This is a major change on how we represent nested name qualifications in
the AST.
* The nested name specifier itself and how it's stored is changed. The
prefixes for types are handled within the type hierarchy, which makes
canonicalization for them super cheap, no memory allocation required.
Also translating a type into nested name specifier form becomes a no-op.
An identifier is stored as a DependentNameType. The nested name
specifier gains a lightweight handle class, to be used instead of
passing around pointers, which is similar to what is implemented for
TemplateName. There is still one free bit available, and this handle can
be used within a PointerUnion and PointerIntPair, which should keep
bit-packing aficionados happy.
* The ElaboratedType node is removed, all type nodes in which it could
previously apply to can now store the elaborated keyword and name
qualifier, tail allocating when present.
* TagTypes can now point to the exact declaration found when producing
these, as opposed to the previous situation of there only existing one
TagType per entity. This increases the amount of type sugar retained,
and can have several applications, for example in tracking module
ownership, and other tools which care about source file origins, such as
IWYU. These TagTypes are lazily allocated, in order to limit the
increase in AST size.
This patch offers a great performance benefit.
It greatly improves compilation time for
[stdexec](https://github.com/NVIDIA/stdexec). For one datapoint, for
`test_on2.cpp` in that project, which is the slowest compiling test,
this patch improves `-c` compilation time by about 7.2%, with the
`-fsyntax-only` improvement being at ~12%.
This has great results on compile-time-tracker as well:
This patch also further enables other optimziations in the future, and
will reduce the performance impact of template specialization resugaring
when that lands.
It has some other miscelaneous drive-by fixes.
About the review: Yes the patch is huge, sorry about that. Part of the
reason is that I started by the nested name specifier part, before the
ElaboratedType part, but that had a huge performance downside, as
ElaboratedType is a big performance hog. I didn't have the steam to go
back and change the patch after the fact.
There is also a lot of internal API changes, and it made sense to remove
ElaboratedType in one go, versus removing it from one type at a time, as
that would present much more churn to the users. Also, the nested name
specifier having a different API avoids missing changes related to how
prefixes work now, which could make existing code compile but not work.
How to review: The important changes are all in
`clang/include/clang/AST` and `clang/lib/AST`, with also important
changes in `clang/lib/Sema/TreeTransform.h`.
The rest and bulk of the changes are mostly consequences of the changes
in API.
PS: TagType::getDecl is renamed to `getOriginalDecl` in this patch, just
for easier to rebasing. I plan to rename it back after this lands.
Fixes#136624
Fixes https://github.com/llvm/llvm-project/issues/43179
Fixes https://github.com/llvm/llvm-project/issues/68670
Fixes https://github.com/llvm/llvm-project/issues/92757
This commit handles the following types:
- clang::ExternalASTSource
- clang::TargetInfo
- clang::ASTContext
- clang::SourceManager
- clang::FileManager
Part of cleanup #151026
Handles clang::DiagnosticsEngine and clang::DiagnosticIDs.
For DiagnosticIDs, this mostly migrates from `new DiagnosticIDs` to
convenience method `DiagnosticIDs::create()`.
Part of cleanup https://github.com/llvm/llvm-project/issues/151026
This switches to `makeIntrusiveRefCnt<FileSystem>` where creating a new
object, and to passing/returning by `IntrusiveRefCntPtr<FileSystem>`
instead of `FileSystem*` or `FileSystem&`, when dealing with existing
objects.
Part of cleanup #151026.
Split out from https://github.com/llvm/llvm-project/pull/148877
This patch prepares `TypeSystemClang` APIs to take `AsmLabel`s which
concatenated strings (hence `std::string`) instead of a plain `const
char*`.
This is a continuation of 68fd102, which did the same thing but only for
StopInfo. Using make_shared is both safer and more efficient:
- With make_shared, the object and the control block are allocated
together, which is more efficient.
- With make_shared, the enable_shared_from_this base class is properly
linked to the control block before the constructor finishes, so
shared_from_this() will be safe to use (though still not recommended
during construction).
Move `RecordDecl::isInjectedClassName` to
`CXXRecordDecl::isInjectedClassName`. C language doesn't have the term
"injected class name".
Co-authored-by: Matheus Izvekov <mizvekov@gmail.com>
`SDKSupportsBuiltinModules` always returns true on newer versions of
Darwin based OS. The only way for this call to return `false` would be
to have a version mismatch between lldb and the SDK (recent lldb
manually installed on macOS 14 for instance).
