Commit Graph

472 Commits

Author SHA1 Message Date
bors
255a4c58f5 Auto merge of #72632 - jonas-schievink:dest-prop, r=oli-obk
Implement a generic Destination Propagation optimization on MIR

This takes the work that was originally started by `@eddyb` in https://github.com/rust-lang/rust/pull/47954, and then explored by me in https://github.com/rust-lang/rust/pull/71003, and implements it in a general (ie. not limited to acyclic CFGs) and dataflow-driven way (so that no additional infrastructure in rustc is needed).

The pass is configured to run at `mir-opt-level=2` and higher only. To enable it by default, some followup work on it is still needed:
* Performance needs to be evaluated. I did some light optimization work and tested against `tuple-stress`, which caused trouble in my last attempt, but didn't go much in depth here.
  * We can also enable the pass only at `opt-level=2` and higher, if it is too slow to run in debug mode, but fine when optimizations run anyways.
* Debuginfo needs to be fixed after locals are merged. I did not look into what is required for this.
* Live ranges of locals (aka `StorageLive` and `StorageDead`) are currently deleted. We either need to decide that this is fine, or if not, merge the variable's live ranges (or remove these statements entirely – https://github.com/rust-lang/rust/issues/68622).

Some benchmarks of the pass were done in https://github.com/rust-lang/rust/pull/72635.
2020-09-20 01:38:26 +00:00
Simon Vandel Sillesen
a875c7a1ea Add assertion for len of vecs 2020-09-19 23:22:54 +02:00
Ralf Jung
9216eb8258 fix some comments 2020-09-19 22:14:17 +02:00
Ralf Jung
7b99c8e1cf never promote non-const operations; revert STATIC promotion change 2020-09-19 22:14:17 +02:00
Ralf Jung
7febd5a257 fix doc comment 2020-09-19 22:14:17 +02:00
Ralf Jung
4d1ef03c9e cleanup promotion const_kind checks
in particular allow a few more promotions for consistency when they were already allowed in other contexts
2020-09-19 22:14:17 +02:00
Simon Vandel Sillesen
738ed9b5ec Fix #76803
Check that the variant index matches the target value from the SwitchInt we came from
2020-09-19 15:40:15 +02:00
Ralf Jung
4831523ac4
Rollup merge of #76757 - matthiaskrgr:clippy_try_into, r=lcnr
don't convert types to the same type with try_into (clippy::useless_conversion)
2020-09-19 11:47:52 +02:00
Ralf Jung
f62ba52f5c
Rollup merge of #75502 - ecstatic-morse:implicit-promotion-in-const-fn, r=RalfJung
Use implicit (not explicit) rules for promotability by default in `const fn`

For crater run. See https://github.com/rust-lang/const-eval/pull/54#discussion_r469995552.

cc #75586
2020-09-19 11:47:35 +02:00
Oliver Scherer
b8e6883a2f Reflect the "do not call this query directly" mentality in its name 2020-09-19 10:57:14 +02:00
Oliver Scherer
c3c8c981a8 Rustfmt 2020-09-19 10:36:36 +02:00
Oliver Scherer
182ed8544d Address review comments 2020-09-19 10:36:36 +02:00
Oliver Scherer
888afd50d9 Unify the names of const eval queries and their return types 2020-09-19 10:36:36 +02:00
Oliver Scherer
69a6be73e6 Rename const eval queries to reflect the validation changes 2020-09-19 10:36:36 +02:00
Oliver Scherer
dd9702a059 Do not call the const_eval query in mir interpretation except for caching of nulary intrinsics 2020-09-19 10:36:36 +02:00
Oliver Scherer
48f366fced Replace and_then map_err and_then chain with a match 2020-09-19 10:36:36 +02:00
Oliver Scherer
b1bd34df0c turn_into_const is infallible 2020-09-19 10:36:36 +02:00
Oliver Scherer
4397d66d42 Document op_to_const's purpose 2020-09-19 10:36:36 +02:00
Oliver Scherer
2d7ac728e4 Stop using the const_eval query for initializers of statics
As a side effect, we now represent most promoteds as `ConstValue::Scalar` again. This is useful because all implict promoteds are just references anyway and most explicit promoteds are numeric arguments to `asm!` or SIMD instructions.
2020-09-19 10:36:36 +02:00
Oliver Scherer
083f1d7a37 Validate constants during const_eval_raw 2020-09-19 10:36:36 +02:00
Jonas Schievink
2f9271b14c Clarify FIXME 2020-09-18 21:23:01 +02:00
Jonas Schievink
ffd9445812 Return Place by value 2020-09-18 21:23:01 +02:00
Jonas Schievink
682de94e31 Move inner items outside 2020-09-18 21:23:01 +02:00
Jonas Schievink
cd5d7201ad Fix rebase fallout 2020-09-18 21:23:01 +02:00
Jonas Schievink
7af964fecf Limit block count 2020-09-18 21:23:01 +02:00
Jonas Schievink
5728834448 Fix rebase fallout 2020-09-18 21:23:01 +02:00
Jonas Schievink
484db5b08a Properly inherit conflicts when merging locals 2020-09-18 21:23:01 +02:00
Jonas Schievink
934634eacc More logging 2020-09-18 21:23:01 +02:00
Jonas Schievink
88538adf9a Record intra-statement/terminator conflicts
Some MIR statements and terminators have an (undocumented...) invariant
that some of their input and outputs must not overlap. This records
conflicts between locals used in these positions.
2020-09-18 21:23:01 +02:00
Jonas Schievink
ddd6930b54 perf: bail out when there's >500 candidate locals 2020-09-18 21:23:01 +02:00
Jonas Schievink
ab26fb140c perf: only calculate conflicts for candidates 2020-09-18 21:23:01 +02:00
Jonas Schievink
402f863d8a perf: walk liveness backwards in Conflicts::build 2020-09-18 21:23:00 +02:00
Jonas Schievink
812d4bbc8d Fix dataflow assert errors 2020-09-18 21:23:00 +02:00
Jonas Schievink
78ff69ba10 Implement a destination propagation pass 2020-09-18 21:23:00 +02:00
Matthias Krüger
40dddd3305 use matches!() macro for simple if let conditions 2020-09-18 20:28:35 +02:00
bors
9f8ac718f4 Auto merge of #76575 - lcnr:abstract-const, r=oli-obk
compare generic constants using `AbstractConst`s

