Lazy normalization of constants (Reprise)
Continuation of #67890 by @skinny121.
Initial implementation of #60471 for constants.
Perform normalization/evaluation of constants lazily, which is known as lazy normalization. Lazy normalization is only enabled when using `#![feature(lazy_normalization_consts)]`, by default constants are still evaluated eagerly as there are currently.
Lazy normalization of constants is achieved with a new ConstEquate predicate which type inferences uses to delay checking whether constants are equal to each other until later, avoiding cycle errors.
Note this doesn't allow the use of generics within repeat count expressions as that is still evaluated during conversion to mir. There are also quite a few other known problems with lazy normalization which will be fixed in future PRs.
r? @nikomatsakis
fixes#71922, fixes#71986
Support coercion between (FnDef | Closure) and (FnDef | Closure)
Fixes#46742, fixes#48109
Inject `Closure` into the `FnDef x FnDef` coercion special case, which makes coercion of `(FnDef | Closure) x (FnDef | Closure)` possible, where closures should be **non-capturing**.
This commit updates the code generation for `#![no_builtins]` to always
produce object files instead of conditionally respecting
`-Clinker-plugin-lto` and sometimes producing bitcode. This is intended
to address rust-lang/cargo#8239.
The issue at hand here is that Cargo has tried to get "smarter" about
codegen in whole crate graph scenarios. When LTO is enabled it attempts
to avoid codegen on as many crates as possible, opting to pass
`-Clinker-plugin-lto` where it can to only generate bitcode. When this
is combined with `-Zbuild-std`, however, it means that
`compiler-builtins` only generates LLVM bitcode instead of object files.
Rustc's own LTO passes then explicitly skip `compiler-builtins` (because
it wouldn't work anyway) which means that LLVM bitcode gets sent to the
linker, which chokes most of the time.
The fix in this PR is to not actually respect `-Clinker-plugin-lto` for
`#![no_builtins]` crates. These crates, even if slurped up by the linker
rather than rustc, will not work with LTO. They define symbols which are
only referenced as part of codegen, so LTO's aggressive internalization
would trivially remove the symbols only to have the linker realize later
that the symbol is undefined. Since pure-bitcode never makes sense for
these libraries, the `-Clinker-plugin-lto` flag is silently ignored.
abort_internal is safe
`sys::abort_internal` is stably exposed as a safe function. Forward that assumption "inwards" to the `sys` module by making the function itself safe, too.
This corresponds to what https://github.com/rust-lang/rust/pull/72204 did for the intrinsic. We should probably wait until that lands because some of the intrinsic calls in this PR might then need adjustments.
use the new interface to initialize conditional variables
HermitCore introduce a new interface to intialize conditional variables.
Consequently, minor changes are required to support this interface.
make offset must_use
https://djugei.github.io/bad-at-unsafe/ describes an error a user had when trying to use offset:
> At first I just assumed that the .add() and .offset() methods on pointers would mutate the pointer. They do not. Instead they return a new pointer, which gets dropped silently if you don't use it. Unlike for example Result, which is must_use annotated.
This PR only adds `offset`, because I wanted to float the idea; I'm imagining that there's more than just `add` and `offset` that could use this. I am also very open to re-wording the warning.
r? @rust-lang/libs
Currently, if you repeatedly push to an empty vector, the capacity
growth sequence is 0, 1, 2, 4, 8, 16, etc. This commit changes the
relevant code (the "amortized" growth strategy) to skip 1 and 2 in most
cases, instead using 0, 4, 8, 16, etc. (You can still get a capacity of
1 or 2 using the "exact" growth strategy, e.g. via `reserve_exact()`.)
This idea (along with the phrase "tiny Vecs are dumb") comes from the
"doubling" growth strategy that was removed from `RawVec` in #72013.
That strategy was barely ever used -- only when a `VecDeque` was grown,
oddly enough -- which is why it was removed in #72013.
(Fun fact: until just a few days ago, I thought the "doubling" strategy
was used for repeated push case. In other words, this commit makes
`Vec`s behave the way I always thought they behaved.)
This change reduces the number of allocations done by rustc itself by
10% or more. It speeds up rustc, and will also speed up any other Rust
program that uses `Vec`s a lot.
move borrow-of-packed-field unsafety check out of loop
Looks like during the place refactoring, this code somehow got into this loop even though it does not actually depend on the loop variables.
Fixes#66898
When we are unable to resolve a reference to `self`, we current assume
that the containing function doesn't have a `self` parameter, and
emit an error message accordingly.
However, if the reference to `self` was created by a macro invocation,
then resolution will correctly fail, due to hygiene. In this case, we
don't want to tell the user that the containing fuction doesn't have a
'self' paramter if it actually has one.
This PR checks for the precense of a 'self' parameter, and adjusts the
error message we emit accordingly.
TODO: The exact error message we emit could probably be improved. Should
we explicitly mention hygiene?
