Mike Larkin <mlar...@nested.page> writes:
> On Wed, Sep 21, 2022 at 08:09:05AM -0400, Christoff Humphries wrote: >> Hello. >> >> I want to help with the virtualization project to get the things that are >> incomplete or missing completed (ie, the "not available at this time" list >> on https://www.openbsd.org/faq/faq16.html). > > Great, always nice to have help. > >> >> Is there a point of contact I should direct questions to, as I'll likely >> have a lot of questions once start diving in to load it into my brain and >> understand the code well enough to do something well and useful. > > tech@ is best, or mlarkin@/dv@. > >> >> Or should I just ask whatever questions I have here? >> >> Asking as I'm not sure how this project handles that stuff (open list or >> pushed more to mentor/owner people for subject areas). >> >> Thanks! >> Christoff Humphries >> > > Generally, find something you want to fix/improve, work on it, send a diff. > That's the best way. Might want to chat with dv or I before you bite on > a task as we may already be working on it (for example, dv is working on > instruction decoding and emulation which is one of the first steps toward > smp). I agree. I would say that as a guiding principal, any new features someone may propose via diffs should land in userland/vmd(8) unless there's a requirement it be in kernel/vmm(4). As for areas we could use help with, off the top of my head, I can think of a few things depending on your current knowledge of VMX/SVM and the general design of vmd(8): * martijn@ just contributed agentx support for snmp monitoring of vmd. It can be improved, but might require we do better accounting of guest vm processes. * I (dv@) am working on instruction decode/emulation as mlarkin@ said, and have two different trees as I reconcile design for x86 port io changes with general read/write mmio. (The emulated devices in userland weren't written with string/rep instructions in mind.) * There are lifecycle issues we could improve, specifically a use case from claudio@ involving auto-installs of guests running -current and rebooting them seemlessly. Too long to explain here. Lastly, I'd say there's a lack of regress tests to help catch slower-to-test issues involving things like send/receive, pause/unpause, etc. It would also be nice to have some level of fuzzing if someone could find a way to fuzz the virtio devices. So...lots to do! -dv