This is a MVP of rust-lang/compiler-team#340. The changes in this PR should only be relevant if `feature(const_evaluatable_checked)` is enabled.

~~currently based on top of #76559, so blocked on that.~~

r? `@oli-obk` cc `@varkor` `@eddyb`
2020-09-18 16:59:50 +00:00
Bastian Kauschke
d327fa112b initial working state 2020-09-18 16:25:25 +02:00
bors
fdc3405c20 Auto merge of #72412 - VFLashM:issue-72408-nested-closures-exponential, r=tmandry
Issue 72408 nested closures exponential

This fixes #72408.

Nested closures were resulting in exponential compilation time.

This PR is enhancing asymptotic complexity, but also increasing the constant, so I would love to see perf run results.
2020-09-18 14:08:39 +00:00
bors
2c69266c06 Auto merge of #76837 - wesleywiser:disable_consideredequal, r=oli-obk
[mir-opt] Disable the `ConsideredEqual` logic in SimplifyBranchSame opt

The logic is currently broken and we need to disable it to fix a beta
regression (see #76803)

r? `@oli-obk`
2020-09-18 10:54:14 +00:00
Valerii Lashmanov
17d2e3b5d2 Better handling for exponential-sized types in misc places
Mostly to fix ui/issues/issue-37311-type-length-limit/issue-37311.rs.

Most parts of the compiler can handle deeply nested types with a lot
of duplicates just fine, but some parts still attempt to naively
traverse type tree.

Before such problems were caught by type length limit check,
but now these places will have to be changed to handle
duplicated types gracefully.
2020-09-17 20:44:11 -05:00
Dylan MacKenzie
e03f4164d9 Default to implicit (not explicit) rules for promotability in const fn 2020-09-17 08:39:11 -07:00
Wesley Wiser
dbd7226d29 [mir-opt] Disable the ConsideredEqual logic in SimplifyBranchSame opt
The logic is currently broken and we need to disable it to fix a beta
regression (see #76803)
2020-09-17 09:36:50 -04:00
bors
7bdb5dee7b Auto merge of #76634 - RalfJung:miri-guaranteed-eq-ne, r=oli-obk
move guaranteed{ne,eq} implementation to compile-time machine

Currently, Miri needs a special hack to avoid using the core engine implementation of these intrinsics. That seems silly, so let's move them to the CTFE machine, which is the only machine that wants to use them.

I also added a reference to https://github.com/rust-lang/rust/issues/73722 as a warning to anyone who wants to adjust `guaranteed_eq`.
2020-09-17 12:15:57 +00:00
Matthias Krüger
012974da7a use strip_prefix over starts_with and manual slicing based on pattern length (clippy::manual_strip) 2020-09-17 10:13:16 +02:00
est31
ebdea01143 Remove redundant #![feature(...)] 's from compiler/ 2020-09-17 07:58:45 +02:00
Hanif Bin Ariffin
f4a7149f49 Don't compile regex at every function call.
Use `SyncOnceCell` to only compile it once.
I believe this still adds some kind of locking mechanism?
2020-09-17 11:03:08 +08:00
Matthias Krüger
0e3414248a don't lazily evaulate some trivial values for Option::None replacements (clippy::unnecessary_lazy_evaluations) 2020-09-17 01:37:35 +02:00
Dylan MacKenzie
81b3b66487 Error if an unstable const eval feature is used in a stable const fn 2020-09-16 14:17:51 -07:00
Dylan MacKenzie
e4edc161f2 Give name to extra Span in LiveDrop error 2020-09-16 14:17:51 -07:00
Dylan MacKenzie
c3607bd7dd Use helper function for searching allow_internal_unstable 2020-09-16 14:17:51 -07:00
Dylan MacKenzie
ed6c7efd87 Use enum for status of non-const ops 2020-09-16 14:17:51 -07:00
Tyler Mandry
3bf66ae25f
Rollup merge of #76794 - richkadel:graphviz-font, r=ecstatic-morse
Make graphviz font configurable

Alternative to PR #76776.

