Create two new semantic names: `ExternalImport` and `ExternalExport`.
These are for symbols which are either imported from an external module
or exported for consumption by external modules.
Now that DLLStorage is only applied when needed, always pass the correct
DLLStorage. The IRLinkage applicator will determine if the DLLStorage
should be applied or not.
Use `ApplyIRLinkage` to the force load thunks to permit multiple
emissions to be COMDATed on Windows. The multi-module tests would emit
the symbols multiply and would fail to link.
This is actually NFC: We should have already deserialized everything
we need at this point, and because of large loadable types and
address lowering, deserializing more stuff in IRGen is not valid,
and in fact we check for this and refuse to deserialize.
This is essentially a long-belated follow-up to Arnold's #12606.
The key observation here is that the enum-tag-single-payload witnesses
are strictly more powerful than the XI witnesses: you can simulate
the XI witnesses by using an extra case count that's <= the XI count.
Of course the result is less efficient than the XI witnesses, but
that's less important than overall code size, and we can work on
fast-paths for that.
The extra inhabitant count is stored in a 32-bit field (always present)
following the ValueWitnessFlags, which now occupy a fixed 32 bits.
This inflates non-XI VWTs on 32-bit targets by a word, but the net effect
on XI VWTs is to shrink them by two words, which is likely to be the
more important change. Also, being able to access the XI count directly
should be a nice win.
- fix code generation for enum types to zext or trunc 32 bit data as appropriate for the platform
- fix IR generation code to use a relative address type of 16 bit width on that platform
- correct order of statements and line up comments
A dynamically replaceable function calls through a global variable that
holds the function pointer.
struct ChainEntry {
void *(funPtr)();
struct ChainEntry *next;
}
ChainEntry dynamicallyReplaceableVar;
void dynamicallyReplaceableFunction() {
dynamicallyReplaceableVar.funPtr()
}
dynamic replacements will be chainable so the global variable also
functions as the root entry in the chain of replacements.
A dynamic replacement functions can call the previous implementation by
going through its chain entry.
ChainEntry chainEntryOf_dynamic_replacement_for_foo;
void dynamic_replacement_for_foo() {
// call the previous (original) implementation.
chainEntryOf_dynamic_replacement_for_foo.funPtr();
}
We were using this just as a convenient way to share an existing
DenseMap, but it's not really related; we don't need to compute
witness table layout just to generate a conformance reference.
I started working on this because the "Cub" source compat project was
hitting issues here, but now I can't reproduce it. Still, this is a
reasonable cleanup.
Previously, when a tuple type had non-fixed layout, we would compute
a layout by building the metadata for that tuple type and then
extracting the layout from the VWT. This can be quite expensive
because it involves constructing the exact metadata for types like
arrays and functions despite those types being fixed-layout across
all instantiations. It also tends to cause unnecessary recursive-type
issues, especially with enums where tuples are currently used to model
cases with mutliple payloads. Since we just need a layout, computing
it directly from element layouts instead of constructing metadata for
the formal type lets us take advantage of all the other fast paths for
layout construction, e.g. for fixed types and single-field aggregates.
This is a good improvement overall, but it also serves to alleviate
some of the problems of rdar://40810002 / SR-7876 in a way that
might be suitable for integration to 4.2.
Because the runtime is compacted into the standard library, functions
which are normally imported are actually local definitions. Use module
level named metadata to identify the module as being the swift standard
library. Refactor the condition slightly to improve code readability.
This addresses SR-7107!
We use dummy symbols to force overlays not to get dropped when
autolinking, even if the user doesn't use anything from them
explicitly. This behavior is triggered by the semi-hidden flag
-autolink-force-load.
(It's semi-hidden because it has few legitimate uses in real life. If
you searched for "how to force autolinking to pick up a library" and
found this commit, don't just do this and move on. Come talk to me on
forums.swift.org.)
Previously we added these dummy symbols to every object file using
"common" linkage, a little-known feature added for C that ensures that
only one definition will actually get used in the final object file.
However, the way we were doing that wouldn't work so well for COFF,
and so in 1025eed64 Saleem changed this to use "weak ODR" linkage.
This has *nearly* the same effect, and avoids some other weirdness,
but has the downside of making the symbol in the final dylib "weak"
itself, meaning that some /other/ library could come along and
override it. That impacts loading time, and an Apple-internal tool
caught that as rdar://39019606.
To avoid this whole mess, "just" emit the symbol into the object file
that corresponds to the first file in the module, which allows us to
mark it as a normal public symbol.
P.S. None of this is actually important at the moment because all of
the overlays are built with single-threaded WMO, which always produces
one object file anyway. But I wanted to get it right once and for all.
Code may end up indirectly using a witness table for a Clang-imported type by inlining code that used the conformance from another module, in which case we need to ensure we have a local definition at hand in the inlining module so we can have something to link against independently. This needs to be fixed from both sides:
- During serialization, serialize not only witness tables from the current module, but from Clang-imported modules too, so that their definitions can be used by other modules that inline code from the current module
- During IRGen, when we emit a reference to a SILWitnessTable or SILFunction declaration with shared linkage, attempt to deserialize the definition on demand
Fixes rdar://problem/38687726.
These will be used as lookup keys for order-independent witness
table instantiation. In the future, a reflective call mechanism
could make use of this metadata as well.
This includes global generic and non-generic global access
functions, protocol associated type access functions,
swift_getGenericMetadata, and generic type completion functions.
The main part of this change is that the functions now need to take
a MetadataRequest and return a MetadataResponse, which is capable
of expressing that the request can fail. The state of the returned
metadata is reported as an second, independent return value; this
allows the caller to easily check the possibility of failure without
having to mask it out from the returned metadata pointer, as well
as allowing it to be easily ignored.
Also, change metadata access functions to use swiftcc to ensure that
this return value is indeed returned in two separate registers.
Also, change protocol associated conformance access functions to use
swiftcc. This isn't really related, but for some reason it snuck in.
Since it's clearly the right thing to do, and since I really didn't
want to retroactively tease that back out from all the rest of the
test changes, I've left it in.
Also, change generic metadata access functions to either pass all
the generic arguments directly or pass them all indirectly. I don't
know how we ended up with the hybrid approach. I needed to change all
the code-generation and calls here anyway in order to pass the request
parameter, and I figured I might as well change the ABI to something
sensible.
The count of the number of witness tables was designed to be an
assertion/check that we've hooked up all the infrastructure
correctly. Everything is now implemented, and the assertion has never
triggered, so it can be removed, saving some work.
Fixes rdar://problem/38038928.
This is yet another waypoint on the path towards the final
generic-metadata design. The immediate goal is to make the
pattern a private implementation detail and to give the runtime
more visibility into the allocation and caching of generic types.
