Particularly for ctfe-stress-4, the hashing of byte slices as part of the
MIR Allocation is quite hot. Previously, we were falling back on byte-by-byte
copying of the slice into the SipHash buffer (64 bytes long) before hashing a 64
byte chunk, and then doing that again and again.
This should hopefully be an improvement for that code.
Inline CStr::from_ptr
Inlining this function is valuable, as it allows LLVM to apply `strlen`-specific optimizations without having to enable LTO.
For instance, the following function:
```rust
pub fn f(p: *const c_char) -> Option<u8> {
unsafe { CStr::from_ptr(p) }.to_bytes().get(0).copied()
}
```
Looks like this if `CStr::from_ptr` is allowed to be inlined.
```asm
before:
push rax
call qword ptr [rip + std::ffi::c_str::CStr::from_ptr@GOTPCREL]
mov rcx, rax
cmp rdx, 1
sete dl
test rax, rax
sete al
or al, dl
jne .LBB1_2
mov dl, byte ptr [rcx]
.LBB1_2:
xor al, 1
pop rcx
ret
after:
mov dl, byte ptr [rdi]
test dl, dl
setne al
ret
```
Note that optimization turned this from O(N) to O(1) in terms of performance, as LLVM knows that it doesn't really need to call `strlen` to determine whether a string is empty or not.
Rollup of 14 pull requests
Successful merges:
- #87537 (Clarify undefined behaviour in binary heap, btree and hashset docs)
- #88624 (Stabilize feature `saturating_div` for rust 1.58.0)
- #89257 (Give better error for `macro_rules name`)
- #89665 (Ensure that pushing empty path works as before on verbatim paths)
- #89895 (Don't mark for loop iter expression as desugared)
- #89922 (Update E0637 description to mention `&` w/o an explicit lifetime name)
- #89944 (Change `Duration::[try_]from_secs_{f32, f64}` underflow error)
- #89991 (rustc_ast: Turn `MutVisitor::token_visiting_enabled` into a constant)
- #90028 (Reject closures in patterns)
- #90069 (Fix const qualification when executed after promotion)
- #90078 (Add a regression test for issue-83479)
- #90114 (Add some tests for const_generics_defaults)
- #90115 (Add test for issue #78561)
- #90129 (triagebot: Treat `I-*nominated` like `I-nominated`)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
triagebot: Treat `I-*nominated` like `I-nominated`
rustbot doesn't allow unauthenticated users to set `I-nominated`; apply the same permissions to the new `I-*nominated` labels.
Add some tests for const_generics_defaults
I think this covers some of the stuff required for stabilisation report, some of these tests are probably covering stuff we already have but it can't hurt to have more :)
r? ````@lcnr````
Fix const qualification when executed after promotion
The const qualification was so far performed before the promotion and
the implementation assumed that it will never encounter a promoted.
With `const_precise_live_drops` feature, checking for live drops is
delayed until after drop elaboration, which in turn runs after
promotion. so the assumption is no longer true. When evaluating
`NeedsNonConstDrop` it is now possible to encounter promoteds.
Use type base qualification for the promoted. It is a sound
approximation in general, and in the specific case of promoteds and
`NeedsNonConstDrop` it is precise.
Fixes#89938.
rustc_ast: Turn `MutVisitor::token_visiting_enabled` into a constant
It's a visitor property rather than something that needs to be determined at runtime
Update E0637 description to mention `&` w/o an explicit lifetime name
Deal with https://github.com/rust-lang/rust/issues/89824#issuecomment-941598647. Another solution would be splitting the error code into two as (I think) it's a bit unclear to users why they have the same error code.
Don't mark for loop iter expression as desugared
We typically don't mark spans of lowered things as desugared. This helps Clippy rightly discern when code is (not) from expansion. This was discovered by ``@flip1995`` at https://github.com/rust-lang/rust-clippy/pull/7789#issuecomment-939289501.
Stabilize feature `saturating_div` for rust 1.58.0
The tracking issue is #89381
This seems like a reasonable simple change(?). The feature `saturating_div` was added as part of the ongoing effort to implement a `Saturating` integer type (see #87921). The implementation has been discussed [here](https://github.com/rust-lang/rust/pull/87921#issuecomment-899357720) and [here](https://github.com/rust-lang/rust/pull/87921#discussion_r691888556). It extends the list of saturating operations on integer types (like `saturating_add`, `saturating_sub`, `saturating_mul`, ...) by the function `fn saturating_div(self, rhs: Self) -> Self`.
