When lowering the arguments for the FFI call on certain targets, LLVM
expects that the parameter be passed as a structure rather than the
extracted values. Enhance the argument lowering for the call and the
function construction to ensure that the type can be flattened according
to the ABI for the target before attempting to explode it for the SIL
IRGen lowering. This allows building for targets such as Linux ARM HF
and Linux ARM SF target with the HF floating point ABI.
Addresses SR-6142!
This reverts commit 25985cb764. For now,
we're trying to avoid spurious non-structural changes to the mangling,
so that the /old/ mangling doesn't appear to change. That doesn't mean
no changes at all, but we can save this one for later.
Use the generic type lowering algorithm described in
"docs/CallingConvention.rst#physical-lowering" to map from IRGen's explosion
type to the type expected by the ABI.
Change IRGen to use the swift calling convention (swiftcc) for native swift
functions.
Use the 'swiftself' attribute on self parameters and for closures contexts.
Use the 'swifterror' parameter for swift error parameters.
Change functions in the runtime that are called as native swift functions to use
the swift calling convention.
rdar://19978563
This allows us to freely pass the address of a variable of such
types to a C function without worrying about it overwriting other
things if e.g. it memcpy's sizeof(T) bytes and thus clobbers the
otherwise-non-existent tail padding.
There are ways to get this optimization back, but they require
things like materializing to a temporary instead of passing the
address directly. We haven't done that work yet, so we don't
get to take advantage of it.
rdar://26828018
This prevents the linker from trying to emit relative relocations to locally-defined public symbols into dynamic libraries, which gives ld.so heartache.
'Ss' appears in manglings tens of thousands of times in the standard library and is also incredibly frequent in other modules. This alone is enough to shrink the standard library by 59KB.
Swift SVN r32409
This brings the David Owens benchmark from http://owensd.io/2015/06/27/performance-xcode7-beta-2.html from parity with simd.h-based C to 3x faster.
Before:
RenderGradient ([UInt32].withUnsafeMutablePointer (SIMD)) │ 7.035851 │ 6.304739 │ 9.815832 │ 1.212 │
After:
RenderGradient ([UInt32].withUnsafeMutablePointer (SIMD)) │ 2.318357 │ 2.223325 │ 2.697981 │ 0.1490 │
This also addresses rdar://problem/21574425, since Builtin.add_VecNxIntM isn't overflow-checked, and overflow checks really aren't wanted when working with vector types directly.
Reapplying now that Nadav's fixed the ARM64 SelectionDAG issue this exposed before, and Arnold's fixed
yet another SelectionDAG issue exposed after that.
Swift SVN r30006
Revert "simd overlay: Use LLVM vector types."
This reverts commit r29922 and r29924.
More arm64 instruction selection errors.
rdar://21703486
Swift SVN r29941
This brings the David Owens benchmark from http://owensd.io/2015/06/27/performance-xcode7-beta-2.html from parity with simd.h-based C to 3x faster.
Before:
RenderGradient ([UInt32].withUnsafeMutablePointer (SIMD)) │ 7.035851 │ 6.304739 │ 9.815832 │ 1.212 │
After:
RenderGradient ([UInt32].withUnsafeMutablePointer (SIMD)) │ 2.318357 │ 2.223325 │ 2.697981 │ 0.1490 │
This also addresses rdar://problem/21574425, since Builtin.add_VecNxIntM isn't overflow-checked, and overflow checks really aren't wanted when working with vector types directly.
Reapplying now that Nadav's fixed the ARM64 SelectionDAG issue this exposed before.
Swift SVN r29922
This brings the David Owens benchmark from http://owensd.io/2015/06/27/performance-xcode7-beta-2.html from parity with simd.h-based C to 3x faster.
Before:
RenderGradient ([UInt32].withUnsafeMutablePointer (SIMD)) │ 7.035851 │ 6.304739 │ 9.815832 │ 1.212 │
After:
RenderGradient ([UInt32].withUnsafeMutablePointer (SIMD)) │ 2.318357 │ 2.223325 │ 2.697981 │ 0.1490 │
This also addresses rdar://problem/21574425, since Builtin.add_VecNxIntM isn't overflow-checked, and overflow checks really aren't wanted when working with vector types directly.
Swift SVN r29891
@inout parameters can be nocapture and dereferenceable. @in, @in_guaranteed, and indirected @direct parameters can be noalias, nocapture, and dereferenceable.
Swift SVN r29353
Insert a block's context argument into the normal parameter set before lowering external arguments, instead of adding it later in a post-pass, so that we assign argument attributes to the correct indices. Fix signature expansion to put the block context parameter in the right place relative to struct returns. Fixes rdar://problem/20035272.
Swift SVN r25763
Upstream clang got better about preserving the type of lowered vector parameters, so float_vec3 lowers to a <3 x float> parameter rather than a generic <2 x double>. We still represent the opaque blob as an aligned i128, since we don't understand padding in imported structs, so we need to generalize coerceValue to deal with size mismatches in both directions. Fixes rdar://problem/20011424.
Swift SVN r25701
Clean up the messy code for putting byval arguments at the right argument index; we don't pretend to handle multiple uncurry levels in IRGen anymore, so we can just use the absolute index of lowered arguments for attribute indices. Add zeroext/signext attributes as needed for small integer arguments in calls. Fixes rdar://problem/19455987.
Swift SVN r25687
The field type generator may end up producing recursive static references to metadata while we're generating metadata. Fixes rdar://problem/19838839.
Swift SVN r25534
If the alloc size of a single iNNN field doesn't match the requested size, then it will use the wrong amount of space inside an LLVM struct. Solve this by using [N x i8] as the storage type for opaque storage, but loading and storing as a single iNNN field.
Swift SVN r24571
Most tests were using %swift or similar substitutions, which did not
include the target triple and SDK. The driver was defaulting to the
host OS. Thus, we could not run the tests when the standard library was
not built for OS X.
Swift SVN r24504
Doing so is safe even though we have mock SDK. The include paths for
modules with the same name in the real and mock SDKs are different, and
the module files will be distinct (because they will have a different
hash).
This reduces test runtime on OS X by 30% and brings it under a minute on
a 16-core machine.
This also uncovered some problems with some tests -- even when run for
iOS configurations, some tests would still run with macosx triple. I
fixed the tests where I noticed this issue.
rdar://problem/19125022
Swift SVN r23683
Specifically, it should not ever have a clang::TypedefDecl as its Clang node,
even if the RecordDecl is anonymous and immediately wrapped in a typedef.
It looks like no one was ever doing this intentionally; it's just left over
from before we looked through typedefs at all. This is important because we
use the clang::RecordDecl for IRGen layout if it's present; before this patch
we would fall back to Swift layout for these structs (clearly wrong).
The one exception here is enums -- NS_OPTIONS and named C enums get imported
as structs as well. If we add any other exceptions, we should be sure they
are dealt with in IRGen as well.
The change in the printed interface output is due to the source location
for the Clang node of an NS_ENUM or NS_OPTIONS decl being inside a macro.
I didn't see a quick fix for this, so I'm going to ignore it for now.
Swift SVN r21748
