Commit Graph

427 Commits

Author SHA1 Message Date
bors
4d45b0745a Auto merge of - cjgillot:mir-cost-visit, r=compiler-errors
Check projection types before inlining MIR

Fixes https://github.com/rust-lang/rust/issues/100550

I'm very unhappy with this solution, having to duplicate MIR validation code, but at least it removes the ICE.

r? `@compiler-errors`
2022-08-25 08:16:43 +00:00
bors
5462da52ba Auto merge of - tmiasko:elide-storage-makers, r=oli-obk
Elide superfluous storage markers

Follow the existing strategy of omitting the storage markers for temporaries
introduced for internal usage when elaborating derefs and deref projections.

Those temporaries are simple scalars which are used immediately after being
defined and never have their address taken. There is no benefit from storage
markers from either liveness analysis or code generation perspective.
2022-08-24 23:51:06 +00:00
nils
8e6c5ad696 Fix typo in UnreachableProp 2022-08-23 08:18:18 +02:00
Tomasz Miąsko
625af2cc75 Elide storage markers for internal locals when inlining 2022-08-23 10:08:48 +02:00
Tomasz Miąsko
162bd16352 Elide storage markers when elaborating deref projections 2022-08-23 10:08:48 +02:00
Tomasz Miąsko
943a380b37 Elide storage markers when elaborating box derefs 2022-08-23 10:08:48 +02:00
Nilstrieb
bc79557311 Enable UnreachablePropagation under mir-opt-level >= 2
It was disabled because of pathological behaviour of LLVM in some
benchmarks. As of , this has been fixed. The problem there was
that it caused pessimizations in some cases. These have now been fixed
as well.
2022-08-21 21:15:28 +02:00
Nilstrieb
eafab66096 UnreachableProp: Preserve unreachable branches for multiple targets
Before, UnreachablePropagation removed all unreachable branches.
This was a pessimization, as it removed information about
reachability that was used later in the optimization pipeline.

For example, this code
```rust
pub enum Two { A, B }
pub fn identity(x: Two) -> Two {
    match x {
        Two::A => Two::A,
        Two::B => Two::B,
    }
}
```

basically has `switchInt() -> [0: 0, 1: 1, otherwise: unreachable]` for the match.
This allows it to be transformed into a simple `x`. If we remove the
unreachable branch, the transformation becomes illegal.
2022-08-21 21:15:28 +02:00
Camille GILLOT
10e71dfdb8 Also validate types before inlining. 2022-08-21 12:54:26 +02:00
Camille GILLOT
e4fd2bd721 Refactor cost computation as a visitor. 2022-08-21 12:54:26 +02:00
Dylan DPC
d83abe8c12
Rollup merge of - cjgillot:inline-polymorphic-recursion, r=tmiasko
Only check the `DefId` for the recursion check in MIR inliner.

The current history check compares `Instance`s, so it cannot detect cases of polymorphic recursion where `Substs` change.
This PR makes it so we only compare `DefId`s, ignoring any change in `Substs`.

According to https://github.com/rust-lang/rust/pull/100522#issuecomment-1214769757, in practice only very few inlining decisions change.

Fixes https://github.com/rust-lang/rust/issues/100476
2022-08-19 12:26:43 +05:30
Dylan DPC
2fe2975391
Rollup merge of - RalfJung:unused-unsafe-in-unsafe-fn, r=jackh726
never consider unsafe blocks unused if they would be required with deny(unsafe_op_in_unsafe_fn)

Judging from https://github.com/rust-lang/rust/issues/71668#issuecomment-1200317370 the consensus nowadays seems to be that we should never consider an unsafe block unused if it was required with `deny(unsafe_op_in_unsafe_fn)`, no matter whether that lint is actually enabled or not. So let's adjust rustc accordingly.

The first commit does the change, the 2nd does some cleanup.
2022-08-19 12:26:40 +05:30
Camille GILLOT
86645c9cf7 Ignore substs when checking inlining history. 2022-08-17 19:25:09 +02:00
Mark Rousskov
154a09dd91 Adjust cfgs 2022-08-12 16:28:15 -04:00
Dylan DPC
392ba5f111
Rollup merge of - RalfJung:extra-const-ub-checks, r=lcnr
add -Zextra-const-ub-checks to enable more UB checking in const-eval

Cc https://github.com/rust-lang/rust/issues/99923
r? `@oli-obk`
2022-08-12 20:39:11 +05:30
Matthias Krüger
8237efc52d
Rollup merge of - nnethercote:simplify-visitors, r=cjgillot
Simplify visitors

By removing some unused arguments.

r? `@cjgillot`
2022-08-11 22:53:08 +02:00
Dylan DPC
32bd147f79
Rollup merge of - tmiasko:rm-duplicated-locals, r=nagisa
Remove duplicated temporaries creating during box derefs elaboration

Temporaries created with `MirPatch::new_temp` will be declared after
patch application. Remove manually created duplicate declarations.

