Before this commit initModelTypes wasn't able to detect the type of the
callee operand for calls to isolated functions.
This commit properly supports this case.
Now extractvalue instruction are replaced by dedicated
OpaqueExtractValue custom opcode, that prevents LLVM from doing strange
things with extractvalues during optimizations (such as e.g. sinking).
This is important since extractvalue instructions and struct-typed
values in general in our LLVM IR are not real first-class citizens, but
only a byproduct of the binary lifting process, and they actually
represent bundles of registers that are returned from isolated
functions.
This commit introduces modelType, which supersedes `llvmIntToModelType`
in order to better handle translation of LLVM types into model types
after introduction of opaque pointer types. The main differences is
that `modelType` accepts the `Value` (instead of just the `Type`), so it
can better handle `AllocaInst` and `GlobalVariable`, which provide
information about the pointee.
This commit does various things oriented at reducing the number of local
variables emitted in C:
- MarkAssignments now know that @Copy and @Assign involving
@LocalVariable only have side effects that affect the local variable
itself; this enables to reduce the number of times we're forced to
emit a local variable due to interfering side effects
- Drop the @AssignmentMarker FunctionTag; AddAssignmentMarkerPass now
doesn't emit @AssignmentMarker anymore; instead it emits groups of
@LocalVariable, @Copy, and @Assign, which benefit from the previous
point
- Drop 2 MarkAssignments::Reasons: HasManyUses and HasUsesOutsideOfBB;
both these have now been aggregated into the AlwaysAssign reason for
simplicity, representing all reasons non involving side effects
- Update BeautifyGHAST and how it reasons about side effects when
beautifying; before this commit it used @AssignmentMarker, now it
looks at @Assign
- Simplify ExitSSA; before this commit it was trying hard to be smart on
where it emitted the store instructions representing the incoming
values of the PHI that was being destroyed; this seemed smart when we
originally did it but it generated C code that was not really better
to read, so this useless complexity is finally gone
This commit adds a new pass, PrettyIntFormatting, that injects calls to
decorator functions print_hex, print_char, and print_bool around
llvm::ConstantInt in various situations.
It also updates the rest of passes of the decompilation pipelin to
understand these new decorator functions and to properly emit decorated
integer literals in the decompiled C code.
Before this commit, the code assumed that whenever we had multiple
return types on the model we had an aggregate return type also in LLVM
IR. This is not true anymore with CABIFunctionType, where many model
types can be stuffed into a single wide integer in LLVM IR.
This commit fixes the code to be able to handle that situation.
This commit removes the bugged addPointerQualifier helper function,
that was wrongly pushing the pointer qualifier at the end.
Instead, we now use the correct model::Binary::getPointerTo method.
1. Add a common helper to traverse ModelGEPs (`traverseModelGEP`)
2. Add a centralized way to deduce the model type of values that
have strong model information attached to them (e.g. isolated
functions and ModelGEPs)
3. Add a similar helper for deducing formal types of operands in known
cases
A simple pass that maps LLVM IR instructions to C operators,
taking into account their precedence and associativity to
emit nice-looking parenthesized expressions.
A pass that strips the casts off from some instructions, including
`ModelGEP`s, general function calls, return and store ones, and
embeds the cast into new dedicated `ModelCast` function calls.
This library replaces the old AddIRSerializationMarkers, cleaning up a
lot of historcal baggage, merging Liveness and MarkAnalysis, and
adopting a more accurate naming across all the codebase.
Alessandro Di Federico