To change the graphviz output to use an alternative `fontname` value,
add a command line option like: `rustc --graphviz-font=monospace`.

r? @ecstatic-morse
2020-09-16 12:24:30 -07:00
Tyler Mandry
cd766c9d09
Rollup merge of #76775 - ecstatic-morse:dataflow-extra-tab-diff, r=Mark-Simulacrum
Strip a single leading tab when rendering dataflow diffs

The `fmt_diff_with` formatter uses a tab to separate additions from subtractions. Strip it when rendering those diffs on separate lines.

r? @Mark-Simulacrum (since you're speedy)
2020-09-16 12:24:27 -07:00
Tyler Mandry
a6c4d30c7b
Rollup merge of #76756 - matthiaskrgr:cl123ppy, r=Dylan-DPC
fix a couple of stylistic clippy warnings

namely:

clippy::redundant_pattern_matching
clippy::redundant_pattern
clippy::search_is_some
clippy::filter_next
clippy::into_iter_on_ref
clippy::clone_on_copy
clippy::needless_return
2020-09-16 12:24:17 -07:00
Rich Kadel
5c29332ace Make graphviz font configurable
Alternative to PR ##76776.

To change the graphviz output to use an alternative `fontname` value,
add a command line option like: `rustc --graphviz-font=monospace`.
2020-09-16 08:10:06 -07:00
Dylan DPC
2c2f1c239e
Rollup merge of #76694 - wesleywiser:partitioning_cx_trait, r=davidtwco
Introduce a PartitioningCx struct

This contains all the data used by the partitioning algorithm and allows that data to be used at each stage of the partitioning. This is useful for other approaches to partitioning which may want different pieces of the data available at each step.

cc @rust-lang/wg-incr-comp
2020-09-16 12:34:18 +02:00
Dylan MacKenzie
c910e038e8 Strip a single leading tab when rendering dataflow diffs 2020-09-15 18:32:12 -07:00
Tomasz Miąsko
ff1a9e406b Fix underflow when calculating the number of no-op jumps folded
When removing unwinds to no-op blocks and folding jumps to no-op blocks,
remove the unwind target first. Otherwise we cannot determine if target
has been already folded or not.

Previous implementation incorrectly assumed that all resume targets had
been folded already, occasionally resulting in an underflow:

remove_noop_landing_pads: removed 18446744073709551613 jumps and 3 landing pads
2020-09-16 00:00:00 +00:00
Dylan DPC
fa4cfeb597
Rollup merge of #75304 - Aaron1011:feature/diag-deref-move-out, r=estebank
Note when a a move/borrow error is caused by a deref coercion

Fixes #73268

When a deref coercion occurs, we may end up with a move error if the
base value has been partially moved out of. However, we do not indicate
anywhere that a deref coercion is occuring, resulting in an error
message with a confusing span.

This PR adds an explicit note to move errors when a deref coercion is
involved. We mention the name of the type that the deref-coercion
resolved to, as well as the `Deref::Target` associated type being used.
2020-09-16 01:30:32 +02:00
Matthias Krüger
f567287f9f don't convert types to the same type with try_into (clippy::useless_conversion) 2020-09-15 22:49:50 +02:00
Matthias Krüger
73d4171ea6 fix a couple of stylistic clippy warnings
namely:

clippy::redundant_pattern_matching
clippy::redundant_pattern
clippy::search_is_some
clippy::filter_next
clippy::into_iter_on_ref
clippy::clone_on_copy
clippy::needless_return
2020-09-15 22:44:54 +02:00
Dylan MacKenzie
0475c365fe Add pass names to some common dataflow analyses 2020-09-14 17:56:39 -07:00
Dylan MacKenzie
8e3ce43de9 Add Engine::pass_name to differentiate dataflow runs 2020-09-14 17:56:21 -07:00
bors
9b4154193e Auto merge of #76541 - matthiaskrgr:unstable_sort, r=davidtwco
use sort_unstable to sort primitive types

It's not important to retain original order if we have &[1, 1, 2, 3] for example.

clippy::stable_sort_primitive
2020-09-14 21:43:17 +00:00
bors
0b65a3d0a6 Auto merge of #76123 - tmiasko:inline-args-storage, r=wesleywiser
inliner: Emit storage markers for introduced arg temporaries

When introducing argument temporaries during inlining, emit storage
marker statements just before the assignment and in the beginning of
the return block.

This ensures that such temporaries will not be considered live across
yield points after inlining inside a generator.

Fixes #71793.
2020-09-14 02:13:02 +00:00
Wesley Wiser
c9686cb31a Introduce a PartitioningCx struct 2020-09-13 21:08:08 -04:00
bors
7402a39447 Auto merge of #76244 - vandenheuvel:remove__paramenv__def_id, r=nikomatsakis
Removing the `def_id` field from hot `ParamEnv` to make it smaller

This PR addresses https://github.com/rust-lang/rust/issues/74865.
2020-09-13 16:28:22 +00:00
Simon Vandel Sillesen
e5447a2222 Fix #76432
Only insert StorageDeads if we actually removed one.
Fixes an issue where we added StorageDead to a place with no StorageLive
2020-09-13 10:47:20 +02:00
bors
498dab0256 Auto merge of #76306 - tmiasko:nrvo-debuginfo, r=ecstatic-morse
NRVO: Allow occurrences of the return place in var debug info

The non-use occurrence of the return place in var debug info does not
currently inhibit NRVO optimization, but it will fail assertion in
`visit_place` when optimization is performed.

