Palacios Public Git Repository

To checkout Palacios execute

  git clone http://v3vee.org/palacios/palacios.web/palacios.git
This will give you the master branch. You probably want the devel branch or one of the release branches. To switch to the devel branch, simply execute
  cd palacios
  git checkout --track -b devel origin/devel
The other branches are similar.


added symbiotic interface and a number of other small changes
[palacios.git] / palacios / src / palacios / vmm_shadow_paging_64.h
index fbd3f3f..421707a 100644 (file)
@@ -1,8 +1,42 @@
+/* 
+ * This file is part of the Palacios Virtual Machine Monitor developed
+ * by the V3VEE Project with funding from the United States National 
+ * Science Foundation and the Department of Energy.  
+ *
+ * The V3VEE Project is a joint project between Northwestern University
+ * and the University of New Mexico.  You can find out more at 
+ * http://www.v3vee.org
+ *
+ * Copyright (c) 2008, Jack Lange <jarusl@cs.northwestern.edu> 
+ * Copyright (c) 2008, The V3VEE Project <http://www.v3vee.org> 
+ * All rights reserved.
+ *
+ * Author: Jack Lange <jarusl@cs.northwestern.edu>
+ *
+ * This is free software.  You are permitted to use,
+ * redistribute, and modify it as specified in the file "V3VEE_LICENSE".
+ */
+
 
 static inline int activate_shadow_pt_64(struct guest_info * info) {
-  //  struct cr3_64 * shadow_cr3 = (struct cr3_64 *)&(info->ctrl_regs.cr3);
-  return -1;
+    struct cr3_64 * shadow_cr3 = (struct cr3_64 *)&(info->ctrl_regs.cr3);
+    struct cr3_64 * guest_cr3 = (struct cr3_64 *)&(info->shdw_pg_state.guest_cr3);
+    struct shadow_page_data * shadow_pt = create_new_shadow_pt(info);
+    addr_t shadow_pt_addr = shadow_pt->page_pa;
+
+    // Because this is a new CR3 load the allocated page is the new CR3 value
+    shadow_pt->cr3 = shadow_pt->page_pa;
+
+    PrintDebug("Top level Shadow page pa=%p\n", (void *)shadow_pt_addr);
+
+    shadow_cr3->pml4t_base_addr = PAGE_BASE_ADDR_4KB(shadow_pt_addr);
+    PrintDebug("Creating new 64 bit shadow page table %p\n", (void *)BASE_TO_PAGE_ADDR(shadow_cr3->pml4t_base_addr));
+
+  
+    shadow_cr3->pwt = guest_cr3->pwt;
+    shadow_cr3->pcd = guest_cr3->pcd;
+
+    return 0;
 }
 
 
@@ -18,27 +52,646 @@ static inline int activate_shadow_pt_64(struct guest_info * info) {
  * *
  */
 
+static int handle_2MB_shadow_pagefault_64(struct guest_info * info, addr_t fault_addr, pf_error_t error_code,
+                                         pte64_t * shadow_pt, pde64_2MB_t * large_guest_pde);
+
+static int handle_pte_shadow_pagefault_64(struct guest_info * info, addr_t fault_addr, pf_error_t error_code,
+                                         pte64_t * shadow_pt, pte64_t * guest_pt);
+
+static int handle_pde_shadow_pagefault_64(struct guest_info * info, addr_t fault_addr, pf_error_t error_code,
+                                         pde64_t * shadow_pd, pde64_t * guest_pd);
+
+static int handle_pdpe_shadow_pagefault_64(struct guest_info * info, addr_t fault_addr, pf_error_t error_code,
+                                          pdpe64_t * shadow_pdp, pdpe64_t * guest_pdp);
+
+
 static inline int handle_shadow_pagefault_64(struct guest_info * info, addr_t fault_addr, pf_error_t error_code) {
-  pt_access_status_t guest_access;
-  pt_access_status_t shadow_access;
-  int ret; 
-  PrintDebug("64 bit shadow page fault\n");
+    pml4e64_t * guest_pml = NULL;
+    pml4e64_t * shadow_pml = CR3_TO_PML4E64_VA(info->ctrl_regs.cr3);
+    addr_t guest_cr3 = CR3_TO_PML4E64_PA(info->shdw_pg_state.guest_cr3);
+    pt_access_status_t guest_pml4e_access;
+    pt_access_status_t shadow_pml4e_access;
+    pml4e64_t * guest_pml4e = NULL;
+    pml4e64_t * shadow_pml4e = (pml4e64_t *)&(shadow_pml[PML4E64_INDEX(fault_addr)]);
+
+    PrintDebug("64 bit Shadow page fault handler: %p\n", (void *)fault_addr);
+    PrintDebug("Handling PML fault\n");
 
