Repository navigation
Fix computed-jump listings, encoded text, and aggregate LLVM C views - #1643
Merged
Merged
Conversation
gmh5225
marked this pull request as ready for review
October 22, 2026 07:30
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Computed jumps with exact scalar targets previously left native listings on a linear decode, allowing a skipped opcode to swallow the real destination. Recover and revalidate those targets in the shared CFG builder, publish its instruction boundaries through an additive SDK query, and use them for listing rows and cross-references. Uncertain inputs, loads, calls, new incoming edges and exhausted evidence retain an indirect transfer. Undefine now works inside an automatic function, with undo/redo and persistence.
Also fix mapped LLVM C views that registered aggregate allocations as oversized integer variables. Both C routes now annotate image blocks with recognized text and encoding while preserving their original accesses and storage. Rejected lifts remain inspectable as explicitly marked linear disassembly. Recovered reference overlays preserve relocated pointer and embedded switch-table references.
Validation on Linux Release: exact-branch tests across x86/x64/AArch64 and ELF/COFF/Mach-O image ownership; public SDK paging/raw-decode checks; real-worker branch, switch-reference and Undefine persistence tests; 632 jump-table, 86 instruction-boundary, 105 LLVM C value, 12 allocation and 28 forwarding/index regressions; 26 GUI tests and real-engine GUI captures. Focused engine and worker checks were repeated after integration with the current dev branch. The installed GUI was opened with its default worker and the matching real engine.
The Python ABI inventory now declares the new query with owned-string lifetime. The Python SDK audit and 218 discovered tests pass locally (116 optional integration tests skipped without their configured fixtures). One resumable-trap test also requires an optional external corpus and is skipped. Windows/macOS native runs and general IDA performance superiority are not claimed.
Post-merge integration at
a3dcf040c: rebuilt the engine, CLI, worker and GUI together. Repeated the branch/C API/source-presentation/LLVM C tests, all six selected real-worker tests including source-synthesis runtime checks, and the four newly integrated Pathfinder chooser cases. The integrated MBA suite passed 70 tests; 10 optional independent SMT proofs were skipped because Z3 is unavailable. Both native listing and LLVM C views were opened with the installed GUI's default worker and engine.