Commit Graph

4649 Commits

Author SHA1 Message Date
Tony Allevato
9907afa0a0 Merge pull request #78463 from allevato/json-ast-new
Add a flag to dump the AST as JSON.
2025-01-24 15:30:24 -08:00
Artem Chikin
5eb471730d Merge pull request #78867 from artemcm/NoPackageOnlyImportOnImplicitBuild
[ModuleInterface] Determine package-only resolution need based on input mode
2025-01-24 14:35:22 -08:00
Artem Chikin
a00c1ee7e5 [ModuleInterface] Determine package-only resolution need based on input mode
Instead of requested action. In implicit builds, implicit interface build sub-invocations inherit their parent invocation's requested action, which the code was failing to detect that we were building an interface, not source, and erroneously resulted in enabling in-package module dependency resolution.

Resolves rdar://143505814
2025-01-23 14:11:16 -08:00
Artem Chikin
ea2c501131 Merge pull request #78672 from artemcm/DepScanInvalidRejectedBinaryModulesWarn
[Dependency Scanning] Emit a warning when failing to load a binary module of a non-resilient dependency
2025-01-22 15:03:43 -08:00
Tony Allevato
a4c64b081d Address most of the simple review feedback. 2025-01-22 14:26:13 -05:00
Tony Allevato
71b0a5fae9 Restrict JSON AST dumps to when -dump-ast is used.
Specifically, this means `-dump-ast-format json` is
incompatible with `-dump-parse`. This is because the JSON
format is meant to export more details about the AST that
require type checking to have been performed.

I'm open to lifting this restriction in the future.
2025-01-22 14:26:13 -05:00
Tony Allevato
a57fb4dfa9 [AST] Add minimal JSON support to ASTDumper.
This only takes the existing AST information and writes it as JSON
instead of S-expressions. Since many of these fields are stringified,
they're not ideal for the kind of analysis clients of the JSON format
would want to do. A future commit will update these values to use a
more structured representation.
2025-01-22 14:26:13 -05:00
Egor Zhdan
752cb36239 Merge pull request #78014 from swiftlang/egorzhdan/libcxx-overlay-linux
[cxx-interop] Support CxxStdlib overlay for libc++ on Linux
2025-01-20 15:09:26 +00:00
Artem Chikin
550538f3b5 [Dependency Scanning] Emit a warning when failing to load a binary module of a non-resilient dependency
This failure will most-likely result in the dependency query failure which will fail the scan. It will be helpful if the scanner emitted diagnostic for each such module it rejected to explain the reason why.

Resolves rdar://142906530
2025-01-17 10:52:10 -08:00
Alastair Houghton
18496c5626 [Backtracing] Remove support for implicit import of _Backtracing.
We're going to rename the module to Runtime, and it isn't going to be an
implicit import, so we don't need any of this.

rdar://124913332
2025-01-17 10:09:36 +00:00
Cassie Jones
7323a752b6 [cxx-interop] Workaround compiler error when importing AppKit/UIKit with C++ interop
The AppKit/UIKit overlay refers to symbols declared via NS_OPTIONS
macro, which is causing issues in C++ language mode due to the macro
definition being different. This teaches the module interface loader to
drop the C++ interop flag when rebuilding the AppKit and UIKit overlay
from its interface.

This is the same fix as #78636, but for different modules.

