immunant / dwarf-writer Goto Github PK

The EntryRef methods in dwarf_entry.rs duplicate some code since the various input sources provide different DWARF function and variable entry attributes. We could deduplicate quite a bit by creating an internal IR for DWARF function and variable entries.

Essentially these structs would have Option<T> fields for the entry attributes. We may possibly want to use a different enum with three variants to distinguish between input sources that cannot provide a certain attribute and those that can, but did not provide it. Then the parsed input data would be transformed into this IR (instead of the current AnvillData, GhidraData, etc...) and we could use a single function to update the DWARF function entries and another for variables. This wouldn't change the behavior of this tool, but would be good to have before adding any more input sources.

We can currently take symbols from anvill json specs to update/create names in DWARF entries for functions and global variables. Consumers of the updated binaries may prioritize reading symbols or be unable to read debug info, so we should support writing/updating symbols as well. This isn't too relevant for the json anvill generates, but it is useful for integrating manual reverse engineering results back into the binaries.

donna doesn't have llvm-dwarfdump in $PATH; it does have llvm-dwarfdump-11, llvm-dwarfdump-12, etc. in $PATH.

Might as well reuse what's already there.