Since object shapes store the capacity of an object, we no longer
need the numiv field on RObjects. This gives us one extra slot which
we can use to give embedded objects one more instance variable (for a
total of 3 ivs). This commit removes the concept of numiv from RObject.
Sidekiq has a method named `❨╯°□°❩╯︵┻━┻`which corrupts
heap dumps.
Normally we could just dump is as is since it's valid UTF-8 and need
no escaping. But our code to escape control characters isn't UTF-8
aware so it's more complicated than it seems.
Ultimately since the overwhelming majority of method names are
pure ASCII, it's not a big loss to just skip it.
I suspect that some shared pages are invalidated because
some static string don't have their coderange set eagerly.
So the first time they are scanned, the entire memory page is
invalidated.
Being able to see the coderange in `ObjectSpace` would help debug
this.
And in addition `dump` currently call `is_broken_string()` and `is_ascii_string()`
which both end up scanning the string and assigning coderange. I think it's
undesirable as `dump` should be read only.
Example:
```
In file included from ../../../include/ruby/defines.h:72,
from ../../../include/ruby/ruby.h:23,
from ../../../gc.h:3,
from ../../../ext/objspace/objspace_dump.c:15:
../../../ext/objspace/objspace_dump.c: In function ‘dump_append_ld’:
../../../ext/objspace/objspace_dump.c:95:26: warning: comparison of integer expressions of different signedness: ‘long unsigned int’ and ‘int’ [-Wsign-compare]
95 | RUBY_ASSERT(required <= width);
| ^~
```
[Feature #17045] ObjectSpace.dump_all should allocate as little as possible in the GC heap
Up until this commit ObjectSpace.dump_all allocates two Hash because of `rb_scan_args`.
It also can allocate a `File` because of `rb_io_get_write_io`.
These allocations are problematic because `dump_all` dumps the Ruby
heap, so it should try modify as little as possible what it is
observing.
This is useful to see what a block of code allocated, e.g.
```
GC.start
GC.disable
ObjectSpace.trace_object_allocations do
# run some code
end
gc_gen = GC.count
allocations = ObjectSpace.dump_all(output: :file, since: gc_gen)
GC.enable
GC.start
retentions = ObjectSpace.dump_all(output: :file, since: gc_gen)
```
It's possible to define methods with any name, even if the parser
doesn't support it and it can only be used with ex. send.
This fixes an issue where invalid JSON was output from ObjectSpace.dump
when a method name needed escaping.
imemo_callcache and imemo_callinfo were not handled by the `objspace`
module and were showing up as "unknown" in the dump. Extract the code for
naming imemos and use that in both the GC and the `objspace` module.
Saves comitters' daily life by avoid #include-ing everything from
internal.h to make each file do so instead. This would significantly
speed up incremental builds.
We take the following inclusion order in this changeset:
1. "ruby/config.h", where _GNU_SOURCE is defined (must be the very
first thing among everything).
2. RUBY_EXTCONF_H if any.
3. Standard C headers, sorted alphabetically.
4. Other system headers, maybe guarded by #ifdef
5. Everything else, sorted alphabetically.
Exceptions are those win32-related headers, which tend not be self-
containing (headers have inclusion order dependencies).
On ar_table, Do not keep a full-length hash value (FLHV, 8 bytes)
but keep a 1 byte hint from a FLHV (lowest byte of FLHV).
An ar_table only contains at least 8 entries, so hints consumes
8 bytes at most. We can store hints in RHash::ar_hint.
On 32bit CPU, we use 4 entries ar_table.
The advantages:
* We don't need to keep FLHV so ar_table only consumes
16 bytes (VALUEs of key and value) * 8 entries = 128 bytes.
* We don't need to scan ar_table, but only need to check hints
in many cases. Especially we don't need to access ar_table
if there is no match entries (in many cases).
It will increase memory cache locality.
The disadvantages:
* This technique can increase `#eql?` time because hints can
conflicts (in theory, it conflicts once in 256 times).
It can introduce incompatibility if there is a object x where
x.eql? returns true even if hash values are different.
I believe we don't need to care such irregular case.
* We need to re-calculate FLHV if we need to switch from ar_table
to st_table (e.g. exceeds 8 entries).
It also can introduce incompatibility, on mutating key objects.
I believe we don't need to care such irregular case too.
Add new debug counters to measure the performance:
* artable_hint_hit - hint is matched and eql?#=>true
* artable_hint_miss - hint is not matched but eql?#=>false
* artable_hint_notfound - lookup counts
The format addresses are printed in are different if you use
`ObjectSpace.dump_all(output: :stdout)` vs.
`ObjectSpace.dump_all(output: :string)` (or `ObjectSpace.dump`) due to
differences in the underlying `vfprintf` implementation.
Use `"%#"PRIxVALUE` to format `VALUE`.
Co-authored-by: Ashe Connor <ashe@kivikakk.ee>
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64974 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
This reverts commit r64970.
Visual C++ 12.0 doesn't have PRIxPTR.
Anyway we have our own vfprintf implementation BSD_vfprintf().
If you want to have portable vfprintf, replace it with BSD_vfprintf like
vsnprintf or just use BSD_vfprintf.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64972 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
The format addresses are printed in are different if you use
`ObjectSpace.dump_all(output: :stdout)` vs.
`ObjectSpace.dump_all(output: :string)` (or `ObjectSpace.dump`) due to
differences in the underlying `vfprintf` implementation.
Use %"PRIxPTR" instead to be consistent across both.
Co-authored-by: Ashe Connor <ashe@kivikakk.ee>
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64970 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
New IMEMO types were introduced, this just fixes the function that
converts the type to support the new types.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@61377 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Now, casting NODE to VALUE is not recommended. This change requires an
explicit cast from VALUE to NODE to use the NODE utility functions such
as `nd_type`.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@60643 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
IMEMO objects have many types. Without this change, we cannot see what
types of IMEMO objects are being used when dumping the heap. Adding the
type to the IMEMO object will allow us to gather statistics about IMEMO
objects being used.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57486 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
The `full` option includes all slots (even `T_NONE`) in the JSON output.
This is to help with debugging heap fragmentation.
Here is an example usage:
```ruby
File.open('heap.json', 'w') do |f|
ObjectSpace.dump_all(output: f, full: true)
end
```
The `heap.json` file contains all slots, including empty slots.
[Feature #13001] [ruby-core:78468]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57260 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
This commit changes full heap dumps back to using JSON lines format
(http://jsonlines.org) so that we can process very large heaps without
loading the entire heap in to memory at once.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56174 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Use PRIuSIZE instead of PRIdSIZE. This fixes the exception message shown
on too large xmalloc2. This commit also fixes other incorrect use of
PRIdSIZE in other functions; though most of them are debug print.
* gc.c (heap_extend_pages, get_envparam_size, ruby_malloc_size_overflow,
gc_profile_dump_on): Use PRIuSIZE instead of PRIdSIZE as the passed
value is size_t, not ssize_t.
* iseq.c (get_line_info, rb_iseq_disasm_insn): Ditto.
* sprintf.c (rb_str_format): Ditto.
* thread_win32.c (native_thread_create): Ditto.
* vm.c (get_param): Ditto.
* ext/objspace/objspace_dump.c (dump_append_string_content,
dump_object): Ditto.
* ext/socket/raddrinfo.c (host_str, port_str): Ditto.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56156 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
