-
Notifications
You must be signed in to change notification settings - Fork 12.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Use ValTree in all type level constants #83234
Conversation
(ty::ConstKind::Value(a_val), ty::ConstKind::Value(b_val)) => { | ||
check_const_value_eq(relation, a_val, b_val, a, b)? | ||
} | ||
(ty::ConstKind::Value(a_val), ty::ConstKind::Value(b_val)) => a_val == b_val, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is the essence of this PR. There is no more special logic for relating constants, we can just compare them with the derived PartialEq impl.
// + val: Value(Slice { data: Allocation { bytes: [101, 120, 112, 108, 105, 99, 105, 116, 32, 112, 97, 110, 105, 99], relocations: Relocations(SortedMap { data: [] }), init_mask: InitMask { blocks: [16383], len: Size { raw: 14 } }, size: Size { raw: 14 }, align: Align { pow2: 0 }, mutability: Not, extra: () }, start: 0, end: 14 }) | ||
// + val: Value(Branch([Leaf(0x65), Leaf(0x78), Leaf(0x70), Leaf(0x6c), Leaf(0x69), Leaf(0x63), Leaf(0x69), Leaf(0x74), Leaf(0x20), Leaf(0x70), Leaf(0x61), Leaf(0x6e), Leaf(0x69), Leaf(0x63)])) | ||
// mir::Constant | ||
// + span: $SRC_DIR/std/src/panic.rs:LL:COL | ||
// + literal: Const { ty: &str, val: Value(Slice { data: Allocation { bytes: [101, 120, 112, 108, 105, 99, 105, 116, 32, 112, 97, 110, 105, 99], relocations: Relocations(SortedMap { data: [] }), init_mask: InitMask { blocks: [16383], len: Size { raw: 14 } }, size: Size { raw: 14 }, align: Align { pow2: 0 }, mutability: Not, extra: () }, start: 0, end: 14 }) } | ||
// + literal: Const { ty: &str, val: Value(Branch([Leaf(0x65), Leaf(0x78), Leaf(0x70), Leaf(0x6c), Leaf(0x69), Leaf(0x63), Leaf(0x69), Leaf(0x74), Leaf(0x20), Leaf(0x70), Leaf(0x61), Leaf(0x6e), Leaf(0x69), Leaf(0x63)])) } |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
we should probably stop printing string constants in mir dumps.. their stringly printed representation is sufficient
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can we tie this to -Zverbose
or so?
- _2 = Eq(move _3, const 17_i8); // scope 0 at $DIR/if-condition-int.rs:44:13: 44:20 | ||
- StorageDead(_3); // scope 0 at $DIR/if-condition-int.rs:44:19: 44:20 | ||
- switchInt(_2) -> [false: bb2, otherwise: bb1]; // scope 1 at $DIR/if-condition-int.rs:46:9: 46:14 | ||
+ _2 = Eq(_3, const 17_i8); // scope 0 at $DIR/if-condition-int.rs:44:13: 44:20 | ||
+ nop; // scope 0 at $DIR/if-condition-int.rs:44:19: 44:20 | ||
+ switchInt(move _3) -> [17_i8: bb1, otherwise: bb2]; // scope 1 at $DIR/if-condition-int.rs:46:9: 46:14 | ||
_2 = Eq(move _3, const 17_i8); // scope 0 at $DIR/if-condition-int.rs:44:13: 44:20 | ||
StorageDead(_3); // scope 0 at $DIR/if-condition-int.rs:44:19: 44:20 | ||
switchInt(_2) -> [false: bb2, otherwise: bb1]; // scope 1 at $DIR/if-condition-int.rs:46:9: 46:14 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This one is interesting... this change by itself enabled an optimization to trigger more often
@@ -1,6 +1,6 @@ | |||
fn main() { | |||
let A = 3; | |||
//~^ ERROR refutable pattern in local binding: `i32::MIN..=1_i32` and | |||
//~^ ERROR refutable pattern in local binding: `i32::MIN..=1` and |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This change is because we're now using ConstInt
instead of ty::Const
printing for these errors
This comment has been minimized.
This comment has been minimized.
☔ The latest upstream changes (presumably #82122) made this pull request unmergeable. Please resolve the merge conflicts. |
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
@bors try @rust-timer queue |
Awaiting bors try build completion. @rustbot label: +S-waiting-on-perf |
⌛ Trying commit 5cf5addca286bfde62041568f38117ceafc84e4c with merge 437ebfc4e9d5c202057e1c2509e25c70d2fc67eb... |
☀️ Try build successful - checks-actions |
Queued 437ebfc4e9d5c202057e1c2509e25c70d2fc67eb with parent b97fd3e, future comparison URL. |
Finished benchmarking try commit (437ebfc4e9d5c202057e1c2509e25c70d2fc67eb): comparison url. Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. Please note that if the perf results are neutral, you should likely undo the rollup=never given below by specifying Importantly, though, if the results of this run are non-neutral do not roll this PR up -- it will mask other regressions or improvements in the roll up. @bors rollup=never |
oof. that's some results... Some good, some very bad results (stress tests). I gotta analyze the stress tests to see which part is problematic. It may be string related, as strings are now encoded as valtree branches (so one valtree leaf per byte in the string), making them rather expensive to be converted back into allocations. Since one of the tests does this a lot, that may just be the root of the regression. Not sure how to address this though. Ideally we would just have a I'll experiment with just fixing strings for now and we'll see whether we can at least not regress our benchmarks, even if we'll still have a regression for byte strings (but no benchmark for them). |
...instead of carrying around ty::Const and erroring later
…aration of value trees making such operations on `ty::Const` obsolete.
…woerister,oli-obk rustc_symbol_mangling: support structural constants and &str in v0. This PR should unblock rust-lang#85530 (except for float `const` generics, which AFAIK should've never worked). (cc `@tmiasko` could the rust-lang#85530 (comment) failures be retried with a quick crater "subset" run of this PR + changing the default to `v0`? Just to make sure I didn't miss anything other than the floats) The encoding is the one suggested before in e.g. rust-lang#61486 (comment), tho this PR won't by itself finish rust-lang#61486, before closing that we'd likely want to move to `@oli-obk's` "valtrees" (i.e. rust-lang#83234 and other associated work). <hr> **EDITs**: 1. switched unit/tuple/braced-with-named-fields `<const-fields>` prefixes from `"u"`/`"T"`/`""` to `"U"`/`"T"`/`"S"` to avoid the ambiguity reported by `@tmiasko` in rust-lang#87194 (comment). 2. `rustc-demangle` PR: rust-lang/rustc-demangle#55 3. RFC amendment PR: rust-lang/rfcs#3161 * also removed the grammar changes included in that PR, from this description 4. added tests (temporarily using my fork of `rustc-demangle`) <hr> r? `@michaelwoerister`
triage: merge conflicts |
I'm going to close this for now. I'm going to need another few months before I get to this, so we don't need to keep it open |
see #96591 |
based on #83207
r? @lcnr
best reviewed commit by commit, as some of the refactoring commits are noisy but don't change anything significant.
cc @rust-lang/wg-const-eval
Type level constants are now fully backed by value trees, so it is impossible for two constants to have the same value but a different representation.