Relax assertion check to allow the return place in var debug info.

This case might be impossible to hit in optimization pipelines as of
now, but can be encountered in customized mir-opt-level=2 pipeline with
copy propagation disabled. For example in:

```rust
pub fn b(s: String) -> String {
    a(s)
}

#[inline]
pub fn a(s: String) -> String {
    let x = s;
    let y = x;
    y
}
```
2020-09-13 00:33:04 +00:00
Ralf Jung
c32127675a move guaranteed{ne,eq} implementation to compile-time machine 2020-09-12 10:10:52 +02:00
Andreas Jonson
b8752fff19 update the version of itertools and parking_lot
this is to avoid compiling multiple version of the crates in rustc
2020-09-12 08:26:53 +02:00
Aaron Hill
d18b4bb7a7
Note when a a move/borrow error is caused by a deref coercion
Fixes #73268

When a deref coercion occurs, we may end up with a move error if the
base value has been partially moved out of. However, we do not indicate
anywhere that a deref coercion is occuring, resulting in an error
message with a confusing span.

This PR adds an explicit note to move errors when a deref coercion is
involved. We mention the name of the type that the deref-coercion
resolved to, as well as the `Deref::Target` associated type being used.
2020-09-10 20:56:20 -04:00
Tyler Mandry
c8f9c728c2
Rollup merge of #76567 - matthiaskrgr:clone_on_copy, r=varkor
use push(char) to add chars (single-char &strs) to strings instead of push_str(&str)
2020-09-10 12:20:12 -07:00
Tyler Mandry
94ae5d1866
Rollup merge of #76565 - matthiaskrgr:box_place, r=oli-obk
take reference to Place directly instead of taking reference to Box<Place>

clippy::borrowed_box
2020-09-10 12:20:11 -07:00
Tyler Mandry
2df1487fc9
Rollup merge of #76548 - tmiasko:validate, r=davidtwco
Validate removal of AscribeUserType, FakeRead, and Shallow borrow

Those statements are removed by CleanupNonCodegenStatements pass
in drop lowering phase, and should not occur afterwards.
2020-09-10 12:20:04 -07:00
Matthias Krüger
9bb10cc907 use push(char) instead of push_str(&str) to add single chars to strings
clippy::single-char-push-str
2020-09-10 13:58:41 +02:00
Matthias Krüger
e2a511fe20 use String::from instead of format!() macro to craft string clippy::useless_format 2020-09-10 13:22:51 +02:00
Matthias Krüger
6bfe132067 take reference to Place directly instead of taking reference to Box<Place>
clippy::borrowed_box
2020-09-10 13:08:28 +02:00
bors
88197214b8 Auto merge of #75573 - Aaron1011:feature/const-mutation-lint, r=oli-obk
Add CONST_ITEM_MUTATION lint

Fixes #74053
Fixes #55721

This PR adds a new lint `CONST_ITEM_MUTATION`.
Given an item `const FOO: SomeType = ..`, this lint fires on:

* Attempting to write directly to a field (`FOO.field = some_val`) or
  array entry (`FOO.array_field[0] = val`)
* Taking a mutable reference to the `const` item (`&mut FOO`), including
  through an autoderef `FOO.some_mut_self_method()`

The lint message explains that since each use of a constant creates a
new temporary, the original `const` item will not be modified.
2020-09-10 05:54:26 +00:00
Tyler Mandry
ba6e2b3a31
Rollup merge of #76500 - richkadel:mir-graphviz-dark, r=tmandry
Add -Zgraphviz_dark_mode and monospace font fix

Many developers use a dark theme with editors and IDEs, but this
typically doesn't extend to graphviz output.

When I bring up a MIR graphviz document, the white background is
strikingly bright. This new option changes the colors used for graphviz
output to work better in dark-themed UIs.

<img width="1305" alt="Screen Shot 2020-09-09 at 3 00 31 PM" src="https://user-images.githubusercontent.com/3827298/92659478-4b9bff00-f2ad-11ea-8894-b40d3a873cb9.png">

Also fixed the monospace font for common graphviz renders (e.g., VS Code extensions), as described in https://github.com/rust-lang/rust/pull/76500#issuecomment-689837948

**Before:**
<img width="943" alt="Screen Shot 2020-09-09 at 2 48 44 PM" src="https://user-images.githubusercontent.com/3827298/92658939-47231680-f2ac-11ea-97ac-96727e4dd622.png">

