Add new `Deinit` statement #95125

JakobDegen · 2022-03-19T22:53:49Z

This rvalue replaces SetDiscriminant for ADTs. This PR is an alternative to #94590 , which only specifies that the behavior of SetDiscriminant is the same as what this rvalue would do. The motivation for this change are discussed in that PR and on Zulip

r? @oli-obk

rust-highfive · 2022-03-19T22:53:52Z

Some changes occured to the CTFE / Miri engine

cc @rust-lang/miri

JakobDegen · 2022-03-20T01:53:13Z

Bot doesn't pick it up, so cc @bjorn3 for the cranelift change and... probably someone for clippy, but not sure

bjorn3 · 2022-03-20T09:19:34Z

cg_clif changes looks fine to me.

oli-obk · 2022-03-20T11:04:35Z

@bors try @rust-timer queue

Will review next week, let's see what perf says 'till then

rust-timer · 2022-03-20T11:04:36Z

Awaiting bors try build completion.

@rustbot label: +S-waiting-on-perf

bors · 2022-03-20T11:04:42Z

⌛ Trying commit 7236347a2324c8d571138f81dafd8b150f17eb5b with merge a920b8a92d5e86353acdc0c28bbcaa945ce3b565...

bors · 2022-03-20T12:55:03Z

☀️ Try build successful - checks-actions
Build commit: a920b8a92d5e86353acdc0c28bbcaa945ce3b565 (a920b8a92d5e86353acdc0c28bbcaa945ce3b565)

rust-timer · 2022-03-20T12:55:06Z

Queued a920b8a92d5e86353acdc0c28bbcaa945ce3b565 with parent 499d4a5, future comparison URL.

rust-timer · 2022-03-20T14:17:06Z

Finished benchmarking commit (a920b8a92d5e86353acdc0c28bbcaa945ce3b565): comparison url.

Summary: This benchmark run did not return any relevant results. 7 results were found to be statistically significant but too small to be relevant.

If you disagree with this performance assessment, please file an issue in rust-lang/rustc-perf.

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. While you can manually mark this PR as fit for rollup, we strongly recommend not doing so since this PR led to changes in compiler perf.

@bors rollup=never
@rustbot label: +S-waiting-on-review -S-waiting-on-perf -perf-regression

bjorn3 · 2022-03-20T14:19:27Z

This seems to be a tiny slowdown overall.

JakobDegen · 2022-03-20T20:06:26Z

Yeah, that might be the result of an additional variant on Rvalue which is matched on quite a bit

compiler/rustc_middle/src/mir/mod.rs

compiler/rustc_const_eval/src/interpret/place.rs

JakobDegen · 2022-03-21T02:11:38Z

The new commits include most of the changes that were also made in #94590 . The only outstanding concern there was Nagisa bringing up that an issue should be opened for the disabled tests - I can do that once either PR merges

compiler/rustc_borrowck/src/invalidation.rs

compiler/rustc_middle/src/mir/mod.rs

JakobDegen · 2022-03-22T01:41:42Z

I had meant to add the validator checks but forgot about it - also addressed the other two concerns

… on deaggregation

oli-obk · 2022-04-11T14:36:28Z

@bors r+ rollup=never

bors · 2022-04-11T14:36:30Z

📌 Commit 2f03767 has been approved by oli-obk

bors · 2022-04-11T14:49:39Z

⌛ Testing commit 2f03767 with merge 625e4dd...

bors · 2022-04-11T17:47:12Z

☀️ Test successful - checks-actions
Approved by: oli-obk
Pushing 625e4dd to master...

rust-timer · 2022-04-11T19:00:49Z

Finished benchmarking commit (625e4dd): comparison url.

Summary:

Primary benchmarks: mixed results
Secondary benchmarks: mixed results

	Regressions 😿 (primary)	Regressions 😿 (secondary)	Improvements 🎉 (primary)	Improvements 🎉 (secondary)	All 😿 🎉 (primary)
count¹	27	39	4	17	31
mean²	0.5%	1.4%	-0.9%	-1.0%	0.3%
max	1.6%	4.1%	-1.4%	-2.2%	1.6%

If you disagree with this performance assessment, please file an issue in rust-lang/rustc-perf.