The stabilization of the feature `saturating_int_impl` (for the `Saturating` type) needs to have this stabilized first.
Closes#89381
Clarify undefined behaviour in binary heap, btree and hashset docs
Previously, it wasn't clear whether "This could include" was referring to logic errors, or undefined behaviour. Tweak wording to clarify this sentence does not relate to UB.
Previously, it wasn't clear whether "This could include" was referring
to logic errors, or undefined behaviour. Tweak wording to clarify this
sentence does not relate to UB.
Fix typo for invalid_null_ptr_usage and missing_inline_in_public_items.
Fix typo for invalid_null_ptr_usage and missing_inline_in_public_items.
changelog: none
Rollup of 14 pull requests
Successful merges:
- #86984 (Reject octal zeros in IPv4 addresses)
- #87440 (Remove unnecessary condition in Barrier::wait())
- #88644 (`AbstractConst` private fields)
- #89292 (Stabilize CString::from_vec_with_nul[_unchecked])
- #90010 (Avoid overflow in `VecDeque::with_capacity_in()`.)
- #90029 (Add test for debug logging during incremental compilation)
- #90031 (config: add the option to enable LLVM tests)
- #90048 (Add test for line-number setting)
- #90071 (Remove hir::map::blocks and use FnKind instead)
- #90074 (2229 migrations small cleanup)
- #90077 (Make `From` impls of NonZero integer const.)
- #90097 (Add test for duplicated sidebar entries for reexported macro)
- #90098 (Add test to ensure that the missing_doc_code_examples is not triggered on foreign trait implementations)
- #90099 (Fix MIRI UB in `Vec::swap_remove`)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Fix MIRI UB in `Vec::swap_remove`
Fixes#90055
I find it weird that `Vec::swap_remove` read the last element to the stack just to immediately put it back in the `Vec` in place of the one at index `index`. It seems much more natural to me to just read the element at position `index` and then move the last element in its place. I guess this might also slightly improve codegen.
Make `From` impls of NonZero integer const.
I also changed the feature gate added to `From` impls of Atomic integer to `const_num_from_num` from `const_convert`.
Tracking issue: #87852
Remove hir::map::blocks and use FnKind instead
The principal tool is `FnLikeNode`, which is not often used and can be easily implemented using `rustc_hir::intravisit::FnKind`.
Add test for line-number setting
The first commit updates the version of the package to be able to have multi-line commands (which looks much nicer for this test).
r? ````@jsha````
config: add the option to enable LLVM tests
I'm working on some LLVM patches in concert with a Rust patch, and it's
helping me quite a bit to have this as an option. It doesn't seem that
hard, so I figured I'd formalize it in x.py and send it upstream.
Add test for debug logging during incremental compilation
Debug logging during incremental compilation had been broken for some
time, until #89343 fixed it (among other things). Add a test so this is
less likely to break without being noticed. This test is nearly a copy
of the `src/test/ui/rustc-rust-log.rs` test, but tests debug logging in
the incremental compliation code paths.
Avoid overflow in `VecDeque::with_capacity_in()`.
The overflow only happens if alloc is compiled with overflow checks enabled and the passed capacity is greater or equal 2^(usize::BITS-1). The overflow shadows the expected "capacity overflow" panic leading to a test failure if overflow checks are enabled for std in the CI.
Unblocks [CI: Enable overflow checks for test (non-dist) builds #89776](https://github.com/rust-lang/rust/pull/89776).
For some reason the overflow is only observable with optimization turned off, but that is a separate issue.
Stabilize CString::from_vec_with_nul[_unchecked]
Closes the tracking issue #73179. I am keeping this in _draft_ mode until the FCP has ended.
This is my first time stabilizing a feature, so I would appreciate any guidance on things I should do differently.
Closes#73179
Remove unnecessary condition in Barrier::wait()
This is my first pull request for Rust, so feel free to call me out if anything is amiss.
After some examination, I realized that the second condition of the "spurious-wakeup-handler" loop in ``std::sync::Barrier::wait()`` should always evaluate to ``true``, making it redundant in the ``&&`` expression.
