Wasmtime GitHub notifications bot (Jan 28 2022 at 04:44): Wasmtime GitHub notifications bot (Jan 28 2022 at 04:44):cfallin commented on issue #3737:
@alexcrichton This is my exploration of what a "dynamic backend choice" might look like, as per your comment in the earlier review. I'm not entirely sure I like it -- it actually creates much more complexity passing through the backend choice everywhere, and it creates additional states to test (one backend included, but user has chosen another).
My intuition here is that simplicity is better found by just reducing the number of backends -- e.g. if/when we decide that memfd supersedes uffd, we can remove that, and then we can do memfd by default on Linux, and all of memfd is already built to support some memories having images and others not, so it should support dynamic fallback to mmap much more gracefully.
I'm also worried that since uffd was built as a static feature choice, I may have missed something somewhere that needs a dynamic branch -- this feels like a somewhat dangerous refactor.
Thoughts?
Wasmtime GitHub notifications bot (Jan 28 2022 at 05:06):github-actions[bot] commented on issue #3737:
Subscribe to Label Action
cc @peterhuene
<details>
This issue or pull request has been labeled: "wasmtime:api"Thus the following users have been cc'd because of the following labels:
- peterhuene: wasmtime:api
To subscribe or unsubscribe from this label, edit the <code>.github/subscribe-to-label.json</code> configuration file.
Learn more.
</details>
Wasmtime GitHub notifications bot (Feb 09 2022 at 21:59):cfallin commented on issue #3737:
This has been superseded by #3778 (for memfd; and I think we're likely to remove uffd), so I'll go ahead and close this. Can bring it back if needed and we decide we want to keep uffd around with a dynamic switch.
Last updated: Nov 22 2024 at 16:03 UTC