Next Steps: If you can justify the regressions found in this perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please open an issue or create a new PR that fixes the regressions, add a comment linking to the newly created issue or PR, and then add the perf-regression-triaged label to this PR.

@rustbot label: +perf-regression

number of relevant changes ↩
the arithmetic mean of the percent change ↩

JakobDegen · 2022-04-11T22:29:41Z

A small perf regression is to be expected here I think. MIR contains more statements now, and even though we don't codegen the new ones, we still have to process them in rustc. The cost of this could definitely be reduced by optimizations as Oli already mentioned, but I don't think this particular case is any more interesting than all the other examples of optimizations we could do but don't yet.

oli-obk · 2022-04-12T07:23:18Z

I ran cachegrind on some of the regressions, adn all of them seem to be in metadata de/serialization except for cranelift, which looks like LLVM noise. Considering that this is groundwork for fixing bugs in our dataflow analyses, which directly impact MIR opt sounness, I think we have to take this hit

@rustbot label: +perf-regression-triaged

RalfJung · 2022-04-17T22:48:46Z

compiler/rustc_const_eval/src/interpret/memory.rs

+
+    /// Mark the entire referenced range as uninitalized
+    pub fn write_uninit(&mut self) {
+        self.alloc.mark_init(self.range, false);


Uh-oh, making mark_init public was a mistake. (Not sure when that happened, that was before this PR.)

This call is wrong, since it forgets to also clear the relocations in that range. Some other uses of mark_init are also wrong. I am working on a fix.

(To be clear, this is not your fault. We did bad API design here.)

We codegen an assignment to non-det value per documentation. See more information here: - rust-lang/rust#95125

* Update rust toolchain to nightly-2022-05-03 This compiles but regression is failing due to unimplemented statement. * Handle change to Box<T> structure Box<T> now users NonNull<T> instead of raw pointer. * Handle new statement kind Deinit We codegen an assignment to non-det value per documentation. See more information here: - rust-lang/rust#95125 * Fix discriminant computation After the merge, the previous wrapping sub logic was triggering a panic due to u128 -> i64 conversion. There were also other overflow issues when trying to convert the `niche_value` to unsigned. For now, I'm disabling the coversion check which I believe is too strict. We should consider implementing a more flexible check later that can be controlled by the user without affecting the internal compiler codegen. * Address PR comments: - Improve comments. - Remove wrong cast to i64. - Fix statement location. - Create util function to create unsigned type.

rust-highfive assigned oli-obk Mar 19, 2022

rustbot added the T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. label Mar 19, 2022

rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Mar 19, 2022

This comment has been minimized.

Sign in to view

JakobDegen force-pushed the uninit-variant-rvalue branch from 33a366c to e1dc823 Compare March 20, 2022 01:16

This comment has been minimized.

Sign in to view

JakobDegen force-pushed the uninit-variant-rvalue branch from e1dc823 to 7236347 Compare March 20, 2022 01:46

This comment has been minimized.

Sign in to view

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Mar 20, 2022

rustbot removed the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Mar 20, 2022

RalfJung reviewed Mar 20, 2022

View reviewed changes

compiler/rustc_middle/src/mir/mod.rs Outdated Show resolved Hide resolved

RalfJung reviewed Mar 20, 2022

View reviewed changes

compiler/rustc_const_eval/src/interpret/place.rs Outdated Show resolved Hide resolved

JakobDegen force-pushed the uninit-variant-rvalue branch from 7236347 to ac33542 Compare March 21, 2022 02:04

JakobDegen marked this pull request as ready for review March 21, 2022 02:13

oli-obk requested changes Mar 21, 2022

View reviewed changes

JakobDegen force-pushed the uninit-variant-rvalue branch from dcf356c to 338c80a Compare March 22, 2022 01:38

JakobDegen added 3 commits April 11, 2022 09:26

Fix tests broken by deaggregation change

2a04028

Add const eval tests ensuring padding gets correctly marked as deinit…

f7ca97a

… on deaggregation

Add new MutatatingUseContexts for deinit and SetDiscriminant

48b01a0

JakobDegen force-pushed the uninit-variant-rvalue branch from 9681a02 to 48b01a0 Compare April 11, 2022 13:45

