Rollup of 7 pull requests
Successful merges:
- #56677 (#[must_use] on traits in stdlib)
- #56679 (overhaul external doc attribute diagnostics)
- #56682 (Update the stdsimd submodule)
- #56691 (fix install broken link)
- #56710 (Always set the RDRAND and RDSEED features on SGX)
- #56713 (Test capacity of ZST vector)
- #56841 (Add some unit tests to compiletest)
Failed merges:
r? @ghost
Add some unit tests to compiletest
Based on #56792, otherwise the tests won't be executed on CI.
Just a small start, I would like to add more testing to compiletest in the future but that will require some refactoring first.
cc #47606
Test capacity of ZST vector
Initially, #50233 accidentally changed the capacity of empty ZST. This was pointed out during code review. This commit adds a test to prevent capacity of ZST vectors from accidentally changing to prevent that from happening again.
Always set the RDRAND and RDSEED features on SGX
Not sure if this is 100% correct.
This [Intel article](https://software.intel.com/en-us/articles/intel-software-guard-extensions-tutorial-part-5-enclave-development) goes in great depth regarding using (untrusted) CPUID to see whether RDRAND/RDSEED is supported, and explains what happens to the enclave if the CPUID result is faked.
I'd say that an implementation of SGX that doesn't make RDRAND available to the enclave is so severely limited/broken that it's ok if you get #UD in that case. The case is less clear for RDSEED, but it so far every processor released by Intel with SGX support also has RDSEED (including Gemini Lake).
cc @briansmith
overhaul external doc attribute diagnostics
This PR improves the error handling and spans for the external doc attribute. Many cases that silently failed before now emit errors, spans are tightened, and the errors have help and suggestions.
I tried to address all the cases that users ran into in the tracking issue.
cc #44732
r? @QuietMisdreavus
#[must_use] on traits in stdlib
Based on #55506.
Adds `#[must_use]` attribute to traits in the stdlib:
- `Iterator`
- `Future`
- `FnOnce`
- `Fn`
- `FnMut`
There may be other traits that should have the attribute, but I couldn't find/think of any.
The (currently) single unit test of the compiletest tool was never
executed on CI. At least I couldn't find any references of it in the
logs. This adds a test suite for compiletest so that our tester is
tested, too.
The compiletest tests can then also be executed with:
./x.py test src/tools/compiletest
Fix docs path to PermissionsExt
Couldn't test the link yet, since I didn't figure out how to build std rustdocs without building the entire compiler itself 🙂
Bootstrap: Add testsuite for compiletest tool
This adds a test suite for compiletest so that the tester is tested, too.
The (currently) single unit test of the compiletest tool was never executed
on CI. At least I couldn't find any references of it in the logs.
The compiletest tests can then also be executed with:
./x.py test src/tools/compiletest --stage 0
cc #47606
Add x86_64-unknown-uefi target
This adds a new rustc target-configuration called 'x86_64-unknown_uefi'.
Furthermore, it adds a UEFI base-configuration to be used with other
targets supported by UEFI (e.g., i386, armv7hl, aarch64, itanium, ...).
UEFI systems provide a very basic operating-system environment, meant
to unify how systems are booted. It is tailored for simplicity and fast
setup, as it is only meant to bootstrap other systems. For instance, it
copies most of the ABI from Microsoft Windows, rather than inventing
anything on its own. Furthermore, any complex CPU features are
disabled. Only one CPU is allowed to be up, no interrupts other than
the timer-interrupt are allowed, no process-separation is performed,
page-tables are identity-mapped, ...
Nevertheless, UEFI has an application model. Its main purpose is to
allow operating-system vendors to write small UEFI applications that
load their kernel and terminate the UEFI system. However, many other
UEFI applications have emerged in the past, including network-boot,
debug-consoles, and more.
This UEFI target allows to compile rust code natively as UEFI
applications. No standard library support is added, but libcore can be
used out-of-the-box if a panic-handler is provided. Furthermore,
liballoc works as well, if a `GlobalAlloc` handler is provided. Both
have been tested with this target-configuration.
Note that full libstd support is unlikely to happen. While UEFI does
have standardized interfaces for networking and alike, none of these
are mandatory and they are unlikely to be shipped in common consumer
firmwares. Furthermore, several features like process-separation are
not available (or only in very limited fashion). Those parts of libstd
would have to be masked.
