Using LLVM large integers to represent enum payloads has been causing compiler performance and code size problems with large types, and has also exposed a long tail of backend bugs. Replace them with an "EnumPayload" abstraction that manages breaking a large opaque binary value into chunks, along with masking, testing, and extracting typed data from the binary blob. For now, use a word-sized chunking schema always, though the architecture here is set up to eventually allow the use of an arbitrary explosion schema, which would benefit single-payload enums by allowing the payload to follow the explosion schema of the contained value.
Swift SVN r28982
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
Preparation to fix <rdar://problem/18151694> Add Builtin.checkUnique
to avoid lost Array copies.
This adds the following new builtins:
isUnique : <T> (inout T[?]) -> Int1
isUniqueOrPinned : <T> (inout T[?]) -> Int1
These builtins take an inout object reference and return a
boolean. Passing the reference inout forces the optimizer to preserve
a retain distinct from what’s required to maintain lifetime for any of
the reference's source-level copies, because the called function is
allowed to replace the reference, thereby releasing the referent.
Before this change, the API entry points for uniqueness checking
already took an inout reference. However, after full inlining, it was
possible for two source-level variables that reference the same object
to appear to be the same variable from the optimizer's perspective
because an address to the variable was longer taken at the point of
checking uniqueness. Consequently the optimizer could remove
"redundant" copies which were actually needed to implement
copy-on-write semantics. With a builtin, the variable whose reference
is being checked for uniqueness appears mutable at the level of an
individual SIL instruction.
The kind of reference count checking that Builtin.isUnique performs
depends on the argument type:
- Native object types are directly checked by reading the
strong reference count:
(Builtin.NativeObject, known native class reference)
- Objective-C object types require an additional check that the
dynamic object type uses native swift reference counting:
(Builtin.UnknownObject, unknown class reference, class existential)
- Bridged object types allow the dymanic object type check to be
bypassed based on the pointer encoding:
(Builtin.BridgeObject)
Any of the above types may also be wrapped in an optional. If the
static argument type is optional, then a null check is also performed.
Thus, isUnique only returns true for non-null, native swift object
references with a strong reference count of one.
isUniqueOrPinned has the same semantics as isUnique except that it
also returns true if the object is marked pinned regardless of the
reference count. This allows for simultaneous non-structural
modification of multiple subobjects.
In some cases, the standard library can dynamically determine that it
has a native reference even though the static type is a bridge or
unknown object. Unsafe variants of the builtin are available to allow
the additional pointer bit mask and dynamic class lookup to be
bypassed in these cases:
isUnique_native : <T> (inout T[?]) -> Int1
isUniqueOrPinned_native : <T> (inout T[?]) -> Int1
These builtins perform an implicit cast to NativeObject before
checking uniqueness. There’s no way at SIL level to cast the address
of a reference, so we need to encapsulate this operation as part of
the builtin.
Swift SVN r27887
Store the number of payload and no-payload cases, the case names, and a lazy case type accessor function for enums, like we do for stored properties of structs and classes. This will be useful for multi-payload runtime support, and should also be enough info to hack together a reflection implementation for enums.
For dynamic multi-payload enums to not be ridiculously inefficient, we'll need to track the size of the payload area in the enum, like we do the field offsets of generic structs and classes, so hack off a byte in the payload case count to track the offset of that field in metadata records. 16 million payloads ought to be enough for anyone, right? (and 256 words between the enum metadata's address point and the payload size offset)
Swift SVN r27789
This can come up now when generic multi-payload enums substitute in empty types for all their payloads. It's also an opportunity to factor things a bit so that they're easier to extend to dynamic cases.
Swift SVN r27783
We don't want to wait to revamp the runtime to track spare bits in order to get multi-payload enum cases working, so when lowering an enum, track when the generic enum type's layout is dependent, and constrain our use of spare bits when it is.
Swift SVN r27781
This is an internal-only affordance for the numerics team to be able to work on SIMD-compatible types. For now, it can only increase alignment of fixed-layout structs and enums; dynamic layout, classes, and other obvious extensions are left to another day when we can design a proper layout control design.
