This is necessary because we need to model its stack-allocation
behavior, although I'm not yet doing that in this patch because
StackNesting first needs to be taught to not try to move the
deallocation.
I'm not convinced that `async let` *should* be doing a stack allocation,
but it undoubtedly *is* doing a stack allocation, and until we have an
alternative to that, we will need to model it properly.
This should make it easier to add new builtins by "following the warnings" and
prevent us from not handling a builtin in IRGen.
When I did this, I discovered that if I did this naively, we would have
AddressOf show up twice in the switch. This turned out to be because:
1. AddressOf is a SIL builtin that semantically is expected to only result in
SIL being emitted instead of having a builtin "addressof" be emitted.
2. For what ever reason, we actually had code in IRGen to emit an AddressOf
BuiltinInst if we saw it (which we never should have)... but also later code
asserted that we would never see it b/c it is a "SIL only builtin".
3. When I converted the if statements to be case statements, helpfully the
compiler told me I had a duplicate case. After investigation, I found the above
meaning that I was able to just delete the IRGen handling.
So now we properly handle AddressOf by asserting. As an additional tactic to
make "SIL only builtins" even more explicit, I added code to the SIL verifier
that validates we never see a builtin inst that is a "SIL only builtin" and
added some comments to Builtins.def that elaborate on this.
So far we only supported this for `load` and `call` arguments. Now we use `llvm.assume` for other kind of argument values.
This helps optimizing bounds checking for `Span`.
Ideally we'd be able to use the llvm interleave2 and deinterleave2
intrinsics instead of adding these, but deinterleave currently isn't
available from Swift, and even if you hack that in, the codegen from
LLVM is worse than what shufflevector produces for both x86 and arm. So
in the medium-term we'll use these builtins, and hope to remove them in
favor of [de]interleave2 at some future point.
It is like `zeroInitializer`, but does not actually initialize the memory.
It only indicates to mandatory passes that the memory is going to be initialized.
This is a value operation that can work just fine on lowered types,
so there's no need to carry along a formal type. Make the value/address
duality clearer, and enforce it in the verifier.
Although it's not used anymore we still have to support it to be able to read old Swift.interface files which still contain the builtin.
rdar://144781646
Although I don't plan to bring over new assertions wholesale
into the current qualification branch, it's entirely possible
that various minor changes in main will use the new assertions;
having this basic support in the release branch will simplify that.
(This is why I'm adding the includes as a separate pass from
rewriting the individual assertions)
* Implement Builtin.freeze for integer and integer-vector types.
https://llvm.org/docs/LangRef.html#freeze-instruction
> If the argument is undef or poison, ‘freeze’ returns an arbitrary, but fixed, value of type ‘ty’. Otherwise, this instruction is a no-op and returns the input argument. All uses of a value returned by the same ‘freeze’ instruction are guaranteed to always observe the same value, while different ‘freeze’ instructions may yield different values.
It's most importation for integer and integer-vector types because floating-point results are generally not poison (except in the case of conversion from poison integer values).
However, we might want to implement this for other types as well in the future.
* Make builtin.freeze TrivialUse
Also fix filecheck patterns for its test to work with asserts build.
The copy operator has been implemented and doesn't use it. Remove
`Builtin.copy` and `_copy` as much as currently possible.
Source compatibility requires that `_copy` remain in the stdlib. It is
deprecated here and just uses the copy operator.
Handling old swiftinterfaces requires that `Builtin.copy` be defined.
Redefine it here as a passthrough--SILGen machinery will produce the
necessary copy_addr.
rdar://127502242
Call `swift_clearSensitive` after destroying or taking "sensitive" struct types.
Also, support calling C-functions with "sensitive" parameters or return values. In SIL, sensitive types are address-only and so are sensitive parameters/return values.
Though, (small) sensitive C-structs are passed directly to/from C-functions. We need re-abstract such parameter and return values for C-functions.
We add the `memory(argmem: readwrite)` attribute to swift_task_create,
which means that the call is only allowed to read or write "pointer
operands". LLVM is smart enough to look through obvious ptrtoint
casts, but not to look through integer selects and so on, which is what
we produce when there's an opaque optional operand that feeds into the
builtin. This was causing miscompiles under optimization when using
`@isolated(any)` function types for task creation, since we're not yet
clever enough to fold the function_extract_isolation for a known function
(and of course it's not necessarily a known function anyway).
LLVM is presumably moving towards `std::string_view` -
`StringRef::startswith` is deprecated on tip. `SmallString::startswith`
was just renamed there (maybe with some small deprecation inbetween, but
if so, we've missed it).
The `SmallString::startswith` references were moved to
`.str().starts_with()`, rather than adding the `starts_with` on
`stable/20230725` as we only had a few of them. Open to switching that
over if anyone feels strongly though.
We've been building up this exponential explosion of task-creation
builtins because it's not currently possible to overload builtins.
As long as all of the operands are scalar, though, it's pretty easy
to peephole optional injections in IRGen, which means we can at
least just use a single builtin in SIL and then break it apart in
IRGen to decide which options to set.
I also eliminated the metadata argument, which can easily be recreated
from the substitutions. I also added proper verification for the builtin,
which required (1) getting `@Sendable` right more consistently and (2)
updating a bunch of tests checking for things that are not actually
valid, like passing a function that returns an Int directly.
Concurrency runtime expects discarding task operation entrypoint
function not to have result type, but the current SILGen
implementation generates reabstraction thunk to convert `() -> Void`
to `() -> T` for the operation function.
Since the `T` is always `Void` for DiscardingTG, the mismatch of result
type expectation does not cause any problem on most platforms, but the
signature mismatch causes a problem on WebAssembly.
This patch introduces new builtin operations for creating discarding
task, which always takes `() -> Void` as the operation function type.
