Hey, looking for feedback on my guard page implementation.
I wasn't sure on how to structure the additional windows specific memory
flags, since the emulation backends won't like additional guard flag. I
opted to create a new `memory_permission_ext` enum to hold the guard
flag, and a `nt_memory_permission` struct to wrap the "common" memory
permission flags, with the new extended flags. This struct implicitly
coerces to the original `memory_permission` to reduce the amount of
changes for the PR.
This however meant that I changed signatures of `map_memory` and
`apply_memory_protection` in `memory_interface` to accommodate this new
structure, and was an afterthought.
The `map_nt_to_emulator_protection` function might also need some
attention now, too. For future reference, windows uses
[MiMakeProtectionMask](https://doxygen.reactos.org/d1/d9a/marea_8c.html#adfb66408771a4df77c1056cc2a99ef21)
in ntoskrnl to map `PAGE_*` flags to [MM PTE
constants](https://reactos.org/wiki/Techwiki:Memory_management_in_the_Windows_XP_kernel).
The test added to the `test-sample` binary seems to be passing.
Fixes#21
Change type of chrono return to be auto
Co-authored-by: Maurice Heumann <MauriceHeumann@gmail.com>
simplify conditional compile and add #else to prevent clang-tidy breaking CI
Co-authored-by: Maurice Heumann <MauriceHeumann@gmail.com>
fix intrinsics, and resolve x86 compiler issues
- fix `time.hpp` compiler warnings (errors) and use proper intrinsic for gcc
- simplify tick calculation logic
- fix: exclude intrinsics include for non x86 builds and fix style
- fix: get() on clock_ unique ptr
Previously, RDTSC in the VM always returned a constant value, which broke any non deterministic timing-based operations, or caused detections in heuristics of malware and ANTI-VM tools.
This patch introduces a spoofed rdtsc_fake counter that tracks and adjusts timing deltas to simulate realistic TSC increments. Can be extended to simulate rdtsc timings based on CPU clock speed.
