This has a number of advantages compared to creating a copy in memory
and passing a pointer. The obvious one is that we don't have to put the
data into memory but can keep it in registers. Since we're currently
passing a pointer anyway (instead of using e.g. a known offset on the
stack, which is what the `byval` attribute would achieve), we only use a
single additional register for each fat pointer, but save at least two
pointers worth of stack in exchange (sometimes more because more than
one copy gets eliminated). On archs that pass arguments on the stack, we
save a pointer worth of stack even without considering the omitted
copies.
Additionally, LLVM can optimize the code a lot better, to a large degree
due to the fact that lots of copies are gone or can be optimized away.
Additionally, we can now emit attributes like nonnull on the data and/or
vtable pointers contained in the fat pointer, potentially allowing for
even more optimizations.
This results in LLVM passes being about 3-7% faster (depending on the
crate), and the resulting code is also a few percent smaller, for
example:
text data filename
5671479 3941461 before/librustc-d8ace771.so
5447663 3905745 after/librustc-d8ace771.so
1944425 2394024 before/libstd-d8ace771.so
1896769 2387610 after/libstd-d8ace771.so
I had to remove a call in the backtrace-debuginfo test, because LLVM can
now merge the tails of some blocks when optimizations are turned on,
which can't correctly preserve line info.
Fixes#22924
Cc #22891 (at least for fat pointers the code is good now)
Loading from and storing to small aggregates happens by casting the
aggregate pointer to an appropriately sized integer pointer to avoid
the usage of first class aggregates which would lead to less optimized
code.
But this means that, for example, a tuple of type (i16, i16) will be
loading through an i32 pointer and because we currently don't provide
alignment information LLVM assumes that the load should use the ABI
alignment for i32 which would usually be 4 byte alignment. But the
alignment requirement for the (i16, i16) tuple will usually be just 2
bytes, so we're overestimating alignment, which invokes undefined
behaviour.
Therefore we must emit appropriate alignment information for
stores/loads through such casted pointers.
Fixes#23431
We provide tools to tell what exact symbols to emit for any fn or static, but
don’t quite check if that won’t cause any issues later on. Some of the issues
include LLVM mangling our names again and our names pointing to wrong locations,
us generating dumb foreign call wrappers, linker errors, extern functions
resolving to different symbols altogether (extern {fn fail();} fail(); in some
cases calling fail1()), etc.
Before the commit we had a function called note_unique_llvm_symbol, so it is
clear somebody was aware of the issue at some point, but the function was barely
used, mostly in irrelevant locations.
Along with working on it I took liberty to start refactoring trans/base into
a few smaller modules. The refactoring is incomplete and I hope I will find some
motivation to carry on with it.
This is possibly a [breaking-change] because it makes dumbly written code
properly invalid.
When this attribute is applied to a function, its return value gets the
noalias attribute, which is how you tell LLVM that the function returns
a "new" pointer that doesn't alias anything accessible to the caller,
i.e. it acts like a memory allocator.
Plain malloc doesn't need this attribute because LLVM already knows
about malloc and adds the attribute itself.
This commit is an implementation of [RFC 592][r592] and [RFC 840][r840]. These
two RFCs tweak the behavior of `CString` and add a new `CStr` unsized slice type
to the module.
[r592]: https://github.com/rust-lang/rfcs/blob/master/text/0592-c-str-deref.md
[r840]: https://github.com/rust-lang/rfcs/blob/master/text/0840-no-panic-in-c-string.md
The new `CStr` type is only constructable via two methods:
1. By `deref`'ing from a `CString`
2. Unsafely via `CStr::from_ptr`
The purpose of `CStr` is to be an unsized type which is a thin pointer to a
`libc::c_char` (currently it is a fat pointer slice due to implementation
limitations). Strings from C can be safely represented with a `CStr` and an
appropriate lifetime as well. Consumers of `&CString` should now consume `&CStr`
instead to allow producers to pass in C-originating strings instead of just
Rust-allocated strings.
