Backport some LLVM compile-time improvements

[nikic]

Pick up two LLVM patches that reduce number of alias analysis queries, to see whether that gives any visible improvement.

r? @ghost

[nikic]

@bors try @rust-timer queue

[rust-timer]

Awaiting bors try build completion.

@rustbot label: +S-waiting-on-perf

[bors]

⌛ Trying commit 3081eafd4373baf97ec55b1d97174714e5cf9ba2 with merge 57645773c360978a7ca9935c2c58f8394fc97e0a...

[bors]

☀️ Try build successful - checks-actions
Build commit: 57645773c360978a7ca9935c2c58f8394fc97e0a (57645773c360978a7ca9935c2c58f8394fc97e0a)

[rust-timer]

Queued 57645773c360978a7ca9935c2c58f8394fc97e0a with parent 45b3c28, future comparison URL.

[rust-timer]

Finished benchmarking try commit (57645773c360978a7ca9935c2c58f8394fc97e0a): comparison url.

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. Please note that if the perf results are neutral, you should likely undo the rollup=never given below by specifying rollup- to bors.

Importantly, though, if the results of this run are non-neutral do not roll this PR up -- it will mask other regressions or improvements in the roll up.

@bors rollup=never
@rustbot label: +S-waiting-on-review -S-waiting-on-perf

[nikic]

Minor improvements to opt builds and a 1.5% improvement on bootstrap.

r? @nagisa Does this seems worthwhile? If so, I'll push this to the branch and update the submodule reference.

[nagisa]

This does seem quite worthwhile yes. I'm comfortable with the LICM changes, since they have been landed upstream, but I wasn't able to find the diffs or commits corresponding to the [Loads] commits upstream. If they have been submitted, can the differential be referenced in the commit messages? If not, can we quantify the impact of the LICM change alone vs the improvements gained from the Loads commits?

[nikic]

@nagisa The upstream commits are llvm/llvm-project@7c706aa, llvm/llvm-project@e0615bc, llvm/llvm-project@3d8f842, llvm/llvm-project@29034f3. I should get into the habit of using git cherry-pick -x...

[nagisa]

Yeah LGTM then, r=me once these commits end up in a proper branch.

[nikic]

@bors r=nagisa

[bors]

📌 Commit c484bfe has been approved by nagisa

[bors]

⌛ Testing commit c484bfe with merge 1d6b0f6...

[bors]

☀️ Test successful - checks-actions
Approved by: nagisa
Pushing 1d6b0f6 to master...

[leonardo-m]

This seems to have fixed some performance regressions (perhaps with the help of other things in the last week) introduced by LLVM12.