**Now with fix:**
<img width="943" alt="Screen Shot 2020-09-09 at 2 49 02 PM" src="https://user-images.githubusercontent.com/3827298/92658959-51451500-f2ac-11ea-9aae-de982d466d6a.png">
2020-09-09 21:02:35 -07:00
Tomasz Miąsko
00e64ba476 Validate removal of AscribeUserType, FakeRead, and Shallow borrow
Those statements are removed by CleanupNonCodegenStatements pass
in drop lowering phase, and should not occur afterwards.
2020-09-10 00:00:00 +00:00
Tyler Mandry
32714eb6bc
Rollup merge of #76523 - tmiasko:non-use-context-coverage, r=wesleywiser
Remove unused PlaceContext::NonUse(NonUseContext::Coverage)

r? @richkadel / @wesleywiser
2020-09-09 15:06:07 -07:00
Tyler Mandry
bab09684b4
Rollup merge of #76313 - richkadel:mir-spanview-2, r=wesleywiser
Improved the MIR spanview output

* Adds missing "tail" spans (spans that continue beyond the end of
overlapping spans)
* Adds a caret to highlight empty spans associated with MIR elements
that have a position, but otherwise would not be visible.
* Adds visual pointing brackets at the beginning and end of each span

<img width="590" alt="Screen Shot 2020-09-03 at 8 38 08 PM" src="https://user-images.githubusercontent.com/3827298/92202571-25510c00-ee34-11ea-89bc-89eea939476d.png">
<img width="1061" alt="Screen Shot 2020-09-03 at 8 41 04 PM" src="https://user-images.githubusercontent.com/3827298/92202629-49145200-ee34-11ea-8fda-fc6e62c80736.png">
<img width="1113" alt="Screen Shot 2020-09-06 at 5 42 57 PM" src="https://user-images.githubusercontent.com/3827298/92339198-ca085f00-f069-11ea-96d1-c01ced50e2ba.png">
<img width="1692" alt="Screen Shot 2020-09-06 at 5 45 54 PM" src="https://user-images.githubusercontent.com/3827298/92339209-d4c2f400-f069-11ea-94c0-b4d36c200878.png">

r? @tmandry
FYI: @wesleywiser
2020-09-09 15:05:49 -07:00
Matthias Krüger
b4935e0726 use sort_unstable to sort primitive types
It's not important to retain original order if we have &[1, 1, 2, 3] for example.

clippy::stable_sort_primitive
2020-09-10 00:03:58 +02:00
Rich Kadel
f7aee330c7 Also fixed monospace font for d3-graphviz engine
VS code graphviz extensions use d3-graphviz, which supports `Courier`
fontname but does not support `monospace`. This caused graphs to render
poorly because the text sizes were wrong.
2020-09-09 14:49:32 -07:00
Tomasz Miąsko
0016405073 Remove unused PlaceContext::NonUse(NonUseContext::Coverage) 2020-09-09 17:02:19 +02:00
Bram van den Heuvel
7dad29d686 Remove def_id field from ParamEnv 2020-09-09 10:14:31 +02:00
Rich Kadel
c19b2370e4 Add -Zgraphviz_dark_mode
Many developers use a dark theme with editors and IDEs, but this
typically doesn't extend to graphviz output.

When I bring up a MIR graphviz document, the white background is
strikingly bright. This new option changes the colors used for graphviz
output to work better in dark-themed UIs.
2020-09-08 17:19:38 -07:00
bors
5a6b426e34 Auto merge of #76308 - wesleywiser:enable_simplifyarmidentity_mir_opt, r=oli-obk
Enable the SimplifyArmIdentity MIR optimization at mir-opt-level=1

r? `@ghost`
2020-09-08 09:27:23 +00:00
bors
e82584a77d Auto merge of #75585 - RalfJung:demotion, r=oli-obk
Do not promote &mut of a non-ZST ever

Since ~pre-1.0~ 1.36, we have accepted code like this:
```rust
static mut TEST: &'static mut [i32] = {
    let x = &mut [1,2,3];
    x
};
```
I tracked it back to https://github.com/rust-lang/rust/pull/21744, but unfortunately could not find any discussion or RFC that would explain why we thought this was a good idea. And it's not, it breaks all sorts of things -- see https://github.com/rust-lang/rust/issues/75556.

To fix https://github.com/rust-lang/rust/issues/75556, we have to stop promoting non-ZST mutable references no matter the context, which is what this PR does. It's a breaking change.

Notice that this still works, since it does not rely on promotion:
```rust
static mut TEST: &'static mut [i32] = &mut [0,1,2];
```

Cc `@rust-lang/wg-const-eval`
2020-09-08 05:13:42 +00:00
bors
0e2c1281e9 Auto merge of #76044 - ecstatic-morse:dataflow-lattice, r=oli-obk
Support dataflow problems on arbitrary lattices

This PR implements last of the proposed extensions I mentioned in the design meeting for the original dataflow refactor. It extends the current dataflow framework to work with arbitrary lattices, not just `BitSet`s. This is a prerequisite for dataflow-enabled MIR const-propagation. Personally, I am skeptical of the usefulness of doing const-propagation pre-monomorphization, since many useful constants only become known after monomorphization (e.g. `size_of::<T>()`) and users have a natural tendency to hand-optimize the rest. It's probably worth exprimenting with, however, and others have shown interest cc `@rust-lang/wg-mir-opt.`

The `Idx` associated type is moved from `AnalysisDomain` to `GenKillAnalysis` and replaced with an associated `Domain` type that must implement `JoinSemiLattice`. Like before, each `Analysis` defines the "bottom value" for its domain, but can no longer override the dataflow join operator. Analyses that want to use set intersection must now use the `lattice::Dual` newtype. `GenKillAnalysis` impls have an additional requirement that `Self::Domain: BorrowMut<BitSet<Self::Idx>>`, which effectively means that they must use `BitSet<Self::Idx>` or `lattice::Dual<BitSet<Self::Idx>>` as their domain.