Remove inlining cost of Deinit statements

2f03767

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Apr 11, 2022

bors added the merged-by-bors This PR was explicitly merged by bors. label Apr 11, 2022

bors merged commit 625e4dd into rust-lang:master Apr 11, 2022

rustbot added this to the 1.62.0 milestone Apr 11, 2022

This was referenced Apr 11, 2022

Fix constants not getting dropped if part of a diverging expression #94775

Merged

replace RawVec with Boxed slice #94421

Closed

More improvents to dest prop #95700

Closed

Rollup of 7 pull requests #95941

Closed

rustbot added the perf-regression Performance regression. label Apr 11, 2022

JakobDegen deleted the uninit-variant-rvalue branch April 11, 2022 20:33

rustbot added the perf-regression-triaged The performance regression has been triaged. label Apr 12, 2022

This was referenced Apr 12, 2022

ICE: broken mir with -Zunsound-mir-opts: simplify arm identity #78628

Closed

ICE: The type checker should prevent reading from a never-written local #69487

Closed

RalfJung reviewed Apr 17, 2022

View reviewed changes

celinval added a commit to celinval/kani-dev that referenced this pull request May 10, 2022

Handle new statement kind Deinit

61993e9

We codegen an assignment to non-det value per documentation. See more information here: - rust-lang/rust#95125

celinval added a commit to celinval/kani-dev that referenced this pull request May 11, 2022

Handle new statement kind Deinit

b378f72

We codegen an assignment to non-det value per documentation. See more information here: - rust-lang/rust#95125

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add new `Deinit` statement #95125

Add new `Deinit` statement #95125

JakobDegen commented Mar 19, 2022 •

edited

Loading

rust-highfive commented Mar 19, 2022

This comment has been minimized.

This comment has been minimized.

JakobDegen commented Mar 20, 2022

This comment has been minimized.

bjorn3 commented Mar 20, 2022

oli-obk commented Mar 20, 2022

rust-timer commented Mar 20, 2022

bors commented Mar 20, 2022

bors commented Mar 20, 2022

rust-timer commented Mar 20, 2022

rust-timer commented Mar 20, 2022

bjorn3 commented Mar 20, 2022

JakobDegen commented Mar 20, 2022

JakobDegen commented Mar 21, 2022

JakobDegen commented Mar 22, 2022

oli-obk commented Apr 11, 2022

bors commented Apr 11, 2022

bors commented Apr 11, 2022

bors commented Apr 11, 2022

rust-timer commented Apr 11, 2022

JakobDegen commented Apr 11, 2022

oli-obk commented Apr 12, 2022 •

edited

Loading

RalfJung Apr 17, 2022 •

edited

Loading

Add new Deinit statement #95125

Add new Deinit statement #95125

Conversation

JakobDegen commented Mar 19, 2022 • edited Loading

rust-highfive commented Mar 19, 2022

This comment has been minimized.

This comment has been minimized.

JakobDegen commented Mar 20, 2022

This comment has been minimized.

bjorn3 commented Mar 20, 2022

oli-obk commented Mar 20, 2022

rust-timer commented Mar 20, 2022

bors commented Mar 20, 2022

bors commented Mar 20, 2022

rust-timer commented Mar 20, 2022

rust-timer commented Mar 20, 2022

bjorn3 commented Mar 20, 2022

JakobDegen commented Mar 20, 2022

JakobDegen commented Mar 21, 2022

JakobDegen commented Mar 22, 2022

oli-obk commented Apr 11, 2022

bors commented Apr 11, 2022

bors commented Apr 11, 2022

bors commented Apr 11, 2022

rust-timer commented Apr 11, 2022

Footnotes

JakobDegen commented Apr 11, 2022

oli-obk commented Apr 12, 2022 • edited Loading

RalfJung Apr 17, 2022 • edited Loading

Choose a reason for hiding this comment

Add new `Deinit` statement #95125

Add new `Deinit` statement #95125

JakobDegen commented Mar 19, 2022 •

edited

Loading

oli-obk commented Apr 12, 2022 •

edited

Loading

RalfJung Apr 17, 2022 •

edited

Loading