-  ret = v3_check_guest_pt_32(info, info->shdw_pg_state.guest_cr3, fault_addr, error_code, &guest_access);
+    if (guest_pa_to_host_va(info, guest_cr3, (addr_t*)&guest_pml) == -1) {
+       PrintError("Invalid Guest PML4E Address: 0x%p\n",  (void *)guest_cr3);
+       return -1;
+    } 
 
-  PrintDebug("Guest Access Check: %d (access=%d)\n", ret, guest_access);
+    guest_pml4e = (pml4e64_t *)&(guest_pml[PML4E64_INDEX(fault_addr)]);
 
-  ret = v3_check_host_pt_32(info->ctrl_regs.cr3, fault_addr, error_code, &shadow_access);
+    PrintDebug("Checking Guest %p\n", (void *)guest_pml);
+    // Check the guest page permissions
+    guest_pml4e_access = v3_can_access_pml4e64(guest_pml, fault_addr, error_code);
 
-  PrintDebug("Shadow Access Check: %d (access=%d)\n", ret, shadow_access);
+    PrintDebug("Checking shadow %p\n", (void *)shadow_pml);
+    // Check the shadow page permissions
+    shadow_pml4e_access = v3_can_access_pml4e64(shadow_pml, fault_addr, error_code);
   
+    /* Was the page fault caused by the Guest's page tables? */
+    if (is_guest_pf(guest_pml4e_access, shadow_pml4e_access) == 1) {
+       PrintDebug("Injecting PML4E pf to guest: (guest access error=%d) (pf error code=%d)\n", 
+                  *(uint_t *)&guest_pml4e_access, *(uint_t *)&error_code);
+       if (inject_guest_pf(info, fault_addr, error_code) == -1) {
+           PrintError("Could not inject guest page fault\n");
+           return -1;
+       }
+       return 0;
+    }
+
+    if (shadow_pml4e_access == PT_ACCESS_USER_ERROR) {
+       //
+       // PML4 Entry marked non-user
+       //      
+       PrintDebug("Shadow Paging User access error (shadow_pml4e_access=0x%x, guest_pml4e_access=0x%x)\n", 
+                  shadow_pml4e_access, guest_pml4e_access);
+       if (inject_guest_pf(info, fault_addr, error_code) == -1) {
+           PrintError("Could not inject guest page fault\n");
+           return -1;
+       }
+       return 0;
+    } else if ((shadow_pml4e_access != PT_ACCESS_NOT_PRESENT) &&
+              (shadow_pml4e_access != PT_ACCESS_OK)) {
+       // inject page fault in guest
+       if (inject_guest_pf(info, fault_addr, error_code) == -1) {
+           PrintError("Could not inject guest page fault\n");
+           return -1;
+       }
+       PrintDebug("Unknown Error occurred (shadow_pde_access=%d)\n", shadow_pml4e_access);
+       PrintDebug("Manual Says to inject page fault into guest\n");
+       return 0;
+    }
+
 
