Fix MIR CopyPropagation errneously propagating assignments to function args

[sinkuu]

Compiling this code with MIR CopyPropagation activated will result in printing 5,
because CopyProp errneously propagates the assignment of 5 to all x:

fn bar(mut x: u8) {
    println!("{}", x);
    x = 5;
}

fn main() {
    bar(123);
}

If a local is propagated, it will result in an ICE at trans due to an use-before-def:

fn dummy(x: u8) -> u8 { x }

fn foo(mut x: u8) {
    x = dummy(x); // this will assign a local to `x`
}

Currently CopyProp conservatively gives up if there are multiple assignments to a local,
but it is not took into account that arguments are already assigned from the beginning.
This PR fixes the problem by preventing propagation of assignments to function arguments.

[rust-highfive]

r? @arielb1

(rust_highfive has picked a reviewer for you, use r? to override)

[arielb1]

I think it's cleaner to have the check with the rest of the destination checks at ~L85:

rust/src/librustc_mir/transform/copy_prop.rs

Lines 85 to 96 in 4efcc66

	let dest_def_count = dest_use_info.def_count_not_including_drop();
	if dest_def_count == 0 {
	debug!(" Can't copy-propagate local: dest {:?} undefined",
	dest_local);
	continue
	}
	if dest_def_count > 1 {
	debug!(" Can't copy-propagate local: dest {:?} defined {} times",
	dest_local,
	dest_use_info.def_count());
	continue
	}

[sinkuu]

@arielb1 I want the copyprop to keep eliminating (arg) = (arg) assignment if possible (it would be ok if lhs and rhs are the same variable, right?), so it needs to check the src of assign statement.

(I noticed I should move the check from the current weird place to just before

rust/src/librustc_mir/transform/copy_prop.rs

Line 121 in 4efcc66

let maybe_action = match *operand {

at least.)

[arielb1]

@sinkuu

Eliminating self-assignment is ok in even more cases (e.g. even if there are multiple definitions of the value, or multiple uses - I actually think that self-assignment can be eliminated in all cases), so I'll rather do its elimination separately. Unless for some reason we generate a lot of arg = arg; assignments, but I doubt that.

[sinkuu]

Depends on #45757

[arielb1]

@bors r+

[bors]

📌 Commit 1142524 has been approved by arielb1

[bors]

⌛ Testing commit 1142524 with merge 3d2dc6e...

[bors]

☀️ Test successful - status-appveyor, status-travis
Approved by: arielb1
Pushing 3d2dc6e to master...

[sinkuu]

Thank you for reviewing.