The issue of passing around SIMD types as values between functions has
seen [quite a lot] of [discussion], and although we thought [we fixed
it][quite a lot] it [wasn't]! This PR is a change to rustc to, again,
try to fix this issue.
The fundamental problem here remains the same, if a SIMD vector argument
is passed by-value in LLVM's function type, then if the caller and
callee disagree on target features a miscompile happens. We solve this
by never passing SIMD vectors by-value, but LLVM will still thwart us
with its argument promotion pass to promote by-ref SIMD arguments to
by-val SIMD arguments.
This commit is an attempt to thwart LLVM thwarting us. We, just before
codegen, will take yet another look at the LLVM module and demote any
by-value SIMD arguments we see. This is a very manual attempt by us to
ensure the codegen for a module keeps working, and it unfortunately is
likely producing suboptimal code, even in release mode. The saving grace
for this, in theory, is that if SIMD types are passed by-value across
a boundary in release mode it's pretty unlikely to be performance
sensitive (as it's already doing a load/store, and otherwise
perf-sensitive bits should be inlined).
The implementation here is basically a big wad of C++. It was largely
copied from LLVM's own argument promotion pass, only doing the reverse.
In local testing this...
Closes#50154Closes#52636Closes#54583Closes#55059
[quite a lot]: https://github.com/rust-lang/rust/pull/47743
[discussion]: https://github.com/rust-lang/rust/issues/44367
[wasn't]: https://github.com/rust-lang/rust/issues/50154
Allow explicit matches on ! without warning
It's now possible to explicitly match on `!` without an unreachable code warning. This seems desirable as promoting explicitness.
Fixes https://github.com/rust-lang/rust/issues/55116.
Prefer unwrap_or_else to unwrap_or in case of function calls/allocations
The contents of `unwrap_or` are evaluated eagerly, so it's not a good pick in case of function calls and allocations. This PR also changes a few `unwrap_or`s with `unwrap_or_default`.
An added bonus is that in some cases this change also reveals if the object it's called on is an `Option` or a `Result` (based on whether the closure takes an argument).
handle underscore bounds in unexpected places
Per the discussion on #54902, I made it a hard error to use lifetime bounds in various places where they used to be permitted:
- `where Foo: Bar<'_>` for example
I also moved error reporting to HIR lowering and added `Error` variants to let us suppress downstream errors that result.
I (imo) improved the error message wording to be clearer, as well.
In the process, I fixed the ICE in #52098.
Fixes#54902Fixes#52098
In order to output a path that could actually be imported (valid and
visible), we need to handle re-exports correctly.
For example, take `std::os::unix::process::CommandExt`, this trait is
actually defined at `std::sys::unix::ext::process::CommandExt` (at time
of writing).
`std::os::unix` rexports the contents of `std::sys::unix::ext`.
`std::sys` is private so the "true" path to `CommandExt` isn't accessible.
In this case, the visible parent map will look something like this:
(child) -> (parent)
`std::sys::unix::ext::process::CommandExt` -> `std::sys::unix::ext::process`
`std::sys::unix::ext::process` -> `std::sys::unix::ext`
`std::sys::unix::ext` -> `std::os`
This is correct, as the visible parent of `std::sys::unix::ext` is in fact
`std::os`.
When printing the path to `CommandExt` and looking at the current
segment that corresponds to `std::sys::unix::ext`, we would normally
print `ext` and then go to the parent - resulting in a mangled path like
`std::os::ext::process::CommandExt`.
Instead, we must detect that there was a re-export and instead print `unix`
(which is the name `std::sys::unix::ext` was re-exported as in `std::os`).
Also, avoid shadowing of the `ty` variable by giving the `cast_ty` and
`var_ty` variables different names. We want to get the user-provided
type from `cast_ty.hir_id`.
The issue of passing around SIMD types as values between functions has
seen [quite a lot] of [discussion], and although we thought [we fixed
it][quite a lot] it [wasn't]! This PR is a change to rustc to, again,
try to fix this issue.
The fundamental problem here remains the same, if a SIMD vector argument
is passed by-value in LLVM's function type, then if the caller and
callee disagree on target features a miscompile happens. We solve this
by never passing SIMD vectors by-value, but LLVM will still thwart us
with its argument promotion pass to promote by-ref SIMD arguments to
by-val SIMD arguments.
This commit is an attempt to thwart LLVM thwarting us. We, just before
codegen, will take yet another look at the LLVM module and demote any
by-value SIMD arguments we see. This is a very manual attempt by us to
ensure the codegen for a module keeps working, and it unfortunately is
likely producing suboptimal code, even in release mode. The saving grace
for this, in theory, is that if SIMD types are passed by-value across
a boundary in release mode it's pretty unlikely to be performance
sensitive (as it's already doing a load/store, and otherwise
perf-sensitive bits should be inlined).
The implementation here is basically a big wad of C++. It was largely
copied from LLVM's own argument promotion pass, only doing the reverse.
In local testing this...
Closes#50154Closes#52636Closes#54583Closes#55059
[quite a lot]: https://github.com/rust-lang/rust/pull/47743
[discussion]: https://github.com/rust-lang/rust/issues/44367
[wasn't]: https://github.com/rust-lang/rust/issues/50154
Add a `copysign` function to f32 and f64
This patch adds a `copysign` function to the float primitive types. It is an exceptionally useful function for writing efficient numeric code, as it often avoids branches, is auto-vectorizable, and there are efficient intrinsics for most platforms.
I think this might work as-is, as the relevant `copysign` intrinsic is already used internally for the implementation of `signum`. It's possible that an implementation might be needed in japaric/libm for portability across all platforms, in which case I'll do that also.
Part of the work towards #55107