This patch removes this check and hardcodes the value of
`BuiltinHeadersInSystemModules` to `false`.
Some inter-plugin dependencies are okay, others are not. Yet others not,
but we're sort of stuck with them. The idea is to be able to prevent
backsliding while making sure that acceptable dependencies are..
accepted. For context, see
https://github.com/llvm/llvm-project/pull/139170 and the attached
changes to the documentation.
Currently when jitting expressions, LLDB scans the IR instructions of
the `$__lldb_expr` and will insert a call to a utility function for each
load/store instruction. The purpose of the utility funciton is to
dereference the load/store operand. If that operand was an invalid
pointer the utility function would trap and LLDB asks the IR checker
whether it was responsible for the trap, in which case it prints out an
error message saying the expression dereferenced an invalid pointer.
This is a lot of setup for not much gain. In fact, creating/running this
utility expression shows up as ~2% of the expression evaluation time
(though we cache them for subsequent expressions). And the error message
we get out of it is arguably less useful than if we hadn't instrumented
the IR. It was also untested.
Before:
```
(lldb) expr int a = *returns_invalid_ptr()
error: Execution was interrupted, reason: Attempted to dereference an invalid pointer..
The process has been returned to the state before expression evaluation.
```
After:
```
(lldb) expr int a = *returns_invalid_ptr()
error: Expression execution was interrupted: EXC_BAD_ACCESS (code=1, address=0x5).
The process has been returned to the state before expression evaluation.
```
This patch removes this IR checker.
This patch ensures we can find decls in submodules during expression
evaluation. Previously, submodules would have all their decls marked as
`Hidden`. When Clang asked LLDB for decls, it would see them in the
submodule but `clang::Sema` would reject them because they weren't
`Visible` (specifically, `getAcceptableDecl` would fail during
`CppNameLookup`). Here we just mark the submodule as visible to work
around this problem.
If we're not touching them, we don't need to do anything special to pass
them along -- with one important caveat: due to how cmake arguments
work, the implicitly passed arguments need to be specified before
arguments that we handle.
This isn't particularly nice, but the alternative is enumerating all
arguments that can be used by llvm_add_library and the macros it calls
(it also relies on implicit passing of some arguments to
llvm_process_sources).
The problem was in calling GetLoadAddress on a value in the error state,
where `ValueObject::GetLoadAddress` could end up accessing the
uninitialized "address type" by-ref return value from `GetAddressOf`.
This probably happened because each function expected the other to
initialize it.
We can guarantee initialization by turning this into a proper return
value.
I've added a test, but it only (reliably) crashes if lldb is built with
ubsan.
A few files of lldb dir & few other files had duplicate headers
included. This patch removes those redundancies.
---------
Co-authored-by: Akash Agrawal <akashag@qti.qualcomm.com>
This reverts commit 57f3151a3144259f4e830fc43a1424e4c1f15985.
LLDB was hitting an assert when compiling the `std` module. The `std`
module was being pulled in because we use `#import <cstdio>` in the test
to set a breakpoint on `puts`. That's redundant and to work around the
crash we just remove that include. The underlying issue of compiling the
`std` module still exists and I'll investigate that separately. The
reason it started failing after the `ClangModulesDeclVendor` patch is that we would previously just fail to load the modulemap (and thus not load any of the modules). Now we do load the modulemap (and modules) when we prepare for parsing the expression.
The TestTemplateWithSameArg.py API test is crashing on macOS with
this change. Michael is going to look into it; reverting for now
to unblock the CI bots.
This reverts commit 052c70451afb7323ef72f321f3b0b5abb024b302.
I've been skimming this code while investigating a bug around module
lookup and this looked like something we could clean up. We don't need
to be carrying around state inside of `NameSearchContext` to tell us to
look into modules. We can signal this via a boolean return from
`LookupFunction`.
`parseAndLoadModuleMapFile` returns `true` on error. This seems to have
always been an issue? This is now preventing me from fixing a different
modules related issue. So this patch checks the return value correctly.
This reverts commit e2a885537f11f8d9ced1c80c2c90069ab5adeb1d. Build failures were fixed right away and reverting the original commit without the fixes breaks the build again.
