In MIR construction, operands need to live exactly until they are used,
which is during the (sub)expression that made the call to `as_operand`.
Before this PR, operands lived until the end of the temporary scope,
which was sometimes unnecessarily longer and sometimes too short.
Fixes#38669.
This reduces the number of dynamic drops in libstd from 1141 to 899.
However, without this change, the next patch would have created much
more dynamic drops.
A basic merge unswitching hack reduced the number of dynamic drops to
644, with no effect on stack usage. I should be writing a more dedicated
drop unswitching pass.
No performance measurements.
Make transmuting from fn item types to pointer-sized types a hard error.
Closes#19925 by removing the future compatibility lint and the associated workarounds.
This is a `[breaking-change]` if you `transmute` from a function item without casting first.
For more information on how to fix your code, see https://github.com/rust-lang/rust/issues/19925.
[12/12] On-demand type-checking, const-evaluation, MIR building & const-qualification.
_This is the last of a series ([prev](https://github.com/rust-lang/rust/pull/38813)) of patches designed to rework rustc into an out-of-order on-demand pipeline model for both better feature support (e.g. [MIR-based](https://github.com/solson/miri) early constant evaluation) and incremental execution of compiler passes (e.g. type-checking), with beneficial consequences to IDE support as well.
If any motivation is unclear, please ask for additional PR description clarifications or code comments._
<hr>
As this contains all of the changes that didn't fit neatly into other PRs, I'll be elaborating a bit:
### User-facing changes
* when determining whether an `impl Trait` type implements an auto-trait (e.g. `Send` or `Sync`), the function the `impl Trait` came from has to be inferred and type-checking, disallowing cycles
* this results from not having an obvious place to put the "deferred obligation" in on-demand atm
* while we could model side-effects like that and "post-processing passes" better, it's still more limiting than being able to know the result in the original function (e.g. specialization) *and* there are serious problems around region-checking (if a `Send` impl required `'static`, it wasn't enforced)
* early const-eval requires type-checking and const-qualification to be performed first, which means:
* you get the intended errors before (if any) constant evaluation error that is simply fallout
* associated consts should always work now, and `const fn` type parameters are properly tracked
* don't get too excited, array lengths still can't depend on type parameters
* #38864 works as intended now, with `Self` being allowed in `impl` bounds
* #32205 is largely improved, with associated types being limited to "exact match" `impl`s (as opposed to traversing the specialization graph to resolve unspecified type parameters to their defaults in another `impl` or in the `trait`) *while* checking for overlaps building the specialization graph for that trait - once all the trait impls' have been checked for coherence (including ahead-of-time/on-demand), it's uniform
* [crater report](https://gist.github.com/eddyb/bbb869072468c7e08d6d808e75938051) looks clean (aside from `clippy` which broke due to `rustc` internal changes)
### Compiler-internal changes
* `ty::Generics`
* no longer contains the actual type parameter defaults, instead they're associated with the type parameter's `DefId`, like associated types in a trait definition
* this allows computing `ty::Generics` as a leaf (reading only its own HIR)
* holds a mapping from `DefIndex` of type parameters to their indices
* `ty::AdtDef`
* only tracks `#[repr(simd)]` in its `ReprOptions` `repr` field
* doesn't contain `enum` discriminant values, but instead each variant either refers to either an explicit value for its discriminant, or the distance from the last explicit discriminant, if any
* the `.discriminants(tcx)` method produces an iterator of `ConstInt` values, looking up explicit discriminants in a separate map, if necessary
* this allows computing `ty::AdtDef` as a leaf (reading only its own HIR)
* Small note: the two above (`Generics`, `AdtDef`), `TraitDef` and `AssociatedItem` should probably end up as part of the HIR, eventually, as they're trivially constructed from it
* `ty::FnSig`
* now also holds ABI and unsafety, alongside argument types, return type and C variadicity
* `&ty::BareFnTy` and `ty::ClosureTy` have been replaced with `PolyFnSig = Binder<FnSig>`
* `BareFnTy` was interned and `ClosureTy` was treated as non-trivial to `Clone` because they had a `PolyFnSig` and so used to contain a `Vec<Ty>` (now `&[Ty]`)
* `ty::maps`
* all the `DepTrackingMap`s have been grouped in a structure available at `tcx.maps`
* when creating the `tcx`, a set of `Providers` (one `fn` pointer per map) is required for the local crate, and one for all other crates (i.e. metadata loading), `librustc_driver` plugging the various crates (e.g. `librustc_metadata`, `librustc_typeck`, `librustc_mir`) into it
* when a map is queried and the value is missing, the appropriate `fn` pointer from the `Providers` of that crate is called with the `TyCtxt` and the key being queried, to produce the value on-demand
* `rustc_const_eval`
* demands both `typeck_tables` and `mir_const_qualif` (in preparation for miri)
* tracks `Substs` in `ConstVal::Function` for `const fn` calls
* returns `TypeckError` if type-checking has failed (or cases that can only be reached if it had)
* this error kind is never reported, resulting in less noisy/redundant diagnostics
* fixes#39548 (testcase by @larsluthman, taken from #39812, which this supersedes)
* on-demand has so far been hooked up to:
* `rustc_metadata::cstore_impl`: `ty`, `generics`, `predicates`, `super_predicates`, `trait_def`, `adt_def`, `variances`, `associated_item_def_ids`, `associated_item`, `impl_trait_ref`, `custom_coerce_unsized_kind`, `mir`, `mir_const_qualif`, `typeck_tables`, `closure_kind`, `closure_type`
* `rustc_typeck::collect`: `ty`, `generics`, `predicates`, `super_predicates`, `type_param_predicates`, `trait_def`, `adt_def`, `impl_trait_ref`
* `rustc_typeck::coherence`: `coherent_trait`, `coherent_inherent_impls`
* `rustc_typeck::check`: `typeck_tables`, `closure_type`, `closure_kind`
* `rustc_mir::mir_map`: `mir`
* `rustc_mir::transform::qualify_consts`: `mir_const_qualif`
Improve backtrace formating while panicking.
