This avoids a break in backward compatibility in the following case:
```
let ptr = std::ptr::null();
let _ = &data as *const *const ();
if data.null() {}
```
Improve error messages on LLVM bitcode parsing failure
The LLVM error causing the parse failure is now printed, in the style
of the other thin LTO error messages. This prevents a flood of
assertion failure messages if the bitcode can’t be parsed.
fix broken assertion in type_param
Nested generics (aka method generics) in trait methods don't have an
*additional* Self parameter in their own type parameter list (they have
a Self parameter in the parent generics), so don't try to check we're
correctly adjusting for it.
Fixes#46568.
r? @eddyb
This simple patch resolves#46756 (which was specifically about the case of
casts, but it would be poor form indeed to fix a reported issue without at
least a cursory attempt at answering the immortal question, "How does this bug
generalize?").
Point at var in short lived borrows instead of drop location
For RLS' sake, point at the borrow location as primary span for short lived borrows, instead of the borrow drop location.
Fix#39268.
After discussing [1] today with @pnkfelix and @Gankro,
we concluded that it’s ok for drop checking not to be much smarter
than the current `#[may_dangle]` design which requires an explicit
unsafe opt-in.
[1] https://github.com/rust-lang/rust/issues/27730#issuecomment-316432083
Loading the dependency graph in the background
Patch is a bit longer than I expected, due to the fact that most of this code relies upon a `Session` value, which is not `Sync`.
incr.comp.: Revert hashing optimization that caused regression.
This PR reverts part of #46562 which caused [a regression in the crossbeam rust-icci](https://travis-ci.org/rust-icci/crossbeam/builds/316574774) test. I don't know what the problem is exactly yet. Fortunately, the problematic part is also the less important one, so reverting should not have much impact on performance.
r? @eddyb
save-analysis: add parents to imports
This PR populates the `parent` field added to `Import` in `rls-data` 0.14.
I'm not quite sure if I handled nested imports' parents correctly: this is a new feature to me.
r? @nrc
cc https://github.com/nrc/rls-analysis/issues/123
The LLVM error causing the parse failure is now printed, in the style
of the other thin LTO error messages. This prevents a flood of
assertion failure messages if the bitcode can’t be parsed.
rustc_trans: approximate ABI alignment for padding/union fillers.
Before #45225 and after this PR, unions and enums are filled with integers of size and alignment matching their alignment (e.g. `Option<u32>` becomes `[u32; 2]`) instead of mere bytes.
Also, the alignment padding between struct fields gets this treatment after this PR.
Partially helps with some reduced testcases in #46449, although it doesn't solve the bug itself.
The input/output types found in `UniversalRegions` are not normalized.
The old code used to assign them directly into the MIR, which would
lead to errors when there was a projection in a argument or return
type. This also led to some special cases in the `renumber` code.
We now renumber uniformly but then pass the input/output types into
the MIR type-checker, which equates them with the types found in MIR.
This allows us to normalize at the same time.
This allows us to re-use the `normalize` method on `TypeCheck`, which
is important since normalization may create fresh region
variables. This is not an ideal solution, though, since the current
representation of "liveness constraints" (a vector of (region, point)
pairs) is rather inefficient. Could do somewhat better by converting
to indices, but it'd still be less good than the older code. Unclear
how important this is.