Removing duplicates exposes another issue. Visitor elaborates
terminator twice and attempts to access new, but not yet available,
local declarations. Remove duplicated call to `visit_terminator`.

Extracted from .
2022-08-11 22:46:59 +05:30
Nicholas Nethercote
8c5303898e Simplify rustc_hir::intravisit::Visitor::visit_variant_data.
It has four arguments that are never used. This avoids lots of argument
passing in functions that feed into `visit_variant_data`.
2022-08-11 10:54:01 +10:00
Ralf Jung
7b2a5f284e dont rely on old macro-in-trait-impl bug 2022-08-09 08:23:16 -04:00
Jakob Degen
7547084ff6 Add option to mir::MutVisitor to not invalidate CFG.
This also applies that option to some uses of the visitor
2022-08-09 01:51:10 -07:00
Tomasz Miąsko
18a21e13b4 Remove duplicated temporaries creating during box derefs elaboration
Temporaries created with `MirPatch::new_temp` will be declared after
patch application. Remove manually created duplicate declarations.

Removing duplicates exposes another issue. Visitor elaborates
terminator twice and attempts to access new, but not yet available,
local declarations. Remove duplicated call to `visit_terminator`.
2022-08-06 11:14:57 +02:00
Matthias Krüger
01ccde5ec8
Rollup merge of - jackh726:early-binder, r=lcnr
More EarlyBinder cleanups

Each commit is independent

r? types
2022-08-04 22:25:04 +02:00
Jack Huey
955fcad758 Add bound_impl_subject and bound_return_ty 2022-08-03 01:02:46 -04:00
Jakob Degen
efa5eaa5d1 Avoid invalidating the CFG in MirPatch.
As a part of this change, we adjust MirPatch to not needlessly create unnecessary resume blocks.
2022-08-02 18:57:54 -07:00
Ralf Jung
35a35d86fd
update comment
Co-authored-by: Léo Lanteri Thauvin <leseulartichaut@gmail.com>
2022-08-02 18:55:43 -04:00
Ralf Jung
3e44ca95dd remove some unused code and types 2022-08-02 17:14:17 -04:00
Ralf Jung
ee3fc9dff8 never consider unsafe blocks unused if they would be required with unsafe_op_in_unsafe_fn 2022-08-02 17:09:41 -04:00
Dylan DPC
403c1b3802
Rollup merge of - camsteffen:closure-localdefid, r=cjgillot
Use LocalDefId for closures more
2022-07-31 17:36:40 +05:30
Cameron Steffen
cf2433a74f Use LocalDefId for closures more 2022-07-30 15:59:17 -05:00
Miguel Guarniz
25bdc8965e Change maybe_body_owned_by to take local def id
Signed-off-by: Miguel Guarniz <mi9uel9@gmail.com>
2022-07-29 18:25:58 -04:00
bors
7dfdd64433 Auto merge of - ouz-a:some_branch, r=oli-obk
Optimize `UnDerefer`

Addresses the performance [issues](https://github.com/rust-lang/rust/pull/98145#issuecomment-1183548597) faced here.

r? `@oli-obk`
2022-07-29 07:11:50 +00:00
David Wood
7bab769b58 lint: add bad opt access internal lint
Some command-line options accessible through `sess.opts` are best
accessed through wrapper functions on `Session`, `TyCtxt` or otherwise,
rather than through field access on the option struct in the `Session`.

Adds a new lint which triggers on those options that should be accessed
through a wrapper function so that this is prohibited. Options are
annotated with a new attribute `rustc_lint_opt_deny_field_access` which
can specify the error message (i.e. "use this other function instead")
to be emitted.

