Custom MIR: Many more improvements #105356

JakobDegen · 2022-12-06T07:09:08Z

Commits are each atomic changes, best reviewed one at a time, with the exception that the last commit includes all the documentation.

First commit

Unsafetyck was not correctly disabled before for dialect = "built" custom MIR. This is fixed and a regression test is added.

Second commit

Implements Discriminant, SetDiscriminant, and SwitchInt.

Third commit

Implements indexing, field, and variant projections.

Fourth commit

Documents the previous commits and everything else.

There is some amount of weirdness here due to having to beat Rust syntax into cooperating with MIR concepts, but it hopefully should not be too much. All of it is documented.

r? @oli-obk

rustbot · 2022-12-06T07:09:16Z

Some changes occurred in compiler/rustc_codegen_cranelift

cc @bjorn3

Hey! It looks like you've submitted a new PR for the library teams!

If this PR contains changes to any rust-lang/rust public library APIs then please comment with @rustbot label +T-libs-api -T-libs to tag it appropriately. If this PR contains changes to any unstable APIs please edit the PR description to add a link to the relevant API Change Proposal or create one if you haven't already. If you're unsure where your change falls no worries, just leave it as is and the reviewer will take a look and make a decision to forward on if necessary.

Examples of T-libs-api changes:

Stabilizing library features
Introducing insta-stable changes such as new implementations of existing stable traits on existing stable types
Introducing new or changing existing unstable library APIs (excluding permanently unstable features / features without a tracking issue)
Changing public documentation in ways that create new stability guarantees
Changing observable runtime behavior of library APIs

Some changes occurred to MIR optimizations

cc @rust-lang/wg-mir-opt

Some changes occurred in src/tools/clippy

cc @rust-lang/clippy

This PR changes MIR

cc @oli-obk, @RalfJung, @JakobDegen, @davidtwco, @celinval, @vakaras

Some changes occurred to the CTFE / Miri engine

cc @rust-lang/miri

JakobDegen · 2022-12-06T07:09:32Z

Oh, did not mean to reping everyone. Sorry about that

oli-obk · 2022-12-06T12:35:06Z

@bors r+

bors · 2022-12-06T12:35:08Z

📌 Commit c3d532e8f2b3036f6b09cc5cab653c952d90822f has been approved by oli-obk

It is now in the queue for this repository.

JakobDegen · 2022-12-06T12:46:27Z

@oli-obk probably shouldn't merge this until the parent is merged

oli-obk · 2022-12-06T12:47:59Z

@bors r- oh right, there was a reason I didn't r+ this earlier ^^ I thought it was jsut CI

bors · 2022-12-08T06:53:46Z

☔ The latest upstream changes (presumably #105425) made this pull request unmergeable. Please resolve the merge conflicts.

library/core/src/intrinsics/mir.rs

saethlin · 2022-12-11T05:07:26Z

library/core/src/intrinsics/mir.rs

+//! extern crate core;
+//! use core::intrinsics::mir::*;
+//!
+//! #[custom_mir(dialect = "built")]


What is a dialect? Perhaps I'm just supposed to know this, but it's not documented anywhere here and somehow I've escaped learning about this concept.

Docs. I've added a link and brief explanation

library/core/src/intrinsics/mir.rs

saethlin · 2022-12-11T05:11:28Z

library/core/src/intrinsics/mir.rs

+//! #### Rvalues
+//!
+//!  - Operands implicitly convert to `Use` rvalues.
+//!  - `&`, `&mut`, `addr_of!`, and `addr_of_mut!` all work to create their associated rvalue.


Why addr_of! instead of &raw? IMO this is going too far in trying to imitate the way people write surface Rust, and &raw makes more sense here.

&raw works just as well. `addr_of! literally just expands to that, the only difference is that you don't need to turn on a feature to use the latter

Yes. addr_of! just seems like a very high-level construct to adopt as custom MIR syntax. And personally I like the symmetry of &, &raw, &mut.

saethlin · 2022-12-11T05:12:32Z

library/core/src/intrinsics/mir.rs

+//! #### Terminators
+//!
+//!  - [`Goto`] and [`Return`] have associated functions.
+//!  - `match some_int_operand` becomes a `SwitchInt`. Each arm should be `literal => basic_block`


The choice to use Rust match syntax makes RET and Return() even more odd. I think in this case it might just be better to at least call it SwitchInt, because otherwise it seems to easy to forget that this match only works on integers.

saethlin · 2022-12-11T05:14:57Z

Jake asked me for comments in general, and I have some comments that are not on lines changed in this PR:

        exit = {
            temp2 = Move(temp1);
            RET = temp2;
            Return()
        }

At this stage, reading this in the docs, it's entirely unclear to me whether exit is a magic block name. Perhaps pick a different name.

