* Completely remove Input as an associated type in multiple traits
* Unify usage of Input as generic instead
* Remove many unused bounds, in particular HasCorpus
* fix multiple generic ordering
* update and fix CONTRIBUTING.md
* update MIGRATION
* use the same generic input type for new / with_max_iterations to make typing easier in most cases.
* Restore libafl_libfuzzer test in CI
(See #2853.)
This commit attempts to improve the [`HasMutatorBytes`] trait to allow it to
be used along with input types that cannot be resizable.
[`HasMutatorBytes`] is split into two traits:
- [`HasMutatorBytes`]: requires `bytes` and `bytes_mut`.
- `HasMutatorResizableBytes`: requires [`HasMutatorBytes`], and also requires
a few other methods for resizing / shrinking the underlying input type.
N.B.: I believe that if merged, this would introduce a breaking change.
[`HasMutatorBytes`]: 198cd5dbc5/libafl/src/inputs/bytes.rs (L26)
* the first step of the last step
* wip
* 99% done
* 99.9% done
* 99.99
* Hello from windows
* aaa
* 99.999
* aa
* 1
* 2
* 3
* 4
* 5
* plz
* plzplzplz
---------
Co-authored-by: Your Name <you@example.com>
* remove UsesInput from libafl_qemu
* Insert I bound instead of ugly Corpus::Input stuff
* Uses a C generic for Commands
* adapt nyx api to new generics
* add qemu linux fuzzers to CI for building.
* more nyx hypercalls implemented, among them:
- panic hypercall
- range filtering hypercall
* fixed some nyx hypercalls behavior.
* added generic read / write to qemu memory
* port linux kernel example to also have nyx API, add better filtering as well.
* make nyx api structs volatile to avoid optimization issues
* Introduce a method create a Vec in place, using a closure.
* use new vec_init function in relevant places.
* removed unused unsafe keywork
* add more allocated memory r/w callbacks
* add more safety notes
* move emulator hooks to separate struct
* update QEMU version
* first step
* wip
* lol
* making progress, slowly
* add
* a bit more
* progress
* 80%done
* 90% done
* done from linux
* Hello from windows
* 1
* 2
* 3
* 4
* 5
* 6
* 7
* Hello from windows
* 9
* 10
* 11
* 11
* 12
* 13
* aa
* fixer
* a
* oops
---------
Co-authored-by: Your Name <you@example.com>
* added change_brk function for correctly handling SYS_brk
* we need to update h.brk with the new brk_val
* map back pages if brk shrunk below the snapshotted value
* fmt and clippy
* use GuestAddr instead of u64
---------
Co-authored-by: Romain Malmain <romain.malmain@pm.me>
* Nyx hypercall API support
* fix linux kernel fuzzer
* hash_me -> hash_64_fast
* fix multiple bug in kernel harness
* do not check libmozjpeg's C files format.
* Qemu config refactoring.
* QEMU error refactoring.
* Single QEMU init function.
* Light refactor of EmulatorModules.
* Qemu is now a parameter to EmulatorModule callbacks and most function hooks.
* EmulatorModules is initialized before QEMU is initialized.
* refactor asan and asanguest modules to avoid custom init of QEMU and use the module interface instead.
* asan fixed size accesses working with generics.
* use pre_syscall_* and post_syscall_* everywhere for consistency.
* adapt qemu_launcher example to fully work with Emulator, since Qemu must now be initialized by Emulator.
* start writing Emulator / EmulatorBuilder / QemuConfig doc.
* fix broken intel pt doc.
* Revert "New year new clippy (#2797)"
This reverts commit deb76555b75ca2ccac83b83a6db1a276c07080b5.
* Mention that program should panic when using InProcessForkExecutor
* Reapply "New year new clippy (#2797)"
This reverts commit 529213ef6334fc18e1898f3cfbd8bed24c312522.
* Added expect error message to TimeFeedback where there used to be an unwrap
* Changed error message handling
* fixed cargo clippy
---------
Co-authored-by: Dominik Maier <domenukk@gmail.com>
* Optimize event serialization with pre-allocated buffer
- Added event_buffer field to LlmpEventManager
- Used to_slice instead of to_allocvec
- Pre-allocated buffer size is 4KB
Fixes#1082
* Fallback to to_allocvec in case of event_buffer overflow
Also combined the shared logic between compressed & uncompressed event
firing while keeping the same behavior
* Made the initial event_buffer size to a const
Also removed the unnecessary event_buffer.clear(), since we are already
resizing it