-  PrintError("64 bit shadow paging not implemented\n");
-  return -1;
+    pdpe64_t * shadow_pdp = NULL;
+    pdpe64_t * guest_pdp = NULL;
+
+    // Get the next shadow page level, allocate if not present
+
+    if (shadow_pml4e_access == PT_ACCESS_NOT_PRESENT) {
+       struct shadow_page_data * shdw_page =  create_new_shadow_pt(info);
+       shadow_pdp = (pdpe64_t *)V3_VAddr((void *)shdw_page->page_pa);
+
+
+       shadow_pml4e->present = 1;
+       shadow_pml4e->user_page = guest_pml4e->user_page;
+       shadow_pml4e->writable = guest_pml4e->writable;
+       shadow_pml4e->cache_disable = guest_pml4e->cache_disable;
+       shadow_pml4e->write_through = guest_pml4e->write_through;
+    
+       guest_pml4e->accessed = 1;
+    
+       shadow_pml4e->pdp_base_addr = PAGE_BASE_ADDR(shdw_page->page_pa);
+    } else {
+       shadow_pdp = (pdpe64_t *)V3_VAddr((void *)(addr_t)BASE_TO_PAGE_ADDR(shadow_pml4e->pdp_base_addr));
+    }
+
+    // Continue processing at the next level
+
+    if (guest_pa_to_host_va(info, BASE_TO_PAGE_ADDR(guest_pml4e->pdp_base_addr), (addr_t *)&guest_pdp) == -1) {
+       // Machine check the guest
+       PrintError("Invalid Guest PDP Address: 0x%p\n", (void *)BASE_TO_PAGE_ADDR(guest_pml4e->pdp_base_addr));
+       v3_raise_exception(info, MC_EXCEPTION);
+       return 0;
+    }
+  
+    if (handle_pdpe_shadow_pagefault_64(info, fault_addr, error_code, shadow_pdp, guest_pdp) == -1) {
+       PrintError("Error handling Page fault caused by PDPE\n");
+       return -1;
+    }
+
+    return 0;
+}
+
+
+
+// For now we are not going to handle 1 Gigabyte pages
+static int handle_pdpe_shadow_pagefault_64(struct guest_info * info, addr_t fault_addr, pf_error_t error_code,
+                                          pdpe64_t * shadow_pdp, pdpe64_t * guest_pdp) {
+    pt_access_status_t guest_pdpe_access;
+    pt_access_status_t shadow_pdpe_access;
+    pdpe64_t * guest_pdpe = (pdpe64_t *)&(guest_pdp[PDPE64_INDEX(fault_addr)]);
+    pdpe64_t * shadow_pdpe = (pdpe64_t *)&(shadow_pdp[PDPE64_INDEX(fault_addr)]);
+    PrintDebug("Handling PDP fault\n");
+
+    if (fault_addr==0) { 
+       PrintDebug("Guest Page Tree for guest virtual address zero fault\n");
+       PrintGuestPageTree(info,fault_addr,(addr_t)(info->shdw_pg_state.guest_cr3));
+       PrintDebug("Host Page Tree for guest virtual address zero fault\n");
+       PrintHostPageTree(info,fault_addr,(addr_t)(info->ctrl_regs.cr3));
+    }
+
+    // Check the guest page permissions
+    guest_pdpe_access = v3_can_access_pdpe64(guest_pdp, fault_addr, error_code);
+
+    // Check the shadow page permissions
+    shadow_pdpe_access = v3_can_access_pdpe64(shadow_pdp, fault_addr, error_code);
+  
+    /* Was the page fault caused by the Guest's page tables? */
+    if (is_guest_pf(guest_pdpe_access, shadow_pdpe_access) == 1) {
+       PrintDebug("Injecting PDPE pf to guest: (guest access error=%d) (pf error code=%d)\n", 
+                  *(uint_t *)&guest_pdpe_access, *(uint_t *)&error_code);
+       if (inject_guest_pf(info, fault_addr, error_code) == -1) {
+           PrintError("Could not inject guest page fault\n");
+           return -1;
+       }
+       return 0;
+    }
+
