Don't emit divide-by-zero panic paths in `StepBy::len`
I happened to notice today that there's actually two such calls emitted in the assembly: <https://rust.godbolt.org/z/1Wbbd3Ts6>
Since they're impossible, hopefully telling LLVM that will also help optimizations elsewhere.
Fix argument ABI for overaligned structs on ppc64le
When passing a 16 (or higher) aligned struct by value on ppc64le, it needs to be passed as an array of `i128` rather than an array of `i64`. This will force the use of an even starting doubleword.
For the case of a 16 byte struct with alignment 16 it is important that `[1 x i128]` is used instead of `i128` -- apparently, the latter will get treated similarly to `[2 x i64]`, not exhibiting the correct ABI. Add a `force_array` flag to `Uniform` to support this.
The relevant clang code can be found here:
fe2119a7b0/clang/lib/CodeGen/Targets/PPC.cpp (L878-L884)fe2119a7b0/clang/lib/CodeGen/Targets/PPC.cpp (L780-L784)
I think the corresponding psABI wording is this:
> Fixed size aggregates and unions passed by value are mapped to as
> many doublewords of the parameter save area as the value uses in
> memory. Aggregrates and unions are aligned according to their
> alignment requirements. This may result in doublewords being
> skipped for alignment.
In particular the last sentence. Though I didn't find any wording for Clang's behavior of clamping the alignment to 16.
Fixes https://github.com/rust-lang/rust/issues/122767.
r? `@cuviper`
Implement minimal, internal-only pattern types in the type system
rebase of https://github.com/rust-lang/rust/pull/107606
You can create pattern types with `std::pat::pattern_type!(ty is pat)`. The feature is incomplete and will panic on you if you use any pattern other than integral range patterns. The only way to create or deconstruct a pattern type is via `transmute`.
This PR's implementation differs from the MCP's text. Specifically
> This means you could implement different traits for different pattern types with the same base type. Thus, we just forbid implementing any traits for pattern types.
is violated in this PR. The reason is that we do need impls after all in order to make them usable as fields. constants of type `std::time::Nanoseconds` struct are used in patterns, so the type must be structural-eq, which it only can be if you derive several traits on it. It doesn't need to be structural-eq recursively, so we can just manually implement the relevant traits on the pattern type and use the pattern type as a private field.
Waiting on:
* [x] move all unrelated commits into their own PRs.
* [x] fix niche computation (see 2db07f94f44f078daffe5823680d07d4fded883f)
* [x] add lots more tests
* [x] T-types MCP https://github.com/rust-lang/types-team/issues/126 to finish
* [x] some commit cleanup
* [x] full self-review
* [x] remove 61bd325da19a918cc3e02bbbdce97281a389c648, it's not necessary anymore I think.
* [ ] ~~make sure we never accidentally leak pattern types to user code (add stability checks or feature gate checks and appopriate tests)~~ we don't even do this for the new float primitives
* [x] get approval that [the scope expansion to trait impls](https://rust-lang.zulipchat.com/#narrow/stream/326866-t-types.2Fnominated/topic/Pattern.20types.20types-team.23126/near/427670099) is ok
r? `@BoxyUwU`
When passing a 16 (or higher) aligned struct by value on ppc64le,
it needs to be passed as an array of `i128` rather than an array
of `i64`. This will force the use of an even starting register.
For the case of a 16 byte struct with alignment 16 it is important
that `[1 x i128]` is used instead of `i128` -- apparently, the
latter will get treated similarly to `[2 x i64]`, not exhibiting
the correct ABI. Add a `force_array` flag to `Uniform` to support
this.
The relevant clang code can be found here:
fe2119a7b0/clang/lib/CodeGen/Targets/PPC.cpp (L878-L884)fe2119a7b0/clang/lib/CodeGen/Targets/PPC.cpp (L780-L784)
I think the corresponding psABI wording is this:
> Fixed size aggregates and unions passed by value are mapped to as
> many doublewords of the parameter save area as the value uses in
> memory. Aggregrates and unions are aligned according to their
> alignment requirements. This may result in doublewords being
> skipped for alignment.
In particular the last sentence.
Fixes https://github.com/rust-lang/rust/issues/122767.
Use unchecked_sub in str indexing
https://github.com/rust-lang/rust/pull/108763 applied this logic to indexing for slices, but of course `str` has its own separate impl.
Found this by skimming over the codegen for https://github.com/oxidecomputer/hubris/; their dist builds enable overflow checks so the lack of `unchecked_sub` was producing an impossible-to-hit overflow check and also inhibiting some inlining.
r? scottmcm
I happened to notice today that there's actually two such calls emitted in the assembly: <https://rust.godbolt.org/z/1Wbbd3Ts6>
Since they're impossible, hopefully telling LLVM that will also help optimizations elsewhere.
CFI: Don't rewrite ty::Dynamic directly
Now that we're using a type folder, the arguments in predicates are processed automatically - we don't need to descend manually.
We also want to keep projection clauses around, and this does so.
r? `@compiler-errors`
Now that we're using a type folder, the arguments in predicates are
processed automatically - we don't need to descend manually.
