TLDR: Was looking at some performance traces and saw that we need to cache the
result of this value.
----
Specifically, I noticed that we were spending a lot of time computing this
operation. When I looked at the code I saw that we already had a cache along the
relevant code paths... but the cache was from equivalence class representative
-> state. Before we hit that cache, we were performing the work to map the value
to the equivalence class representative... so the work to perform the relevant
lookup from value -> state (which goes through the equivalence class
representative) was not just a hash table lookup. This operation makes it
cheaper by making it two cache lookups.
It may be possible to make this cheaper by redoing the actual mapping of
information so that we can go straight from value to state. I think it would be
slightly different since we would probably need to represent the state in a
separate array and map with indices... which is really just a more efficient
hash table. We could also use malloc/etc but lets not even talk about that.
rdar://139520959
This makes it so that one does not need to deal with the differences in text in
between the task isolated case and the actor isolated case. This is done by
swallowing the entire part of this message in one method rather than having the
caller do the work.
This is going to let me just pass through the error struct to the diagnostic
rather than having the CRTP and then constructing an info object per CRTP.
Currently, to make it easier to refactor, I changed the code in
TransferNonSendable to just take in the new error and call the current CRTP
routines. In the next commit, I am going to refactor TransferNonSendable.cpp
itself. This just makes it easier to test that I did not break anything.
The optimization replaces a `load [copy]` with a `load_borrow` if possible.
```
%1 = load [copy] %0
// no writes to %0
destroy_value %1
```
->
```
%1 = load_borrow %0
// no writes to %0
end_borrow %1
```
The new implementation uses alias-analysis (instead of a simple def-use walk), which is much more powerful.
rdar://115315849
In Embedded Swift, witness method lookup is done from specialized witness tables.
For this to work, the type of witness_method must be specialized as well.
Otherwise the method call would be done with wrong parameter conventions (indirect instead of direct).
As the optimizer uses more and more AST stuff, it's now time to create an "AST" module.
Initially it defines following AST datastructures:
* declarations: `Decl` + derived classes
* `Conformance`
* `SubstitutionMap`
* `Type` and `CanonicalType`
Some of those were already defined in the SIL module and are now moved to the AST module.
This change also cleans up a few things:
* proper definition of `NominalTypeDecl`-related APIs in `SIL.Type`
* rename `ProtocolConformance` to `Conformance`
* use `AST.Type`/`AST.CanonicalType` instead of `BridgedASTType` in SIL and the Optimizer
MandatoryPerformanceOptimizations already did most of the vtable specialization work.
So it makes sense to remove the VTableSpecializerPass completely and do everything in MandatoryPerformanceOptimizations.
* add missing APIs
* bridge the entries as values and not as pointers
* add lookup functions in `Context`
* make WitnessTable.Entry.Kind enum cases lower case
The reason why is that we want to distinguish inbetween SILFunction's that are
marked as unspecified by SILGen and those that are parsed from textual SIL that
do not have any specified isolation. This will make it easier to write nice
FileCheck tests against SILGen output on what is the inferred isolation for
various items.
NFCI.
`SWIFT_IMPORT_UNSAFE` is an escape hatch that can be used to make the Swift compiler ignore its usual safety heuristics for C++ types.
`BridgedOwnedString` fits into the definition of a self-contained C++ type in Swift: it manages the lifetimes of its own fields.
This removes the usages of `SWIFT_IMPORT_UNSAFE` for C++ functions that return `BridgedOwnedString`, and annotates `BridgedOwnedString` as a self-contained type.
This only occurs specifically for async nonisolated functions with an isolated
parameter that passes a disconnected value in its body off to a nonisolated
async function as a sending parameter.
rdar://134409359
The old analysis pass doesn't take into account profile data, nor does
it consider post-dominance. It primarily dealt with _fastPath/_slowPath.
A block that is dominated by a cold block is itself cold. That's true
whether it's forwards or backwards dominance.
We can also consider a call to any `Never` returning function as a
cold-exit, though the block(s) leading up to that call may be executed
frequently because of concurrency. For now, I'm ignoring the concurrency
case and assuming it's cold. To make use of this "no return" prediction,
use the `-enable-noreturn-prediction` flag, which is currently off by
default.
If there is no read from an indirect argument, this argument has to be dropped.
At the call site the store to the argument's memory location could have been removed (based on the callee's memory effects).
Therefore, converting such an unused indirect argument to a direct argument, would load an uninitialized value at the call site.
This would lead to verifier errors and in worst case to a miscompile because IRGen can implicitly use dead arguments, e.g. for getting the type of a class reference.
Instead of adding a "flag" (`m` in `...Tgm5`) make it more generic to allow to drop any unused argument.
Add all dropped arguments with a `t<n-1>` (where `<n-1>` is empty for n === 0). For example `...Ttt2g5`.
These do not specifically have to do with PartitionUtils... they are really
logging options for the whole infrastructure, so it makes sense to have them in
the a different file.