Swift SVN r27323
come up when pattern matching on optional. Instead of expanding to a
switch with a default and one case, expand to an icmp and condbr.
Swift SVN r26090
Per the previous commit we are no longer using this. Minor save in
simplicity and maybe a bit of compilation time as well.
In the long run IRGen probably shouldn't be pulling information from the
AST at all; the SILModule should be able to tell it what types it needs
to emit information for. But this is still an improvement for now.
No functionality change (that was the previous commit).
Swift SVN r24840
I thought. Apparently there is no way to directly test whether a general
enum contains the payload without testing against each of the potential
inhabitant cases with a switch. Instead of solving this open problem, just
optimize the case where we're testing for a single non-payload case, and
special case enums with one payload and one non-payload case (hello optional!)
by testing for the non-payload case and inverting the result. This resolves
the miscompilations I was seeing before with my full patch set, but is still
NFC since the code that uses this isn't included yet.
Swift SVN r24365
specific case and returning an i1 result. This can be done a lot more efficiently (in
terms of generated LLVM IR) than doing a general switch over the cases. This will be used
to implement rdar://19404937, but there is a miscompilation here that I'm tracking down.
Until it is working, I'm not committing the code to use these entrypoints, which wires it
into IRGen of SelectEnumInst and SelectEnumAddrInst. Since that part isn't included, this
is NFC.
Swift SVN r24362
IRGen uses a typedef, SpareBitVector, for its principal
purpose of tracking spare bits. Other uses should not
use this typedef, and I've tried to follow that, but I
did this rewrite mostly with sed and may have missed
some fixups.
This should be almost completely NFC. There may be
some subtle changes in spare bits for witness tables
and other off-beat pointer types. I also fixed a bug
where IRGen thought that thin functions were two
pointers wide, but this wouldn't have affected anything
because we never store thin functions anyway, since
they're not a valid AST type.
This commit repplies r24305 with two fixes:
- It fixes the computation of spare bits for unusual
integer types to use the already-agreed-upon type
size instead of recomputing it. This fixes the
i386 stdlib build. Joe and I agreed that we should
also change the size to use the LLVM alloc size
instead of the next power of 2, but this patch
does not do that yet.
- It changes the spare bits in function types back
to the empty set. I'll be changing this in a
follow-up, but it needs to be tied to runtime
changes. This fixes the regression test failures.
Swift SVN r24324
IRGen uses a typedef, SpareBitVector, for its principal
purpose of tracking spare bits. Other uses should not
use this typedef, and I've tried to follow that, but I
did this rewrite mostly with sed and may have missed
some fixups.
This should be almost completely NFC. There may be
some subtle changes in spare bits for witness tables
and other off-beat pointer types. I also fixed a bug
where IRGen thought that thin functions were two
pointers wide, but this wouldn't have affected anything
because we never store thin functions anyway, since
they're not a valid AST type.
Swift SVN r24305
llvm::Optional lives in "llvm/ADT/Optional.h". Like Clang, we can get
Optional in the 'swift' namespace by including "swift/Basic/LLVM.h".
We're now fully switched over to llvm::Optional!
Swift SVN r22477
In preparation for the switch to llvm::Optional, which doesn't have a 'cache'
method. Given how long we spent bikeshedding over the name and how few places
we ended up using it, I didn't feel like trying to push it through on the
LLVM side.
Swift SVN r22471
- A spot fix in SILGen for reabstracting the result of a downcast, which fixes checked casts to function types.
- Associate the layout information in type metadata records with the most abstract representation of the type. This is the correct thing to do in cases where we need the metadata as a tag for an opaque value--if we store a value in an Any, or pass it as an unconstrained generic parameter, we must maximally reabstract it. This fixes the value semantics of existentials containing trivial metatypes.
