Skip to content

Fix KeyPath failure on s390x #25467

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
mundaym merged 1 commit into from
Jul 1, 2019
Merged

Fix KeyPath failure on s390x #25467

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 7 additions & 0 deletions include/swift/ABI/Metadata.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1495,7 +1495,11 @@ struct TargetTupleTypeMetadata : public TargetMetadata<Runtime> {
ConstTargetMetadataPointer<Runtime, swift::TargetMetadata> Type;

/// The offset of the tuple element within the tuple.
#if __APPLE__
StoredSize Offset;
#else
uint32_t Offset;
Copy link
Contributor

@jckarter jckarter Jun 20, 2019

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good! For ABI compatibility on Apple platforms, we'd want to conditionalize this, because existing binaries will try to load the full 64 bits, and we want to keep those zeroed:

#if __APPLE__
  StoredSize Offset;
#else
  uint32_t Offset;
#endif

I think the IRGen side is fine, though, because in practice nobody is going to have a 4GB tuple (they'll run into other practical limits first), and new binaries doing a 32-bit little-endian load of the 64-bit field should be safe, and would open the door to us possibly being able to drop this compatibility hack in time.

Copy link
Contributor Author

@levivic levivic Jun 24, 2019

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@jckarter Change made. Thanks.

#endif

OpaqueValue *findIn(OpaqueValue *tuple) const {
return (OpaqueValue*) (((char*) tuple) + Offset);
Expand All @@ -1506,6 +1510,9 @@ struct TargetTupleTypeMetadata : public TargetMetadata<Runtime> {
}
};

static_assert(sizeof(Element) == sizeof(StoredSize) * 2,
"element size should be two words");

Element *getElements() {
return reinterpret_cast<Element*>(this + 1);
}
Expand Down
2 changes: 1 addition & 1 deletion lib/IRGen/IRGenModule.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -266,7 +266,7 @@ IRGenModule::IRGenModule(IRGenerator &irgen,
// A tuple type metadata record has a couple extra fields.
auto tupleElementTy = createStructType(*this, "swift.tuple_element_type", {
TypeMetadataPtrTy, // Metadata *Type;
SizeTy // size_t Offset;
Int32Ty // int32_t Offset;
});
TupleTypeMetadataPtrTy = createStructPointerType(*this, "swift.tuple_type", {
TypeMetadataStructTy, // (base)
Expand Down
2 changes: 1 addition & 1 deletion test/IRGen/generic_tuples.swift
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@
// CHECK: [[TYPE:%swift.type]] = type {
// CHECK: [[OPAQUE:%swift.opaque]] = type opaque
// CHECK: [[TUPLE_TYPE:%swift.tuple_type]] = type { [[TYPE]], i64, i8*, [0 x %swift.tuple_element_type] }
// CHECK: %swift.tuple_element_type = type { [[TYPE]]*, i64 }
// CHECK: %swift.tuple_element_type = type { [[TYPE]]*, i32 }

func dup<T>(_ x: T) -> (T, T) { var x = x; return (x,x) }
// CHECK: define hidden swiftcc void @"$s14generic_tuples3dupyx_xtxlF"(%swift.opaque* noalias nocapture, %swift.opaque* noalias nocapture, %swift.opaque* noalias nocapture, %swift.type* %T)
Expand Down