A simpler alternative would be to simply rename the options in the
option type so that it is clear they should not be used, however this
doesn't prevent uses, just discourages them. Another alternative would
be to make the option fields private, and adding accessor functions on
the option types, however the wrapper functions sometimes rely on
additional state from `Session` or `TyCtxt` which wouldn't be available
in an function on the option type, so the accessor would simply make the
field available and its use would be discouraged too.

Signed-off-by: David Wood <david.wood@huawei.com>
2022-07-27 11:24:27 +01:00
Dylan DPC
962da8bdce
Rollup merge of - nnethercote:rm-E0133, r=Dylan-DPC
Remove erroneous E0133 code from an error message.

This error message is about `derive` and `packed`, but E0133 is for
"Unsafe code was used outside of an unsafe function or block".

r? ``@estebank``
2022-07-26 14:26:59 +05:30
Yuki Okushi
3c1eef2e91
Rollup merge of - tmiasko:coverage, r=wesleywiser
Remove reachable coverage without counters

Remove reachable coverage without counters to maintain invariant that
either there is no coverage at all or there is a live coverage counter
left that provides the function source hash.

The motivating example would be a following closure:

```rust
    let f = |x: bool| {
        debug_assert!(x);
    };
```

Which, with span changes from , with disabled debug assertions,
after the final CFG simplifications but before removal of dead blocks,
gives rise to MIR:

```rust
fn main::{closure#0}(_1: &[closure@a.rs:2:13: 2:22], _2: bool) -> () {
    debug x => _2;
    let mut _0: ();

    bb0: {
        Coverage::Expression(4294967295) = 1 - 2;
        return;
    }

    ...
}
```

Which also makes the initial instrumentation quite suspect, although
this pull request doesn't attempt to address that aspect directly.

Fixes .

r? ``@wesleywiser`` ``@richkadel``
2022-07-26 13:12:22 +09:00
Nicholas Nethercote
6a4b1572f9 Remove erroneous E0133 code from an error message.
This error message is about `derive` and `packed`, but E0133 is for
"Unsafe code was used outside of an unsafe function or block".
2022-07-26 12:54:24 +10:00
Yuki Okushi
74be487ca0
Rollup merge of - Dajamante:clean_up, r=oli-obk
Lighten up const_prop_lint, reusing const_prop

r? `@oli-obk`
2022-07-26 07:14:48 +09:00
Yuki Okushi
29892759f6
Rollup merge of - ouz-a:Optimize-backend, r=oli-obk
Simplify some code that depend on Deref

Now that we can assume  works, it's safe to expect Deref is always in the first place of projections. With this, I was able to simplify some code that depended on Deref's place in projections. When we are able to move Derefer before `ElaborateDrops` successfully we will be able to optimize more places.

r? `@oli-obk`
2022-07-26 07:14:44 +09:00
Aïssata
e6518296dc removed CanConstProp + Visitor 2022-07-25 13:54:49 +00:00
Aïssata
68b433a089 Lighten up const_prop_lint, reusing const_prop 2022-07-25 13:33:44 +00:00
Tomasz Miąsko
5f40a4f7a0 Remove reachable coverage without counters
Remove reachable coverage without counters to maintain invariant that
either there is no coverage at all or there is a live coverage counter
left that provides the function source hash.

The motivating example would be a following closure:

```rust
    let f = |x: bool| {
        debug_assert!(x);
    };
```

Which, with span changes from , with disabled debug assertions,
after the final CFG simplifications but before removal of dead blocks,
gives rise to MIR:

```rust
fn main::{closure#0}(_1: &[closure@a.rs:2:13: 2:22], _2: bool) -> () {
    debug x => _2;
    let mut _0: ();

    bb0: {
        Coverage::Expression(4294967295) = 1 - 2;
        return;
    }

    ...
}
```
2022-07-25 14:14:49 +02:00
Yuki Okushi
921cfbe56f
Rollup merge of - nnethercote:improve-derive-packed-errors, r=estebank
Improve error messages involving `derive` and `packed`.

There are two errors involving `derive` and `packed`.

```
`#[derive]` can't be derived on a `#[repr(packed)]` struct with type or const parameters
`#[derive]` can't be derived on a `#[repr(packed)]` struct that does not derive Copy
```
The second one overstates things. It is possible to use derive on a
repr(packed) struct that doesn't derive Copy in two cases.
- If all the fields within the struct meet the required alignment: 1 for
  `repr(packed)`, or `N` for `repr(packed(N))`.
- If `Default` is the only trait derived.