- To ensure that we get runtime layout of structs and enums correct when they contain reabstractable types, introduce a "metadata for layout" concept, which doesn't need to describe the canonical metadata for the type, but only needs to describe a type with equivalent layout and value semantics. This is a correctness fix that allows us to correctly lay out generic types containing dependent tuples and functions, and although we don't really take advantage of it here, it's also a potential runtime performance win down the road, because we could potentially produce direct metadata for a primitive type that's layout-equivalent with a runtime-instantiated type. To aid in type safety here, push SILType deeper into IRGen in places where we potentially care about specific representations of types.
- Finally, fix an inconsistency between the runtime and IRGen's concept of what spare bits unmanaged references and thick metatypes have.
Together, these fixes address rdar://problem/16406907, rdar://problem/17822208, rdar://problem/18189508, and likely many other related issues, and also fixes crash suite cases 012 and 024.
Swift SVN r21963
I introduced a function swift_keepAlive2() which has a different signature from
swift_keepAlive() until I can verify that the stdlib is using the new
infrastructure.
The difference in signature is that swift_keepAlive2 takes just a pointer while
swift_keepAlive also takes a metadata value that is not necessary for our
purposes anymore.
Swift SVN r21718
The simple cases after the first one end up getting assigned invalid non-null pointer representations that *_retain and *_release aren't written to expect. We checked for this case incorrectly by checking the size of vector arguments that had been std::move'd from. Oops. Fixes rdar://problem/18118396.
Swift SVN r21612
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
This allows IRGen to complain about types it doesn't know how to lower yet, while still recovering well enough not to take the compiler down with it. This reduces the common "unimplemented enum layout" error to be a mere error instead of a compiler crash.
Swift SVN r20773
LLDB needs this in order to accurately test enums that use extra inhabitants of aggregates, for which we don't guarantee that we set all of the bits to an exact value. Allows lldb to address <rdar://problem/17787682>. Now with no test crashes.
Swift SVN r20545
LLDB needs this in order to accurately test enums that use extra inhabitants of aggregates, for which we don't guarantee that we set all of the bits to an exact value. Allows lldb to address <rdar://problem/17787682>.
Swift SVN r20527
Don't use spare bits on platforms that use ObjC tagged pointers when an enum payload involves a class-constrained existential, archetype, or ObjC-defined class type. If a payload is of a Swift-defined class type, we can still assume it's a real pointer and use its spare bits. Add an @unsafe_no_objc_tagged_pointer attribute that can be applied to protocols to denote that existentials bounded by that protocol can use spare bits; this is necessary to preserve the layout of bridged Array and Dictionary types, which should not be bound to tagged pointer types in practice (fingers crossed). Fixes <rdar://problem/16270219>.
Swift SVN r18781
We were miscalculating the number of empty-payload cases representable by each tag value and failing to lower cases after the first two. Fixes <rdar://problem/17025341>.
Swift SVN r18620
for extra inhabitants.
For structs in particular, this eliminates a major source
of abstraction penatlies. For example, an optional struct
containing an object pointer is now represented the same
way as an optional object pointer, which is critical for
correctly importing CF types as Unmanaged<T>!.
In time, we should generalize this to consider all elements
as sources for extra inhabitants, as well as exploiting
spare bits in the representation, but getting the
single-element case right really provides the bulk of the
benefit.
This commit restores r17242 and r17243 with a fix to use
value witnesses that actually forward the right type metadata
down. We were already generating these value witnesses in
the dependent struct VWT pattern, but I was being too clever
and trying to use the underlying value witness directly.
Swift SVN r17267
This reverts commit r17242. We can't simply forward tuple element extra
inhabitant witnesses for the same reason laid out in the previous commit.
Swift SVN r17252
This reverts commit r17243. We can't just forward the extra inhabitant payloads
from a field, because they will end up receiving metadata for the incorrect
type and crashing.
Swift SVN r17251
extra inhabitants.
Obviously this should eventually be generalized to
take from any element, but this is good enough to
give us zero-cost abstraction via single-field structs.
Contains some bugfixes for the tuple-extra-inhabitant
changes as well, because test coverage for optional
structs is obviously quite a bit richer than for
optional tuples.
All of this is leading towards unblocking IRGen for
importing CFStringRef as Unmanaged<CFString>!.
Swift SVN r17243