The `DiagnosticOptions` class is currently intrusively
reference-counted, which makes reasoning about its lifetime very
difficult in some cases. For example, `CompilerInvocation` owns the
`DiagnosticOptions` instance (wrapped in `llvm::IntrusiveRefCntPtr`) and
only exposes an accessor returning `DiagnosticOptions &`. One would
think this gives `CompilerInvocation` exclusive ownership of the object,
but that's not the case:
```c++
void shareOwnership(CompilerInvocation &CI) {
llvm::IntrusiveRefCntPtr<DiagnosticOptions> CoOwner = &CI.getDiagnosticOptions();
// ...
}
```
This is a perfectly valid pattern that is being actually used in the
codebase.
I would like to ensure the ownership of `DiagnosticOptions` by
`CompilerInvocation` is guaranteed to be exclusive. This can be
leveraged for a copy-on-write optimization later on. This PR changes
usages of `DiagnosticOptions` across `clang`, `clang-tools-extra` and
`lldb` to not be intrusively reference-counted.
Add an overloaded `GetTypeSystem` to specify the expected type system subclass. Changes code from `GetTypeSystem().dyn_cast_or_null<TypeSystemClang>()` to `GetTypeSystem<TypeSystemClang>()`.
Instead of eagerly populating the `clang::ModuleMap` when looking up a
module by name, this patch changes `HeaderSearch` to only load the
modules that are actually used.
This introduces `ModuleMap::findOrLoadModule` which will load modules
from parsed but not loaded module maps. This cannot be used anywhere
that the module loading code calls into as it can create infinite
recursion.
This currently just reparses module maps when looking up a module by
header. This is fine as redeclarations are allowed from the same file,
but future patches will also make looking up a module by header lazy.
This patch changes the shadow.m test to use explicitly built modules and
`#import`. This test and the shadow feature are very brittle and do not
work in general. The test relied on pcm files being left behind by prior
failing clang invocations that were then reused by the last invocation.
If you clean the cache then the last invocation will always fail. This
is because the input module map and the `-fmodule-map-file=` module map
are parsed in the same module scope, and `-fmodule-map-file=` is
forwarded to implicit module builds. That means you are guaranteed to
hit a module redeclaration error if the TU actually imports the module
it is trying to shadow.
This patch changes when we load A2's module map to after the `A` module
has been loaded, which sets the `IsFromModuleFile` bit on `A`. This
means that A2's `A` is skipped entirely instead of creating a shadow
module, and we get textual inclusion. It is possible to construct a case
where this would happen before this patch too.
An upcoming patch in this series will rework shadowing to work in the
general case, but that's only possible once header -> module lookup is
lazy too.
This fixes a regression caused by us starting to parse types from
declarations. The code in TypeSystemClang was assuming that the value
held in ClangASTMetadata was authoritative, but this isn't (and cannot)
be the case when the type is parsed from a forward-declaration.
For the fix, I add a new "don't know" state to ClangASTMetadata, and
make sure DWARFASTParserClang sets it only when it encounters a forward
declaration. In this case, the type system will fall back to completing
the type.
This does mean that we will be completing more types than before, but
I'm hoping this will offset by the fact that we don't search for
definition DIEs eagerly. In particular, I don't expect it to make a
difference in -fstandalone-debug scenarios, since types will nearly
always be present as definitions.
To avoid this cost, we'd need to create some sort of a back channel to
query the DWARFASTParser about the dynamicness of the type without
actually completing it. I'd like to avoid that if it is not necessary.
This PR makes it so that `CompilerInvocation` needs to be provided to
`CompilerInstance` on construction. There are a couple of benefits in my
view:
* Making it impossible to mis-use some `CompilerInstance` APIs. For
example there are cases, where `createDiagnostics()` was called before
`setInvocation()`, causing the `DiagnosticEngine` to use the
default-constructed `DiagnosticOptions` instead of the intended ones.
* This shrinks `CompilerInstance`'s state space.
* This makes it possible to access **the** invocation in
`CompilerInstance`'s constructor (to be used in a follow-up).
This PR hides the reference-counted pointer that holds `TargetOptions`
from the public API of `CompilerInvocation`. This gives
`CompilerInvocation` an exclusive control over the lifetime of this
member, which will eventually be leveraged to implement a copy-on-write
behavior.