+    if (shadow_pdpe_access == PT_ACCESS_USER_ERROR) {
+       //
+       // PML4 Entry marked non-user
+       //      
+       PrintDebug("Shadow Paging User access error (shadow_pdpe_access=0x%x, guest_pdpe_access=0x%x)\n", 
+                  shadow_pdpe_access, guest_pdpe_access);
+       if (inject_guest_pf(info, fault_addr, error_code) == -1) {
+           PrintError("Could not inject guest page fault\n");
+           return -1;
+       }
+       return 0;
+    } else if ((shadow_pdpe_access != PT_ACCESS_NOT_PRESENT) &&
+              (shadow_pdpe_access != PT_ACCESS_OK)) {
+       // inject page fault in guest
+       if (inject_guest_pf(info, fault_addr, error_code) == -1) {
+           PrintError("Could not inject guest page fault\n");
+           return -1;
+       }
+       PrintDebug("Unknown Error occurred (shadow_pde_access=%d)\n", shadow_pdpe_access);
+       PrintDebug("Manual Says to inject page fault into guest\n");
+       return 0;
+    }
+
+
+    pde64_t * shadow_pd = NULL;
+    pde64_t * guest_pd = NULL;
+
+    // Get the next shadow page level, allocate if not present
+
+    if (shadow_pdpe_access == PT_ACCESS_NOT_PRESENT) {
+       struct shadow_page_data * shdw_page = create_new_shadow_pt(info);
+       shadow_pd = (pde64_t *)V3_VAddr((void *)shdw_page->page_pa);
+
+
+       shadow_pdpe->present = 1;
+       shadow_pdpe->user_page = guest_pdpe->user_page;
+       shadow_pdpe->writable = guest_pdpe->writable;
+       shadow_pdpe->write_through = guest_pdpe->write_through;
+       shadow_pdpe->cache_disable = guest_pdpe->cache_disable;
+
+    
+       guest_pdpe->accessed = 1;
+    
+       shadow_pdpe->pd_base_addr = PAGE_BASE_ADDR(shdw_page->page_pa);
+    } else {
+       shadow_pd = (pde64_t *)V3_VAddr((void *)(addr_t)BASE_TO_PAGE_ADDR(shadow_pdpe->pd_base_addr));
+    }
+
+    // Continue processing at the next level
+
+    if (guest_pa_to_host_va(info, BASE_TO_PAGE_ADDR(guest_pdpe->pd_base_addr), (addr_t *)&guest_pd) == -1) {
+       // Machine check the guest
+       PrintError("Invalid Guest PTE Address: 0x%p\n", (void *)BASE_TO_PAGE_ADDR(guest_pdpe->pd_base_addr));
+       v3_raise_exception(info, MC_EXCEPTION);
+       return 0;
+    }
+  
+    if (handle_pde_shadow_pagefault_64(info, fault_addr, error_code, shadow_pd, guest_pd) == -1) {
+       PrintError("Error handling Page fault caused by PDE\n");
+       return -1;
+    }
+
+    return 0;
+}
+
+
+static int handle_pde_shadow_pagefault_64(struct guest_info * info, addr_t fault_addr, pf_error_t error_code,
+                                         pde64_t * shadow_pd, pde64_t * guest_pd) {
+    pt_access_status_t guest_pde_access;
+    pt_access_status_t shadow_pde_access;
+    pde64_t * guest_pde = (pde64_t *)&(guest_pd[PDE64_INDEX(fault_addr)]);
+    pde64_t * shadow_pde = (pde64_t *)&(shadow_pd[PDE64_INDEX(fault_addr)]);
+
+    PrintDebug("Handling PDE fault\n");
+    // Check the guest page permissions
+    guest_pde_access = v3_can_access_pde64(guest_pd, fault_addr, error_code);
+
+    // Check the shadow page permissions
+    shadow_pde_access = v3_can_access_pde64(shadow_pd, fault_addr, error_code);
+  
+    /* Was the page fault caused by the Guest's page tables? */