A new constructor was added to `CString`, `new`, which takes `T: IntoBytes`
instead of separate `from_slice` and `from_vec` methods (both have been
deprecated in favor of `new`). The `new` method returns a `Result` instead of
panicking. The error variant contains the relevant information about where the
error happened and bytes (if present). Conversions are provided to the
`io::Error` and `old_io::IoError` types via the `FromError` trait which
translate to `InvalidInput`.
This is a breaking change due to the modification of existing `#[unstable]` APIs
and new deprecation, and more detailed information can be found in the two RFCs.
Notable breakage includes:
* All construction of `CString` now needs to use `new` and handle the outgoing
`Result`.
* Usage of `CString` as a byte slice now explicitly needs a `.as_bytes()` call.
* The `as_slice*` methods have been removed in favor of just having the
`as_bytes*` methods.
Closes#22469Closes#22470
[breaking-change]
As the function comment already says, the types generated in the
foreign_signture function don't necessarily match the types used for a
corresponding rust function. Therefore we can't just use these types to
guide the translation of the wrapper function that bridges between the
external ABI and the rust ABI. Instead, we can query LLVM about the
types used in the rust function and use those to generate an appropriate
wrapper.
Fixes#21454
I don't know if this handling of SIMD types is correct for the C ABI on
all platforms, so lets add an even finer feature gate than just the
`simd` one.
The `simd` one can be used with (relatively) little risk of complete
nonsense, the reason for it is that it is likely that things will
change. Using the types in FFI with an incorrect ABI will at best give
absolute nonsense results, but possibly cause serious breakage too, so
this is a step up in badness, hence a new feature gate.
Currently, small aggregates are passed to functions as immediate values
as is. This has two consequences.
One is that aggregates are passed component-wise by LLVM, so e.g. a
struct containing four u8 values (e.g. an RGBA struct) will be passed as
four individual values.
The other is that LLVM isn't very good at optimizing loads/stores of
first class attributes. What clang does is converting the aggregate to
an appropriately sized integer type (e.g. i32 for the four u8 values),
and using that for the function argument. This allows LLVM to create
code that is a lot better.
Fixes#20450#20149#16506#13927
This commit is an implementation of [RFC 494][rfc] which removes the entire
`std::c_vec` module and redesigns the `std::c_str` module as `std::ffi`.
[rfc]: https://github.com/rust-lang/rfcs/blob/master/text/0494-c_str-and-c_vec-stability.md
The interface of the new `CString` is outlined in the linked RFC, the primary
changes being:
* The `ToCStr` trait is gone, meaning the `with_c_str` and `to_c_str` methods
are now gone. These two methods are replaced with a `CString::from_slice`
method.
* The `CString` type is now just a wrapper around `Vec<u8>` with a static
guarantee that there is a trailing nul byte with no internal nul bytes. This
means that `CString` now implements `Deref<Target = [c_char]>`, which is where
it gains most of its methods from. A few helper methods are added to acquire a
slice of `u8` instead of `c_char`, as well as including a slice with the
trailing nul byte if necessary.
* All usage of non-owned `CString` values is now done via two functions inside
of `std::ffi`, called `c_str_to_bytes` and `c_str_to_bytes_with_nul`. These
functions are now the one method used to convert a `*const c_char` to a Rust
slice of `u8`.
Many more details, including newly deprecated methods, can be found linked in
the RFC. This is a:
[breaking-change]
Closes#20444
This commit is an implementation of [RFC 503][rfc] which is a stabilization
story for the prelude. Most of the RFC was directly applied, removing reexports.
Some reexports are kept around, however:
* `range` remains until range syntax has landed to reduce churn.
* `Path` and `GenericPath` remain until path reform lands. This is done to
prevent many imports of `GenericPath` which will soon be removed.
* All `io` traits remain until I/O reform lands so imports can be rewritten all
at once to `std::io::prelude::*`.
This is a breaking change because many prelude reexports have been removed, and
the RFC can be consulted for the exact list of removed reexports, as well as to
find the locations of where to import them.
[rfc]: https://github.com/rust-lang/rfcs/blob/master/text/0503-prelude-stabilization.md
[breaking-change]
Closes#20068
