Initial implementation of transmutability trait. #92268

jswrenn · 2021-12-25T01:01:06Z

T'was the night before Christmas and all through the codebase, not a miri was stirring — no hint of unsafe!

This PR provides an initial, incomplete implementation of MCP 411: Lang Item for Transmutability. The core::mem::BikeshedIntrinsicFrom trait provided by this PR is implemented on-the-fly by the compiler for types Src and Dst when the bits of all possible values of type Src are safely reinterpretable as a value of type Dst.

What this PR provides is:

support for transmutations involving primitives
support for transmutations involving arrays
support for transmutations involving structs
support for transmutations involving enums
support for transmutations involving unions
support for weaker validity checks (i.e., Assume::VALIDITY)
visibility checking

What isn't yet implemented:

transmutability options passed using the Assume struct
support for references
smarter error messages

These features will be implemented in future PRs.

rust-highfive · 2021-12-25T01:01:09Z

r? @petrochenkov

(rust-highfive has picked a reviewer for you, use r? to override)

oli-obk

I did not review the new crate at all yet

oli-obk · 2021-12-25T10:42:32Z

compiler/rustc_trait_selection/src/traits/select/candidate_assembly.rs

+        if obligation.potentially_has_param_types_or_consts() {
+            return candidates.ambiguous = false;


What's going on here? Why can we set the ambiguity to false if there are generics involved? Because we know that all impls must be from trait bounds? Not sure that is true. There may be user impls on generic types giving us an ambiguous result, and now you're overwriting it as unambiguous. I don't know if this can happen or if it's fine as we'll later error out due to ambiguity, but this early return should be documented

I'm also deeply unsure about this code. If I comment it out, the compiler panics here while building stage 1 artifacts. I think I recall putting this check here after discovering that the trait definition itself was causing problems.

Really, I just stumbled my way through assemble_candidates_for_transmutability and would be very grateful for insight of what should go there.

compiler/rustc_trait_selection/src/traits/select/candidate_assembly.rs

compiler/rustc_transmute/src/nfa.rs

petrochenkov · 2021-12-25T15:20:48Z

r? @oli-obk

jswrenn · 2021-12-26T01:11:15Z

compiler/rustc_transmute/src/lib.rs

+    Yes,
+
+    /// `Src` is NOT transmutable into `Dst`.
+    No,


Re. better diagnostics: I suspect eventually we'll make this No carry information about the incompatibility that makes the transmute invalid. The challenge with this is going to be ensure that the single counter-example that's eventually surfaced in the diagnostic is the most useful one.

bors · 2022-01-15T08:00:24Z

☔ The latest upstream changes (presumably #92915) made this pull request unmergeable. Please resolve the merge conflicts.

apiraino · 2022-03-03T12:37:15Z

Discussed in T-compiler meeting (on zulip).

I'm flipping the switch as it seems this PR is still a bit in progress (but don't hesitate to ask for comments or suggestions).

@rustbot author

JohnCSimon · 2022-04-11T01:55:32Z

@jswrenn

Ping from triage: Can you please post the status of this PR?
Thank you.

jswrenn · 2022-06-14T18:48:47Z

Hey @JohnCSimon, so sorry for missing your comment (I had just defended my thesis, went offline for a bit, and evidently missed a lot of notifications). This PR needs some minor work before it's ready to graduate out of draft status, namely: implementing the NFA lowering for enums and unions. I've budgeted that work for this Thursday. Basic support for references, error messages, and recursive types will come in follow-up PRs.

jswrenn · 2022-06-16T20:58:28Z

Hm, I don't get the diagnostics about TyKind when I build locally. Anyone know why that might be?

m1el · 2022-06-17T05:38:06Z

compiler/rustc_transmute/src/nfa.rs

+
+    /// Constructs an `Nfa` that describes the layout of the given discriminant.
+    pub fn from_disr(discr: Discr<'tcx>, tcx: TyCtxt<'tcx>) -> Self {
+        // FIXME(@jswrenn): I'm certain this is missing needed endian nuance.


Maybe rustc_middle::mir::interpret::write_target_uint?

m1el · 2022-06-17T05:39:28Z

compiler/rustc_transmute/src/nfa.rs

+                    }
+                    AdtKind::Enum => {
+                        // is the layout well-defined?
+                        if !(adt_def.repr().c() || adt_def.repr().int.is_some()) {


should be && here.

Doesn't this have a well-defined layout?

#[repr(u8)] enum Foo { A, B, C }

Yes, it does.
However, #[repr(u8)] enum Foo { A(u8,u32,u8) } ~~does not~~
has a different representation from #[repr(u8, C)].

