Add initial version of codegen unit partitioning for incremental compilation.
The task of the partitioning module is to take the complete set of translation items of a crate and produce a set of codegen units from it, where a codegen unit is a named set of (translation-item, linkage) pairs. That is, this module decides which translation item appears in which codegen units with which linkage.
This version only handles the case of partitioning for incremental compilation, not the regular N-codegen units case. In the future the regular case should be handled too, maybe even doing a bit more analysis to intelligently figure out a good partitioning.
One thing that could be improved is the syntax of the codegen unit tests. Right now they still use the compile-fail error specification infrastructure, so everything has to be on one line. Would be nice to be able to format things in a more readable way.
Deduplicate libraries on hash instead of filename.
Removes the need for canonicalization to prevent #12459.
(Now with passing tests!)
Canonicalization breaks certain environments where the libraries are symlinks to files that don't end in .rlib (e.g. /remote/cas/$HASH).
Replace consider_unification_despite_ambiguity with new obligation variant
Is work towards #32730. Addresses part one of #32286. Addresses #24210 and #26046 to some degree.
r? @nikomatsakis
Save/load incremental compilation dep graph
Contains the code to serialize/deserialize the dep graph to disk between executions. We also hash the item contents and compare to the new hashes. Also includes a unit test harness. There are definitely some known limitations, such as https://github.com/rust-lang/rust/issues/32014 and https://github.com/rust-lang/rust/issues/32015, but I am leaving those for follow-up work.
Note that this PR builds on https://github.com/rust-lang/rust/pull/32007, so the overlapping commits can be excluded from review.
r? @michaelwoerister
diagnostics: make paths to external items more visible
This PR changes the reported path for an external item so that it is visible from at least one local module (i.e. it does not use any inaccessible external modules) if possible. If the external item's crate was declared with an `extern crate`, the path is guarenteed to use the `extern crate`.
Fixes#23224, fixes#23355, fixes#26635, fixes#27165.
r? @nrc
We used to track, for each crate, a path that led to the extern-crate
that imported it. Instead of that, track the def-id of the extern crate,
along with a bit more information, and derive the path on the fly.
We want to prevent compiling something against one version
of a dynamic library and then, at runtime accidentally
using a different version of the dynamic library. With the
old symbol-naming scheme this could not happen because every
symbol had the SVH in it and you'd get an error by the
dynamic linker when using the wrong version of a dylib. With
the new naming scheme this isn't the case any more, so this
patch adds the "link-guard" to prevent this error case.
This is implemented as follows:
- In every crate that we compile, we emit a function called
"__rustc_link_guard_<crate-name>_<crate-svh>"
- The body of this function contains calls to the
"__rustc_link_guard" functions of all dependencies.
- An executable contains a call to it's own
"__rustc_link_guard" function.
As a consequence the "__rustc_link_guard" function call graph
mirrors the crate graph and the dynamic linker will fail if a
wrong dylib is loaded somewhere because its
"__rustc_link_guard" function will contain a different SVH in
its name.
This hack has long since outlived its usefulness; the transition to
trans passing around full substitutions is basically done. Instead of
`ErasedRegions`, just supply substitutions with a suitable number of
`'static` entries, and invoke `erase_regions` when needed (the latter of
which we already do).
Automated conversion using the untry tool [1] and the following command:
```
$ find -name '*.rs' -type f | xargs untry
```
at the root of the Rust repo.
[1]: https://github.com/japaric/untry
projection sensitive to "mode" (most importantly, trans vs middle).
This commit introduces several pieces of iteration infrastructure in the
specialization graph data structure, as well as various helpers for
finding the definition of a given item, given its kind and name.
In addition, associated type projection is now *mode-sensitive*, with
three possible modes:
- **Topmost**. This means that projection is only possible if there is a
non-`default` definition of the associated type directly on the
selected impl. This mode is a bit of a hack: it's used during early
coherence checking before we have built the specialization
graph (and therefore before we can walk up the specialization
parents to find other definitions). Eventually, this should be
replaced with a less "staged" construction of the specialization
graph.
- **AnyFinal**. Projection succeeds for any non-`default` associated
type definition, even if it is defined by a parent impl. Used
throughout typechecking.
- **Any**. Projection always succeeds. Used by trans.
The lasting distinction here is between `AnyFinal` and `Any` -- we wish
to treat `default` associated types opaquely for typechecking purposes.
In addition to the above, the commit includes a few other minor review fixes.
- Rewrites the overlap checker to instead build up a specialization
graph, checking for overlap errors in the process.
- Use the specialization order during impl selection.
This commit does not yet handle associated types correctly, and assumes
that all items are `default` and are overridden.
typestrong const integers
~~It would be great if someone could run crater on this PR, as this has a high danger of breaking valid code~~ Crater ran. Good to go.
----
So this PR does a few things:
1. ~~const eval array values when const evaluating an array expression~~
2. ~~const eval repeat value when const evaluating a repeat expression~~
3. ~~const eval all struct and tuple fields when evaluating a struct/tuple expression~~
4. remove the `ConstVal::Int` and `ConstVal::Uint` variants and replace them with a single enum (`ConstInt`) which has variants for all integral types
* `usize`/`isize` are also enums with variants for 32 and 64 bit. At creation and various usage steps there are assertions in place checking if the target bitwidth matches with the chosen enum variant
5. enum discriminants (`ty::Disr`) are now `ConstInt`
6. trans has its own `Disr` type now (newtype around `u64`)
This obviously can't be done without breaking changes (the ones that are noticable in stable)
We could probably write lints that find those situations and error on it for a cycle or two. But then again, those situations are rare and really bugs imo anyway:
```rust
let v10 = 10 as i8;
let v4 = 4 as isize;
assert_eq!(v10 << v4 as usize, 160 as i8);
```
stops compiling because 160 is not a valid i8
```rust
struct S<T, S> {
a: T,
b: u8,
c: S
}
let s = S { a: 0xff_ff_ff_ffu32, b: 1, c: 0xaa_aa_aa_aa as i32 };
```
stops compiling because `0xaa_aa_aa_aa` is not a valid i32
----
cc @eddyb @pnkfelix
related: https://github.com/rust-lang/rfcs/issues/1071
