Optimize EPT Structure for Space and Speed

[memN0ps]

Description:

This PR refines our Extended Page Tables (EPT) setup, moving away from the initial broad pre-allocation of Page Tables (Pt) for every Page Directory (Pd) entry (option A) to a better allocation model (option B). This adjustment, coupled with the introduction of split_2mb_to_4kb, remap_gpa_to_hpa, and modify_page_permissions functions, really helps boost the hypervisor's efficiency and flexibility.

Option A:

#[repr(C, align(4096))]
pub struct Ept {
    pml4: Pml4,
    pdpt: Pdpt,
    pd: [Pd; 512],
    pt: [[Pt; 512]; 512],
}

Option B:

#[repr(C, align(4096))]
pub struct Ept {
    pml4: Pml4,
    pdpt: Pdpt,
    pd: [Pd; 512],
    pt: Pt,
}

Option C:

#[repr(C, align(4096))]
pub struct Ept {
    pml4: Pml4,
    pdpt: Pdpt,
    pd: Pd,
    pt: Pt,
}

Key Changes:

• Moved from pre-allocating a Pt for each Pd (option A) to a targeted allocation strategy (option B), optimizing memory usage.
• Introduced split_2mb_to_4kb, remap_gpa_to_hpa, and modify_page_permissions functions to enable fine-grained memory control and manipulation.

Benefits:

• Space Efficiency: Reduces the EPT's memory footprint by avoiding unnecessary Pt allocations.
• Performance: Minimizes translation traversal depth, speeding up memory operations.
• Flexibility: Enables precise memory access control for tasks like stealth hooking.
• Platform Compatibility: Adapting to VMware's limitations on large page support in PDPT for compatibility and performance.