Ruby: extension types should define a proper memsize function

[casperisfine]

Here's an snippet from a Ruby heap dump (generated with ObjectSpace.dump_all

{"address":"0x7fa2dadd7320", "type":"DATA", "class":"0x7fa3eb7ab408", "struct":"Google::Protobuf::Internal::Arena", "memsize":40}
{"address":"0x7fa2dadd1678", "type":"DATA", "class":"0x7fa3eb7ab408", "struct":"Google::Protobuf::Internal::Arena", "memsize":40}
{"address":"0x7fa2dadd3c70", "type":"DATA", "class":"0x7fa3eb7ab408", "struct":"Google::Protobuf::Internal::Arena", "memsize":40}
{"address":"0x7fa2dadcfaa8", "type":"DATA", "class":"0x7fa3eb7ab408", "struct":"Google::Protobuf::Internal::Arena", "memsize":40}
{"address":"0x7fa2dadc9c98", "type":"DATA", "class":"0x7fa3eb7ab408", "struct":"Google::Protobuf::Internal::Arena", "memsize":40}
{"address":"0x7fa2dadcb390", "type":"DATA", "class":"0x7fa3eb7ab408", "struct":"Google::Protobuf::Internal::Arena", "memsize":40}

As you can see, all these Google::Protobuf::Internal::Arena instances claim to be 40 bytes, which is obviously a lie.

This is because the memsize function is not defined:

protobuf/ruby/ext/google/protobuf_c/protobuf.c

Line 192 in 520c601

{Arena_mark, Arena_free, NULL},

What language does this apply to?

Ruby

Describe the problem you are trying to solve.

I'd like to have a semi-accurate idea of how much memory the google-protobuf gem is holding onto.

Describe the solution you'd like

I'd like these functions to be defined (Arena, Message, etc). They don't have to be 100% accurate, not particularly fast. It's a debugging facility.

Describe alternatives you've considered

I don't see any.

cc @zhangskz because #9384 (comment)

[zhangskz]

This makes sense and would be helpful as a whole for debugging and understanding memory usage.

We previously investigated implementing dsize which similarly allows extensions to report the object's memory per ruby extension documentation when addressing memory leak #9467.

Implementing a trivial dsize that reported a large number didn't help trigger more frequent GC so we didn't pursue this further at the time, but this is still useful for profiling memory used by Arena objects.

[casperisfine]

We previously investigated implementing dsize which similarly allows extensions to report the object's memory

That's the same thing.

Implementing a trivial dsize that reported a large number didn't help trigger more frequent GC

Yes it's purely for introspection. To notify the GC of allocations, you need to use the ruby malloc functions.

[github-actions]

We triage inactive PRs and issues in order to make it easier to find active work. If this issue should remain active or becomes active again, please add a comment.

This issue is labeled inactive because the last activity was over 90 days ago.

[github-actions]

We triage inactive PRs and issues in order to make it easier to find active work. If this issue should remain active or becomes active again, please reopen it.

This issue was closed and archived because there has been no new activity in the 14 days since the inactive label was added.