When producing TypeInfo for a box, try to reuse instantiations for common type structures:
- any POD type with the same stride and alignment can share a fixed HeapLayout;
- any single-refcounted-pointer type can share a fixed HeapLayout with types that have the same ReferenceCounting;
- dynamically-sized types can share a runtime-based box implementation.
For the runtime implementation, use new to-be-implemented variants of allocBox/deallocBox that will produce polymorphically-projectable boxes using instantiated metadata.
Swift SVN r29612
Don't project every value witness from the metadata every time we need one; this wastes code size in a way LLVM can't really optimize since it doesn't know the metadata is immutable. The code size wins on the standard library are disappointingly small (stdlib only shrinks by 4KB), but this makes generic IR a lot more compact and easier to read.
Swift SVN r28095
As part of this, re-arrange the argument order so that
generic arguments come before the context, which comes
before the error result. Be more consistent about always
adding a context parameter on thick functions, even
when it's unused. Pull out the witness-method Self
argument so that it appears last after the error
argument.
Swift SVN r26667
This avoids redundant calls of swift_get*Metadata() if the metadata is reused in the same basic block
or if it is already availabe in the function entry block.
There is still room for improvement, e.g. by not limiting the reuse to a single block.
This change reduces the dylib codesize by about 11%, mostly in protocoll witnesses (-34%).
It also results in some improvements for the -Onone performance:
RangeAssignment: +31%
StdlibSort: +20%
Swift SVN r25123
If a type has to be passed or returned resiliently, it
will necessarily be passed indirectly, which is already
represented in SILFunctionType. There is no need to
represent this as a separate channel of information.
NFC. Also fixes a problem where the signature cache
for ExtraData::Block was writing past the end of an
array (but into the storage for an adjacent array
which was fortunately never used).
ExtraData should also disappear as a concept, but we're
still relying on that for existential protocol witnesses.
Swift SVN r21548
Failable initializers will need to clean up an uninitialized 'self' box in the case where a class initializer calls out to a 'super.init' or 'self.init' fails. Because some framework classes assume that they have a refcount of exactly one during initialization, we have to take 'self' from the box during the delegation, and reinitialize the box with the result of the delegation on success. This means we actually have to implement lowering for dealloc_box.
Swift SVN r21493
The allocator's crimes include:
* It uses OS SPI that must not be used by non-OS apps.
* It does not play well with memory debugging tools like Instruments.
* It does not return memory to the OS in response to memory pressure.
* It is less tested than we would like because many configurations
inadvertently turn it off (such as running from Xcode).
* Its per-thread magazine implementation does not actually work.
* Its "try alloc" flag is incompletely implemented and never used.
* Its "zero fill" flag is unimplemented and inconsistently used.
Swift SVN r20757
Introduces a new flag in SILLocation: InPrologue to mark instructions
that setup the stack and allocate storage for local variables/arguments.
Fixes rdar://problem/15290023: Breakpoint set on prologue - crash ensues.
Swift SVN r9686
Add overloads of getFragileTypeInfo and getFunctionType that take a SILType instead of a Swift CanType, and use them where it's easy to do so. Right now they just forward to the CanType versions, but we'll want to do SILType-specific type conversion soon. Clean up some IRGenSILFunction interfaces now that SILFunction carries most of the information IRGen needs intrinsically. No functionality change.
Swift SVN r5141
dead methods, and moving emitInvoke to CallInvocation.
An interesting semantic change of this is that we're now calling getResultType
on the function type every time IGF is constructed, which exposed some latent
bugs. Specifically two places in GenObjC are trying to extra curry level two
from function types like "SomeObject -> (value : SomeObject) -> ()" which
doesn't make sense. I switched them to get curry level 1, but this definitely
needs some close review.
Swift SVN r4852
Per John's comments, make a GenericBox struct to represent the layout of a generic box allocation, move all the layout calculations into methods on GenericBox, and alter swift_deallocBox to take the type as an argument so that it's future-proofed for when we start special-casing certain types in allocBox.
Swift SVN r4613
Add a runtime function that, given a generic type metadata pointer, allocates a heap object capable of containing a value of that type. This is a first-pass implementation that always does the worst case thing of stuffing the type metadata into the box with the value and using its value witness table to size, align, and destroy the box. Use swift_allocBox to implement ArchetypeTypeInfo::allocate correctly for heap object allocations. This means SIL's alloc_box $T now works for archetypes, and a simple generics test now (almost) compiles through SIL!
Swift SVN r4599
Push LLVM attribute generation from expandAbstractCC into getFunctionSignature and CallEmission so that they can generate sret and/or byval attributes per-argument according to the calling convention. Copy our bogus rule for emitting sret returns (more than three elements in the explosion) and reuse it to pass large struct values as byvals rather than as explosions. This should be good enough to get both 'NSRect' and
'NSRange', 'NSSize' etc. to pass correctly to ObjC methods. Next step is to set the AbstractCC correctly for imported func decls so that standalone C functions follow the same bogus rule.
Swift SVN r3993
Implement SIL-to-IR lowering for allocation, deallocation, load, store, and branching instructions so that local variables and branching control flow can be used. Add a Fibonacci loop test to exercise the new instructions.
Swift SVN r3767
Add a path through IRGenModule to optionally codegen FuncDecls using their corresponding SIL Functions when constructed with a SILModule. Jury-rig an IRGenSILFunction subclass of IRGenFunction that does the bare minimum necessary to compile "hello world" from SIL. There are some impedance mismatches between irgen and SIL that need to be smoothed out, particularly the AST-dependent way irgen currently handles function calls. Nonetheless, `swift -sil-i hello.swift` works!
Swift SVN r3759
This is kindof a pain in a few places where the type system
doesn't propagate canonicality. Also, member initializations
are always direct-initializations and so are allowed to use
explicit constructors, which is a hole in our canonicality
tracking. But overall I like the idea of always working
with canonical types.
Swift SVN r2893
match the actual entrypoints vended by the runtime:
1) there is no swift_slowRawAlloc
2) swift_deallocObject takes two arguments
Also, make all calls to swift_allocObject go through a common
function so that we can easily use optimized entrypoints if the
runtime provides them.
Swift SVN r1993
value witnesses goes.
There are three major remaining things to do to support protocols:
- laying out the actual protocol members
- emitting witnesse for the actual protocol members
- detecting uses of the actual protocol members and funnelling
them through the witnesses as appropriate
All this work was just to let us treat protocol types as
first-class values.
Swift SVN r1899
