Allow MIR borrowck to catch unused mutable locals

[KiChjang]

Fixes #47279.

r? @nikomatsakis

[nikomatsakis]

This looks about right. There may be some complications around closures:

let mut x = &mut vec![]; 
//  ^^^ not needed
let closure = || {
    x.push(22);
};
closure();

Also, we have to disable the other lint, from old AST borrowck

[bors]

☔ The latest upstream changes (presumably #48208) made this pull request unmergeable. Please resolve the merge conflicts.

[nikomatsakis]

This looks great! My main concern is the tests, left some suggestions below.

[nikomatsakis]

I don't think that this will necessarily test the way you want it. How about this instead?

// revisions: lexical nll
#![cfg_attr(nll, feature(nll))]

This will run the test twice, once in normal mode and once in nll mode.

[nikomatsakis]

Also, can we add another test case for nested closures? I see this one:

 let mut a = Vec::new();
    callback(|| {
        a.push(3);
});

but I think I want:

 let mut a = Vec::new();
    callback(|| {
        callback(|| a.push(3)); // nested call
    });
});

as well

[nikomatsakis]

We're getting travis errors. The problem is that we have to update the //~ ERROR declarations in this file to account for the revisions.

For each line:

foo //~ ERROR bar

we have to change it to

foo //[lexical]~ ERROR bar
 //[nll]~^ ERROR bar

[KiChjang]

Ok, so I figured out why Travis CI is complaining. Given the following code:

fn main() {
    let mut a = 3;
}

The current implementation sees an assignment to a during initialization, which makes it erroneously believe that the value has been mutated. We need to make it so that it understands that initialization is not equivalent with mutation.

[KiChjang]

Ok, so the latest commit still doesn't fix everything yet. It's quite over-eager in its job and flags too much fn args as being unnecessarily mutable. It also doesn't catch the following case:

    fn mut_ref_arg(mut arg : &mut [u8]) -> &mut [u8] {
        &mut arg[..] //[lexical]~^ ERROR: variable does not need to be mutable
                     //[nll]~^ ERROR: variable does not need to be mutable
    }

... which tells me that we need additional logic in consume_rvalue.

[bors]

☔ The latest upstream changes (presumably #48770) made this pull request unmergeable. Please resolve the merge conflicts.

[nikomatsakis]

@KiChjang

We need to make it so that it understands that initialization is not equivalent with mutation.

We should be able to use the initialization state for this. Presumably we are doing a similar analysis already to report errors in the case that the variable is not mut...

Ah, indeed we do, right here:

rust/src/librustc_mir/borrow_check/mod.rs

Lines 1279 to 1303 in 8c4ff22

	fn check_if_reassignment_to_immutable_state(
	&mut self,
	context: Context,
	(place, span): (&Place<'tcx>, Span),
	flow_state: &Flows<'cx, 'gcx, 'tcx>,
	) {
	debug!("check_if_reassignment_to_immutable_state({:?})", place);
	// determine if this path has a non-mut owner (and thus needs checking).
	if let Ok(()) = self.is_mutable(place, LocalMutationIsAllowed::No) {
	return;
	}
	debug!(
	"check_if_reassignment_to_immutable_state({:?}) - is an imm local",
	place
	);
	for i in flow_state.ever_inits.elems_incoming() {
	let init = self.move_data.inits[i];
	let init_place = &self.move_data.move_paths[init.path].place;
	if self.places_conflict(&init_place, place, Deep) {
	self.report_illegal_reassignment(context, (place, span), init.span);
	break;
	}
	}
	}

[nikomatsakis]

I think logic is too simplistic. For example, consider this case:

let mut x = (1, 2);
x.0 = 22;

Here, the mutated place will be a field projection, but we still need to record x as a "used mut". We want to integrate more with the is_mutable code, I think -- in particular, maybe it should return the "root reason" that a path was mutable, even on success (right now it does that only on error). Then, if that root reason is a Local, we can add it to the used_mut set? etc

[Zoxc]

@bors r=nikomatsakis

[bors]

📌 Commit 0a1cb9b has been approved by nikomatsakis

[bors]

⌛ Testing commit 0a1cb9b with merge 4073a87ff295c3262735383970366c4f2d12d6ec...

