use PlaceRef abstractions more consistently

[henryboisdequin]

Addresses this comment
Associated issue: #80647

r? @RalfJung

[henryboisdequin]

@rustbot label: +C-cleanup +A-mir +T-compiler

[henryboisdequin]

@RalfJung Update:

• Addressed all of your comments (but not pushed yet)
• I'm currently changing visit_projection and all of its associated functions to take a PlaceRef

[henryboisdequin]

@rustbot label: +S-waiting-on-author -S-waiting-on-review

[henryboisdequin]

@RalfJung Nearly finished but I still have some errors and don't know how to resolve them. The current code is pushed, as you can see. Here are the errors:

error[E0308]: mismatched types
    --> compiler/rustc_middle/src/mir/visit.rs:154:34
     |
68   | / macro_rules! make_mir_visitor {
69   | |     ($visitor_trait_name:ident, $($mutability:ident)?) => {
70   | |         pub trait $visitor_trait_name<'tcx> {
71   | |             // Override these, and call `self.super_xxx` to revert back to the
...    |
154  | |                 self.super_place(place_ref, context, location);
     | |                                  ^^^^^^^^^ expected struct `mir::PlaceRef`, found `&mut mir::PlaceRef<'tcx>`
...    |
929  | |     }
930  | | }
     | |_- in this expansion of `make_mir_visitor!`
...
1084 |   make_mir_visitor!(Visitor,);
     |   ---------------------------- in this macro invocation

error[E0308]: mismatched types
    --> compiler/rustc_middle/src/mir/visit.rs:383:29
     |
68   | / macro_rules! make_mir_visitor {
69   | |     ($visitor_trait_name:ident, $($mutability:ident)?) => {
70   | |         pub trait $visitor_trait_name<'tcx> {
71   | |             // Override these, and call `self.super_xxx` to revert back to the
...    |
383  | |                             &mut place.as_ref(),
     | |                             ^^^^^^^^^^^^^^^^^^^ expected struct `mir::PlaceRef`, found `&mir::Place<'_>`
...    |
929  | |     }
930  | | }
     | |_- in this expansion of `make_mir_visitor!`
...
1084 |   make_mir_visitor!(Visitor,);
     |   ---------------------------- in this macro invocation
     |
     = note: expected mutable reference `&mut mir::PlaceRef<'tcx>`
                found mutable reference `&mut &mir::Place<'_>`

error[E0308]: mismatched types
    --> compiler/rustc_middle/src/mir/visit.rs:390:29
     |
68   | / macro_rules! make_mir_visitor {
69   | |     ($visitor_trait_name:ident, $($mutability:ident)?) => {
70   | |         pub trait $visitor_trait_name<'tcx> {
71   | |             // Override these, and call `self.super_xxx` to revert back to the
...    |
390  | |                             &mut place.as_ref(),
     | |                             ^^^^^^^^^^^^^^^^^^^ expected struct `mir::PlaceRef`, found `&mir::Place<'_>`
...    |
929  | |     }
930  | | }
     | |_- in this expansion of `make_mir_visitor!`
...
1084 |   make_mir_visitor!(Visitor,);
     |   ---------------------------- in this macro invocation
     |
     = note: expected mutable reference `&mut mir::PlaceRef<'tcx>`
                found mutable reference `&mut &mir::Place<'_>`

error[E0308]: mismatched types
    --> compiler/rustc_middle/src/mir/visit.rs:383:29
     |
68   | / macro_rules! make_mir_visitor {
69   | |     ($visitor_trait_name:ident, $($mutability:ident)?) => {
70   | |         pub trait $visitor_trait_name<'tcx> {
71   | |             // Override these, and call `self.super_xxx` to revert back to the
...    |
383  | |                             &mut place.as_ref(),
     | |                             ^^^^^^^^^^^^^^^^^^^ expected struct `mir::PlaceRef`, found `&mir::Place<'_>`
...    |
929  | |     }
930  | | }
     | |_- in this expansion of `make_mir_visitor!`
...
1085 |   make_mir_visitor!(MutVisitor, mut);
     |   ----------------------------------- in this macro invocation
     |
     = note: expected mutable reference `&mut mir::PlaceRef<'tcx>`
                found mutable reference `&mut &mir::Place<'_>`

error[E0308]: mismatched types
    --> compiler/rustc_middle/src/mir/visit.rs:390:29
     |
68   | / macro_rules! make_mir_visitor {
69   | |     ($visitor_trait_name:ident, $($mutability:ident)?) => {
70   | |         pub trait $visitor_trait_name<'tcx> {
71   | |             // Override these, and call `self.super_xxx` to revert back to the
...    |
390  | |                             &mut place.as_ref(),
     | |                             ^^^^^^^^^^^^^^^^^^^ expected struct `mir::PlaceRef`, found `&mir::Place<'_>`
...    |
929  | |     }
930  | | }
     | |_- in this expansion of `make_mir_visitor!`
...
1085 |   make_mir_visitor!(MutVisitor, mut);
     |   ----------------------------------- in this macro invocation
     |
     = note: expected mutable reference `&mut mir::PlaceRef<'tcx>`
                found mutable reference `&mut &mir::Place<'_>`

error: aborting due to 5 previous errors

For more information about this error, try `rustc --explain E0308`.
error: could not compile `rustc_middle`

I'm not really sure why they are happening or how to fix them. Help would be appreciated!

[RalfJung]

@henryboisdequin thanks! I'll take a look, but I can't promise I'll do that before next weekend. In the mean time, you could also ask on the Rust Zulip; there might be people there that can help you.

[henryboisdequin]

@RalfJung on Zulip I and a couple of others discussed that it would be impossible to replace &mut Place with &mut PlaceRef in a mutable visitor. To make such a change, we would require a lot more refactoring which I think should be done in another PR. In this PR, it just addresses your comment. I'll start working on the other PR!

Oops: #80647 (comment)

[henryboisdequin]

@rustbot label -S-waiting-on-author +S-waiting-on-review

[henryboisdequin]

@RalfJung visit_projection now takes a PlaceRef

[RalfJung]

Awesome, thanks a lot. :)
Let's see if CI passes, then I'll r+.

[RalfJung]

@bors r+

[bors]

📌 Commit a9c6188 has been approved by RalfJung

[camelid]

Associated issue: #80647

Should this PR close that issue, or no?

[henryboisdequin]

Associated issue: #80647

Should this PR close that issue, or no?

@RalfJung What do you think?

[RalfJung]

No it should not close the issue, there's still more cleanup left to be done.