fix: use scratch base GPA instead of scratch size in page table walker

The access_gpa and SharedMemoryPageTableBuffer::new functions now take
a scratch_base_gpa (u64) instead of scratch_size (usize). Update the
read_guest_memory_by_gva caller to pass the correct value from
layout.get_scratch_base_gpa().

Author: danbugs

src/hyperlight_host/src/mem/mgr.rs

@@ -435,11 +435,11 @@ impl SandboxMemoryManager<HostSharedMemory> {
 
         use crate::sandbox::snapshot::{SharedMemoryPageTableBuffer, access_gpa};
 
-        let scratch_size = self.scratch_mem.mem_size();
+        let scratch_base_gpa = self.layout.get_scratch_base_gpa();
 
         self.shared_mem.with_exclusivity(|snap| {
             self.scratch_mem.with_exclusivity(|scratch| {
-                let pt_buf = SharedMemoryPageTableBuffer::new(snap, scratch, scratch_size, root_pt);
+                let pt_buf = SharedMemoryPageTableBuffer::new(snap, scratch, scratch_base_gpa, root_pt);
 
                 // Walk page tables to get all mappings that cover the GVA range
                 let mappings: Vec<_> = unsafe {
@@ -479,7 +479,7 @@ impl SandboxMemoryManager<HostSharedMemory> {
 
                     // Translate the GPA to host memory
                     let gpa = mapping.phys_base + page_offset as u64;
-                    let (mem, offset) = access_gpa(snap, scratch, scratch_size, gpa)
+                    let (mem, offset) = access_gpa(snap, scratch, scratch_base_gpa, gpa)
                         .ok_or_else(|| {
                             new_error!(
                                 "Failed to resolve GPA {:#x} to host memory (GVA {:#x})",