Trivial conflict caused by the line above the
`IGM.constructInitialFnAttributes` change in `lib/IRGen/GenDecl.cpp`
having an extra argument passed in rebranch (due to the new LLVM API).
llvm/llvm-project d0262c2394f46bb7da2a75529413d625c70908e5 added a new
default bool param to the two constructors in `SmallVectorMemoryBuffer`.
Since `options.OutputPath` is a `const char *` and that can be promoted
to a `bool`, the constructor being called was changed to the first
constructor (with a default buffer name) - promotion is preferred over
conversion.
Convert the various output paths to a `StringRef` - all their uses
converted to `StringRef` anyway. Also specify the default parameter in
order to maintain the old behaviour, which didn't require a null
terminator.
When looking for a Swift module on disk, we were scanning all module search paths if they contain the module we are searching for. In a setup where each module is contained in its own framework search path, this scaled quadratically with the number of modules being imported. E.g. a setup with 100 modules being imported form 100 module search paths could cause on the order of 10,000 checks of `FileSystem::exists`. While these checks are fairly fast (~10µs), they add up to ~100ms.
To improve this, perform a first scan of all module search paths and list the files they contain. From this, create a lookup map that maps filenames to the search paths they can be found in. E.g. for
```
searchPath1/
Module1.framework
searchPath2/
Module1.framework
Module2.swiftmodule
```
we create the following lookup table
```
Module1.framework -> [searchPath1, searchPath2]
Module2.swiftmodule -> [searchPath2]
```
Instead of checking that the stdlib can be loaded in a variety of places, check it when setting up the compiler instance. This required a couple more checks to avoid loading the stdlib in cases where it’s not needed.
To be able to differentiate stdlib loading failures from other setup errors, make `CompilerInstance::setup` return an error message on failure via an inout parameter. Consume that error on the call side, replacing a previous, more generic error message, adding error handling where appropriate or ignoring the error message, depending on the context.
Ideally, module interface verification should fail the build when fatal error occurs when
type checking emitted module interfaces. However, we found it's hard to stage this phase in
because the ideal case requires all Swift adopters to have valid interfaces. This new front-end flag allows
driver to downgrade all interface verification errors to warnings as an intermediate step.
* Fix unnecessary one-time recompile of stdlib with -enable-ossa-flag
This includes a bit in the module format to represent if the module was
compiled with -enable-ossa-modules flag. When compiling a client module
with -enable-ossa-modules flag, all dependent modules are checked for this bit,
if not on, recompilation is triggered with -enable-ossa-modules.
* Updated tests
Serialize the canonical name of the SDK used when building a swiftmodule
file and use it to ensure that the swiftmodule file is loaded only with
the same SDK. The SDK name must be passed down from the frontend.
This will report unsupported configurations like:
- Installing roots between incompatible SDKs without deleting the
swiftmodule files.
- Having multiple targets in the same project using different SDKs.
- Loading a swiftmodule created with a newer SDK (and stdlib) with an
older SDK.
All of these lead to hard to investigate deserialization failures and
this change should detect them early, before reaching a deserialization
failure.
rdar://78048939
We have implemented a libSwiftDriver-based tool to generate prebuilt module cache for
entire SDKs. Anchored on the same infrastructure, we could also generate ABI baselines
for entire SDKs.
This mechanism allows the compiler to use a backup interface file to build into a binary module when
a corresponding interface file from the SDK is failing for whatever reasons. This mechansim should be entirely opaque
to end users except several diagnostic messages communicating backup interfaces are used.
Part of rdar://77676064
This allows library authors to pass down a project version number so that library users can conditionally
import that library based on the available version in the search paths.
Needed for rdar://73992299
Previously, when the standard library module interface was broken, Swift would try to rebuild it repeatedly during -compile-module-from-interface jobs because `ASTContext::getStdlibModule()` would try to load the standard library again each time it was called. This led to extremely slow compilations that repeatedly emitted the same errors.
To avoid this, we make ModuleInterfaceBuilder try to load the standard library right away and bail out if it can’t.
Fixes rdar://75669548.
When rebuilding a module interface from the textual interface, lock the
destination path of the created swiftmodule instead of the source
swiftinterface. The swiftinterface files are likely to be in the SDK and
may be on a read-only filesystem.
rdar://60247977
The original remark message “could not acquire lock file for module interface” sounds too severe.
It may confuse users that this is a serious issue. We should rephrase it to a less obtrusive one.
rdar://70055223
This refactoring allows us to drop ModuleInterfaceLoader when explicit modules
are enabled. Before this change, the dependencies scanner needs the loader to be
present to access functionalities like collecting prebuilt module candidates.
-compile-module-from-interface action now takes arguments of -candidate-module-file.
If one of the candidate module files is up-to-date, the action emits a forwarding
module pointing to the candidate module instead of building a binary module.
And also rename the underlying request and
descriptor.
This rename is motivated by the fact that the
operation may instead perform parsing of SIL files
and/or deserialization of SIB files.
Module interface builder used to maintain a separate compiler instance for
building Swift modules. The configuration of this compiler instance is also
useful for dependencies scanner because it needs to emit front-end compiler invocation
for building Swift modules explicitly.
This patch refactor the configuration out to a delegate class, and the
delegate class is also used by the dependency scanner.
Additional flags in interface files may change parsing behavior like #if
statements. We should use a fresh ASTContext with these additional
flags when parsing interface files to collect imports.
rdar://62612027
When a swiftinterface fails to build for any of various reasons, we try to diagnose the failure at the site of the `import` declaration. But if the import is implicitly added—which happens for many SDK modules, like the standard library and ClangImporter overlays—there is no source location for the import, so the error ends up being diagnosed at <unknown>:0. This causes a number of issues; most notably, Xcode doesn’t display the diagnostic as prominently as others.
This change falls back to diagnosing the error at line 1, column 1 of the swiftinterface file itself. This is perhaps not an ideal location, and it won’t help with I/O errors where we can’t open the swiftinterface file (and therefore can’t diagnose an error in it), but it should improve the way we display most module interface building errors.
Currently, when a swiftinterface file fails to load, we emit the specific diagnostics for the failures, followed by a generic “failed to load module ‘Foo’” message. This PR improves that final diagnostic, particularly when the cause may be that the interface was emitted by a newer compiler using backwards-incompatible syntax.
Because we block the calling thread, this should be safe for the existing
implementation of PrettyStackTrace; however, it would be much nicer if
LLVM provided direct support for this.
The most important thing from the trace that we want to print is the
original command line, so an alternative would be to just make a
PrettyStackTrace that printed all of the CompilerInvocation context;
I think that's accessible somewhere. But this is a nice, simple
improvement.
Remove the option to switch off nested types tables. In a world where
re-entrant direct lookup will cause deserialization to fail (or worse),
disabling these tables will only lead to further instability in the
compiler.
