This new SIL di-expression represents the dereference on the SSA value.
Similar to DW_OP_deref in DWARF. It is also going to replace the
existing `debug_value_addr`. Namely, replacing the following
instruction:
```
debug_value_addr %a : $*T, name "my_var"
```
with this one:
```
debug_value %a : $*T, name "my_var", expr op_deref
```
Just for convenicence.
* Replace `llvm::isa_and_nonnull` with imported `isa_and_nonnull`
* Repalce some `EXPR && isa<T>(EXPR)` with `isa_and_nonnull<T>(EXPR)`
The symptom here was a crash in getConformanceAccessPath(), but the root
cause was that the ASTMangler used an incorrect generic signature when
looking up the conformance path for a type parameter.
The mangler requires a generic signature to be set to properly mangle types
containing retroactive conformances. However, DebugInfo just used the
IRGenModule's current signature, which wasn't always the same as the
signature describing the archetypes found inside the SILFunction.
Fixes rdar://problem/81683650 / https://bugs.swift.org/browse/SR-15046.
In https://reviews.llvm.org/D95617 LLVM stopped to emit debug values without a
location as an optimization for inlined functions. Because Swift is emitting
dbg.values(undef) for sizeless variables even at -Onone, we need to make sure
all local variables are preserved even at -Onone.
rdar://82038523
We were using compiler-generated source location (i.e. line number
0) on `debug_value` instructions were emitted by salvage debug info.
But it turned out that IRGen will attach DW_AT_artificial on the
associated debug variables, which are hidden in debugger by default.
This patch prevent this issue by using the source location from the
just-deleted instruction for these `debug_value` instructions.
This indirectly triggered another bug where some of the LLVM
!DIExpression-s we generated are actually invalid -- more specifically,
the fragment inside is covering the whole debug variable. The reason it was
not previously caught is because LLVM verifier skips any debug variable
that is marked artificial, and all debug variables that have illegal fragment
are falling under this category. Thus, this patch also fixes this issue
by not generating the DW_OP_LLVM_fragment part if it's illegal.
We used to represent the interface type of variadic parameters directly
with ArraySliceType. This was awfully convenient for the constraint
solver since it could just canonicalize and open [T] to Array<$T>
wherever it saw a variadic parameter. However, this both destroys the
sugaring of T... and locks the representation to Array<T>. In the
interest of generalizing this in the future, introduce
VariadicSequenceType. For now, it canonicalizes to Array<T> just like
the old representation. But, as you can guess, this is a new staging
point for teaching the solver how to munge variadic generic type bindings.
rdar://81628287
SROA and Mem2Reg now can leverage DIExpression -- op_fragment, more
specifically -- to generate correct debug info for optimized SIL. Some
important highlights:
- The new swift::salvageDebugInfo, similar to llvm::salvageDebugInfo,
tries to restore / transfer debug info from a deleted instruction.
Currently I only implemented this for store instruction whose
destination is an alloc_stack value.
- Since we now have source-variable-specific SIL location inside a
`debug_value` instruction (and its friends), this patch teaches
SILCloner and SILInliner to remap the debug scope there in addition
to debug scope of the instruction.
- DCE now does not remove `debug_value` instruction whose associating
with a function argument SSA value that is not used elsewhere. Since
that SSA value will not disappear so we should keep the debug info.
This patch is primarily doing two things:
- Teach IRGen for some of the instructions to generate debug info based
on SILDebugVariable rather than AST node, which is not always
available when reading from SIL.
- Generate corresponding `llvm.dbg.*` intrinsics and `!DIExpression`
from the newly added SIL DIExpression.
Remove the default argument for the `memberType`
parameter and enforce that GenericFunctionType is
not passed. Also add a defaulted overload for the
property case, as they should never have a
GenericFunctionType interface type.
Currently IRGen requires the AST node of a variable declaration to
generate debug location. However, this will fail if the input is SIL due
to the lack of AST reconstruction. Plus, it's unnecessary since we can
just use the `SILLocation` attached on `debug_value` (and its friends) SIL
instruction to generate the correct LLVM debug metadata.
Resolves SR-14868
IRGen tried to decode various types of `SILLocation` into
`FilenameAndLocation` object. But it couldn't handle the case where
`SILLocation` already uses `FilenameAndLocationKind` storage kind. This
patch fixes this issue.
