Remove dead code in `rustc_session::Options`
- Don't recompile the same functions for each debugging option
This reduces the amount of items in the crate by quite a lot.
- Remove unused `parse_opt_list` and `parse_pathbuf_push` functions
- Remove unused macro parameters
- Remove `allow(dead_code)`.
Don't download cargo twice when download-rustc is set
Previously, this caused a bug on NixOS:
1. bootstrap.py would download and patch stage0/cargo
2. bootstrap.py would download nightly cargo, but extract it to
stage0/cargo instead of ci-rustc/cargo. It would still try (and fail) to patch ci-rustc/cargo.
3. bootstrap.py would fail to build rustbuild because stage0/cargo
wasn't patched.
The "proper" fix is to extract nightly cargo to ci-rustc instead, but it
doesn't seem to be necessary at all, so this just skips downloading it
instead.
Fixes https://github.com/rust-lang/rust/issues/84702
[Arm64] use isb instruction instead of yield in spin loops
On arm64 we have seen on several databases that ISB (instruction synchronization
barrier) is better to use than yield in a spin loop. The yield instruction is a
nop. The isb instruction puts the processor to sleep for some short time. isb
is a good equivalent to the pause instruction on x86.
Below is an experiment that shows the effects of yield and isb on Arm64 and the
time of a pause instruction on x86 Intel processors. The micro-benchmarks use
https://github.com/google/benchmark.git
```
$ cat a.cc
static void BM_scalar_increment(benchmark::State& state) {
int i = 0;
for (auto _ : state)
benchmark::DoNotOptimize(i++);
}
BENCHMARK(BM_scalar_increment);
static void BM_yield(benchmark::State& state) {
for (auto _ : state)
asm volatile("yield"::);
}
BENCHMARK(BM_yield);
static void BM_isb(benchmark::State& state) {
for (auto _ : state)
asm volatile("isb"::);
}
BENCHMARK(BM_isb);
BENCHMARK_MAIN();
$ g++ -o run a.cc -O2 -lbenchmark -lpthread
$ ./run
--------------------------------------------------------------
Benchmark Time CPU Iterations
--------------------------------------------------------------
AWS Graviton2 (Neoverse-N1) processor:
BM_scalar_increment 0.485 ns 0.485 ns 1000000000
BM_yield 0.400 ns 0.400 ns 1000000000
BM_isb 13.2 ns 13.2 ns 52993304
AWS Graviton (A-72) processor:
BM_scalar_increment 0.897 ns 0.874 ns 801558633
BM_yield 0.877 ns 0.875 ns 800002377
BM_isb 13.0 ns 12.7 ns 55169412
Apple Arm64 M1 processor:
BM_scalar_increment 0.315 ns 0.315 ns 1000000000
BM_yield 0.313 ns 0.313 ns 1000000000
BM_isb 9.06 ns 9.06 ns 77259282
```
```
static void BM_pause(benchmark::State& state) {
for (auto _ : state)
asm volatile("pause"::);
}
BENCHMARK(BM_pause);
Intel Skylake processor:
BM_scalar_increment 0.295 ns 0.295 ns 1000000000
BM_pause 41.7 ns 41.7 ns 16780553
```
Tested on Graviton2 aarch64-linux with `./x.py test`.
Allow running `x.py test --stage 2 src/tools/linkchecker` with `download-rustc = true`
Previously, the LD_LIBRARY_PATH for the linkchecker looked like
`build/x86_64-unknown-linux-gnu/stage1/lib/rustlib/x86_64-unknown-linux-gnu/lib`, because the linkchecker depends on the master copy of the standard library. This is true, but doesn't include the library path for the compiler libraries:
```
/home/joshua/src/rust/rust/build/x86_64-unknown-linux-gnu/stage1-tools-bin/error_index_generator: error while loading shared libraries: libLLVM-12-rust-1.53.0-nightly.so: cannot open shared object file: No such file or directory
```
That file is in
`build/x86_64-unknown-linux-gnu/stage1/lib/libLLVM-12-rust-1.53.0-nightly.so`,
which wasn't included in the dynamic path. This adds `build/x86_64-unknown-linux-gnu/stage1/lib` to the dynamic path for the linkchecker.
Move HIR parenting information out of hir_owner
Split out of #82681.
The parent of a HIR node and its content are currently bundled together, but are rarely used together.
This PR separates both information in two distinct queries for HIR owners.
This reduces incremental invalidation for HIR items that appear within a function body when this body (and the local ids) changes.
added --no-run option for rustdoc
resolve#59053
add `--no-run` option for `rustdoc` for compiling doc test but not running them.
Intended for use with `--persist-doctests`.
Be stricter about rejecting LLVM reserved registers in asm!