Most of these changes were mechanical. However, because a `Domain` is no longer always a powerset of some index type, we can no longer use an `IndexVec<BasicBlock, GenKillSet<A::Idx>>>` to store cached block transfer functions. Instead, we use a boxed `dyn Fn` trait object. I discuss a few alternatives to the current approach in a commit message.

The majority of new lines of code are to preserve existing Graphviz diagrams for those unlucky enough to have to debug dataflow analyses. I find these diagrams incredibly useful when things are going wrong and considered regressing them unacceptable, especially the pretty-printing of `MovePathIndex`s, which are used in many dataflow analyses. This required a parallel `fmt` trait used only for printing dataflow domains, as well as a refactoring of the `graphviz` module now that we cannot expect the domain to be a `BitSet`. Some features did have to be removed, such as the gen/kill display mode (which I didn't use but existed to mirror the output of the old dataflow framework) and line wrapping. Since I had to rewrite much of it anyway, I took the opportunity to switch to a `Visitor` for printing dataflow state diffs instead of using cursors, which are error prone for code that must be generic over both forward and backward analyses. As a side-effect of this change, we no longer have quadratic behavior when writing graphviz diagrams for backward dataflow analyses.

r? `@pnkfelix`
2020-09-07 21:29:43 +00:00
Aaron Hill
f422ef141a
Add CONST_ITEM_MUTATION lint
Fixes #74053
Fixes #55721

This PR adds a new lint `CONST_ITEM_MUTATION`.
Given an item `const FOO: SomeType = ..`, this lint fires on:

* Attempting to write directly to a field (`FOO.field = some_val`) or
  array entry (`FOO.array_field[0] = val`)
* Taking a mutable reference to the `const` item (`&mut FOO`), including
  through an autoderef `FOO.some_mut_self_method()`

The lint message explains that since each use of a constant creates a
new temporary, the original `const` item will not be modified.
2020-09-07 08:44:35 -04:00
Rich Kadel
9046a9343b Improved the MIR spanview output
* Adds missing "tail" spans (spans that continue beyond the end of
overlapping spans)
* Adds a caret to highlight empty spans associated with MIR elements
that have a position, but otherwise would not be visible.
* Adds visual pointing brackets at the beginning and end of each span
2020-09-06 19:04:08 -07:00
Ralf Jung
720293b640 do not premote non-ZST mutable references ever 2020-09-06 14:14:27 +02:00
Simon Vandel Sillesen
9b0fc6202b Generalize to Eq(true, _place) and Eq(_place, true) 2020-09-06 11:51:44 +02:00
Simon Vandel Sillesen
c2693db264 Add peephold optimization that simplifies Ne(_1, false) and Ne(false, _1) into _1
This was observed emitted from the MatchBranchSimplification pass.
2020-09-06 11:51:44 +02:00
Dylan DPC
b4d3873024
Rollup merge of #76263 - tmiasko:inline-codegen-fn-attrs, r=ecstatic-morse
inliner: Check for codegen fn attributes compatibility

* Check for target features compatibility
* Check for no_sanitize attribute compatibility

Fixes #76259.
2020-09-05 16:28:34 +02:00
LeSeulArtichaut
3e14b684dd Change ty.kind to a method 2020-09-04 17:47:51 +02:00
David Wood
f8376b59d1
shim: monomorphic FnPtrShims during construction
This commit adjusts MIR shim construction so that substitutions are
applied to function pointer shims during construction, rather than
during codegen (as determined by `substs_for_mir_body`) - as
substitutions will no longer occur during codegen, function pointer
shims can now be polymorphic without incurring double substitutions.

Signed-off-by: David Wood <david@davidtw.co>
2020-09-04 13:00:49 +01:00
bors
4ffb5c5954 Auto merge of #76004 - richkadel:llvm-coverage-map-gen-6b.5, r=tmandry
Tools, tests, and experimenting with MIR-derived coverage counters

Leverages the new mir_dump output file in HTML+CSS (from #76074) to visualize coverage code regions
and the MIR features that they came from (including overlapping spans).

See example below.

The `run-make-fulldeps/instrument-coverage` test has been refactored to maximize test coverage and reduce code duplication. The new tests support testing with and without `-Clink-dead-code`, so Rust coverage can be tested on MSVC (which, currently, only works with `link-dead-code` _disabled_).

New tests validate coverage region generation and coverage reports with multiple counters per function. Starting with a simple `if-else` branch tests, coverage tests for each additional syntax type can be added by simply dropping in a new Rust sample program.

Includes a basic, MIR-block-based implementation of coverage injection,
available via `-Zexperimental-coverage`. This implementation has known
flaws and omissions, but is simple enough to validate the new tools and
tests.

The existing `-Zinstrument-coverage` option currently enables
function-level coverage only, which at least appears to generate
accurate coverage reports at that level.

