| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
| |
The jitter cpu/vm modules used an unecessary vm_ prefix for various api.
jitter.cpu.vm_get_gpreg() => jitter.cpu.get_gpreg()
jitter.vm.vm_get_mem... => jitter.vm.get_mem...
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
| |
Fat patch here: some API have changed.
Each irbloc now affects a special "IRDst" register which is used to
describe the destination irbloc. It allows simple description of
architectures using delay slots. Architectures semantic and tcc/python
jitter are modified in consequence. LLVM jitter is disabled for now,
but should be patch soon.
|
| | |
|
| | |
|
| |
|
|
|
| |
- Rewrite updates (engine and JitCpu) in subfunctions
- Explode eval block in an eval_ir loop to gain more control
|
| | |
|
| | |
|
|
|
This way, we hope to:
- make symbexec more reliable
- provide a full Python based Jitter (but *very* slow)
Currently, memory writes, self-modiying code and exceptions aren't handled
|