* Add a custom typed builder for Emulator
* Unify qemu_init for usermode and systemmode
* Remove env from qemu init args (it is unused in QEMU in practice)
* expose thread hooks to systemmode
* rename qemu_config to config
* Replace ExitHandler by EmulatorDriver
* Reorder generics alphabetically for Qemu{,Fork}Executor
* Moved snapshot manager to Emulator to continue centralizing mains objects in the same structure
* Reimplementation of CommandManager working with enums instead of tables
* Macro has been adapted to do this work automatically
* Moved snapshot stuff to dedicated module
* Removed many Rc<RefCell<...>>, now useless with the removal of vtables
* Builder given by Emulator via `Emulator::builder`. Reduced trait bound overhead
* Remove stateless qemu executor
* All harnesses take a reference to an emulator as parameter now
* harness takes an emulator as first parameter, and input as second parameter (opposite of previous definition)
* bump libafl qemu dependencies to the latest version
* Update LibAFL QEMU to the latest version (V9.0.2 update, important bug fixes, ... - check the dedicated repo for more info)
* fix bug in hook execution, causing first execution hooks to be run multiple times.
* 0.13.2 (for real this time)
* More 0.13.2
* Make sure this never bites us
* Update ahash
* Fix CI after rust 1.80
* Does 18 work?
* Trying to fix this?
* lol regex
* noble
* works?
* fix llvm-config
* fmt
* Windows?
* more fix?
* more fix
* remove pip
* Add find_python3_version support to libafl_cc
* llvm-config
* more curl
* windows has no apt lol
* more
use of LLVM 19 by default
LLVM version should be much easier to switch from now on
a lot of code repetition has been deleted
removed llvm-related github action
fix format check
other small things
---------
Co-authored-by: Toka <tokazerkje@outlook.com>
* Helper is now called Module.
* Emulator now contains hooks state.
* Emulator is managed by QemuExecutor.
* QEMU hooks have been completely refactored on the rust side.
* Generics cleanup.
* aa
* why??????????????
* ppp
* aa
* aa
* abcde
* fixer
* ppp
* aa
* fix from windows
* sugar
* ff
* ??
* a
* to log::error
* exclude
* exclude libafl_qemu clippy on windows
* pp
* aa
---------
Co-authored-by: Your Name <you@example.com>
* Fix publish errors
* publish.sh
* remove include_str readme in libafl_qemu
* execute libafl_qemu_build::maybe_generate_stub_bindings() only if in cargo doc
* try this thing
* this?
* use new mmap signature in nix.
* fix import
* fix to work with addr2line v0.23.
* remove unused flag
* update strum also for dependency...
* clippy
* clippy
* update lain git hash
* bump many outdated crates
* fix hash
* introduce feedbacks hit tracking for testcases
* make Testcase::hit_feedbacks into Cow<&str> instead of String
rename get_hit_feedbacks to append_hit_feedbacks
update documentation
* simplify ConstFeedback
* rename Feedback::last_result to prev_result
* impl TODO prev_result for NewHashFeedback, ListFeedback, TransferredFeedback, NautilusFeedback
* rename prev_result to last_result
* add docs
* introduce Objectives hit tracking
* update docs
* update Cargo.toml docs
* update docs
* track Feedbacks & Objectives hit in Fuzzer::add_input
* fmt
* clippy
* fix type error in OomFeedback::last_result
* impl last_result for AsanErrorsFeedback
* add track_hit_feedbacks as a feature to libafl_libfuzzer_runtime
* fix clippy
* change return type of Feedback::last_result to a Result
* remove expect in NewHashFeedback::is_interesting
* move Error::premature_last_result to libafl from libafl_bolts
* Add BytesSubMutator that allows us to mutate sub-parts of a bytes-backed input
* no_std
* fix string mutator
* make build
* Fix clippy on macOS
* Docs
* More docs
* Better docs
* --amend
* Renamed bsi to sub_input. Too much BSI
* More more
* balance backticks
* Make splicing usable with sub_input (not that it makes sense)
* More annotations
* more input annotations?
* Implement HasMutatorBytes for &mut Vec
* clippy
* Use a wrapper type instead
* Add wrapper type for Vec as well
* Remove the duplicate BytesInput... lol