Experimental coverage is not accurate at this time. When branch coverage
works as intended, the `-Zexperimental-coverage` option should be
removed.

This PR replaces the bulk of PR #75828, with the remaining parts of
that PR distributed among other separate and indentpent PRs.

This PR depends on two of those other PRs: #76002, #76003 and #76074

Rust compiler MCP rust-lang/compiler-team#278

Relevant issue: #34701 - Implement support for LLVMs code coverage
instrumentation

![Screen-Recording-2020-08-21-at-2](https://user-images.githubusercontent.com/3827298/90972923-ff417880-e4d1-11ea-92bb-8713c6198f6d.gif)

r? @tmandry
FYI: @wesleywiser
2020-09-04 01:31:07 +00:00
Wesley Wiser
903c0399e7 Enable the SimplifyArmIdentity MIR optimization at mir-opt-level=1 2020-09-03 21:16:30 -04:00
Tomasz Miąsko
01510612ee NRVO: Allow occurrences of the return place in var debug info
The non-use occurrence of the return place in var debug info does not
currently inhibit NRVO optimization, but it will fail assertion in
`visit_place` when optimization is performed.

Relax assertion check to allow the return place in var debug info.

This case might be impossible to hit in optimization pipelines as of
now, but can be encountered in customized mir-opt-level=2 pipeline with
copy propagation disabled. For example in:

```
pub fn b(s: String) -> String {
    a(s)
}

#[inline]
pub fn a(s: String) -> String {
    let x = s;
    let y = x;
    y
}
```
2020-09-04 02:33:37 +02:00
Tomasz Miąsko
326b772609 inliner: Check for no_sanitize attribute compatibility 2020-09-04 20:01:15 +02:00
Tomasz Miąsko
cbc396fdfb inliner: Check for target features compatibility 2020-09-04 19:36:45 +02:00
bors
af3c6e733a Auto merge of #73996 - da-x:short-unique-paths, r=petrochenkov
diagnostics: shorten paths of unique symbols

This is a step towards implementing a fix for #50310, and continuation of the discussion in [Pre-RFC: Nicer Types In Diagnostics - compiler - Rust Internals](https://internals.rust-lang.org/t/pre-rfc-nicer-types-in-diagnostics/11139). Impressed upon me from previous discussion in #21934 that an RFC for this is not needed, and I should just come up with code.

The recent improvements to `use` suggestions that I've contributed have given rise to this implementation. Contrary to previous suggestions, it's rather simple logic, and I believe it only reduces the amount of cognitive load that a developer would need when reading type errors.

-----

If a symbol name can only be imported from one place, and as long as it was not glob-imported anywhere in the current crate, we can trim its printed path to the last component.

This has wide implications on error messages with types, for example, shortening `std::vec::Vec` to just `Vec`, as long as there is no other `Vec` importable from anywhere.
2020-09-03 23:27:45 +00:00
Tomasz Miąsko
9daf8fd5b1 inliner: Emit storage markers for introduced arg temporaries
When introducing argument temporaries during inlining, emit storage
marker statements just before the assignment and in the beginning of
the return block.

This ensures that such temporaries will not be considered live across
yield points after inlining inside a generator.
2020-09-03 17:01:27 +02:00
Rich Kadel
51d692cf77 Tools, tests, and experimenting with MIR-derived coverage counters
Adds a new mir_dump output file in HTML/CSS to visualize code regions
and the MIR features that they came from (including overlapping spans).
See example below:

Includes a basic, MIR-block-based implementation of coverage injection,
available via `-Zexperimental-coverage`. This implementation has known
flaws and omissions, but is simple enough to validate the new tools and
tests.

The existing `-Zinstrument-coverage` option currently enables
function-level coverage only, which at least appears to generate
accurate coverage reports at that level.

Experimental coverage is not accurate at this time. When branch coverage
works as intended, the `-Zexperimental-coverage` option should be
removed.

This PR replaces the bulk of PR #75828, with the remaining parts of
that PR distributed among other separate and indentpent PRs.

This PR depends on three of those other PRs: #76000, #76002, and

Rust compiler MCP rust-lang/compiler-team#278

Relevant issue: #34701 - Implement support for LLVMs code coverage
instrumentation

![Screen-Recording-2020-08-21-at-2](https://user-images.githubusercontent.com/3827298/90972923-ff417880-e4d1-11ea-92bb-8713c6198f6d.gif)
2020-09-03 00:20:29 -07:00
Dan Aloni
07e7823c01 pretty: trim paths of unique symbols
If a symbol name can only be imported from one place for a type, and
as long as it was not glob-imported anywhere in the current crate, we
can trim its printed path and print only the name.

This has wide implications on error messages with types, for example,
shortening `std::vec::Vec` to just `Vec`, as long as there is no other
`Vec` importable anywhere.

This adds a new '-Z trim-diagnostic-paths=false' option to control this
feature.

On the good path, with no diagnosis printed, we should try to avoid
issuing this query, so we need to prevent trimmed_def_paths query on
several cases.

