ci: test cross compiling core for the thumb targets

[japaric]

this commit adds a new path, src/libcore, to bootstrap that (cross) compiles
only the core crate for the target, and a thumb image that will test cross
compiling core for the 4 existing thumb targets.

closes #38828

r? @alexcrichton

I'm not sure if the .travis.yml changes are correct / enough but
./src/ci/docker/run.sh thumb does the core cross compilation test

[rust-highfive]

Thanks for the pull request, and welcome! The Rust team is excited to review your changes, and you should hear from @alexcrichton (or someone else) soon.

If any changes to this PR are deemed necessary, please add them as extra commits. This ensures that the reviewer can see what has changed since they last reviewed the code. Due to the way GitHub handles out-of-date commits, this should also make it reasonably obvious what issues have or haven't been addressed. Large or tricky changes may require several passes of review and changes.

Please see the contribution instructions for more information.

[alexcrichton]

Thanks @japaric! Would it be possible to avoid duplicating the std function used in rustbuild to compile libstd? The goal was in theory to only have one function that invokes cargo but the std/test/rustc split necessitated three unfortunately, and I was hoping to not grow beyond that.

[alexcrichton]

Oh right, we're also running near the limit of our capacity right now, so it may be a moment before we can land this unfortunately :(

[alexcrichton]

Ok so we're currently using 41/45 of our capacity on Travis on each PR. I think we need to leave some breathing room for projects like Cargo/libc and PRs to have capacity as well, so until we get more capacity or shuffle around what we have already I don't think we'll be able to expand our test coverage unfortunately. If we can manage to slot this in somewhere else that'd be great, though!

cc @aturon

[aturon]

To echo what @alexcrichton is saying: while we can potentially grow our capacity further, we need to figure out a sustainable model going forward as we add an increasing number of platforms; the costs add up quickly. I'll be working with the core team and Mozilla to develop a strategy here, but it will take some time. If we can find other avenues in the meantime, that'd be great.

[japaric]

If we can manage to slot this in somewhere else that'd be great, though!

Do you mean running these tests on an existing builder?

I'm OK with postponing this if there's no capacity atm.

[alexcrichton]

Yeah if we can get an existing builder to run this and stay under 2hrs then we can land this immediately

[shepmaster]

@japaric do you think you will have time to make the changes to reuse an existing builder?

[shepmaster]

Thank you for the PR @japaric, but we are going to close this due to a few weeks of inactivity. Please feel free to reopen when you are able to make the requested changes.