[bors]

💔 Test failed - status-travis

[rust-highfive]

The job arm-android of your PR failed on Travis (raw log). Through arcane magic we have determined that the following fragments from the build log may contain information about the problem.

Click to expand the log.

[00:10:15] Looks like docker image is the same as before, not uploading
[00:10:15] travis_fold:end:build_docker

[00:10:16] + export SHELL=/bin/bash
[00:10:16] + nohup+  nohup emulatorexec @armeabi-v7a-18 -engine /checkout/src/ci/run.sh classic
[00:10:16]  -no-window -partition-size 2047
[00:10:16] travis_time:end:343bda7e:start=1524986800823055427,finish=1524987405636164840,duration=604813109413
[CI_JOB_NAME=arm-android]
[00:10:16] [CI_JOB_NAME=arm-android]
[00:10:16] Starting sccache server...
---
[01:24:15] /checkout/obj/build/x86_64-unknown-linux-gnu/test/debuginfo/var-captured-in-nested-closure.stage2-arm-linux-androideabi: 1 file pushed. 6.4 MB/s (29380 bytes in 0.004s)
[01:24:17] ok
[01:24:17] test [debuginfo-gdb] debuginfo/var-captured-in-sendable-closure.rs ... [100%] /data/tmp/work/var-captured-in-sendable-closure.stage2-arm-linux-androideabi
[01:24:17] /checkout/obj/build/x86_64-unknown-linux-gnu/test/debuginfo/var-captured-in-sendable-closure.stage2-arm-linux-androideabi: 1 file pushed. 3.6 MB/s (29428 bytes in 0.008s)
No output has been received in the last 30m0s, this potentially indicates a stalled build or something wrong with the build itself.
Check the details on how to adjust your build configuration on: https://docs.travis-ci.com/user/common-build-problems/#Build-times-out-because-no-output-was-received
The build has been terminated

I'm a bot! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact @TimNN. (Feature Requests)

[rust-highfive]

The job arm-android of your PR failed on Travis (raw log). Through arcane magic we have determined that the following fragments from the build log may contain information about the problem.

Click to expand the log.

[00:10:15] Looks like docker image is the same as before, not uploading
[00:10:15] travis_fold:end:build_docker

[00:10:16] + export SHELL=/bin/bash
[00:10:16] + nohup+  nohup emulatorexec @armeabi-v7a-18 -engine /checkout/src/ci/run.sh classic
[00:10:16]  -no-window -partition-size 2047
[00:10:16] travis_time:end:343bda7e:start=1524986800823055427,finish=1524987405636164840,duration=604813109413
[CI_JOB_NAME=arm-android]
[00:10:16] [CI_JOB_NAME=arm-android]
[00:10:16] Starting sccache server...
---
[01:24:15] /checkout/obj/build/x86_64-unknown-linux-gnu/test/debuginfo/var-captured-in-nested-closure.stage2-arm-linux-androideabi: 1 file pushed. 6.4 MB/s (29380 bytes in 0.004s)
[01:24:17] ok
[01:24:17] test [debuginfo-gdb] debuginfo/var-captured-in-sendable-closure.rs ... [100%] /data/tmp/work/var-captured-in-sendable-closure.stage2-arm-linux-androideabi
[01:24:17] /checkout/obj/build/x86_64-unknown-linux-gnu/test/debuginfo/var-captured-in-sendable-closure.stage2-arm-linux-androideabi: 1 file pushed. 3.6 MB/s (29428 bytes in 0.008s)
No output has been received in the last 30m0s, this potentially indicates a stalled build or something wrong with the build itself.
Check the details on how to adjust your build configuration on: https://docs.travis-ci.com/user/common-build-problems/#Build-times-out-because-no-output-was-received
The build has been terminated

I'm a bot! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact @TimNN. (Feature Requests)

[kennytm]

@bors retry

Assuming the 30-minute timeout is spurious.

[bors]

⌛ Testing commit 0a1cb9b with merge 79252ff...

[bors]

☀️ Test successful - status-appveyor, status-travis
Approved by: nikomatsakis
Pushing 79252ff to master...