Fixes#37783.
Done:
- Fix alignment of file paths for better readability
- `RUST_BACKTRACE=full` prints all the informations (current behaviour)
- `RUST_BACKTRACE=(short|yes)` is the default and does:
- Skip irrelevant frames at the beginning and the end
- Remove function address
- Remove the current directory from the absolute paths
- Remove `::hfabe6541873` at the end of the symbols
- `RUST_BACKTRACE=(0|no)` disables the backtrace.
- `RUST_BACKTRACE=<everything else>` is equivalent to `short` for
backward compatibility.
- doc
- More uniform printing across platforms.
Removed, TODO in a new PR:
- Remove path prefix for libraries and libstd
Example of short backtrace:
```rust
fn fail() {
panic!();
}
fn main() {
let closure = || fail();
closure();
}
```
Short:
```
thread 'main' panicked at 'explicit panic', t.rs:2
Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
stack backtrace:
0: t::fail
at ./t.rs:2
1: t::main::{{closure}}
at ./t.rs:6
2: t::main
at ./t.rs:7
```
Full:
```
thread 'main' panicked at 'This function never returns!', t.rs:2
stack backtrace:
0: 0x558ddf666478 - std::sys:👿:backtrace::tracing:👿:unwind_backtrace::hec84c9dd8389cc5d
at /home/yamakaky/dev/rust/rust/src/libstd/sys/unix/backtrace/tracing/gcc_s.rs:49
1: 0x558ddf65d90e - std::sys_common::backtrace::_print::hfa25f8b31f4b4353
at /home/yamakaky/dev/rust/rust/src/libstd/sys_common/backtrace.rs:71
2: 0x558ddf65cb5e - std::sys_common::backtrace::print::h9b711e11ac3ba805
at /home/yamakaky/dev/rust/rust/src/libstd/sys_common/backtrace.rs:60
3: 0x558ddf66796e - std::panicking::default_hook::{{closure}}::h736d216e74748044
at /home/yamakaky/dev/rust/rust/src/libstd/panicking.rs:355
4: 0x558ddf66743c - std::panicking::default_hook::h16baff397e46ea10
at /home/yamakaky/dev/rust/rust/src/libstd/panicking.rs:371
5: 0x558ddf6682bc - std::panicking::rust_panic_with_hook::h6d5a9bb4eca42c80
at /home/yamakaky/dev/rust/rust/src/libstd/panicking.rs:559
6: 0x558ddf64ea93 - std::panicking::begin_panic::h17dc549df2f10b99
at /home/yamakaky/dev/rust/rust/src/libstd/panicking.rs:521
7: 0x558ddf64ec42 - t::diverges::he6bc43fc925905f5
at /tmp/p/t.rs:2
8: 0x558ddf64ec5a - t::main::h0ffc20356b8a69c0
at /tmp/p/t.rs:6
9: 0x558ddf6687f5 - core::ops::FnOnce::call_once::hce41f19c0db56f93
10: 0x558ddf667cde - std::panicking::try::do_call::hd4c8c97efb4291df
at /home/yamakaky/dev/rust/rust/src/libstd/panicking.rs:464
11: 0x558ddf698d77 - __rust_try
12: 0x558ddf698c57 - __rust_maybe_catch_panic
at /home/yamakaky/dev/rust/rust/src/libpanic_unwind/lib.rs:98
13: 0x558ddf667adb - std::panicking::try::h2c56ed2a59ec1d12
at /home/yamakaky/dev/rust/rust/src/libstd/panicking.rs:440
14: 0x558ddf66cc9a - std::panic::catch_unwind::h390834e0251cc9af
at /home/yamakaky/dev/rust/rust/src/libstd/panic.rs:361
15: 0x558ddf6809ee - std::rt::lang_start::hb73087428e233982
at /home/yamakaky/dev/rust/rust/src/libstd/rt.rs:57
16: 0x558ddf64ec92 - main
17: 0x7fecb869e290 - __libc_start_main
18: 0x558ddf64e8b9 - _start
19: 0x0 - <unknown>
```
trans: don't ICE when trying to create ADT trans-items
ADTs are translated in-place from rustc_trans::callee, so no trans-items
are needed.
