This commit updates the LLVM submodule in use to the current HEAD of the LLVM
repository. This is primarily being done to start picking up unwinding support
for MSVC, which is currently unimplemented in the revision of LLVM we are using.
Along the way a few changes had to be made:
* As usual, lots of C++ debuginfo bindings in LLVM changed, so there were some
significant changes to our RustWrapper.cpp
* As usual, some pass management changed in LLVM, so clang was re-scrutinized to
ensure that we're doing the same thing as clang.
* Some optimization options are now passed directly into the
`PassManagerBuilder` instead of through CLI switches to LLVM.
* The `NoFramePointerElim` option was removed from LLVM, favoring instead the
`no-frame-pointer-elim` function attribute instead.
Additionally, LLVM has picked up some new optimizations which required fixing an
existing soundness hole in the IR we generate. It appears that the current LLVM
we use does not expose this hole. When an enum is moved, the previous slot in
memory is overwritten with a bit pattern corresponding to "dropped". When the
drop glue for this slot is run, however, the switch on the discriminant can
often start executing the `unreachable` block of the switch due to the
discriminant now being outside the normal range. This was patched over locally
for now by having the `unreachable` block just change to a `ret void`.
* Removes `RustJITMemoryManager` from public API.
This was really sort of an implementation detail to begin with.
* `__morestack` is linked to C++ wrapper code and this pointer
is used when resolving the symbol for `ExecutionEngine` code.
* `__morestack_addr` is also resolved for `ExecutionEngine` code.
This function is sometimes referenced in LLVM-generated code,
but was not able to be resolved on Mac OS systems.
* Added Windows support to `ExecutionEngine` API.
* Added a test for basic `ExecutionEngine` functionality.
As per RFC#520 the syntax for arrays has changed,
this changes the remaining comments to reflect
the new syntax.
I checked for existing occurences of this with the following command:
`ag "\[., \.\..\]"` which by now should only return a single occurence.
For a user following the path of reading Chapter 5: Syntax & Symantics
prior to Chapter 4: Learn Rust, this will be the first time they have
encountered executable tests inside documentation comments.
The test will fail because the `add_one` function is not defined in
the context of the doctest. This might not be the optimal place to
introduce and explain the `/// #` notation but I think it is important
that this snippet pass as a test when `rustdoc --test` is run against
it.
The doc indicates that you can replace 'before' with 'after' showing the use of try!. The two examples should be equivalent, but they are not.
In the File::create we were inducing a panic before in case of error, not propagating. It is important for newbies (like myself) to understand that try! propagates failures, while unwrap can induce a panic.
The other alternative is to make the 'before' File::create also manually handle Err like the other calls. Either way it would be consistent.
As mentioned in #25893 the copy trait is not very well explained for beginners. There is no clear mention that all primitive types implement the copy trait and there are not a lot of examples.
With this change I try to make it more visible and understandable for new users.
I myself have struggled with this, see [my question on stackoverflow](http://stackoverflow.com/questions/30540419/why-are-booleans-copyable-even-though-the-documentation-doesnt-indicate-that). And I want to make it more transparent for others.
I filed issue #25893 but I thought that I could give it a shot myself to relieve some of the work from the devs :)
If it is not well written or there are some changes to be made before it can be merged, let me know.
Cheers,
Mathieu
* Removes `RustJITMemoryManager` from public API.
This was really sort of an implementation detail to begin with.
* `__morestack` is linked to C++ wrapper code and this pointer
is used when resolving the symbol for `ExecutionEngine` code.
* `__morestack_addr` is also resolved for `ExecutionEngine` code.
This function is sometimes referenced in LLVM-generated code,
but was not able to be resolved on Mac OS systems.
* Added Windows support to `ExecutionEngine` API.
* Added a test for basic `ExecutionEngine` functionality.
With the latter is provided by the `From` conversion trait, the former is now completely redundant. Their code is identical. Let’s deprecate now and plan to remove in the next cycle. (It’s `#[unstable]`.)
r? @alexcrichton
CC @nagisa
I had to use `impl<'a, V: Copy> Extend<(usize, &'a V)> for VecMap<V>` instead of `impl<'a, V: Copy> Extend<(&'a usize, &'a V)> for VecMap<V>` as that's what is needed for doing
```rust
let mut a = VecMap::new();
let b = VecMap::new();
b.insert(1, "foo");
a.extend(&b)
```
I can squash the commits after review.
r? @Gankro
Closes#19826
r? @nikomatsakis
There is still some work to do until parallel codegen is perfect - we don't pass make check due to missing the symbol for registering a macro. But we do bootstrap with codegen_units=4 and pass make check without codegen_units, so I think it should land.
For the curious, make rustc-stage2 -j8:
```
codegen_units=1:
real 9m18.074s
user 11m59.858s
sys 0m13.281s
codegen_units=4:
real 6m3.672s
user 13m5.474s
sys 0m12.146s
```
Which is a 33% speedup :-)
Based on the patch from Luca Bruno.
Instead of creating an empty C function in the rt, this version creates an shim
noop function using llvm. This function is declared as internal, and the
unsupported assume intrinsic and the shim gets completly removed by the
optimizer.
This commit introduce a third parameter for compatible_ifn!, as new
intrinsics are being added in recent LLVM releases and there is no
need to hardcode a specific case.
Signed-off-by: Luca Bruno <lucab@debian.org>
This makes the maximum edit distance of typo suggestions a function of the typo'd name's length. FWIW, clang uses this same hueristic, and I've found their suggestions to be better than rustc's. Without something like this, you end up with suggestions that aren't related at all when there are short variable names.
See also https://github.com/rust-lang/rust/issues/20028#issuecomment-109767159
