After we drop legacy mode, the only user of LocalVariableBuilder will be
SegregateStackAccesses. This is possible thanks to the fact that
dropping the legacy mode will demistify local variables and turn them
into plain allocas accessed via loads/stores. This will end the need for
a centralized place to deal with local variable creation, that is caused
by the various ad-hoc manipulations we've been doing for a long time in
legacy mode.
Drop the !revng.variable_type metadata and the matching helpers,
along with the only place in the Clifter that still consumed them.
After the previous LVB cleanup nothing emits this metadata, and the
model variable type can be reconstructed from the alloca's array
type alone.
Reorganise LocalVariableBuilder so that the legacy and non-legacy
code paths are separated into two specialisations, and so that the
non-legacy specialisation does not need a model::Binary to operate.
This is a prerequisite for using LocalVariableBuilder from passes
that do not know about the model, e.g. unit tests for the new-PM
SwitchToStatements and the upcoming non-legacy decompilation
pipeline.
* Make the following private headers public:
* Lift/CPUStateAccessAnalysisPass.h
* Lift/CSVOffsets.h
* Lift/PTCDump.h
* Lift/VariableManager.h
* Move from revngSupport to revngLift:
* IRAnnotators.{h,cpp}
* SelfReferencingDbgAnnotationWriter.{h,cpp}
* Move from revngSupport to revngModel:
* FunctionTags.{h,cpp}
* ProgramCounterHandler.{h,cpp}
* Move from revngSupport to revngRecompile:
* OriginalAssemblyAnnotationWriter.{h,cpp}
This commit further centralizes the creation of allocas + ptrtoint. The
creation of these pair of instructions is problematic when the size of a
pointer in the input and target architecture differ.
In fact, in such cases, LLVM turns trunc into masking the integer to
suppresse the high bits. This results in problematic code.
This commit introduces an assumption that ensures LLVM does not do that.
Pietro Fezzardi