RET as a magic name for the return place is odd. We only use shouty case for consts and statics in Rust, and I don't think it is used anywhere in MIR. I don't have a suggestion for a better name. It's probably a step up from just having to know that _0 is very different from _1 but I hope we can do better?

Return() is an odd choice amongst all this other syntax; declaration and assignment exactly mirrors Rust, why not the syntax for returning from a function?

JakobDegen · 2022-12-11T06:07:12Z

RET as a magic name for the return place is odd. We only use shouty case for consts and statics in Rust, and I don't think it is used anywhere in MIR. I don't have a suggestion for a better name.

I don't have a better idea either. I suppose ret in lower case is an option? The uppercase might help let people know that they should not be looking for an associated let-statement.

I think in this case it might just be better to at least call it SwitchInt

declaration and assignment exactly mirrors Rust, why not the syntax for returning from a function?

The mir! macro has to expand to something that parses and type checks normally. That means any new syntax we introduce has to be re-written by the macro (keeping in mind that this is a macro_rules!). I could do that for switch_int foo { ... } => match foo { ... }, but that might not be easier to remember, and it's likely to degrade error messages as spans get mangled.

return syntax has the same problem. I'd have to re-write it to Return() in the macro (since return; won't type check in a function returning i32). The problem is that I then don't have anything great to do with return 5;. I can try and detect that in the macro too, but the compiler_error! I generate won't have a good span. I'm not so sure that's better

saethlin · 2022-12-12T00:23:48Z

Thanks for the improvements 👍

The limitations of the syntax make a bit more sense knowing a bit about the implementation 😉 though I wonder about the wisdom of using macro_rules! for something so complicated, I understand the tradeoff you're juggling and I'm not sure I would do it differently.

JakobDegen · 2022-12-12T07:42:28Z

though I wonder about the wisdom of using macro_rules! for something so complicated

I would love to write a proc macro for this instead, but then it wouldn't be available in std and that seems like it defeats most of the point

library/core/src/intrinsics/mir.rs

oli-obk · 2022-12-14T10:04:59Z

@bors r+

bors · 2022-12-14T10:05:02Z

📌 Commit b580f29 has been approved by oli-obk

It is now in the queue for this repository.

bors · 2022-12-15T19:59:52Z

⌛ Testing commit b580f29 with merge ec56537...

bors · 2022-12-15T22:52:56Z

☀️ Test successful - checks-actions
Approved by: oli-obk
Pushing ec56537 to master...

rust-timer · 2022-12-16T00:12:26Z

Finished benchmarking commit (ec56537): comparison URL.

Overall result: ❌ regressions - no action needed

@rustbot label: -perf-regression

Instruction count

This is a highly reliable metric that was used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	0.6%	[0.6%, 0.6%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	2.2%	[2.2%, 2.2%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-2.3%	[-2.4%, -2.2%]	2
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-0.8%	[-2.4%, 2.2%]	3

Cycles

This benchmark run did not return any relevant results for this metric.

Custom MIR: Many more improvements Commits are each atomic changes, best reviewed one at a time, with the exception that the last commit includes all the documentation. ### First commit Unsafetyck was not correctly disabled before for `dialect = "built"` custom MIR. This is fixed and a regression test is added. ### Second commit Implements `Discriminant`, `SetDiscriminant`, and `SwitchInt`. ### Third commit Implements indexing, field, and variant projections. ### Fourth commit Documents the previous commits and everything else. There is some amount of weirdness here due to having to beat Rust syntax into cooperating with MIR concepts, but it hopefully should not be too much. All of it is documented. r? `@oli-obk`

rustbot assigned oli-obk Dec 6, 2022

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Dec 6, 2022

This comment has been minimized.

Sign in to view

JakobDegen force-pushed the more-custom-mir branch from 5c2b44f to e17bc54 Compare December 6, 2022 08:10

This comment has been minimized.