LLVM will silently produce incorrect code if these registers are used as operands.
cc `@rust-lang/wg-inline-asm`
Rollup of 8 pull requests
Successful merges:
- #84601 (rustdoc: Only store locations in Cache::extern_locations and calculate the other info on-demand)
- #84704 (platform-support.md: Update for consistency with Target Tier Policy)
- #84724 (Replace llvm::sys::fs::F_None with llvm::sys::fs::OF_None)
- #84740 (Reset the docs' copy path button after 1 second)
- #84749 (Sync `rustc_codegen_cranelift`)
- #84756 (Add a ToC to the Target Tier Policy documentation)
- #84765 (Update cargo)
- #84774 (Fix misspelling)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Add a ToC to the Target Tier Policy documentation
The policy document is quite lengthy, I figured it might be good to have a quick way to jump to the specific tier policies.
Sync `rustc_codegen_cranelift`
Retrying #84746
r? ``@bjorn3``
---
Edit(bjorn3): Since the last sync there have been some refactorings around the driver code in preparation for a planned new feature. In addition ``@mominul`` implemented `-Ctarget-cpu` support and ``@XAMPPRocky`` fixed compilation of cg_clif itself for Windows with the MSVC toolchain.
Reset the docs' copy path button after 1 second
I like that this copy path button on the top next to the type/module's name changes to a check mark when you successfully clicked and copied the path but I find it really weird how the icon stays that check mark forever after the first time of clicking it. Imagine you leave that documentation tab open and come back after 2 hours and you still see that check mark in that box because you copied the path 2 hours ago. You will probably be confused and you might've forgotten what that button even does (even more so currently where this is a new feature, or when you simply don't use it often), so I really think at some point it should go back to the ⎘ icon which, at least to me, pretty clearly indicates copying, whereas the check mark (if it stays there for so long) could falsely look like a verification mark indicating "this module is verified" or something like that.
I believe after a longer period of time it's not logical to still tell the user "yes you've copied this successful".
In addition to this timeout, maybe it could be made so that you can't copy again until this cooldown of 1 second is over, but I'm not sure how useful or user-friendly that feature would be so maybe it's fine the way it is now.
Also the timeout is cleared every time you click again so if you constantly click it, it won't reset during that.
platform-support.md: Update for consistency with Target Tier Policy
Split into five sections to match the tiers: "Tier 1 with Host Tools",
"Tier 1", "Tier 2 with Host Tools", "Tier 2", and "Tier 3". Explain each
tier briefly in prose, and link to the corresponding section of the
policy for full requirements.
Drop the `host` columns from the first four, since the different
sections distinguish that. (Keep the `host` column for "Tier 3", since
it's a single list and the `host` column just indicates if host tools
are expected to work.)
Targets with host tools always have full support for std, so drop the
`std` column from those.
Move the explanations of the `std` column next to the appropriate
tables, and drop the unknown/WIP case for tier 2 targets.
Use "target" terminology consistently throughout.
Sort each table by target name.
Vastly improves coverage spans for macros
Fixes: #84561
This resolves problems where macros like `trace!(...)` would show zero coverage if tracing was disabled, and `assert_eq!(...)` would show zero coverage if the assertion did not fail, because only one coverage span was generated, for the branch.
This PR started with an idea that I could just drop branching blocks with same span as expanded macro. (See the fixed issue for more details.)
That did help, but it didn't resolve everything.
I also needed to add a span specifically for the macro name (plus `!`) to ensure the macro gets coverage even if it's internal expansion adds conditional branching blocks that are retained, and would otherwise drop the outer span. Now that outer span is _only_ the `(argument, list)`, which can safely be dropped now), because the macro name has its own span.
While testing, I also noticed the spanview debug output can cause an ICE on a function with no body. The
workaround for this is included in this PR (separate commit).
r? `@tmandry`
cc? `@wesleywiser`
Move iter_results to dyn FnMut rather than a generic
This means that we're no longer generating the iteration/locking code for each invocation site of iter_results, rather just once per query (roughly), which seems much better: this is a 15% win in instruction counts when compiling the rustc_query_impl crate. The code where this is used also is pretty cold, I suspect; the old solution didn't fully monomorphize either.
Update LLVM for more wasm simd updates
This fixes the temporary regression introduced in #84339 where the wasm
target uses `fpto{s,u}i` intrinsics but the codegen for those intrinsics
with the `+nontrapping-fptoint` LLVM feature wasn't very good (aka it
didn't use the wasm instruction). The fixes brought in here fix that and
also implement the second-to-last simd instruction in LLVM.
This fixes the temporary regression introduced in #84339 where the wasm
target uses `fpto{s,u}i` intrinsics but the codegen for those intrinsics
with the `+nontrapping-fptoint` LLVM feature wasn't very good (aka it
didn't use the wasm instruction). The fixes brought in here fix that and
also implement the second-to-last simd instruction in LLVM.
