Teach rustc about DW_AT_noreturn and a few more DIFlags
We achieve two small things with this PR:
1. We provide definitions for a few additional llvm debuginfo flags
1. We _use_ one of these new flags, `FlagNoReturn`, and add it to debuginfo for functions with the never return type (`!`).
This commit changes the ABI of SIMD types in the "Rust" ABI to unconditionally
be passed via pointers instead of being passed as immediates. This should fix a
longstanding issue, #44367, where SIMD-using programs ended up showing very odd
behavior at runtime because the ABI between functions was mismatched.
As a bit of a recap, this is sort of an LLVM bug and sort of an LLVM feature
(today's behavior). LLVM will generate code for a function solely looking at the
function it's generating, including calls to other functions. Let's then say
you've got something that looks like:
```llvm
define void @foo() { ; no target features enabled
call void @bar(<i64 x 4> zeroinitializer)
ret void
}
define void @bar(<i64 x 4>) #0 { ; enables the AVX feature
...
}
```
LLVM will codegen the call to `bar` *without* using AVX registers becauase `foo`
doesn't have access to these registers. Instead it's generated with emulation
that uses two 128-bit registers. The `bar` function, on the other hand, will
expect its argument in an AVX register (as it has AVX enabled). This means we've
got a codegen problem!
Comments on #44367 have some more contexutal information but the crux of the
issue is that if we want SIMD to work in general we'll need to ensure that
whenever a function calls another they ABI of the arguments being passed is in
agreement.
One possible solution to this would be to insert "shim functions" where whenever
a `target_feature` mismatch is detected the compiler inserts a shim function
where you pass arguments via memory to the shim and then the shim loads the
values and calls the target function (where the shim and the target have the
same target features enabled). This unfortunately is quite nontrivial to
implement in rustc today (especially when accounting for function pointers and
such).
This commit takes a different solution, *always* passing SIMD arguments through
memory instead of passing as immediates. This strategy solves the problem at the
LLVM layer because the ABI between two functions never uses SIMD registers. This
also shouldn't be a hit to performance because SIMD performance is thought to
often rely on inlining anyway, where a `call` instruction, even if using SIMD
registers, would be disastrous to performance regardless. LLVM should then be
more than capable of fixing all our memory usage to use registers instead after
enough inlining has been performed.
Note that there's a few caveats to this commit though:
* The "platform intrinsic" ABI is omitted from "always pass via memory". This
ABI is used to define intrinsics like `simd_shuffle4` where LLVM and rustc
need to have the arguments as an immediate.
* Additionally this commit does *not* fix the `extern` ("C") ABI. This means
that the bug in #44367 can still happen when using non-Rust-ABI functions. My
hope is that before stabilization we can ban and/or warn about SIMD types in
these functions (as AFAIK there's not much motivation to belong there anyway),
but I'll leave that for a later commit and if this is merged I'll file a
follow-up issue.
All in all this...
Closes#44367
Implement repr(transparent)
r? @eddyb for the functional changes. The bulk of the PR is error messages and docs, might be good to have a doc person look over those.
cc #43036
cc @nox
These are already removed in the normal optimization pipeline - so this
should slightly improve codegen performance, as these cleanup blocks are
known to hurt LLVM.
This un-regresses and is therefore a fix for #47442. However, the
reporter of that issue should try using `-C panic=abort` instead of
carefully avoiding panics.
It looks like many of these tests are already disabled on emscripten,
which also doesn't seem to support environment variables and subprocess
spawning. Just add a similar tag for CloudABI. While there, sort some of
the lists of operating systems alphabetically.
This commit adds compiler support for two basic operations needed for binding
SIMD on x86 platforms:
* First, a `nontemporal_store` intrinsic was added for the `_mm_stream_ps`, seen
in rust-lang-nursery/stdsimd#114. This was relatively straightforward and is
quite similar to the volatile store intrinsic.
* Next, and much more intrusively, a new type to the backend was added. The
`x86_mmx` type is used in LLVM for a 64-bit vector register and is used in
various intrinsics like `_mm_abs_pi8` as seen in rust-lang-nursery/stdsimd#74.
This new type was added as a new layout option as well as having support added
to the trans backend. The type is enabled with the `#[repr(x86_mmx)]`
attribute which is intended to just be an implementation detail of SIMD in
Rust.
I'm not 100% certain about how the `x86_mmx` type was added, so any extra eyes
or thoughts on that would be greatly appreciated!
Emit better debugging information for a trait object pointer. In
particular, now:
* The fields are explicitly represented in the DWARF;
* DWARF for the vtable itself is emitted; and
* The DWARF for the vtable's type has a DW_AT_containing_type which
points to the concrete type for which the vtable was emitted. This is
a small DWARF extension, that allows debuggers to determine the real
type of the object to which a trait object points.
I'll submit the gdb patch to take advantage of this new debuginfo once
this lands.
The vtable type is not currently complete -- it doesn't include members
for the pointers it contains. This information was not needed for this
feature.
This addresses part 1 of #1563.
... rather than being gated by -Z saturating-float-casts.
There are several reasons for this:
1. Const eval already implements this behavior.
2. Unlike with float->int casts, this behavior is uncontroversially the
right behavior and it is not as performance critical. Thus there is no
particular need to make the bug fix for u128->f32 casts opt-in.
3. Having two orthogonal features under one flag is silly, and never
should have happened in the first place.
4. Benchmarking float->int casts with the -Z flag should not pick up
performance changes due to the u128->f32 casts (assuming there are any).
Fixes#41799
Saturating casts between integers and floats
Introduces a new flag, `-Z saturating-float-casts`, which makes code generation for int->float and float->int casts safe (`undef`-free), implementing [the saturating semantics laid out by](https://github.com/rust-lang/rust/issues/10184#issuecomment-299229143) @jorendorff for float->int casts and overflowing to infinity for `u128::MAX` -> `f32`.
Constant evaluation in trans was changed to behave like HIR const eval already did, i.e., saturate for u128->f32 and report an error for problematic float->int casts.
Many thanks to @eddyb, whose APFloat port simplified many parts of this patch, and made HIR constant evaluation recognize dangerous float casts as mentioned above.
Also thanks to @ActuallyaDeviloper whose branchless implementation served as inspiration for this implementation.
cc #10184#41799fixes#45134
This affects regular code generation as well as constant evaluation in trans,
but not the HIR constant evaluator because that one returns an error for
overflowing casts and NaN-to-int casts. That error is conservatively
correct and we should be careful to not accept more code in constant
expressions.
The changes to code generation are guarded by a new -Z flag, to be able
to evaluate the performance impact. The trans constant evaluation changes
are unconditional because they have no run time impact and don't affect
type checking either.
Initially MIR differentiated between arguments and locals, which
introduced a need to add extra copies assigning the argument to a
local, even for simple bindings. This differentiation no longer exists,
but we're still creating those copies, bloating the MIR and LLVM IR we
emit.
Additionally, the current approach means that we create debug info for
both the incoming argument (marking it as an argument), and then
immediately shadow it a local that goes by the same name. This can be
confusing when using e.g. "info args" in gdb, or when e.g. a debugger
with a GUI displays the function arguments separately from the local
variables, especially when the binding is mutable, because the argument
doesn't change, while the local variable does.
The necessary changes were only in upstream LLVM in 4.0, but they were
for a while backported to Rust LLVM. Now that Rust LLVM is also 4.0, we
can make the test conditional here more accurate.
This informs LLVM that functions can't unwind, which while it should typically
have already been inferred when necessary or otherwise not impact codegen is
apparently needed on targets like ARM to avoid references to unnecessary
symbols.
Closes#44992