rdar://143033209
2025-01-16 15:19:23 -08:00
Slava Pestov
d1d2411bf3 Merge pull request #78267 from slavapestov/incremental-bindings-part-0
Sema: Some small cleanups in CSBindings.cpp and related code
2025-01-16 09:03:32 -05:00
Slava Pestov
cfcc73f164 Sema: Add -solver-disable-splitter flag for debugging
We're not planning on removing the splitter because it is a big win
in some cases, but we want to run it less often since it can also
be a source of overhead. This flag allows us to compare performance
to understand the tradeoffs better.
2025-01-15 22:30:33 -05:00
Allan Shortlidge
24f5632ca1 Frontend: Implement optional parsing diagnostics for enabled language features.
Parsing for `-enable-upcoming-feature` and `-enable-experimental-feature` is
lenient by default because some projects need to be compatible with multiple
language versions and compiler toolchains simultaneously, and strict
diagnostics would be a nuisance. On the other hand, though, it would be useful
to get feedback from the compiler when you attempt to enable a feature that
doesn't exist. This change splits the difference by introducing new diagnostics
for potential feature enablement misconfigurations but leaves those diagnostics
ignored by default. Projects that wish to use them can specify `-Wwarning
StrictLanguageFeatures`.
2025-01-15 16:34:32 -08:00
Allan Shortlidge
4ea157efdb Frontend: Parse diagnostic options earlier.
Diagnostics may be emitted while parsing command line arguments. This implies
that the options which affect how diagnostics are emitted and presented need to
be parsed first.
2025-01-15 16:34:32 -08:00
Allan Shortlidge
250479177e Merge pull request #78653 from tshortli/suppress-option-parsing-warnings-in-swiftinterfaces
Frontend: Honor warning suppression when parsing arguments from swiftinterfaces
2025-01-15 09:45:40 -08:00
Egor Zhdan
8c07ca90f9 Merge pull request #78636 from swiftlang/egorzhdan/coregraphics-workaround
[cxx-interop] Workaround compiler error when importing CoreGraphics with C++ interop
2025-01-15 12:59:07 +00:00
Allan Shortlidge
3c5ae232dc Frontend: Honor warning suppression when parsing arguments from swiftinterfaces.
Diagnostics are suppressed when parsing swiftinterface files, since the
warnings emitted from compiling the swiftinterface of a dependency would just
be a nuisance. It follows that warnings generated when parsing the arguments in
a swiftinterface file should also be suppressed, but that wasn't happening
because the diagnostic engine of the main compile was used for parsing. Pass
the diagnostic engine of the compiler subinstance instead, and proactively
suppress warnings before parsing begins.

Resolves rdar://142814164.
2025-01-14 23:07:24 -08:00
Egor Zhdan
d72006e265 [cxx-interop] Workaround compiler error when importing CoreGraphics with C++ interop
The CoreGraphics overlay refers to symbols declared via `CF_OPTIONS` macro, which is causing issues in C++ language mode due to the macro definition being different.

This teaches the module interface loader to drop the C++ interop flag when rebuilding the CoreGraphics overlay from its interface.

rdar://142762174
2025-01-14 19:25:40 +00:00
Hiroshi Yamauchi
4a94fc95aa Merge pull request #78556 from hjyamauchi/compilerperf
Sink unnecessary steps for Windows in CompilerInvocation::computeCXXS…
2025-01-13 13:23:34 -08:00
Ellie Shin
727fb8c32d Merge pull request #78258 from swiftlang/elsh/disallow-bypass-deser-check
Package CMO: add deserialization checks to ensure correct memory layout
2025-01-11 05:40:49 -08:00
Doug Gregor
ede7bfe1e8 Merge pull request #78554 from DougGregor/unsafe-effect 2025-01-10 23:51:29 -08:00
Meghana Gupta
53280d730e Merge pull request #78509 from meg-gupta/fixrecompile
Allow recompilation of modules in the resource directory when importing non-ossa module to ossa module
2025-01-10 11:49:37 -08:00
Hiroshi Yamauchi
e767b74790 Sink unnecessary steps for Windows in CompilerInvocation::computeCXXStdlibOptions 2025-01-10 11:20:11 -08:00
Doug Gregor
1b7707d2cc Remove the now-unused @safe(unchecked) 2025-01-10 10:39:16 -08:00
Meghana Gupta
453369517a Allow recompilation of modules in the resource directory when importing non-ossa module to ossa module
This effects local swift development only and is needed now to use -enable-ossa-modules in local tests that
import Synchronization and Distributed which are non-ossa.
2025-01-08 20:45:37 -08:00
Ian Anderson
e73c08c773 Merge pull request #78378 from ian-twilightcoder/nostdlibimport
[Driver][Frontend] Add a -nostdlibimport argument
2025-01-08 15:15:05 -08:00
Doug Gregor
9bd48e0177 Merge pull request #78491 from DougGregor/strict-safety-more-cases
Strict safety more cases
2025-01-08 13:49:46 -08:00
Doug Gregor
c9b2d8ba5f Diagnose the use of -disable-access-checking with strict safety checking 2025-01-08 09:47:17 -08:00
elsh
c03abed00d Package optimization allows bypassing resilience, but that assumes the memory layout of the
decl being accessed is correct. When this assumption fails due to a deserialization error
of its members, the use site accesses the layout with a wrong field offset, resulting in
UB or a crash. The deserialization error is currently not caught at compile time due to
LangOpts.EnableDeserializationRecovery being enabled by default to allow for recovery of some
of the deserialization errors at a later time. In case of member deserialization, however,
it's not necessarily recovered later on.