This commit improves things in a few ways.
- Changes the errors to say `this trait can't be derived on this ...`.
  This is more accurate, because it's just *this* trait and *this*
  packed struct that are a problem, not *all* derived traits on *all*
  packed structs.
- Adds more details to the "ERROR" lines in the test case, enough to
  distinguish between the two error messages.
- Adds more cases to the test case that don't cause errors, e.g. `Default`
  derives.
- Uses a wider variety of builtin traits in the test case, for better coverage.

r? `@estebank`
2022-07-25 18:46:51 +09:00
Nicholas Nethercote
168c5b1839 Improve error messages involving derive and packed.
There are two errors involving `derive` and `packed`.

```
`#[derive]` can't be derived on a `#[repr(packed)]` struct with type or const parameters
`#[derive]` can't be derived on a `#[repr(packed)]` struct that does not derive Copy
```
The second one overstates things. It is possible to use derive on a
repr(packed) struct that doesn't derive Copy in two cases.
- If all the fields within the struct meet the required alignment: 1 for
  `repr(packed)`, or `N` for `repr(packed(N))`.
- If `Default` is the only trait derived.

This commit improves things in a few ways.
- Changes the errors to say `$TRAIT can't be derived on this ...`.
  This is more accurate, because it's just $TRAIT and *this* packed
  struct that are a problem, not *all* derived traits on *all* packed
  structs.
- Adds more details to the "ERROR" lines in the test case, enough to
  distinguish between the two error messages.
- Adds more cases to the test case that don't cause errors, e.g. `Default`
  derives.
- Uses a wider variety of builtin traits in the test case, for better coverage.
2022-07-25 10:30:43 +10:00
ouz-a
09134982e5 optimize un_derefer 2022-07-24 14:40:43 +03:00
ouz-a
447aaceed7 has_deref: simpler comparison, ty fix 2022-07-22 17:35:28 +03:00
ouz-a
c3e1e7a947 simplify more, ret_deref -> has_deref 2022-07-22 17:35:26 +03:00
ouz-a
c0e4230bf5 simplify some code that depend on Deref 2022-07-22 17:32:50 +03:00
Dylan DPC
6e3dd69e36
Rollup merge of - tmiasko:unreachable-coverage, r=wesleywiser
Fix unreachable coverage generation for inlined functions

To generate a function coverage we need at least one coverage counter,
so a coverage from unreachable blocks is retained only when some live
counters remain.

The previous implementation incorrectly retained unreachable coverage,
because it didn't account for the fact that those live counters can
belong to another function due to inlining.

Fixes .
2022-07-22 11:53:40 +05:30
bors
aa01891700 Auto merge of - RalfJung:vtable, r=oli-obk
make vtable pointers entirely opaque

This implements the scheme discussed in https://github.com/rust-lang/unsafe-code-guidelines/issues/338: vtable pointers should be considered entirely opaque and not even readable by Rust code, similar to function pointers.

- We have a new kind of `GlobalAlloc` that symbolically refers to a vtable.
- Miri uses that kind of allocation when generating a vtable.
- The codegen backends, upon encountering such an allocation, call `vtable_allocation` to obtain an actually dataful allocation for this vtable.
- We need new intrinsics to obtain the size and align from a vtable (for some `ptr::metadata` APIs), since direct accesses are UB now.

I had to touch quite a bit of code that I am not very familiar with, so some of this might not make much sense...
r? `@oli-obk`
2022-07-22 01:33:49 +00:00
Wesley Wiser
9473141253
Update compiler/rustc_mir_transform/src/simplify.rs 2022-07-21 11:51:40 -04:00
bors
be9cfb307e Auto merge of - michaelwoerister:remove-stable-set-and-map, r=nagisa
Remove the unused StableSet and StableMap types from rustc_data_structures.

The current implementation is not "stable" in the same sense that `HashStable` and `StableHasher` are stable, i.e. across compilation sessions. So, in my opinion, it's better to remove those types (which are basically unused anyway) than to give the wrong impression that these are safe for incr. comp.

I plan to provide new "stable" collection types soon that can be used to replace `FxHashMap` and `FxHashSet` in query results (see [draft](69d03ac7a7)). It's unsound that `HashMap` and `HashSet` implement `HashStable` (see https://github.com/rust-lang/rust/issues/98890 for a recent P-critical bug caused by this) -- so we should make some progress there.
2022-07-20 22:19:30 +00:00