[rebranch] Fix compilation failures #40168
Conversation
llvm-project `ErrorHandling.h` was updated to remove std::string. This added a new `report_fatal_error` overload taking a `const Twine &`, removed the overload that took `const std::string &`, and updated `fatal_error_handler_t` to use `const char *` rather than `const std::string &`. Fix uses of these functions to take into account these updates. Note that without the `const std::string &` overload, passing a `std::string` into `report_fatal_error` now results in an ambiguous match between the `StringRef` and `Twine` overloads so we need to be explicit about one or the other.
llvm-project 601102d282d5e9a1429fea52ee17303aec8a7c10 renamed various functions in `CharInfo.h` and `Lexer.h`. Rename uses in Swift.
llvm-project updated `hashExtension` in 655bea4226b401a11164f99c6344e38d8742b8e4 to use a `HashBuilder` rather than `hash_code`. Update use in ClangImporter.
|
@swift-ci please test |
| auto attrs = llvm::AttributeList::get(IGM.getLLVMContext(), | ||
| llvm::AttributeList::FunctionIndex, | ||
| llvm::Attribute::NoUnwind); | ||
|
|
||
| llvm::CallInst *call = | ||
| Builder.CreateCall(IGM.getAllocBoxFn(), typeMetadata); | ||
| call->setAttributes(attrs); | ||
| call->addFnAttr(llvm::Attribute::NoUnwind); |
There was a problem hiding this comment.
@jckarter was this specifically replacing all attributes added in CreateCall or is using add okay (there's a few other cases like this)? The LLVM commits mention possibly hiding AttributeList::AttrIndex in the future so I tried to use the new add(Fn|Param|Ret)* methods where possible. There's just a single FirstArgIndex remaining, used in a replace* method that doesn't have a (Fn|Param|Ret) equivalent at the moment.
If setAttributes is used here on purpose I can just change this to llvm::AttributeList().addFnAttr... instead.
| @@ -38,11 +38,11 @@ struct WidthPreservingAPIntDenseMapInfo { | |||
| } | |||
|
|
|||
| static unsigned getHashValue(const APInt &Key) { | |||
| return static_cast<unsigned>(hash_value(Key)); | |||
| return llvm::DenseMapInfo<APInt>::getHashValue(Key); | |||
There was a problem hiding this comment.
@rjmccall any objections to this change? I'm actually not sure what header is no longer including Hashing.h but this avoids including any extra entirely 🤷
The functions in llvm-project `AttributeList` have been renamed/refactored to help remove uses of `AttributeList::*Index`. Update to use these new functions where possible. There's one use of `AttrIndex` remaining as `replaceAttributeTypeAtIndex` still takes the index and there is no `param` equivalent. We could add one locally, but presumably that will be added eventually.
`WidthPreservingAPIntDenseMapInfo` uses the same `getHashValue` and `isEqual` as the normal `DenseMapInfo` but with different empty and tombstone keys. `hash_value` is no longer a complete type, so just delegate to `DenseMapInfo<APInt>` rather than add another `#include`.
`CoverageFilenamesSectionWriter` now takes an `ArrayRef<std::string>` rather than `ArrayRef<StringRef>`. stable/20210726 reverted that change so that the coverage formats were the same. stable/20211026 doesn't have that revert though.
bnbarham
force-pushed
the
rebranch-failures
branch
from
f48cb2b to
381d053
Compare
| @@ -188,22 +188,22 @@ getAccessorForComputedComponent(IRGenModule &IGM, | |||
| case Getter: | |||
| // Original accessor's args should be @in or @out, meaning they won't be | |||
| // captured or aliased. | |||
| accessorThunk->addAttribute(1, llvm::Attribute::NoCapture); | |||
There was a problem hiding this comment.
@jckarter I'm fairly sure this is correct, but if you want to double check that'd be great :). addParamAttr eventually adds FirstArgIndex, which is 1.
|
@swift-ci please test |
|
Build failed |
|
Build failed |
|
@swift-ci please test |
|
Build failed |
|
Build failed |
|
@swift-ci please test |
|
Build failed |
|
Build failed |
|
@swift-ci please test |
|
Build failed |
|
Build failed |
|
Going to merge since this is now compiling at least. I can do any fixes to the above later. |
Haven't run any tests yet, but this at least fixes all the macOS compilation failures.