This PR tracks whether member deserialization had an error by recursively loading members and
checking for deserialization error, and fails and emits a diagnostic. It provides a way to
prevent resilience bypassing when the deserialized decl's layout is incorrect.

Resolves rdar://132411524
2025-01-07 21:51:49 -08:00
Allan Shortlidge
f9f8cfb7ee ModuleInterface: Use SemanticAvailableAttr. 2025-01-07 07:31:13 -08:00
Ian Anderson
87d6979dae Merge pull request #78303 from ian-twilightcoder/clang-importer-search-paths
[ClangImporter] clang's -iframework comes before builtin usr/local/include, but Swift's -Fsystem comes after
2025-01-06 13:05:25 -08:00
Steven Wu
9d43e7c429 Merge pull request #78406 from cachemeifyoucan/eng/PR-141961161
[CAS] Allow empty filename for diagnostics location
2025-01-06 10:42:43 -08:00
Allan Shortlidge
3f0eb8ce2c Frontend: Fix -target-variant subarch normalization.
In https://github.com/swiftlang/swift/pull/77156, normalization was introduced
for -target-variant triples. That PR also caused -target-variant arguments to
be inherited from the main compilation options whenever building dependency
modules from their interfaces, which is incorrect. The -target-variant option
must only be specified when compiling a "zippered" module, but the dependencies
of zippered modules are not necessarily zippered themselves and
indiscriminantly propagating the option can cause miscompilation.

The new, more targeted approach to normalizing arm64e triples simply uses the
arch and subarch of the -target argument of the main compile to decide whether
the subarch of both the -target and -target-variant arguments of a dependency
need adjustment.

Resolves rdar://135322077 and rdar://141640919.
2025-01-02 13:51:26 -08:00
Steven Wu
ca87615b4a [CAS] Allow empty filename for diagnostics location
Fix a bug that an empty filename for diagnostics can trigger an error
in the cached diagnostics. The empty filename can be a virtual file
synthesized, for example, to hold an implicit attribute from clang
importer.

The cached diagnostics can correctly handle such filename now and
cached/replay those diagnostics.

rdar://141961161
2025-01-02 13:18:43 -08:00
Xi Ge
eb6933bbed Merge pull request #78337 from nkcsgexi/141850408
FineModuleTracing/CAS: move the computation of whether fine module trace will be emitted to an early stage.
2025-01-02 09:26:45 -08:00
Ian Anderson
ae753e61a1 [Driver][Frontend] Add a -nostdlibimport argument
Add a -nostdlibimport (analagous to clang's -nostdlibinc) to remove the SDK paths from the import search paths, but leave the toolchain paths.

rdar://139322299
2024-12-28 22:04:56 -05:00
Doug Gregor
d593442cc4 Add module trace information for strict memory safety
Extend the module trace format with a field indicating whether a given
module, or any module it depends on, was compiled with strict memory
safety enabled. This separate output from the compiler can be used as
part of an audit to determine what parts of Swift programs are built
with strict memory safety checking enabled.
2024-12-24 12:27:35 -08:00
Ian Anderson
cdb42c3535 [ClangImporter] clang's -iframework comes before builtin usr/local/include, but Swift's -Fsystem comes after
When Swift passes search paths to clang, it does so directly into the HeaderSearch. That means that those paths get ordered inconsistently compared to the equivalent clang flag, and causes inconsistencies when building clang modules with clang and with Swift. Instead of touching the HeaderSearch directly, pass Swift search paths as driver flags, just do them after the -Xcc ones.

