Use vm_memory crate instead of in-house

[jounathaen]

Extracted from #643

Honestly, I'm not a 100% fan of this, but we need this in order to use vmm-mem's virtqueue abstractions in #643

[codecov]

Codecov Report

❌ Patch coverage is 89.16667% with 13 lines in your changes missing coverage. Please review.
✅ Project coverage is 77.57%. Comparing base (c0666c5) to head (712d9cf).
⚠️ Report is 3 commits behind head on main.

Files with missing lines	Patch %	Lines
src/linux/x86_64/kvm_cpu.rs	56.25%	7 Missing ⚠️
src/mem.rs	95.34%	4 Missing ⚠️
src/arch/x86_64/paging/mod.rs	75.00%	1 Missing ⚠️
src/vm.rs	91.66%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1199      +/-   ##
==========================================
+ Coverage   76.82%   77.57%   +0.75%     
==========================================
  Files          26       26              
  Lines        3667     3675       +8     
==========================================
+ Hits         2817     2851      +34     
+ Misses        850      824      -26     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[fogti]

Having three address types for guest addresses certainly makes stuff more confusing... Would it be possible to get rid of one?

[jounathaen]

Having three address types for guest addresses certainly makes stuff more confusing... Would it be possible to get rid of one?

Yeah. I really hate it. But the vm_memory crate uses a different type than the kernel. So we can't use the same type in uhyve-interface without changing it in the kernel, and I would rather not include a new dependency in the kernel just for the type alias, and the current address type is "better" than the vm_memory one.

[fogti]

re: my proposal of the vm_memory dependency in uhyve-interface: One could also make the dependency optional, and provide From/Into conversion implementations there (and perhaps a few utility functions, albeit I haven't looked (yet) into which ones would be useful).

[jounathaen]

@fogti So now I've changed this PR so that we still use our MmapMemory struct within Uhyve, but it uses the vmm-mem types internally. What do you think? Does this resolve the issue with the three address types?

[fogti]

It does solve the issue with the address types.

[fogti]

That are a lot of force pushes...

[jounathaen]

May the force be with me...

I guess this is ready now?

All should be implemented by now. As we can still improve stuff later, let's merge it!