Add short emoji status to toolstate updates
I get a lot of these emails and it's good to know which ones I should be paying closer attention to -- i.e. the ones where clippy breaks. This adds a short emoji status report to the first line of the commit message, which shows up in notifications directly
I haven't been able to test it, and the actual emoji are just suggestions.
r? @kennytm
cc @rust-lang/infra @rust-lang/devtools
Add test of current behavior (infer free region within closure body)
This behavior was previously not encoded in our test suite.
it is pretty important that we test this behavior. In particular, in #56537 I had proposed expanding the lifetime elision rules so that they would apply to some of the cases encoded in this test, which would cause them to start failing to compile successfully (because the lifetime attached to the return type would start being treated as connected to the lifetime on the input parameter to the lambda expression, which is explicitly *not* what the code wants in this particular case).
In other words, I am trying to ensure that anyone who tries such experiments with lifetime elision in the future quickly finds out why we don't support lifetime elision on lambda expressions (at least not in the naive manner described on #56537).
Fix private_no_mangle_fns message grammar
Simply changes "an warning" to "a warning" in the `private_no_mangle_fns` warning. I started getting this in some code after upgrading to 1.31.0.
fix rust-lang/rust issue #50583
Rationale for the fix is in #50583. I've verified that before the fix /musl-armhf/lib/libc.a is riddled with the illegal variant of vmov.f64 and after the fix the version built doesn't contain any of these illegal instructions.
I originally thought that the arm-linux-gnueabi version also needed fixing - to add a -mfloat-abi-soft but that's unnecessary as it's compiled with the gnueabi (not hf) compiler (I've some a quick check that the libc.a produced doesn't include VFP instructions).
r? @alexcrichton
2018 edition - confusing error message when declaring unnamed parameters
Fixes#53990.
This PR adds a note providing context for the change to argument
names being required in the 2018 edition for trait methods and a
suggestion for the fix.
Greatly improve rustdoc rendering speed issues
Fixes#55900.
So a few improvements here:
* we're switching to `DOMTokenList` API when available providing a replacement if it isn't (should only happen on safari and IE I think...)
* hide doc sections by default to allow the whole HTML generation to happen in the background to avoid triggering DOM redraw all the times (which killed the performances)
r? @QuietMisdreavus
std: Activate compiler_builtins `mem` feature for no_std targets
This was an accidental regression from #56092, but for `no_std` targets
being built and distributed we want to be sure to activate the
compiler-builtins `mem` feature which demangles important memory-related
intrinsics.
The (currently) single unit test of the compiletest tool was never
executed on CI. At least I couldn't find any references of it in the
logs. This adds a test suite for compiletest so that our tester is
tested, too.
The compiletest tests can then also be executed with:
./x.py test src/tools/compiletest
Contexually dependent error message for E0424 when value is assigned to "self"
This is an improvement for pull request #54495 referencing issue #54369. If the "self" keyword is assigned a value as though it were a valid identifier, it will now report:
```
let self = "self";
^^^^ `self` value is a keyword and may not be bound to variables or shadowed
```
instead of
```
let self = "self";
^^^^ `self` value is a keyword only available in methods with `self` parameter
```
If anyone has a better idea for what the error should be I'd be happy to modify it appropriately.
This was an accidental regression from #56092, but for `no_std` targets
being built and distributed we want to be sure to activate the
compiler-builtins `mem` feature which demangles important memory-related
intrinsics.
rustc: Add an unstable `simd_select_bitmask` intrinsic
This is going to be required for binding a number of AVX-512 intrinsics
in the `stdsimd` repository, and this intrinsic is the same as
`simd_select` except that it takes a bitmask as the first argument
instead of a SIMD vector. This bitmask is then transmuted into a `<NN x
i8>` argument, depending on how many bits it is.
cc rust-lang-nursery/stdsimd#310
Disable btree pretty-printers on older gdbs
gdb versions before 8.1 have a bug that prevents the BTreeSet and
BTreeMap pretty-printers from working. This patch disables the test
on those versions, and also disables the pretty-printers there as
well.
Closes#56730
Update panic message to be clearer about env-vars
Esteban Kuber requested that the panic message make it clear
that `RUST_BACKTRACE=1` is an environment variable. This change
makes that clear.
I understand that this may simply be closed if the concept isn't accepted, and I'd be fine with that :-)
Fixes#56734
Remove unneeded extra chars to reduce search-index size
Before:
```
2013782 Dec 11 10:16 build/x86_64-unknown-linux-gnu/doc/search-index.js
```
After:
```
1736597 Dec 11 10:50 build/x86_64-unknown-linux-gnu/doc/search-index.js
```
No changes in the output of the search.
r? @QuietMisdreavus
Use a `newtype_index!` within `Symbol`.
This shrinks `Option<Symbol>` from 8 bytes to 4 bytes, which shrinks
`Token` from 24 bytes to 16 bytes. This reduces instruction counts by up
to 1% across a range of benchmarks.
r? @oli-obk