There are two clients that currently share ownership of that pointer:
* `TargetInfo` - This was refactored to hold a non-owning reference to
`TargetOptions`. The options object is typically owned by the
`CompilerInvocation` or by the new `CompilerInstance::AuxTargetOpts` for
the auxiliary target. This needed a bit of care in `ASTUnit::Parse()` to
keep the `CompilerInvocation` alive.
* `clangd::PreambleData` - This was refactored to exclusively own the
`TargetOptions` that get moved out of the `CompilerInvocation`.
This PR reland https://github.com/llvm/llvm-project/pull/135808, fixed
some missed changes in LLDB.
I found this issue when I working on
https://github.com/llvm/llvm-project/pull/107168.
Currently we have many similiar data structures like:
- std::pair<IdentifierInfo *, SourceLocation>.
- Element type of ModuleIdPath.
- IdentifierLocPair.
- IdentifierLoc.
This PR unify these data structures to IdentifierLoc, moved
IdentifierLoc definition to SourceLocation.h, and deleted other similer
data structures.
---------
Signed-off-by: yronglin <yronglin777@gmail.com>
Reverts llvm/llvm-project#132274
Broke a test on LLDB Widows on Arm:
https://lab.llvm.org/buildbot/#/builders/141/builds/7726
```
FAIL: test_dwarf (lldbsuite.test.lldbtest.TestExternCSymbols.test_dwarf)
<...>
self.assertTrue(self.res.Succeeded(), msg + output)
AssertionError: False is not true : Command 'expression -- foo()' did not return successfully
Error output:
error: Couldn't look up symbols:
int foo(void)
Hint: The expression tried to call a function that is not present in the target, perhaps because it was optimized out by the compiler.
```
The `TestMemoryHistory.py`/`TestReportData.py` are currently failing on
the x86 macOS CI (started after we upgraded the Xcode SDK on that
machien). The LLDB ASAN utility expression is failing to run with
following error:
```
(lldb) image lookup -n __asan_get_alloc_stack
1 match found in /usr/lib/system/libsystem_sanitizers.dylib:
Address: libsystem_sanitizers.dylib[0x00007ffd11e673f7] (libsystem_sanitizers.dylib.__TEXT.__text + 11287)
Summary: libsystem_sanitizers.dylib`__asan_get_alloc_stack
1 match found in /Users/michaelbuch/Git/lldb-build-main-no-modules/lib/clang/21/lib/darwin/libclang_rt.asan_osx_dynamic.dylib:
Address: libclang_rt.asan_osx_dynamic.dylib[0x0000000000009ec0] (libclang_rt.asan_osx_dynamic.dylib.__TEXT.__text + 34352)
Summary: libclang_rt.asan_osx_dynamic.dylib`::__asan_get_alloc_stack(__sanitizer::uptr, __sanitizer::uptr *, __sanitizer::uptr, __sanitizer::u32 *) at asan_debugging.cpp:132
(lldb) memory history 'pointer'
Assertion failed: ((uintptr_t)addr == report.access.address), function __asan_get_alloc_stack, file debugger_abi.cpp, line 62.
warning: cannot evaluate AddressSanitizer expression:
error: Expression execution was interrupted: signal SIGABRT.
The process has been returned to the state before expression evaluation.
```
The reason for this is that the system sanitizer dylib and the locally
built libclang_rt contain the same symbol `__asan_get_alloc_stack`, and
depending on the order in which they're loaded, we may pick the one from
the wrong dylib (this probably changed during the buildbot upgrade and
is why it only now started failing). Based on discussion with @wrotki we
always want to pick the one that's in the libclang_rt dylib if it was
loaded, and libsystem_sanitizers otherwise.
This patch addresses this by adding a "preferred lookup context list" to
the expression evaluator. Currently this is only exposed in the
`EvaluateExpressionOptions`. We make it a `SymbolContextList` in case we
want the lookup contexts to be contexts other than modules (e.g., source
files, etc.). In `IRExecutionUnit` we make it a `ModuleList` because it
makes the symbol lookup implementation simpler and we only do module
lookups here anyway. If we ever need it to be a `SymbolContext`, that
transformation shouldn't be too difficult.
This patch pushes the error handling boundary for the GetBitSize()
methods from Runtime into the Type and CompilerType APIs. This makes it
easier to diagnose problems thanks to more meaningful error messages
being available. GetBitSize() is often the first thing LLDB asks about a
type, so this method is particularly important for a better user
experience.
rdar://145667239
