liballoc: mark str.to_owned() and String::from(&str) as #[inline].

[matthiaskrgr]

Fixes #53681

[rust-highfive]

r? @bluss

(rust_highfive has picked a reviewer for you, use r? to override)

[matthiaskrgr]

I didn't notice any benchmarks regressing and it size increase of binaries seem to be below 1% in average but perhaps this should get a perf run anyway, just to be extra sure.

[nagisa]

@bors try

[bors]

⌛ Trying commit d24070b with merge 02d7f80...

[bors]

☀️ Test successful - status-travis
State: approved= try=True

[nagisa]

@rust-timer build 02d7f80

[rust-timer]

Insufficient permissions to issue commands to rust-timer.

[Mark-Simulacrum]

@rust-timer build 02d7f80

[rust-timer]

Success: Queued 02d7f80 with parent e999ebd, comparison URL.

[matthiaskrgr]

Hmm, still no perf results? 🤔

[Mark-Simulacrum]

It looks like something went wrong during collection; @bors try

[nagisa]

@rust-timer build 02d7f80

[rust-timer]

Success: Queued 02d7f80 with parent e999ebd, comparison URL.

[nagisa]

Lets see if we can repeat rust-timer builds with results of previous try builds :)

[nagisa]

Seems like not.

@bors retry

[bors]

⌛ Trying commit d24070b with merge 9aac7ca572e8ba96e99249a1a9d026331374f228...

[bors]

☀️ Test successful - status-travis
State: approved= try=True

[nagisa]

@rust-timer build 9aac7ca572e8ba96e99249a1a9d026331374f228

[rust-timer]

Success: Queued 9aac7ca572e8ba96e99249a1a9d026331374f228 with parent 6310be4, comparison URL.

[nagisa]

Overall a very slight increase in number of executed instructions, but that says very little.

[TimNN]

Ping from triage @bluss / @nagisa: What is the status of this PR?

[nagisa]

@bors r+

I’ll r+ this for the following reasons: while the slight increase in number of executed instructions exists, I believe it is a incorrect statistic to look at here. Inlining can easily go one way or the other on that, while also enabling other optimisations (like eliding allocations) that are important for naive code to be more efficient. The specific case in the code(expected) provided in the issue is a prime example of such enablement – LLVM elides the code for the allocation altogether, because the information necessary to do so is available in that context.

[bors]

📌 Commit d24070b has been approved by nagisa

[bors]

⌛ Testing commit d24070b with merge 96cafc5...

[bors]

☀️ Test successful - status-appveyor, status-travis
Approved by: nagisa
Pushing 96cafc5 to master...