[stephenduong1004]

Add a MIR pass PrepareColdBlocks that keeps hot Copy locals in registers when they are only borrowed on cold panic paths (like panic!("{x}") or assert!(c, "{x}")).

Currently, borrowing &x inside a formatting panic forces non_ssa_locals to allocate a stack slot for x across the entire function, spilling x on the hot path. This pass fixes this by copying x into a fresh temporary and borrowing &copy right before a cold &x instead.

AI Usage Disclosure: Used an LLM to generate the initial prototype (#164821) and help draft tests. I implemented this split change myself based on the prototype and design feedback, and reviewed and verified all code and test outputs.

[rustbot]

Some changes occurred to MIR optimizations

cc @rust-lang/wg-mir-opt

[rustbot]

r? @khyperia

rustbot has assigned @khyperia.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: compiler, mir, mir-opt
• compiler, mir, mir-opt expanded to 77 candidates
• Random selection from 21 candidates

[maxdexh]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: 8e92508 (8e9250848565f47d620195e660a85a293ae0d759)
Base parent: ae36fe2 (ae36fe22c7e590df40ffd53ac25fea35597c3f3d)

[rust-timer]

Finished benchmarking commit (8e92508): comparison URL.

Overall result: ✅ improvements - no action needed

Benchmarking means the PR may be perf-sensitive. It's automatically marked not fit for rolling up. Overriding is possible but disadvised: it risks changing compiler perf.

@bors rollup=never rustc-perf
@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.3%	[-0.8%, -0.1%]	24
Improvements ✅ (secondary)	-0.3%	[-0.8%, -0.1%]	15
All ❌✅ (primary)	-0.3%	[-0.8%, -0.1%]	24

Max RSS (memory usage)

Results (primary -3.0%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-3.0%	[-4.3%, -1.7%]	2
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-3.0%	[-4.3%, -1.7%]	2

Cycles

Results (primary -2.2%, secondary -2.4%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-2.2%	[-2.2%, -2.2%]	1
Improvements ✅ (secondary)	-2.4%	[-2.4%, -2.4%]	1
All ❌✅ (primary)	-2.2%	[-2.2%, -2.2%]	1

Binary size

Results (secondary 0.0%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	0.0%	[0.0%, 0.0%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Bootstrap: 483.826s -> 485.567s (0.36%)
Artifact size: 406.63 MiB -> 406.68 MiB (0.01%)

[khyperia]

I'm very much not knowledgeable about mir opts, so, reassigning (I think mir-opt is the right group, lmk if I should have rolled mir or just compiler or something instead):

r? mir-opt