Fixes for LLVM change 0f45c16f2caa7c035e5c3edd40af9e0d51ad6ba7
More details in the individual commit messages, but the summary is: LLVM deleted an unused-to-them method that we used, we worked around it to avoid annoying cleanup/restructuring in the Rust-side code.
Fix debugger stepping behavior with `match` expressions
Previously, we would set up the source lines for `match` expressions so
that the code generated to perform the test of the scrutinee was matched
to the line of the arm that required the test and then jump from the arm
block to the "next" block was matched to all of the lines in the `match`
expression.
While that makes sense, it has the side effect of causing strange
stepping behavior in debuggers.
I've changed the source information so that all of the generated tests
are sourced to `match {scrutinee}` and the jumps are sourced to the last
line of the block they are inside. This resolves the weird stepping
behavior in all debuggers and resolves some instances of "ambiguous
symbol" errors in WinDbg preventing the user from setting breakpoints at
`match` expressions.
Before:
https://user-images.githubusercontent.com/831192/128577421-ee0c9c03-da28-4d16-997a-d57988a7bb7f.mp4
After:
https://user-images.githubusercontent.com/831192/128577433-2ceab04d-953e-4e31-9387-93f049c71ff3.mp4Fixes#87817
- All attributes for an item need to be considered at once, they can't
be considered a line at a time.
- The top-level crate was not being visited. This bug was caught by
`extern-crate-used-only-in-link`, which I'm very glad I added.
- Make the loader private to the module, so that only one function is
exposed.
Morph `layout_raw` query into `layout_of`.
Before this PR, `LayoutCx::layout_of` wrapped the `layout_raw` query, to:
* normalize the type, before attempting to compute the layout
* pass the layout to `record_layout_for_printing`, for `-Zprint-type-sizes`
Moving those two responsibilities into the query may reduce overhead (due to cached calls skipping those steps), but I want to do a perf run to know.
One of the changes I had to make was changing the return type of the query, to be able to both get out the type produced by normalizing inside the query *and* to match the signature of the old `TyCtxt::layout_of`. This change may be worse, perf-wise, so that's another reason I want to check.
r? `@nagisa` cc `@oli-obk`
The above-mentioned commit (part of the LLVM 14 development cycle)
removes a method that rustc uses somewhat extensively. We mostly switch
to lower-level methods that exist in all versions of LLVM we use, so no
new ifdef logic is required in most cases.
Link the static libraries with "-bundle" modifier from upstream rust crate
right after linking this rust crate. Some linker such as GNU linker
`ld.bdf` treat order of linking as order of dependency. After this change,
static libraries with "-bundle" modifier is linked in the same order as
"+bundle" modifier. So we can change the value of "bundle" modifier without
causing linking error.
Use undef for uninitialized bytes in constants
Fixes#83657
This generates good code when the const is fully uninit, e.g.
```rust
#[no_mangle]
pub const fn fully_uninit() -> MaybeUninit<[u8; 10]> {
const M: MaybeUninit<[u8; 10]> = MaybeUninit::uninit();
M
}
```
generates
```asm
fully_uninit:
ret
```
as you would expect.
There is no improvement, however, when it's partially uninit, e.g.
```rust
pub struct PartiallyUninit {
x: u64,
y: MaybeUninit<[u8; 10]>
}
#[no_mangle]
pub const fn partially_uninit() -> PartiallyUninit {
const X: PartiallyUninit = PartiallyUninit { x: 0xdeadbeefcafe, y: MaybeUninit::uninit() };
X
}
```
generates
```asm
partially_uninit:
mov rax, rdi
mov rcx, qword ptr [rip + .L__unnamed_1+16]
mov qword ptr [rdi + 16], rcx
movups xmm0, xmmword ptr [rip + .L__unnamed_1]
movups xmmword ptr [rdi], xmm0
ret
.L__unnamed_1:
.asciz "\376\312\357\276\255\336\000"
.zero 16
.size .L__unnamed_1, 24
```
which copies a bunch of zeros in place of the undef bytes, the same as before this change.
Edit: generating partially-undef constants isn't viable at the moment anyways due to #84565, so it's disabled
Use if-let guards in the codebase and various other pattern cleanups
Dogfooding if-let guards as experimentation for the feature.
Tracking issue #51114. Conflicts with #87937.
This can happen when a tag is opened after the length limit is reached;
the tag will not end up being added to `unclosed_tags` because the queue
will never be flushed. So, now, if the `unclosed_tags` stack is empty,
`close_tag()` does nothing.
This change fixes a panic in the `limit_0` unit test.
PGO for LLVM builds on x86_64-unknown-linux-gnu in CI
This shows up to 6% less instruction counts with larger - up to 18% - wins on cycles
on multiple benchmarks, and up to 19% wins on the -j1 wall times for rustc self-compilation.
We can afford to spend the extra cycles building LLVM essentially once more for
the x86_64-unknown-linux-gnu CI build today. The builder finishes in around 50
minutes on average, and this adds just 10 more minutes. Given the sizeable
improvements in compiler performance, this is definitely worth it.
