tree 0bb6165dd4bb6fac57490d290b4e7d3d8d577fbb
parent 4751950ccc948c07047e62c20adf423d7e5f668c
parent 0cfdbc04721e0ca6706051c84ffd51b7f024af78
author bors <bors@rust-lang.org> 1654730027 +0000
committer bors <bors@rust-lang.org> 1654730027 +0000

Auto merge of #10737 - ehuss:revert-num-cpus, r=weihanglo

[beta] Revert #10427: switch from num_cpus

This temporarily reverts #10427 (Use available_parallelism instead of num_cpus) per the discussion at https://github.com/rust-lang/rust/issues/97549. `available_parallelism` does not handle cgroups v1 on Linux unlike num_cpus. I am concerned that this potentially affects a significant percentage of users. For example, Docker just added cgroups v2 support last year. Various Linux distributions have only recently switched to it as the default. The following is what I can find on the web:

* Fedora (since 31)
* Arch Linux (since April 2021)
* openSUSE Tumbleweed (since c. 2021)
* Debian GNU/Linux (since 11)
* Ubuntu (since 21.10)
* RHEL and RHEL-like distributions (since 9)

This also appears to affect CircleCI.

The consequence is that Cargo ends up using too much parallelism and can run out of memory.

I'm not sure what to do about 1.63.  If std adds support for cgroups v1, then I don't think there is anything to do there. Otherwise I think we should revert similarly if that doesn't happen.