This change also relies on a previous commit that differentiates
between `Debug` and `Display` on various rustc types, where the latter
is trimmed and presented to the user and the former is not.
2020-09-02 22:26:37 +03:00
Mark Rousskov
990a39596c Prevent ICE on uninhabited MIR interpretation 2020-09-02 08:37:48 -04:00
Tomasz Miąsko
6c51ec96bb inliner: Avoid query cycles when optimizing generators
The HIR Id trick is insufficient to prevent query cycles when optimizing
generators, since merely requesting a layout of a generator also
computes its `optimized_mir`.

Make no attempts to inline functions into generators within the same
crate to avoid query cycles.
2020-09-02 00:00:00 +00:00
Rich Kadel
6b5869a0ae Add new -Z dump-mir-spanview option
Similar to `-Z dump-mir-graphviz`, this adds the option to write
HTML+CSS files that allow users to analyze the spans associated with MIR
elements (by individual statement, just terminator, or overall basic
block).

This PR was split out from PR #76004, and exposes an API for spanview
HTML+CSS files that is also used to analyze code regions chosen for
coverage instrumentation (in a follow-on PR).

Rust compiler MCP rust-lang/compiler-team#278

Relevant issue: #34701 - Implement support for LLVMs code coverage
instrumentation
2020-08-31 22:57:55 -07:00
Tyler Mandry
6d834a4046
Rollup merge of #76002 - richkadel:llvm-coverage-map-gen-6b.3, r=tmandry
Fix `-Z instrument-coverage` on MSVC

Found that `-C link-dead-code` (which was enabled automatically
under `-Z instrument-coverage`) was causing the linking error that
resulted in segmentation faults in coverage instrumented binaries. Link
dead code is now disabled under MSVC, allowing `-Z instrument-coverage`
to be enabled under MSVC for the first time.

More details are included in Issue #76038 .

Note this PR makes it possible to support `Z instrument-coverage` but
does not enable instrument coverage for MSVC in existing tests. It will be
enabled in another PR to follow this one (both PRs coming from original
PR #75828).

r? @tmandry
FYI: @wesleywiser
2020-08-31 19:18:14 -07:00
Rich Kadel
ddb054aee8 Fix -Z instrument-coverage on MSVC
Found that -C link-dead-code (which was enabled automatically
under -Z instrument-coverage) was causing the linking error that
resulted in segmentation faults in coverage instrumented binaries. Link
dead code is now disabled under MSVC, allowing `-Z instrument-coverage`
to be enabled under MSVC for the first time.

More details are included in Issue #76038.

(This PR was broken out from PR #75828)
2020-08-31 18:41:13 -07:00
Dylan MacKenzie
b015109ba9 Add documentation to the Analysis traits 2020-08-30 14:26:04 -07:00
Dylan MacKenzie
e178a87036 Expand documentation for the lattice module 2020-08-30 13:27:07 -07:00
Dylan MacKenzie
c03eba2d08 Add FIXME for faster cached block transfer functions
I've tried a few ways of implementing this, but each fell short.

Adding an auxiliary `_Idx` associated type to `Analysis` that defaults
to `!` but is overridden in the blanket impl of `Analysis` for `A:
GenKillAnalysis` to `A::Idx` seems promising, but the trait solver is
unable to prove equivalence between `A::Idx` and `A::_Idx` within the
overridden version of `into_engine`. Without full-featured
specialization, removing `into_engine` or splitting it into a different
trait would have a significant ergonomic penalty.

Alternatively, we could erase the index type and store a
`GenKillSet<u32>` as well as a function pointer for transmuting between
`&mut A::Domain` and `&mut BitSet<u32>` in the hopes that LLVM can
devirtualize a simple function pointer better than the boxed closure.
However, this is brittle, requires `unsafe` code, and doesn't work for
index types that aren't the same size as a `u32` (e.g. `usize`) since
`GenKillSet` stores a `HybridBitSet`, which may be a `Vec<I>`. Perhaps
safe transmute could help here?
2020-08-30 11:15:25 -07:00
Dylan MacKenzie
b19b8ea611 Update dataflow analyses to use new interface 2020-08-30 11:15:25 -07:00
Dylan MacKenzie
3233fb18a8 Extend dataflow framework to support arbitrary lattices 2020-08-30 11:15:24 -07:00
Dylan MacKenzie
9e45e90596 Allow access to the underlying Results from a ResultsCursor 2020-08-30 11:15:24 -07:00
Dylan MacKenzie
a88dc37c54 Add regex dependency to librustc_mir 2020-08-30 11:15:21 -07:00
David Wood
6ff471b1cf
ty: remove obsolete printer
This commit removes the obsolete printer and replaces all uses of it
with `FmtPrinter`. Of the replaced uses, all but one use was in `debug!`
logging, two cases were notable:

- `MonoItem::to_string` is used in `-Z print-mono-items` and therefore
  affects the output of all codegen-units tests.
- `DefPathBasedNames` was used in `librustc_codegen_llvm/type_of.rs`
  with `LLVMStructCreateNamed` and that'll now get different values, but
  this should result in no functional change.

Signed-off-by: David Wood <david@davidtw.co>
2020-08-30 18:59:07 +01:00
mark
9e5f7d5631 mv compiler to compiler/ 2020-08-30 18:45:07 +03:00