Edit: It was wrong

jswrenn · 2022-07-28T18:59:32Z

@bors r=oli-obk

bors · 2022-07-28T18:59:33Z

📌 Commit e8a1925 has been approved by oli-obk

It is now in the queue for this repository.

bors · 2022-07-30T20:11:35Z

⌛ Testing commit e8a1925 with merge 6ac6e3903baf610c3dfe8071e0919b0e5c1a480d...

bors · 2022-07-30T20:45:37Z

💔 Test failed - checks-actions

jswrenn · 2022-08-02T15:38:59Z

@bors r=oli-obk

bors · 2022-08-02T15:39:00Z

📌 Commit 965ffb0 has been approved by oli-obk

It is now in the queue for this repository.

matthiaskrgr · 2022-08-02T21:09:29Z

@bors p=15 (moving before rollup, also pr is older than half a year already)

bors · 2022-08-02T21:17:34Z

⌛ Testing commit 965ffb0 with merge e4417cf...

bors · 2022-08-03T00:03:46Z

☀️ Test successful - checks-actions
Approved by: oli-obk
Pushing e4417cf to master...

rust-timer · 2022-08-03T01:20:28Z

Finished benchmarking commit (e4417cf): comparison url.

Instruction count

Primary benchmarks: no relevant changes found
Secondary benchmarks: 😿 relevant regressions found

	mean¹	max	count²
Regressions 😿 (primary)	N/A	N/A	0
Regressions 😿 (secondary)	0.6%	1.4%	4
Improvements 🎉 (primary)	N/A	N/A	0
Improvements 🎉 (secondary)	N/A	N/A	0
All 😿🎉 (primary)	N/A	N/A	0

Max RSS (memory usage)

Results

Primary benchmarks: no relevant changes found
Secondary benchmarks: 😿 relevant regressions found

	mean¹	max	count²
Regressions 😿 (primary)	N/A	N/A	0
Regressions 😿 (secondary)	4.0%	4.8%	2
Improvements 🎉 (primary)	N/A	N/A	0
Improvements 🎉 (secondary)	N/A	N/A	0
All 😿🎉 (primary)	N/A	N/A	0

Cycles

Results

Primary benchmarks: 🎉 relevant improvement found
Secondary benchmarks: 🎉 relevant improvement found

	mean¹	max	count²
Regressions 😿 (primary)	N/A	N/A	0
Regressions 😿 (secondary)	N/A	N/A	0
Improvements 🎉 (primary)	-2.8%	-2.8%	1
Improvements 🎉 (secondary)	-4.1%	-4.1%	1
All 😿🎉 (primary)	-2.8%	-2.8%	1

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

@rustbot label: -perf-regression

the arithmetic mean of the percent change ↩ ↩² ↩³
number of relevant changes ↩ ↩² ↩³

This was left as a TODO in rust-lang#92268, and brings the trait more in line with what was defined in MCP411. `Assume::visibility` has been renamed to `Assume::safety`, as library safety is what's actually being assumed; visibility is just the mechanism by which it is currently checked (this may change). ref: rust-lang/compiler-team#411 ref: rust-lang#99571

safe transmute: use `Assume` struct to provide analysis options This task was left as a TODO in rust-lang#92268; resolving it brings [`BikeshedIntrinsicFrom`](https://doc.rust-lang.org/nightly/core/mem/trait.BikeshedIntrinsicFrom.html) more in line with the API defined in [MCP411](rust-lang/compiler-team#411). **Before:** ```rust pub unsafe trait BikeshedIntrinsicFrom< Src, Context, const ASSUME_ALIGNMENT: bool, const ASSUME_LIFETIMES: bool, const ASSUME_VALIDITY: bool, const ASSUME_VISIBILITY: bool, > where Src: ?Sized, {} ``` **After:** ```rust pub unsafe trait BikeshedIntrinsicFrom<Src, Context, const ASSUME: Assume = { Assume::NOTHING }> where Src: ?Sized, {} ``` `Assume::visibility` has also been renamed to `Assume::safety`, as library safety invariants are what's actually being assumed; visibility is just the mechanism by which it is currently checked (and that may change). r? `@oli-obk` --- Related: - rust-lang/compiler-team#411 - rust-lang#99571

ref: rust-lang/rust#92268 (comment)

rust-highfive assigned petrochenkov Dec 25, 2021

rustbot added the T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. label Dec 25, 2021

rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Dec 25, 2021

jswrenn force-pushed the transmute branch from ea23b2c to 1d3fe43 Compare December 25, 2021 01:19

jswrenn mentioned this pull request Dec 25, 2021

project-safe-transmute rust-lang/lang-team#21

Open

oli-obk requested changes Dec 25, 2021

View reviewed changes

fbstj reviewed Dec 25, 2021

View reviewed changes

compiler/rustc_transmute/src/nfa.rs Outdated Show resolved Hide resolved

rust-highfive assigned oli-obk and unassigned petrochenkov Dec 25, 2021

jswrenn commented Dec 26, 2021

View reviewed changes

jswrenn force-pushed the transmute branch 2 times, most recently from 1d8176e to f848e36 Compare January 13, 2022 19:18

This comment has been minimized.

