This is already an assumption in SILFunctionType:
bool hasIndirectResult() const {
return !getParameters().empty()
&& getParameters()[0].isIndirectResult();
}
SILParameterInfo getIndirectResult() const {
assert(hasIndirectResult());
return getParameters()[0];
}
This just bakes the invariant into the IR so it is clear when writing tests that
this is not an issue.
I also updated function_ref and partial_apply to have the same such checks. This
should handle the majority of the cases.
Swift SVN r22119
It is defaulted to nullptr. When it is set, we increment the SILFunction's
ref count to keep it alive.
It will be used in followon patches for globals that can be statically
initialized.
Swift SVN r21983
Now that we rely on type lowering to catch recursive value types, this optimization is no longer valid. Fixes rdar://problem/18232757 (crash suite #015).
Swift SVN r21967
Currently, the pass just calls a local version of that function. After OzU, I
will enable the full pass (which is currently disabled behind a flag).
Swift SVN r21894
The cache is needed to ensure we do not run into compile time problems once we
start looking through Phi Nodes.
The analysis is currently disabled and just returns
SILValue::stripRCIdentityPreservingOps. I am going to thread it through the rest
of the passes that use that call. Then I am going to hide
stripRCIdentityPreservingArgs. Finally post OzU, I am going to enable the pass.
rdar://18300069
Swift SVN r21891
Update SILGen to create SILGlobalVariable and SILGlobalAddrInst instead of
GlobalAddrInst. When we see a definition for a global variable, we create
the corrsponding SILGlobalVariable definition.
When creating SILGlobalVariable from a global VarDecl, we mangle the global
VarDecl in the same way as we mangle it at IRGen. The SILLinkage is also
set in the same way as we set it at IRGen.
At IRGen, we use the associated VarDecl for SILGlobalVariable if it exists,
to have better debugging information.
We set the initializer for SILGlobalVariable definition only.
We also handle SILGlobalAddrInst in various SILPasses, in the similar way
as we handle GlobalAddrInst.
rdar://15493694
Swift SVN r21887
We add two more fields to SILGlobalVariable: a VarDecl and a flag to see if this
is a declaration. VarDecl is mainly used for debugger support, it is also used
to check if the variable is weak imported.
We also modify the serializer to serialize the extra two fields.
Swift SVN r21883
This follows the model of dominance info and allows me to create reachability
methods on SILBasicBlock without creating dependencies from swiftSIL to
swiftSILAnalysis.
Swift SVN r21866
SILArgument::getIncomingValues() takes in an out array parameter and attempts to
gather up all values from the SILArguments parents predecessors whose value the
SILArgument could take on.
This will let me refactor the single predecessor handling code to also handle
multiple predecessors in a simple way.
Swift SVN r21864
This is necessary to be able to properly stash values with nontrivial lowerings, such as metatypes and functions, inside existential containers. Modify SILGen to lower values to the proper abstraction level before storing them in an existential container. Part of the fix for rdar://problem/18189508, though runtime problems still remain when trying to actually dynamicCast out a metatype from an Any container.
Swift SVN r21830
This fix should also extend to the general case of witnesses that return non-optional for optional-returning requirements, when/if we support that.
Swift SVN r21721
We want to be able to work around problems with non-failable
Objective-C initializers actually failing, which can happen when the
API audit data incorrectly marks an initializer as non-failable.
Swift SVN r21711
so add these verifier checks back. The rest of the changes I made are
still correct/useful, as they remove pointless unchecked_ref_cast's to
Builtin.UnknownObject, which generates cleaner SIL.
Swift SVN r21631
Originally I was using the notion of layout compatible to define the notion of a
ref count identity preserving operation. This is too restrictive of a definition
since layout compatible implies that the operation does not change the memory
location being pointed at. On the other hand, we just care that the ref counts
are the same.
rdar://18189329
Swift SVN r21613
Any single-refcounted-pointer type can be cast from a raw pointer. Also generalize unchecked_ref_cast to cast among arbitrary single-refcounted-pointer types. Fixes rdar://problem/18184883.
Swift SVN r21609
- Split getSelfTypeForDynamicLookup into two pieces, and generalize it
to work on non-loadable protocols.
- Change dynamic_method_branch to take its argument as a protocol of any
protocol type, instead of as something of UnownObject type.
- Teach emitForcedDynamicMemberRef to only do its peephole optimization for
@objc cases, since it is special behavior of objc_msgSend.
- enhance emitDynamicPartialApply & emitDynamicMemberRefExpr to emit the
proper project_existential instruction (not a _ref) when dealing with a
non-classbound protocol.
Change the verifier to allow DynamicMethodBranchInst to work on non-@objc
protocol members.
This eliminates a bunch of pointless unchecked_ref_cast's in the generated
SIL for existing code, but this got squashed at IRGen time anyway, so no
real change for anything that sema permits.
Swift SVN r21519
The AST printer can't handle non-type-checked decls, so for now just only
print SIL if we're in -primary-file mode. That doesn't result in parseable
SIL, but it's better than crashing because we tried to print something
irrelevant from another file.
Swift SVN r21500
It should definitely work with AnyObject, but it can also apply to any class existential by just shedding the witness tables. This lets us eliminate a bunch of unsafeBitCasts from the alimentary canals of the stdlib.
Swift SVN r21472
This will let the performance inliner inline a function even if the costs are too high.
This attribute is only a hint to the inliner.
If the inliner has other good reasons not to inline a function,
it will ignore this attribute. For example if it is a recursive function (which is
currently not supported by the inliner).
Note that setting the inline threshold to 0 does disable performance inlining at all and in
this case also the @inline(__always) has no effect.
Swift SVN r21452
I also cleaned up some code in InstructionIdentityComparer. We compare operands
before we hit the comparer, so we don't need to check operand equality there.
Swift SVN r21359
This matches the name for SwitchEnumInst::getCaseDestination() and includes the word
'unique' so that the name self documents.
I also removed a local function doing the same work in SimplifyCFG and changed
its user to use getUniqueCaseForDestination instead.
Swift SVN r21339
