Add unstable -Zdefault-hidden-visibility cmdline flag for rustc.

[anforowicz]

The new flag has been described in the Major Change Proposal at
rust-lang/compiler-team#656

[rustbot]

r? @TaKO8Ki

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

[rustbot]

Some changes occurred in compiler/rustc_codegen_gcc

cc @antoyo

These commits modify compiler targets.
(See the Target Tier Policy.)

[anforowicz]

/cc @bridiver, @danakj

[anforowicz]

@rustbot label +S-waiting-on-author -S-waiting-on-review

I need to fix the test failures reported above. AFAICT they didn't show up when I've run ./x.py test, but I can indeed repro them with ./x.py test compiler/rustc_interface.

[bridiver]

Thanks @anforowicz ! Not sure when I was ever going to finally get around to this if you didn't pick it up.

[anforowicz]

@rustbot label -S-waiting-on-author +S-waiting-on-review

[bridiver]

@TaKO8Ki do you have some time to look at this? It's going to solve some important issues for us (Brave/Chromium)

[TaKO8Ki]

@bridiver Sorry. I missed it. I'll take a look.

[TaKO8Ki]

Sorry for the late review! @bors r+ rollup

[bors]

📌 Commit 1a0829c has been approved by TaKO8Ki

It is now in the queue for this repository.

[rustbot]

This PR modifies config.example.toml.

If appropriate, please update CONFIG_CHANGE_HISTORY in src/bootstrap/src/utils/change_tracker.rs and change-id in config.example.toml.

[anforowicz]

This PR modifies config.example.toml.

If appropriate, please update CONFIG_CHANGE_HISTORY in src/bootstrap/src/utils/change_tracker.rs and change-id in config.example.toml.

That was accidental and should be fixed now.

[anforowicz]

@TaKO8Ki, could you PTAL again?

The only change is in tests/codegen/default-hidden-visibility.rs - in the latest revision the test only applies to only-linux and only-x86_64. This is sufficient to verify that the new command-line flag works and has the desirable effect. I think it's okay that the verification is only done on a single target, because the implementation is target-agnostic (and because test expectations differ on other targets - the outcome on those targets is still acceptable, but difficult to cover with a test).

@rustbot label -S-waiting-on-author +S-waiting-on-review

---

Let me also discuss some alternative testing approaches that I have considered for this PR.

If there is a desire to cover additional target triples (now or (preferably) in a separate follow-up PR), then:

• One way to proceed is to duplicate the test file into additional only-X flavours, with different expectations (e.g. expecting internal constant instead of hidden constant - both are okay). If needed I can ask teammates with a Mac to generate test expectations for their target triple. Additional triples can also be covered, although I would have to lean on others to generate the test expectations.
• Another way to proceed may be to mimic tests/codegen/align-byval.rs which selects target triples via [PREFIX] compile-flags: --target …. I have low confidence in my ability to follow this approach, because it fails on my machine for some triples - the test complains about missing matching standard library, but trying to build the library with ./x.py build library --target i686-unknown-linux-gnu fails saying: “/usr/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-linux-gnu/13/libgcc.a when searching for -lgcc”.

Instead of per-triple test expectations, we could also tweak and relax the patterns being matched against (e.g. saying {{hidden|internal}} in the test expectations. This alternative feels less robust: the relaxed pattern may match accidentally and I have low confidence in my ability to construct a pattern that will work on all triples (given the previous failed attempts in this PR + given that IIUC tier3 targets may not immediately exercise the new test).

[anforowicz]

Actually, I think it would be safest if I specified the full triple, because i686-unknown-linux-gnu passes for me locally, but we've already seen that i686, linux, musl combination has different outcome.

[anforowicz]

Actually, I think it would be safest if I specified the full triple, because i686-unknown-linux-gnu passes for me locally, but we've already seen that i686, linux, musl combination has different outcome.

Ok - switched to only-x86_64-unknown-linux-gnu...

[TaKO8Ki]

@anforowicz

TaKO8Ki, could you PTAL again?

Sure. @bors r+

[bors]

📌 Commit 981c4e3 has been approved by TaKO8Ki

It is now in the queue for this repository.

[bors]

⌛ Testing commit 981c4e3 with merge 9d49eb7...

[bors]

☀️ Test successful - checks-actions
Approved by: TaKO8Ki
Pushing 9d49eb7 to master...

[rust-timer]

Finished benchmarking commit (9d49eb7): comparison URL.

Overall result: no relevant changes - no action needed

@rustbot label: -perf-regression

Instruction count

This benchmark run did not return any relevant results for this metric.

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	0.7%	[0.7%, 0.7%]	1
Regressions ❌ (secondary)	0.7%	[0.7%, 0.7%]	3
Improvements ✅ (primary)	-0.7%	[-0.8%, -0.6%]	2
Improvements ✅ (secondary)	-2.4%	[-2.4%, -2.4%]	1
All ❌✅ (primary)	-0.3%	[-0.8%, 0.7%]	3

Cycles

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.9%	[-0.9%, -0.9%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-0.9%	[-0.9%, -0.9%]	1

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 670.867s -> 670.776s (-0.01%)
Artifact size: 312.41 MiB -> 312.39 MiB (-0.01%)