Open stefano-garzarella opened 1 month ago
@endeneer @TimePrinciple we have several PRs opened. Can you try to summarize below which ones we need, if there are dependencies between them, and which ones are duplicated?
It's a little confusing for reviewers which ones to start with.
As of now, these are the open PR:
In #101, I proposed using qemu-user-static
to run cargo test
, which might work with some crates, but will eventually fail with kvm-ioctls
and crates need the presence of /dev/kvm
, so I moved on to the second draft.
In #104, I proposed a standard way just like other architectures (x86_64
, arm64
) did, introduced riscv64
support for buildkite-agent
, and that should be the "right" way when AWS has riscv64
machines available. Therefore, I have worked out my third draft.
In #106, the third draft bears some similarity with #91 in image building stage, and thanks to @endeneer's work which I found very heuristic and helpful. In my proposal, I used ssh
to produce the original output that cargo
and any other commands may produce. And this approach essentially runs everything related to riscv64
inside qemu-system-riscv64
instead of cross compilation, enabling us to finally come to use build_container.sh
just like other architectures did.
In short, my proposals are completely independent, while one may not seem to be appropriate at this moment. Thus, I'm closing #101, leaving #104 open until riscv64
machines are available to AWS, #106 are the one we worth discussing and reviewing at the time being.
In short, my proposals are completely independent, while one may not seem to be appropriate at this moment. Thus, I'm closing https://github.com/rust-vmm/rust-vmm-container/pull/101, leaving https://github.com/rust-vmm/rust-vmm-container/pull/104 open until riscv64 machines are available to AWS, https://github.com/rust-vmm/rust-vmm-container/pull/106 are the one we worth discussing and reviewing at the time being.
@TimePrinciple thanks for the recap!
Should we mark #104 as Draft
while waiting AWS support for riscv64
?
Are #91 and #106 in conflict?
In short, my proposals are completely independent, while one may not seem to be appropriate at this moment. Thus, I'm closing #101, leaving #104 open until riscv64 machines are available to AWS, #106 are the one we worth discussing and reviewing at the time being.
@TimePrinciple thanks for the recap!
Should we mark #104 as
Draft
while waiting AWS support forriscv64
? Are #91 and #106 in conflict?
Yes, #91 and #106 are one way or the other.
I'll mark #104 now :)
There are several requests to support RISC-V, so let's create this issue to monitor the current status.