Swift doesn't have a way to pass a search path to clang as -isystem, only as -I which usually isn't the right flag. Add an -Isystem Swift flag so that those paths can be passed to clang as -isystem.

rdar://93951328
2024-12-23 22:15:52 -08:00
Doug Gregor
5b90b8851a Generalize protocol conformance options spelled via attribute and incorporate @unsafe
Protocol conformances have a handful attributes that can apply to them
directly, including @unchecked (for Sendable), @preconcurrency, and
@retroactive. Generalize this into an option set that we carry around,
so it's a bit easier to add them, as well as reworking the
serialization logic to deal with an arbitrary number of such options.

Use this generality to add support for @unsafe conformances, which are
needed when unsafe witnesses are used to conform to safe requirements.
Implement general support for @unsafe conformances, including
producing a single diagnostic per missing @unsafe that provides a
Fix-It and collects together all of the unsafe witnesses as notes.
2024-12-20 23:16:23 -08:00
Doug Gregor
ba3e197841 Diagnose the combination of -Ounchecked with strict memory safety
Aligns with the current proposal and fixes rdar://127128995
2024-12-20 23:15:41 -08:00
Xi Ge
c8fa52b96d FineModuleTracing/CAS: move the computation of whether fine module trace will be emitted to an early stage.
CAS needs to determine if an output is needed very early (when computing supplementary outputs) in the pipeline so we can do caching and replays.

Resolves: rdar://141850408
2024-12-20 19:09:35 -08:00
Steven Wu
2d2dc39047 Merge pull request #78302 from cachemeifyoucan/eng/stevenwu/cache-replay-rewrite
[Caching][NFC] Unify the two cache replay code path
2024-12-20 15:02:18 -08:00
Slava Pestov
3f1491e2b4 Merge pull request #78298 from slavapestov/remove-one-way-bind
Sema: Remove -experimental-one-way-closure-params
2024-12-19 19:37:10 -05:00
Steven Wu
627c64f391 [Caching][NFC] Unify the two cache replay code path
There were two cache replay code exists, one for cache replay from
swift-frontend, the other for replay using C API from libSwiftScan. It
is easy to forget to update one copy when new specialized cache replay
logic is added for some output kinds. Now unify the replay logics to a
single location to avoid confusion. This is a rewrite of the existing
logic and NFCI.
2024-12-19 14:10:46 -08:00
Slava Pestov
bb55d9c59a Sema: Remove -experimental-one-way-closure-params 2024-12-19 13:14:11 -05:00
Allan Shortlidge
0288eaa3b7 AST: Introduce Decl::getSemanticAvailableAttrs().
This new attribute iterator returned from the query makes it simpler to
implement algorithms that need access to both the `AvailableAttr *` and its
corresponding `AvailabilityDomain`. This is also work towards making it
possible to return an optional `AvailabilityDomain` from
`Decl::getDomainForAvailableAttr()`.
2024-12-19 08:40:00 -08:00
Steven Wu
58d71379aa [Caching] Fix cache replay logics for makefile dependency file
Make sure the driver replay logics from C API can replay makefile
dependency file correctly.

rdar://141567785
2024-12-18 14:44:00 -08:00
Xi Ge
64e8f66e4e Merge pull request #78178 from nkcsgexi/fine-trace-opt-out
FineModuleTrace: add a blocklist action for skipping fine module tracing
2024-12-13 19:52:26 -08:00
Xi Ge
c965343f54 Frontend: add a flag to disable fine-grained module tracing 2024-12-13 16:07:15 -08:00