Here is the affected function before the fix:
```rust
#[stable(feature = "rust1", since = "1.0.0")]
pub fn wait(&self) -> BarrierWaitResult {
let mut lock = self.lock.lock().unwrap();
let local_gen = lock.generation_id;
lock.count += 1;
if lock.count < self.num_threads {
// We need a while loop to guard against spurious wakeups.
// https://en.wikipedia.org/wiki/Spurious_wakeup
while local_gen == lock.generation_id && lock.count < self.num_threads { // fixme
lock = self.cvar.wait(lock).unwrap();
}
BarrierWaitResult(false)
} else {
lock.count = 0;
lock.generation_id = lock.generation_id.wrapping_add(1);
self.cvar.notify_all();
BarrierWaitResult(true)
}
}
```
At first glance, it seems that the check that ``lock.count < self.num_threads`` would be necessary in order for a thread A to detect when another thread B has caused the barrier to reach its thread count, making thread B the "leader".
However, the control flow implicitly results in an invariant that makes observing ``!(lock.count < self.num_threads)``, i.e. ``lock.count >= self.num_threads`` impossible from thread A.
When thread B, which will be the leader, calls ``.wait()`` on this shared instance of the ``Barrier``, it locks the mutex in the first line and saves the ``MutexGuard`` in the ``lock`` variable. It then increments the value of ``lock.count``. However, it then proceeds to check if ``lock.count < self.num_threads``. Since it is the leader, it is the case that (after the increment of ``lock.count``), the lock count is *equal* to the number of threads. Thus, the second branch is immediately taken and ``lock.count`` is zeroed. Additionally, the generation ID is incremented (with wrap). Then, the condition variable is signalled. But, the other threads are waiting at the line ``lock = self.cvar.wait(lock).unwrap();``, so they cannot resume until thread B's call to ``Barrier::wait()`` returns, which drops the ``MutexGuard`` acquired in the first ``let`` statement and unlocks the mutex.
The order of events is thus:
1. A thread A calls `.wait()`
2. `.wait()` acquires the mutex, increments `lock.count`, and takes the first branch
3. Thread A enters the ``while`` loop since the generation ID has not changed and the count is less than the number of threads for the ``Barrier``
3. Spurious wakeups occur, but both conditions hold, so the thread A waits on the condition variable
4. This process repeats for N - 2 additional times for non-leader threads A'
5. *Meanwhile*, Thread B calls ``Barrier::wait()`` on the same barrier that threads A, A', A'', etc. are waiting on. The thread count reaches the number of threads for the ``Barrier``, so all threads should now proceed, with B being the leader. B acquires the mutex and increments the value ``lock.count`` only to find that it is not less than ``self.num_threads``. Thus, it immediately clamps ``self.num_threads`` back down to 0 and increments the generation. Then, it signals the condvar to tell the A (prime) threads that they may continue.
6. The A, A', A''... threads wake up and attempt to re-acquire the ``lock`` as per the internal operation of a condition variable. When each A has exclusive access to the mutex, it finds that ``lock.generation_id`` no longer matches ``local_generation`` **and the ``&&`` expression short-circuits -- and even if it were to evaluate it, ``self.count`` is definitely less than ``self.num_threads`` because it has been reset to ``0`` by thread B *before* B dropped its ``MutexGuard``**.
Therefore, it my understanding that it would be impossible for the non-leader threads to ever see the second boolean expression evaluate to anything other than ``true``. This PR simply removes that condition.
Any input would be appreciated. Sorry if this is terribly verbose. I'm new to the Rust community and concurrency can be hard to explain in words. Thanks!
Reject octal zeros in IPv4 addresses
This fixes#86964 by rejecting octal zeros in IP addresses, such that `192.168.00.00000000` is rejected with a parse error, since having leading zeros in front of another zero indicates it is a zero written in octal notation, which is not allowed in the strict mode specified by RFC 6943 3.1.1. Octal rejection was implemented in #83652, but due to the way it was implemented octal zeros were still allowed.
Don't emit a warning for empty rmeta files.
This avoids displaying a warning when attempting to load an empty rmeta file. Warnings were enabled via #89634 which can cause a lot of noise (for example, running `./x.py check`). rustc generates empty rmeta files for things like binaries, which can happen when checking libraries as unittests.
Closes#89795
