Large tuples of values (e.g char[32]) can be passed directly at the abi
boundry but expand to a big explosion of values.
Peephole this explosion at argument passing and return value passing
points to avoid code size growth associated with the explosion.
In GenCall, fix the IR gen for C++ method calls as under MSVC as the
calling conventions for free functions and C++ methods can be
different. This also fixes the missing inreg (on sret arguments)
issues on Windows ARM64. Also refactor to use CGFunctionInfo
returnInfo isSretAfterThis to detect when to reorder the sret and the
this arguments under MSVC.
In ClagImporter, don't drop the return type for the compound
assignment operators such as operator+= when the return value is a
reference so that the CGFunctionInfo will be correctly indicate an
indirect return for the compound assignment operators.
When its operand has coroutine kind `yield_once_2`, a `begin_apply`
instruction produces an additional value representing the storage
allocated by the callee. This storage must be deallocated by a
`dealloc_stack` on every path out of the function. Like any other stack
allocation, it must obey stack discipline.
Add a setting to IRGenOptions and key off of it to emit yield_once_2
coroutines using either (1) the same code-path as yield_once coroutines
or (2) a new, not-yet implemented code-path.
Add flags to set the value in both directions. During bringup, by
default, use the existing caller-allocated ABI.
Unless there is some meaningful code on the same line as an await call (e.g.
other parts of an expression), there should be no additional line entries
associated with that line on the continuation funclet.
This patch changes codegen to avoid emitting debug location for the "prologue" code
at the start of a funclet, instead of simply copying the debug location of the
split point.
On Windows/AArch64, a different register is used between when an
arugment is both inreg and sret (X0 or X1) and when it is just sret
(X8) as the following comment indicates:
46fe36a429/llvm/lib/Target/AArch64/AArch64CallingConvention.td (L42)
```
// In AAPCS, an SRet is passed in X8, not X0 like a normal pointer parameter.
// However, on windows, in some circumstances, the SRet is passed in X0 or X1
// instead. The presence of the inreg attribute indicates that SRet is
// passed in the alternative register (X0 or X1), not X8:
// - X0 for non-instance methods.
// - X1 for instance methods.
// The "sret" attribute identifies indirect returns.
// The "inreg" attribute identifies non-aggregate types.
// The position of the "sret" attribute identifies instance/non-instance
// methods.
// "sret" on argument 0 means non-instance methods.
// "sret" on argument 1 means instance methods.
CCIfInReg<CCIfType<[i64],
CCIfSRet<CCIfType<[i64], CCAssignToReg<[X0, X1]>>>>>,
CCIfSRet<CCIfType<[i64], CCAssignToReg<[X8]>>>,
```
So missing/dropping inreg can cause a codegen bug.
This is a partial fix for #74866
Adds sections `__TEXT,__swift_as_entry`, and `__TEXT,__swift_as_ret` that
contain relative pointers to async functlets modelling async function entries,
and function returns, respectively.
Emission of the sections can be trigger with the frontend option
`-Xfrontend -enable-async-frame-push-pop-metadata`.
This is done by:
* IRGen adding a `async_entry` function attribute to async functions.
* LLVM's coroutine splitting identifying continuation funclets that
model the return from an async function call by adding the function
attribute `async_ret`. (see #llvm-project/pull/9204)
* An LLVM pass that keys off these two function attribute and emits the
metadata into the above mention sections.
rdar://134460666
This fixes a runtime crash that occurs when a pointer to a foreign
reference type is passed to objc_retainAutoreleasedReturnValue.
This reverts 335cec0e8d.
rdar://117353222