Resolves SR-14856
This patch removes a heuristic to promote all debug intrinsics pointing into
allocas to llvm.dbg.declare() intrinsics and instead more accurate classifies
variables in async contexts by adding the missing cases alloc_box and
alloc_stack cases.
rdar://78977132
DebugInfoTypes didn't properly distinguish between types with an unknown size
and used 0 as the default value. This can cause types to show up in DWARF with a
size of 0, which will prevent LLDB from looking at the type further. In
practice, this meant that many resilient types (for example, SwiftUI Views)
couldn't be displayed in the debugger and just showed up as empty.
This patch does three things:
0. Make size in DebugTypeInfo optional.
1. Introduce CompletedDebugTypeInfo, a subclass of DebugTypeInfo where
size is guaranteed to be non-empty.
2. Change some APIs that only make sense with a size to use
CompletedDebugTypeInfo. This caused some churn because I needed to refactor
the get*Elements() functions back into their callers to facilitate the new
early-exit path. These functions are only called when compiling with
-gdwarf-types.
3. Change createOpaqueStructWithSizedContainer() to create a size-less forward
declaration instead of a zero-sized type for the inner type.
rdar://76973844
(cherry picked from commit b8e6eebdbc)
Usually debug info only ever describes the *result* of a projectBox
call. To display a boxed parameter of an async continuation object,
however, the debug info can only describe the box itself and thus also
needs to emit a box type for it so the debugger knows to call into
Remote Mirrors to unbox the value.
SR-14059
rdar://problem/73358988
My goal was to reduce the size of SILLocation. It now contains only of a storage union, which is basically a pointer and a bitfield containing the Kind, StorageKind and flags. By far, most locations are only single pointers to an AST node. For the few cases where more data needs to be stored, this data is allocated separately: with the SILModule's bump pointer allocator.
While working on this, I couldn't resist to do a major refactoring to simplify the code:
* removed unused stuff
* The term "DebugLoc" was used for 3 completely different things:
- for `struct SILLocation::DebugLoc` -> renamed it to `FilePosition`
- for `hasDebugLoc()`/`getDebugSourceLoc()` -> renamed it to `hasASTNodeForDebugging()`/`getSourceLocForDebugging()`
- for `class SILDebugLocation` -> kept it as it is (though, `SILScopedLocation` would be a better name, IMO)
* made SILLocation more "functional", i.e. replaced some setters with corresponding constructors
* replaced the hand-written bitfield `KindData` with C bitfields
* updated and improved comments
Previously, the name of the entry point function was always main. Here,
a new frontend flag is added to enable an arbitrary name to be
specified.
rdar://58275758
The generic signature isn't used for a whole lot, so this all mostly
worked before; the test case I have hits the code path for mangling
a retroactive conformance.
Fixes <https://bugs.swift.org/browse/SR-14016> / <rdar://problem/72865083>.
Since these types have an implicit stored property, this requires
adding an abstraction over fields to IRGen, at least throughout
the class code. In some ways I think this significantly improves
the code, especially in how we approach missing members.
Fixes rdar://72202671.
This commit depends on changes to the coroutine-splitting pass in LLVM. Shadow
copies are also turned off for async function arguments, because they make it
impossible to track debug info during coroutine splitting. Instead we are
relying on LLVM's CoroSplit.cpp to emit shadow copies. The Swift frontend gives
CoroSplit license to move do this by describing the arguments using a
dbg.declare intrinsic, even though it points to chain of load/GEP/bitcase
instructions into the Swift context function argument.
rdar://71866936
To help consolidate our various types describing imports, this commit moves the following types and methods to Import.h:
* ImplicitImports
* ImplicitStdlibKind
* ImplicitImportInfo
* ModuleDecl::ImportedModule
* ModuleDecl::OrderImportedModules (as ImportedModule::Order)
* ModuleDecl::removeDuplicateImports() (as ImportedModule::removeDuplicates())
* SourceFile::ImportFlags
* SourceFile::ImportOptions
* SourceFile::ImportedModuleDesc
This commit is large and intentionally kept mechanical—nothing interesting to see here.
