Tagged pointers, now with strict provenance!
This is a big refactor of tagged pointers in rustc, with three main goals:
1. Porting the code to the strict provenance
2. Cleanup the code
3. Document the code (and safety invariants) better
This PR has grown quite a bit (almost a complete rewrite at this point...), so I'm not sure what's the best way to review this, but reviewing commit-by-commit should be fine.
r? `@Nilstrieb`
Bypass the varint path when encoding InitMask
The data in a `InitMask` is stored as `u64` but it is a large bitmask (not numbers) so varint encoding doesn't make sense.
tests: adapt for LLVM change 5b386b864c7619897c51a1da97d78f1cf6f3eff6
The above-mentioned change modified the output of thread-local.rs by changing some variable names. Rather than assume things get put in %0, we capture the variable so the test passes in both the old and new version.
Check freeze with right param-env in `deduced_param_attrs`
We're checking if a trait (`Freeze`) holds in a polymorphic function, but not using that function's own (reveal-all) param-env. This causes us to try to eagerly normalize a specializable projection type that has no default value, which causes an ICE.
Fixes#110171
Various minor Idx-related tweaks
Nothing particularly exciting here, but a couple of things I noticed as I was looking for more index conversions to simplify.
cc https://github.com/rust-lang/compiler-team/issues/606
r? `@WaffleLapkin`
Correct default value for default-linker-libraries
This setting is false by default according to rustc code here: https://github.com/rust-lang/rust/blob/master/compiler/rustc_session/src/options.rs#L1236
I tested on a project and confirmed the behavior described. First, with no value, the `-nodefaultlibs` linker flag is present. Setting this to false has no effect, as well. The linker flag still appears. Setting it to true removes the linker flag as expected.
The above-mentioned change modified the output of thread-local.rs by
changing some variable names. Rather than assume things get put in %0,
we capture the variable so the test passes in both the old and new
version.
Remove some suspicious cast truncations
These truncations were added a long time ago, and as best I can tell without a perf justification. And with rust-lang/rust#110410 it has become perf-neutral to not truncate anymore. We worked hard for all these bits, let's use them.
Rollup of 7 pull requests
Successful merges:
- #110038 (Erase regions when confirming transmutability candidate)
- #110341 (rustdoc: stop passing a title to `replaceState` second argument)
- #110388 (Add a message for if an overflow occurs in `core::intrinsics::is_nonoverlapping`.)
- #110404 (fix clippy::toplevel_ref_arg and ::manual_map)
- #110421 (Spelling librustdoc)
- #110423 (Spelling srcdoc)
- #110433 (Windows: map a few more error codes to ErrorKind)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Windows: map a few more error codes to ErrorKind
NotFound errors:
* `ERROR_INVALID_DRIVE`: The system cannot find the drive specified
* `ERROR_BAD_NETPATH`: The network path was not found
* `ERROR_BAD_NET_NAME`: The network name cannot be found.
InvalidFilename:
* `ERROR_BAD_PATHNAME`: The specified path is invalid.
Source: [System Error Codes (0-499)](https://learn.microsoft.com/en-us/windows/win32/debug/system-error-codes--0-499-)
Spelling librustdoc
This is split from https://github.com/rust-lang/rust/pull/110392
There's one change to src/tools/rustdoc-gui/tester.js which feels like a reasonable thing to piggy-back here.
rustdoc: stop passing a title to `replaceState` second argument
As described on [MDN's replaceState page], this parameter is not currently used, and the empty string is "safe against future changes to the method."
[MDN's replaceState page]: https://developer.mozilla.org/en-US/docs/Web/API/History/replaceState
Alloc `hir::Lit` in an arena to remove the destructor from `Expr`
This allows allocating `Expr`s into a dropless arena, which is useful for using length prefixed thing slices in HIR, since these can only be allocated in the dropless arena and not in a typed arena.
Permit MIR inlining without #[inline]
I noticed that there are at least a handful of portable-simd functions that have no `#[inline]` but compile to an assign + return.
I locally benchmarked inlining thresholds between 0 and 50 in increments of 5, and 50 seems to be the best. Interesting. That didn't include check builds though, ~maybe perf will have something to say about that~.
Perf has little useful to say about this. We generally regress all the check builds, as best as I can tell, due to a number of small codegen changes in a particular hot function in the compiler. Probably this is because we've nudged the inlining outcomes all over, and uses of `#[inline(always)]`/`#[inline(never)]` might need to be adjusted.
NotFound errors:
* `ERROR_INVALID_DRIVE`: The system cannot find the drive specified
* `ERROR_BAD_NETPATH`: The network path was not found
* `ERROR_BAD_NET_NAME`: The network name cannot be found.
InvalidFilename:
* `ERROR_BAD_PATHNAME`: The specified path is invalid.
Make cargo a workspace
8 commits in 7bf43f028ba5eb1f4d70d271c2546c38512c9875..39116ccc9b420a883a98a960f0597f9cf87414b8
2023-04-10 16:01:41 +0000 to 2023-04-15 20:24:15 +0000
- Make cargo a workspace (rust-lang/cargo#11851)
- Fix flaky not_found_permutations test. (rust-lang/cargo#11976)
- Use restricted Damerau-Levenshtein algorithm (rust-lang/cargo#11963)
- Correct the bug report for `cargo clippy --fix` (rust-lang/cargo#11882)
- Stabilize `cargo logout` (rust-lang/cargo#11950)
- Add more information to HTTP errors to help with debugging. (rust-lang/cargo#11878)
- Use registry.default for login/logout (rust-lang/cargo#11949)
- Change -C to be unstable (rust-lang/cargo#11960)
---
### What does this PR try to resolve?
Making cargo a workspace.
Why doing this?
* `rustc-workspace-hack` is primarily for sharing dependencies between rls and cargo, as rls previously depends on cargo. After rls retired, it is no longer the case sharing dependencies.
* It's q bit painful that cargo needs to deal with some dependency and licensing complexities. For example, #108665 failed because of the interaction bewteen `windows-sys` and `raw-dylib`. It currenctly blocks cargo's feature `-Zgitxodie` from moving forward.
* See rust-lang/cargo#11851
### Benchmark result
I've done a simple benchmark on both keeping or removing entire `rustc-workspace-hack`. It had no significant difference. Both took ~2m30s to finish `./x.py build -j8 src/tools/cargo src/tools/rls src/tools/clippy src/tools/miri src/tools/rustfmt`. Environment info:
```
host: aarch64-apple-darwin
os: Mac OS 13.2.1 [64-bit]
```
A sophisticated benchmark may be needed.
### Additional information
This depends on prior works from `@Muscraft` and `@ehuss.` Credits to them!
Like #107044, this will let us track compatibility with LLVM 16 going
forward, especially after we eventually upgrade our own to the next.
This also drops `tidy` here and in `x86_64-gnu-llvm-15`, syncing with
that change in #106085.