This fix will be superseded by the shimmir branch, but I prefer not to
backport that to beta.
Fixes#39823.
Beta-nominating because regression.
r? @michaelwoerister
add `-C overflow-checks` option
In addition to defining and handling the new option, we also add a method on librustc::Session for determining the necessity of overflow checks. This method provides a single point to sort out the three (!) different ways for turning on overflow checks: -C debug-assertions, -C overflow-checks, and -Z force-overflow-checks.
I was seeing a [run-pass/issue-28950.rs](b1363a73ed/src/test/run-pass/issue-28950.rs) failure on my machine with these patches, but I was also seeing the failure without the changes to the core compiler. We'll see what travis says.
Fixes#33134. r? @alexcrichton
check_match: don't treat privately uninhabited types as uninhabited
Fixes#38972, which is a regression in 1.16 from @canndrew's patchset.
r? @nikomatsakis
beta-nominating because regression.
ADTs are translated in-place from rustc_trans::callee, so no trans-items
are needed.
This fix will be superseded by the shimmir branch, but I prefer not to
backport that to beta.
Fixes#39823.
In addition to defining and handling the new option, we also add a
method on librustc::Session for determining the necessity of overflow
checks. This method provides a single point to sort out the three (!)
different ways for turning on overflow checks: -C debug-assertions, -C
overflow-checks, and -Z force-overflow-checks.
Fixes#33134.
Report full details of inference errors
When the old suggestion machinery was removed by @brson in https://github.com/rust-lang/rust/pull/37057, it was not completely removed. There was a bit of code that had the job of going through errors and finding those for which suggestions were applicable, and it remained, causing us not to emit the full details of such errors. This PR removes that.
I've also added various lifetime tests to the UI test suite (so you can also see the before/after there). I have some concrete thoughts on how to improve these cases and am planning on writing those up in some mentoring issues (@CengizIO has expressed interest in working on those changes, so I plan to work with him on it, at least to start).
cc @jonathandturner
erase late bound regions in `get_vtable_methods()`
Higher-ranked object types can otherwise cause late-bound regions to
sneak into the substs, leading to the false conclusion that some method
is unreachable.
r? @arielb1, who wrote the heart of this patch anyhow
Fixes#39292
Higher-ranked object types can otherwise cause late-bound regions to
sneak into the substs, leading to the false conclusion that some method
is unreachable. The heart of this patch is from @arielb1.
- `RUST_BACKTRACE=full` prints all the informations (old behaviour)
- `RUST_BACKTRACE=(0|no)` disables the backtrace.
- `RUST_BACKTRACE=<everything else>` (including `1`) shows a simplified
backtrace, without the function addresses and with cleaned filenames
and symbols. Also removes some unneded frames at the beginning and the
end.
Fixes#37783.
PR is #38165.
remove wrong packed struct test
This UB was found by running the test under [Miri](https://github.com/solson/miri) which rejects these unsafe unaligned loads. 😄
Fix unsafe unaligned loads in test.
r? @eddyb
cc @Aatch @nikomatsakis
The `#[derive(PartialEq, Debug)]` impls on a packed struct contain undefined behaviour. Both generated impls take references to unaligned fields, which will fail to compile once we correctly treat that as unsafe (see https://github.com/rust-lang/rust/issues/27060).
This UB was found by running the test under [Miri](https://github.com/solson/miri/) which rejects these unsafe unaligned loads. 😄
Here's a simpler example:
```rust
struct Packed {
a: u8,
b: u64,
}
```
It expands to:
```rust
fn fmt(&self, __arg_0: &mut ::std::fmt::Formatter) -> ::std::fmt::Result {
match *self {
Packed { a: ref __self_0_0, b: ref __self_0_1 } => { // BAD: these patterns are unsafe
let mut builder = __arg_0.debug_struct("Packed");
let _ = builder.field("a", &&(*__self_0_0));
let _ = builder.field("b", &&(*__self_0_1));
builder.finish()
}
}
}
```
and
```rust
fn eq(&self, __arg_0: &Packed) -> bool {
match *__arg_0 {
Packed { a: ref __self_1_0, b: ref __self_1_1 } => // BAD: these patterns are unsafe
match *self {
Packed { a: ref __self_0_0, b: ref __self_0_1 } => // BAD: these patterns are unsafe
true && (*__self_0_0) == (*__self_1_0) &&
(*__self_0_1) == (*__self_1_1),
},
}
}
```
Stabilize static lifetime in statics
Stabilize the "static_in_const" feature. Blockers before this PR can be merged:
* [x] The [FCP with inclination to stabilize](https://github.com/rust-lang/rust/issues/35897#issuecomment-270441437) needs to be over. FCP lasts roughly three weeks, so will be over at Jan 25, aka this thursday.
* [x] Documentation needs to be added (#37928)
Closes#35897.