Sign in to view

JakobDegen force-pushed the more-custom-mir branch from e17bc54 to 9706162 Compare December 6, 2022 09:03

This comment has been minimized.

Sign in to view

JakobDegen force-pushed the more-custom-mir branch from 9706162 to b998485 Compare December 6, 2022 10:33

This comment has been minimized.

Sign in to view

JakobDegen force-pushed the more-custom-mir branch from b998485 to c3d532e Compare December 6, 2022 11:55

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 Dec 6, 2022

bors 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-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. labels Dec 6, 2022

JakobDegen force-pushed the more-custom-mir branch from c3d532e to eaa31d3 Compare December 11, 2022 00:02

saethlin reviewed Dec 11, 2022

View reviewed changes

library/core/src/intrinsics/mir.rs Outdated Show resolved Hide resolved

saethlin reviewed Dec 11, 2022

View reviewed changes

library/core/src/intrinsics/mir.rs Outdated Show resolved Hide resolved

saethlin reviewed Dec 11, 2022

View reviewed changes

library/core/src/intrinsics/mir.rs Outdated Show resolved Hide resolved

saethlin reviewed Dec 11, 2022

View reviewed changes

oli-obk reviewed Dec 12, 2022

View reviewed changes

library/core/src/intrinsics/mir.rs Outdated Show resolved Hide resolved

JakobDegen added 5 commits December 14, 2022 01:02

Fix unsafetyck disabling for custom MIR

c1b27ee

Support common enum operations in custom mir

409f4d2

Support more projections in custom mir

e598394

Add documentation for custom mir

aca1bc5

Address documentation suggestions

b580f29

JakobDegen force-pushed the more-custom-mir branch from b5bbec1 to b580f29 Compare December 14, 2022 09:12

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 Dec 14, 2022

bors added the merged-by-bors This PR was explicitly merged by bors. label Dec 15, 2022

bors merged commit ec56537 into rust-lang:master Dec 15, 2022

rustbot added this to the 1.68.0 milestone Dec 15, 2022

JakobDegen deleted the more-custom-mir branch December 16, 2022 13:44

JakobDegen mentioned this pull request Dec 16, 2022

Bump master bootstrap compiler #105741

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Custom MIR: Many more improvements #105356

Custom MIR: Many more improvements #105356

JakobDegen commented Dec 6, 2022 •

edited

Loading

rustbot commented Dec 6, 2022

JakobDegen commented Dec 6, 2022

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

oli-obk commented Dec 6, 2022

bors commented Dec 6, 2022

JakobDegen commented Dec 6, 2022

oli-obk commented Dec 6, 2022

bors commented Dec 8, 2022

saethlin Dec 11, 2022

JakobDegen Dec 11, 2022

saethlin Dec 11, 2022

JakobDegen Dec 11, 2022

saethlin Dec 11, 2022

saethlin Dec 11, 2022

saethlin commented Dec 11, 2022

JakobDegen commented Dec 11, 2022

saethlin commented Dec 12, 2022

JakobDegen commented Dec 12, 2022

oli-obk commented Dec 14, 2022

bors commented Dec 14, 2022

bors commented Dec 15, 2022

bors commented Dec 15, 2022

rust-timer commented Dec 16, 2022

Custom MIR: Many more improvements #105356

Custom MIR: Many more improvements #105356

Conversation

JakobDegen commented Dec 6, 2022 • edited Loading

First commit

Second commit

Third commit

Fourth commit

rustbot commented Dec 6, 2022

JakobDegen commented Dec 6, 2022

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

oli-obk commented Dec 6, 2022

bors commented Dec 6, 2022

JakobDegen commented Dec 6, 2022

oli-obk commented Dec 6, 2022

bors commented Dec 8, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

saethlin commented Dec 11, 2022

JakobDegen commented Dec 11, 2022

saethlin commented Dec 12, 2022

JakobDegen commented Dec 12, 2022

oli-obk commented Dec 14, 2022

bors commented Dec 14, 2022

bors commented Dec 15, 2022

bors commented Dec 15, 2022

rust-timer commented Dec 16, 2022

Overall result: ❌ regressions - no action needed

JakobDegen commented Dec 6, 2022 •

edited

Loading