Sign in to view

jswrenn force-pushed the transmute branch from f848e36 to b071fc0 Compare January 13, 2022 22:02

JohnCSimon added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Feb 27, 2022

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Mar 3, 2022

jswrenn force-pushed the transmute branch from b071fc0 to d29263c Compare June 16, 2022 16:46

This comment has been minimized.

Sign in to view

jswrenn force-pushed the transmute branch from d29263c to c82092e Compare June 16, 2022 21:58

m1el reviewed Jun 17, 2022

View reviewed changes

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-review Status: Awaiting review from the assignee but also interested parties. labels Jul 28, 2022

bors added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. labels Jul 30, 2022

This comment has been minimized.

Sign in to view

safe transmute: fix broken intradoc link

965ffb0

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-review Status: Awaiting review from the assignee but also interested parties. labels Aug 2, 2022

bors added the merged-by-bors This PR was explicitly merged by bors. label Aug 3, 2022

bors merged commit e4417cf into rust-lang:master Aug 3, 2022

rustbot added this to the 1.64.0 milestone Aug 3, 2022

jswrenn mentioned this pull request Aug 18, 2022

safe transmute: use Assume struct to provide analysis options #100726

Merged

workingjubilee pushed a commit to tcdi/postgrestd that referenced this pull request Sep 15, 2022

safe transmute: add rustc_on_unimplemented to BikeshedIntrinsicFrom

cf4bdd0

ref: rust-lang/rust#92268 (comment)

workingjubilee pushed a commit to tcdi/postgrestd that referenced this pull request Sep 15, 2022

safe transmute: reference tracking issue

2d04c50

ref: rust-lang/rust#92268 (comment)

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Initial implementation of transmutability trait. #92268

Initial implementation of transmutability trait. #92268

jswrenn commented Dec 25, 2021 •

edited

Loading

rust-highfive commented Dec 25, 2021

oli-obk left a comment

oli-obk Dec 25, 2021

jswrenn Dec 25, 2021

petrochenkov commented Dec 25, 2021

jswrenn Dec 26, 2021

This comment has been minimized.

bors commented Jan 15, 2022

apiraino commented Mar 3, 2022 •

edited

Loading

JohnCSimon commented Apr 11, 2022

jswrenn commented Jun 14, 2022

This comment has been minimized.

jswrenn commented Jun 16, 2022

m1el Jun 17, 2022

m1el Jun 17, 2022

jswrenn Jun 17, 2022

m1el Jun 17, 2022 •

edited

Loading

jswrenn commented Jul 28, 2022

bors commented Jul 28, 2022

bors commented Jul 30, 2022

bors commented Jul 30, 2022

This comment has been minimized.

jswrenn commented Aug 2, 2022

bors commented Aug 2, 2022

matthiaskrgr commented Aug 2, 2022

bors commented Aug 2, 2022

bors commented Aug 3, 2022

rust-timer commented Aug 3, 2022

		if obligation.potentially_has_param_types_or_consts() {
		return candidates.ambiguous = false;

Initial implementation of transmutability trait. #92268

Initial implementation of transmutability trait. #92268

Conversation

jswrenn commented Dec 25, 2021 • edited Loading

rust-highfive commented Dec 25, 2021

oli-obk left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

petrochenkov commented Dec 25, 2021

Choose a reason for hiding this comment

This comment has been minimized.

bors commented Jan 15, 2022

apiraino commented Mar 3, 2022 • edited Loading

JohnCSimon commented Apr 11, 2022

jswrenn commented Jun 14, 2022

This comment has been minimized.

jswrenn commented Jun 16, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

m1el Jun 17, 2022 • edited Loading

Choose a reason for hiding this comment

jswrenn commented Jul 28, 2022

bors commented Jul 28, 2022

bors commented Jul 30, 2022

bors commented Jul 30, 2022

This comment has been minimized.

jswrenn commented Aug 2, 2022

bors commented Aug 2, 2022

matthiaskrgr commented Aug 2, 2022

bors commented Aug 2, 2022

bors commented Aug 3, 2022

rust-timer commented Aug 3, 2022

Footnotes

jswrenn commented Dec 25, 2021 •

edited

Loading

apiraino commented Mar 3, 2022 •

edited

Loading

m1el Jun 17, 2022 •

edited

Loading