Add a null check before invoking finalizer. Before, we would do a callvirt Finalize() on the object and if it didn't have a finalizer, it would call Object::Finalize(). However, since now we use mono_method_invoke with a concrete method, we crash if it's null. So we just

Use mono_runtime_invoke to invoke finalizers. This way managed runtime can do whatever is necessary to transition to executing managed code, like marking JIT memory executable on macOS ARM64.

Exposing functions: mono_custom_attrs_construct_by_type and mono_unit…

* mono_custom_attrs_construct_by_type allows unity to have a chance at handling the error if one arises instead of fatally asserting
* mono_unity_error_convert_to_exception allows unity to take the provided MonoError and fetch the MonoException object within to log into the editor as the exception that was thrown.

This change has associated unity changes in order to fix (case 1255935)

Pickup BDWGC fix for GC_retry_signals performance regression (case 1255866)

Retry running bee several times if it fails

Fixing a hang that would occur when there is incorrectly set time zon…

Fixing a hang that would occur when there is incorrectly set time zone data in the windows registry. (case 1256569)

Add Apple silicon slice to Mono runtime built for macOS

Signed-off-by: NTautvydas Žilys <tautvydas.zilys@gmail.com>

Signed-off-by: NTautvydas Žilys <tautvydas.zilys@gmail.com>

Do not use MONO_MMAP_32BIT on Apple platforms MMAP_32BITS used to be undefined on darwin but was added recently which leads to unexpected behavior
Signed-off-by: NTautvydas Žilys <tautvydas.zilys@gmail.com>

Signed-off-by: NTautvydas Žilys <tautvydas.zilys@gmail.com>

Signed-off-by: NTautvydas Žilys <tautvydas.zilys@gmail.com>

Signed-off-by: NTautvydas Žilys <tautvydas.zilys@gmail.com>

Correct file sharing for exFAT partitions on macOS (case 1253812)

This change applies a similar fix for exFAT partitions as was applied
for FAT32 partitions earlier. For exFAT partitions, the invalid inode
value seems to be 1 though.

[debugger] Fixing two crashes while debugging an Android app. (#13373) (unity case 1249172)

* [debugger] Fixing two crashes while debugging an Android app.

-> Doing stepping into in VSM in some situation the frame saved in TLS is not synchronised with what is really been executed in the main thread. This makes the debugger-agent crashes because it tries to get variable info in some memory that is not available anymore.
	-> To fix it I forced the update of stack when CMD_THREAD_GET_FRAME_INFO is called.

-> Doing step over in Visual Studio for Windows, if you have the threads debugger window enabled, VSW calls frame_commands for each thread that is showed, and if the thread is not really_suspended it tries to get variable info from a memory that is not available anymore because the thread is not suspended yet.
	-> To fix it I don't send variable info of a frame if the thread is not really_suspended and doesn't have an async_state valid.

Cherry-picked with some light massaging by Alex Thibodeau (case 1249172)

Update readme

Gather callstack even if instruction pointer is null.

Valuetype ref return x64jit fix

Add `mono_unity_gc_set_mode` API to replace `mono_unity_gc_enable`/`mono_unity_gc_disable`/`mono_unity_gc_is_disabled` (#1302)

Same method that ldloc.s uses, which incorporates the byref test.

When we have the following sequence:

    ldloc n
    ldfld valuetype [Unity.Mathematics]Unity.Mathematics.int2 Foo::bar

and local n is a reference to value, when walking the instructions in
mono_method_to_ir, we generally try to avoid loading a whole value
type just to load one of the fields.  I believe this is an
optimization.

The test here is not quite complete, though, as we have ref valuetypes
where the object stored at the location is actually an address, in
which case we still need to treat it as we would a normal reference.

This optimization isn't present in the ldloc.n coded forms and is
written in a slightly more robust way in ldloc.s, so no changes are
necessary there.

As an alternative to this commit, we might adopt the ldloc.s test for
ldloc as well, or merge in
https://github.com/mono/mono/commit/29428d951637ff85b593ea5c9cc5d5236579906d/mono/mini/method-to-ir.c
from master which encapsulates much the same intent.

only build 32bit osx on build server. Catalina doesn't support buildi…

only build 32bit osx on build server. Catalina doesn't support building 32bit so we don't want to do this by default anymore.

Avoid exceeding the mark stack limit (case 1235202)

* update bdwgc

* Expose mono_gc_start_incremental_collection API

* switch bdwgc repo back to unity-master branch