We also want to keep projection clauses around, and this does so.
CFI: Restore typeid_for_instance default behavior
Restore typeid_for_instance default behavior of performing self type erasure, since it's the most common case and what it does most of the time. Using concrete self (or not performing self type erasure) is for assigning a secondary type id, and secondary type ids are only assigned when they're unique and to methods, and also are only tested for when methods are used as function pointers.
Restore typeid_for_instance default behavior of performing self type
erasure, since it's the most common case and what it does most of the
time. Using concrete self (or not performing self type erasure) is for
assigning a secondary type id, and secondary type ids are only assigned
when they're unique and to methods, and also are only tested for when
methods are used as function pointers.
CFI: Support function pointers for trait methods
Adds support for both CFI and KCFI for function pointers to trait methods by attaching both concrete and abstract types to functions.
KCFI does this through generation of a `ReifyShim` on any function pointer for a method that could go into a vtable, and keeping this separate from `ReifyShim`s that are *intended* for vtable us by setting a `ReifyReason` on them.
CFI does this by setting both the concrete and abstract type on every instance.
This should land after #123024 or a similar PR, as it diverges the implementation of CFI vs KCFI.
r? `@compiler-errors`
Rename `UninhabitedEnumBranching` to `UnreachableEnumBranching`
Per [#120268](https://github.com/rust-lang/rust/pull/120268#discussion_r1517492060), I rename `UninhabitedEnumBranching` to `UnreachableEnumBranching` .
I solved some nits to add some comments.
I adjusted the workaround restrictions. This should be useful for `a <= b` and `if let Some/Ok(v)`. For enum with few variants, `early-tailduplication` should not cause compile time overhead.
r? RalfJung
Add `Ord::cmp` for primitives as a `BinOp` in MIR
Update: most of this OP was written months ago. See https://github.com/rust-lang/rust/pull/118310#issuecomment-2016940014 below for where we got to recently that made it ready for review.
---
There are dozens of reasonable ways to implement `Ord::cmp` for integers using comparison, bit-ops, and branches. Those differences are irrelevant at the rust level, however, so we can make things better by adding `BinOp::Cmp` at the MIR level:
1. Exactly how to implement it is left up to the backends, so LLVM can use whatever pattern its optimizer best recognizes and cranelift can use whichever pattern codegens the fastest.
2. By not inlining those details for every use of `cmp`, we drastically reduce the amount of MIR generated for `derive`d `PartialOrd`, while also making it more amenable to MIR-level optimizations.
Having extremely careful `if` ordering to μoptimize resource usage on broadwell (#63767) is great, but it really feels to me like libcore is the wrong place to put that logic. Similarly, using subtraction [tricks](https://graphics.stanford.edu/~seander/bithacks.html#CopyIntegerSign) (#105840) is arguably even nicer, but depends on the optimizer understanding it (https://github.com/llvm/llvm-project/issues/73417) to be practical. Or maybe [bitor is better than add](https://discourse.llvm.org/t/representing-in-ir/67369/2?u=scottmcm)? But maybe only on a future version that [has `or disjoint` support](https://discourse.llvm.org/t/rfc-add-or-disjoint-flag/75036?u=scottmcm)? And just because one of those forms happens to be good for LLVM, there's no guarantee that it'd be the same form that GCC or Cranelift would rather see -- especially given their very different optimizers. Not to mention that if LLVM gets a spaceship intrinsic -- [which it should](https://rust-lang.zulipchat.com/#narrow/stream/131828-t-compiler/topic/Suboptimal.20inlining.20in.20std.20function.20.60binary_search.60/near/404250586) -- we'll need at least a rustc intrinsic to be able to call it.
As for simplifying it in Rust, we now regularly inline `{integer}::partial_cmp`, but it's quite a large amount of IR. The best way to see that is with 8811efa88b (diff-d134c32d028fbe2bf835fef2df9aca9d13332dd82284ff21ee7ebf717bfa4765R113) -- I added a new pre-codegen MIR test for a simple 3-tuple struct, and this PR change it from 36 locals and 26 basic blocks down to 24 locals and 8 basic blocks. Even better, as soon as the construct-`Some`-then-match-it-in-same-BB noise is cleaned up, this'll expose the `Cmp == 0` branches clearly in MIR, so that an InstCombine (#105808) can simplify that to just a `BinOp::Eq` and thus fix some of our generated code perf issues. (Tracking that through today's `if a < b { Less } else if a == b { Equal } else { Greater }` would be *much* harder.)
---
r? `@ghost`
But first I should check that perf is ok with this
~~...and my true nemesis, tidy.~~
This is just one part of the MCP, but it's the one that IMHO removes the most noise from the standard library code.
Seems net simpler this way, since MIR already supported heterogeneous shifts anyway, and thus it's not more work for backends than before.
Remove len argument from RawVec::reserve_for_push
Removes `RawVec::reserve_for_push`'s `len` argument since it's always the same as capacity.
Also makes `Vec::insert` use `RawVec::reserve_for_push`.
