I put in a simple fixup pass (MarkUninitializedFixup) for staging purposes. I
don't expect it to be in tree long. I just did not feel comfortable fixing up in
1 commit all of the passes up to DI.
rdar://31521023
Today the ownership checker has two different behaviors on detecting an error:
1. In normal operating mode, it prints a message to stderr and asserts.
2. In testing mode, it prints a message to stderr and returns false.
This refactoring will allow me to add a third mode:
1. In optimization mode, it does not print a message and just returns false.
This will allow me to use the verifier in an early ARC optimization that I need
to prevent COW copies due to refactoring in SILGenPattern.
rdar://29870610
Introduce a new runtime entry point,
`swift_objc_swift3ImplicitObjCEntrypoint`, which is called from any
Objective-C method that was generated due to `@objc` inference rules
that were removed by SE-0160. Aside from being a central place where
users can set a breakpoint to catch when this occurs, this operation
provides logging capabilities that can be enabled by setting the
environment variable SWIFT_DEBUG_IMPLICIT_OBJC_ENTRYPOINT:
SWIFT_DEBUG_IMPLICIT_OBJC_ENTRYPOINT=0 (default): do not log
SWIFT_DEBUG_IMPLICIT_OBJC_ENTRYPOINT=1: log failed messages
SWIFT_DEBUG_IMPLICIT_OBJC_ENTRYPOINT=2: log failed messages with
backtrace
SWIFT_DEBUG_IMPLICIT_OBJC_ENTRYPOINT=3: log failed messages with
backtrace and abort the process.
The log messages look something like:
***Swift runtime: entrypoint -[t.MyClass foo] generated by
implicit @objc inference is deprecated and will be removed in
Swift 4
Previously often times when casting a value, we would just pass along the
cleanup of the uncasted value. With semantic SIL this is no longer correct since
the cleanup now needs to be on the cast result.
This caused problems for certain usages of Builtin.castToNativeObject(...) by
the stdlib. Specifically, the stdlib was using this on AnyObject values that
were not necessarily native. Since we were recreating the cleanup on the native
value, a swift native release was being used =><=.
In this commit I solve this problem by:
1. Adding an assert in Builtin.castToNativeObject(...) that ensures that any value
passed to Builtin.castToNativeObject() is known conservatively to use swift
native reference counting.
2. I changed all uses where we do not have a precondition of a native ref
counting type to use Builtin.castToUnknownObject(...).
3. I added a new Builtin called Builtin.unsafeCastToNativeObject(...) that does
not have the compile time check. I used this to rewrite callsites in the stdlib
where we know via preconditions that an AnyObject will dynamically always be
native.
rdar://29791263
...and IRGen it into a call to __tsan_write1 in compiler-rt. This is
preparatory work for a later patch that will add an experimental
option to treat Swift inout accesses as TSan writes.
This is the lifetime ending variant of fix_lifetime. It is a lie to the
ownership verifier that a value is being consumed along a path. Its intention is
to be used to allow for the static verification of ownership in deallocating
deinits which for compatibility with objective-c have weird ownership behavior.
See the commit merged with this commit for more information.
Previously, we would put a destroy_value directly on the value that we tried to
cast. Since checked_cast_br is consuming, this would cause the destroy_value on
the failure path to be flagged as a double consume.
This commit causes SILGen to emit the value consumed by checked_cast_br as an
@owned argument to the failure BB, allowing semantic arc rules to be respected.
As an additional benefit, I also upgraded the ownership_model_eliminator test to
use semantic sil verification.
One issue that did come up though is that I was unable to use the new code in
all locations in the compiler. Specifically, there is one location in
SILGenPattern that uses argument unforwarding. I am going to need to undo
argument unforwarding in SILGenPattern in order to completely eliminate the old
code path.
Rememebering that the verifier ensures that any edge that propagates ownership
along a cond_br can not be critical, we do this by sinking the use by the
cond_br into the destination blocks.
rdar://29791263
There are a few different use cases here:
1. In Raw SIL, no return folding may not have been run yet implying that a call
to a no-return function /can/ have arbitrary control flow after it (consider
mandatory inlined functions). We need to recognize that the region of code that
is strictly post dominated by the no-return function is "transitively
unreachable" and thus leaking is ok from that point. *Footnote 1*.
2. In Canonical and Raw SIL, we must recognize that unreachables and no-return
functions constitute places where we are allowed to leak.
rdar://29791263
----
*Footnote 1*: The reason why this is done is since we want to emit unreachable
code diagnostics when we run no-return folding. By leaving in the relevant code,
we have preserved all of the SILLocations on that code allowing us to create
really nice diagnostics.
This allows for an unchecked_enum_data to be either a consumed instruction or a
borrowed instruction. The reason why this makes sense in contrast to other value
projection operations like struct_extract and tuple_extract is that an enum
payload is essentially a tuple. This means that we are extracting the entire
value when we perform a struct_extract. So forwarding is viable from a semantic
perspective since if we destroy the payload, there is nothing left to destroy.
This contrasts with struct_extract and tuple_extract where we may have other
parts of the struct/tuple to destroy.
rdar://29791263
This caused a crasher when running the ownership verifier. I don't have a test
case right now, since it happened several weeks ago.
The bug can not happen again since I eliminated the nullptr default argument.
rdar://29791263
I implemented the mapping from SILValue -> ValueOwnershipKind for builtins a
long time ago, but did not implement the stub for what ValueOwnershipKind's can
be accepted by builtins. It is a pretty simple mapping (most things are
trivial).
The only interesting cases are:
1. ErrorInMain, UnexpectedError, WillThrow: These take in an @owned error
parameter, but are just normal uses of the value, not lifetime ending uses.
2. UnsafeGuaranteed. This is a bonafide actual +1 lifetime ending use. It is the
only one in the builtins.
rdar://29791263
The logic here is that a function argument can be viewed as a PHI node in the
program graph (i.e. cfg for each function + call graph). Just like with normal
PHI nodes, we need to treat enums passed into these PHI nodes as true sum types,
but at the same time, do not want to constrain optimization when we know that
the payloaded type is trivial.
rdar://29791263
Storing this separately is unnecessary since we already
serialize the enum element's interface type. Also, this
eliminates one of the few remaining cases where we serialize
archetypes during AST serialization.
