: Shopify/yjit#313

* RJIT: Fix autosplat miscomp for blocks with optionals

This is mirrors the same issue as YJIT. See previous commit.
 2023-07-04 10:45:29 -04:00
Peter Zhu 7577c101ed
Unify length field for embedded and heap strings (#7908) 
* Unify length field for embedded and heap strings

The length field is of the same type and position in RString for both
embedded and heap allocated strings, so we can unify it.

* Remove RSTRING_EMBED_LEN
 2023-06-06 10:19:20 -04:00
Takashi Kokubun d4dc149c3c Update an Intel SDM link [ci skip] 2023-04-28 10:09:52 -07:00
Takashi Kokubun cbfb5fbb25 RJIT: Fix unspecified_bits with locals 2023-04-26 18:12:53 -07:00
Aaron Patterson bdffcd6df3 Update RJIT to support newarray_send 
This also adds max / hash support
 2023-04-18 17:16:22 -07:00
John Hawthorn be08e1b14c RJIT: argc check in known cfuncs 2023-04-12 17:48:34 -07:00
Takashi Kokubun 65d27d3c0a RJIT: Skip a class guard if known to be T_STRING 2023-04-05 00:46:10 -07:00
Takashi Kokubun 9bef39de74 RJIT: Handle include_all argument of respond_to? 2023-04-05 00:38:38 -07:00
Takashi Kokubun 66fe64b269 RJIT: Remove unused variables 2023-04-04 23:55:19 -07:00
Takashi Kokubun 2ddf6079f1 RJIT: Always use guard_two_fixnums 2023-04-04 23:52:14 -07:00
Takashi Kokubun 4e9c3b4518 RJIT: Eliminate known-result guards for blockarg 2023-04-04 23:42:14 -07:00
Takashi Kokubun 4f77d1cdb8 RJIT: Eliminate known-result branches 2023-04-04 23:30:26 -07:00
Takashi Kokubun e06bebff5e RJIT: Propagate argument types on method calls 2023-04-04 22:13:25 -07:00
Takashi Kokubun 6d8875b5d7 RJIT: Fix mapping offsets in stack_swap 2023-04-04 22:05:45 -07:00
Peter Zhu 1da2e7fca3
[Feature #19579] Remove !USE_RVARGC code (#7655) 
Remove !USE_RVARGC code

[Feature #19579]

The Variable Width Allocation feature was turned on by default in Ruby
3.2. Since then, we haven't received bug reports or backports to the
non-Variable Width Allocation code paths, so we assume that nobody is
using it. We also don't plan on maintaining the non-Variable Width
Allocation code, so we are going to remove it.
 2023-04-04 17:30:06 -04:00
Takashi Kokubun 51b2f4c0f5 RJIT: Fix the argument of shift_stack 2023-04-04 01:17:23 -07:00
Takashi Kokubun 98f0e2c330 RJIT: Fix the argument for defined 2023-04-04 00:41:58 -07:00
Takashi Kokubun 19506650ef RJIT: Add --rjit-verify-ctx option 2023-04-04 00:35:29 -07:00
Takashi Kokubun 2c560b976e RJIT: Fix arguments to SPECIAL_CONST_P 2023-04-04 00:30:57 -07:00
Takashi Kokubun 6a4087702b RJIT: Update type information on setlocal 2023-04-03 22:42:21 -07:00
Takashi Kokubun 6ab86e4626 RJIT: Fix arguments for shift_stack 2023-04-03 21:26:40 -07:00
Takashi Kokubun 3cb5e7558d Fix a test_rubyoptions failure 2023-04-03 08:44:16 -07:00
Takashi Kokubun 46e93ef53d RJIT: Propagate self's type information 2023-04-03 00:49:10 -07:00
Takashi Kokubun ca8102355a RJIT: Upgrade type on jit_guard_known_class 2023-04-03 00:39:29 -07:00
Takashi Kokubun 1d452c2cf8 RJIT: Upgrade type to Fixnum after guard 2023-04-03 00:02:45 -07:00
Takashi Kokubun 0abe9d8b32 RJIT: Upgrade type to String after guard 2023-04-02 23:54:08 -07:00
Takashi Kokubun f0b69eb14f RJIT: Upgrade type to Array after guard 2023-04-02 23:49:55 -07:00
Takashi Kokubun d876c008ec RJIT: Limit the number of versions per block 2023-04-02 23:37:15 -07:00
Takashi Kokubun 3bacc3877a RJIT: Find a best matching block version 2023-04-02 23:24:14 -07:00
Takashi Kokubun eb51248c4c RJIT: Upgrade type to UnknownHeap after guard 2023-04-02 23:11:18 -07:00
Takashi Kokubun be4e16bc1a RJIT: Update type information on setn insn 2023-04-02 23:07:58 -07:00
Takashi Kokubun 50285d2864 RJIT: Swap type information on swap insn 2023-04-02 23:06:30 -07:00
git 0276c1d935 * remove trailing spaces. [ci skip] 2023-04-03 05:32:32 +00:00
Takashi Kokubun d546f8c518 RJIT: Store type information in Context 2023-04-02 22:32:16 -07:00
Takashi Kokubun 8968b47ba8 RJIT: Refactor getlocal and setlocal 2023-04-02 16:50:54 -07:00
Takashi Kokubun 6002b12611 RJIT: Support entry with different PCs 2023-04-02 15:27:40 -07:00
Takashi Kokubun 4fc336127e RJIT: Support has_opt ISEQs 2023-04-02 14:47:23 -07:00
Takashi Kokubun ad2b719fc2 RJIT: C::FL_TEST_RAW returns RBOOL 2023-04-02 14:32:51 -07:00
Takashi Kokubun 57ee627929 RJIT: Lazily guard block arg 
to simplify the implementation
 2023-04-02 14:22:42 -07:00
Takashi Kokubun 66f8efc342 RJIT: Simplify cfunc implementation 2023-04-02 13:58:39 -07:00
Takashi Kokubun 284a0ff96c RJIT: Remove an unneeded branch 2023-04-02 12:28:49 -07:00
Takashi Kokubun 3fe134759c Skip assert_linear_performance for RJIT 2023-04-02 12:23:23 -07:00
Takashi Kokubun bf7587748d RJIT: Simplify invokesuper implementation 2023-04-02 11:42:16 -07:00
Takashi Kokubun 5cc644b147 RJIT: Group blockarg exit reasons 2023-04-02 11:01:23 -07:00
Takashi Kokubun cd1cd8030c RJIT: Support splat args 2023-04-02 10:55:03 -07:00