CFI: Fix methods as function pointer cast
Fix casting between methods and function pointers by assigning a secondary type id to methods with their concrete self so they can be used as function pointers.
This was split off from #116404.
cc `@compiler-errors` `@workingjubilee`
Codegen const panic messages as function calls
This skips emitting extra arguments at every callsite (of which there
can be many). For a librustc_driver build with overflow checks enabled,
this cuts 0.7MB from the resulting shared library (see [perf]).
A sample improvement from nightly:
```
leaq str.0(%rip), %rdi
leaq .Lalloc_d6aeb8e2aa19de39a7f0e861c998af13(%rip), %rdx
movl $25, %esi
callq *_ZN4core9panicking5panic17h17cabb89c5bcc999E@GOTPCREL(%rip)
```
to this PR:
```
leaq .Lalloc_d6aeb8e2aa19de39a7f0e861c998af13(%rip), %rdi
callq *_RNvNtNtCsduqIKoij8JB_4core9panicking11panic_const23panic_const_div_by_zero@GOTPCREL(%rip)
```
[perf]: https://perf.rust-lang.org/compare.html?start=a7e4de13c1785819f4d61da41f6704ed69d5f203&end=64fbb4f0b2d621ff46d559d1e9f5ad89a8d7789b&stat=instructions:u
Fix casting between methods and function pointers by assigning a
secondary type id to methods with their concrete self so they can be
used as function pointers.
CFI: Fix drop and drop_in_place
Fix drop and drop_in_place by transforming self of drop and drop_in_place methods into a Drop trait objects.
This was split off from https://github.com/rust-lang/rust/pull/116404.
cc `@compiler-errors` `@workingjubilee`
CFI: Support self_cell-like recursion
Current `transform_ty` attempts to avoid cycles when normalizing `#[repr(transparent)]` types to their interior, but runs afoul of this pattern used in `self_cell`:
```
struct X<T> {
x: u8,
p: PhantomData<T>,
}
#[repr(transparent)]
struct Y(X<Y>);
```
When attempting to normalize Y, it will still cycle indefinitely. By using a types-visited list, this will instead get expanded exactly one layer deep to X<Y>, and then stop, not attempting to normalize `Y` any further.
This PR was split off from #121962 as part of fixing the larger vtable compatibility issues.
r? ``````@workingjubilee``````
Let codegen decide when to `mem::swap` with immediates
Making `libcore` decide this is silly; the backend has so much better information about when it's a good idea.
Thus this PR introduces a new `typed_swap` intrinsic with a fallback body, and replaces that fallback implementation when swapping immediates or scalar pairs.
r? oli-obk
Replaces #111744, and means we'll never need more libs PRs like #111803 or #107140
Current `transform_ty` attempts to avoid cycles when normalizing
`#[repr(transparent)]` types to their interior, but runs afoul of this
pattern used in `self_cell`:
```
struct X<T> {
x: u8,
p: PhantomData<T>,
}
#[repr(transparent)]
struct Y(X<Y>);
```
When attempting to normalize Y, it will still cycle indefinitely. By
using a types-visited list, this will instead get expanded exactly
one layer deep to X<Y>, and then stop, not attempting to normalize `Y`
any further.
This skips emitting extra arguments at every callsite (of which there
can be many). For a librustc_driver build with overflow checks enabled,
this cuts 0.7MB from the resulting binary.
CFI: Skip non-passed arguments
Rust will occasionally rely on fn((), X) -> Y being compatible with fn(X) -> Y, since () is a non-passed argument. Relax CFI by choosing not to encode non-passed arguments.
This PR was split off from #121962 as part of fixing the larger vtable compatibility issues.
r? `@workingjubilee`
Rust will occasionally rely on fn((), X) -> Y being compatible with
fn(X) -> Y, since () is a non-passed argument. Relax CFI by choosing not
to encode non-passed arguments.
CFI: Break tests into smaller files
Break type metadata identifiers tests into smaller set of tests/files, and move CFI (and KCFI) codegen tests to a cfi (and kcfi) subdirectory,
Making `libcore` decide this is silly; the backend has so much better information about when it's a good idea.
So introduce a new `typed_swap` intrinsic with a fallback body, but replace that implementation for immediates and scalar pairs.
Tests added in cast-target-abi.rs, covering the single element, array,
and prefix cases in `CastTarget::llvm_type`, and the Rust-is-larger/smaller
cases in the Rust<->ABI copying code.
ffi-out-of-bounds-loads.rs was overhauled to be runnable on any
platform. Its alignment also increases due to the removal of a `min` in
the previous commit; this was probably an insufficient workaround for
this issue or similar. The higher alignment is fine, since the alloca is
actually aligned to 8 bytes, as the test checks now confirm.
Stop walking the bodies of statics for reachability, and evaluate them instead
cc `@saethlin` `@RalfJung`
cc #119214
This reuses the `DefIdVisitor` from `rustc_privacy`, because they basically try to do the same thing.
This PR's changes can probably be extended to constants, too, but let's tackle that separately, it's likely more involved.
