Exception raising can now be controlled by the caller.
The main purpose on this patch is:
- Suppress exceptions from `obj_free()` in `src/gc.c` with `mrb_env_unshare()`.
- Consider the possibility that calls to `mrb_malloc()` may cause `e` objects to be subject to GC.
When control is returned to `mrb_env_unshare()`, `struct free_obj::next` in the same offset as `struct REnv::stack` is rewritten.
Unexpected results then occur when the object is reused.
Also, if `mrb_heap_page` containing an `e` object is freed, it may cause `SIGSEGV` at that point.
- Protects the value of the stack on `callinfo` that just exits if GC occurs inside `mrb_env_unshare()`.
```ruby
def m
b = -> { b }
end
p m.call
# => print block object, not nil
```
This patch does not raise a `NoMemoryError` exception in `mrb_env_unshare()` and can detect that error.
Thus, the problem fixed in # 3087 is not resurrected.
Also, it may seem that this patch should suppress exceptions raised by `cipop()` during `mrb_protect_error()` and `mrb_vm_exec()` unwinds.
However, `mrb_callinfo::u.env` by `CINFO_DIRECT` is not seen to be set.
So in that case `mrb_env_unshare()` is assumed to be originally exception-free.
Internal functions can only be called from within the library.
Functions listed in `mruby/internal.h` can be called from:
* core (src/*.c)
* gems (mrbgems/**/*.c)
But not from the application linked with `libmruby`.
The `__id__` method implemented in the C function has `MRB_ARGS_NONE()` specified, but it is also effective in the following cases.
```ruby
p nil.__id__ opts: 1 rescue p :a
p nil.method(:__id__).call 1 rescue p :b
p nil.method(:__id__).call opts: 1 rescue p :c
p nil.method(:__id__).to_proc.call 1 rescue p :d
p nil.method(:__id__).to_proc.call opts: 1 rescue p :e
p nil.method(:__id__).unbind.bind_call nil, 1 rescue p :f
p nil.method(:__id__).unbind.bind_call nil, opts: 1 rescue p :g
p nil.__send__ :__id__, 1 rescue p :h
p nil.__send__ :__id__, opts: 1 rescue p :i
```
After applying this patch, all items will output symbols in the same way as CRuby.
For this purpose, add `MRB_PROC_NOARG` to `struct RProc::flags`.
This enhances self-containment.
- Changed the `mrb_callinfo::pc` field to point to itself.
Previously it indicated the return destination of the previous call level.
`mrb_callinfo::pc` will now hold the address to its own `proc->body.irep->iseq`.
- Removed `mrb_callinfo::err` field.
This is because `mrb_callinfo::pc - 1` is semantically the same as the previous `err`.
- The `pc0` and `pc_save` variables in `mrb_vm_exec()` are no longer needed and have been deleted.
- It removes the argument because `cipush()` doesn't need to save the previous `pc`.
`MRB_METHOD_TABLE_INLINE` was fragile. It requires `-falign-functions=n`.
On platform that uses higher bits of function pointers, you can use new
`MRB_METHOD_T_STRUCT` configuration macro.
So you don't need `-falign-functions=2` anymore. Instead your platform
must not use higher bits of the pointer (true for most platforms).
If not, you have to use `struct mrb_method_t` version.
Now the method tables (in classes/modules and caches) keeps C function
pointers without wrapping in `struct RProc` objects. For the sake of
portability, `mrb_method_t` is represented by the struct and union, but
if the most significant bit of the pointer is not used by the platform,
`mrb_method_t` should be packed in `uintptr_t` to reduce memory usage.
`MRB_METHOD_TABLE_INLINE` is turned on by default for linux.
Instead of `irep` links, we added a `upper` link to `struct RProc`.
To make a space for the `upper` link, we moved `target_class` reference.
If a `Proc` does not have `env`, `target_class` is saved in an `union`
shared with `env` (if a `Proc` has env, you can tell it by `MRB_PROC_ENV_P()).
Otherwise `target_class` is referenced from `env->c`. We removed links
in `env` as well.
This change removes 2 members from `mrb_irep` struct, thus saving 2
words per method/proc/block. This also fixes potential memory leaks
due to the circular references caused by a link from `mrb_irep`.
Instead of `irep -> proc` chain, we use `irep -> irep` chain to
avoid GC bugs like #3804. We added `target_class` reference to
`mrb_irep` struct. That means one more word consumption per `irep`.
Yukihiro "Matz" Matsumoto