+    if (is_guest_pf(guest_pde_access, shadow_pde_access) == 1) {
+       PrintDebug("Injecting PDE pf to guest: (guest access error=%d) (pf error code=%d)\n", 
+                  *(uint_t *)&guest_pde_access, *(uint_t *)&error_code);
+       if (inject_guest_pf(info, fault_addr, error_code) == -1) {
+           PrintError("Could not inject guest page fault\n");
+           return -1;
+       }
+       return 0;
+    }
+
+    if (shadow_pde_access == PT_ACCESS_USER_ERROR) {
+       //
+       // PDE Entry marked non-user
+       //      
+       PrintDebug("Shadow Paging User access error (shadow_pdpe_access=0x%x, guest_pdpe_access=0x%x)\n", 
+                  shadow_pde_access, guest_pde_access);
+       if (inject_guest_pf(info, fault_addr, error_code) == -1) {
+           PrintError("Could not inject guest page fault\n");
+           return -1;
+       }
+       return 0;
+
+    } else if ((shadow_pde_access == PT_ACCESS_WRITE_ERROR) && 
+              (guest_pde->large_page == 1)) {
+
+       ((pde64_2MB_t *)guest_pde)->dirty = 1;
+       shadow_pde->writable = guest_pde->writable;
+
+       //PrintDebug("Returning due to large page Write Error\n");
+       //PrintHostPageTree(info, fault_addr, info->ctrl_regs.cr3);
+
+       return 0;
+    } else if ((shadow_pde_access != PT_ACCESS_NOT_PRESENT) &&
+              (shadow_pde_access != PT_ACCESS_OK)) {
+       // inject page fault in guest
+       if (inject_guest_pf(info, fault_addr, error_code) == -1) {
+           PrintError("Could not inject guest page fault\n");
+           return -1;
+       }
+       PrintDebug("Unknown Error occurred (shadow_pde_access=%d)\n", shadow_pde_access);
+       PrintDebug("Manual Says to inject page fault into guest\n");
+       return 0;
+    }
+
+
+    pte64_t * shadow_pt = NULL;
+    pte64_t * guest_pt = NULL;
+
+    // Get the next shadow page level, allocate if not present
+
+    if (shadow_pde_access == PT_ACCESS_NOT_PRESENT) {
+       struct shadow_page_data * shdw_page = create_new_shadow_pt(info);
+       shadow_pt = (pte64_t *)V3_VAddr((void *)shdw_page->page_pa);
+
+       PrintDebug("Creating new shadow PT: %p\n", shadow_pt);
+
+       shadow_pde->present = 1;
+       shadow_pde->user_page = guest_pde->user_page;
+
+
+       if (guest_pde->large_page == 0) {
+           shadow_pde->writable = guest_pde->writable;
+       } else {
+           // This large page flag is temporary until we can get a working cache....
+           ((pde64_2MB_t *)guest_pde)->vmm_info = V3_LARGE_PG;
+
+           if (error_code.write) {
+               shadow_pde->writable = guest_pde->writable;
+               ((pde64_2MB_t *)guest_pde)->dirty = 1;  
+           } else {
+               shadow_pde->writable = 0;
+               ((pde64_2MB_t *)guest_pde)->dirty = 0;
+           }
+       }
+    
+       // VMM Specific options
+       shadow_pde->write_through = guest_pde->write_through;
+       shadow_pde->cache_disable = guest_pde->cache_disable;
+       shadow_pde->global_page = guest_pde->global_page;
+       //
+    
+       guest_pde->accessed = 1;
+    
+       shadow_pde->pt_base_addr = PAGE_BASE_ADDR(shdw_page->page_pa);
+    } else {
+       shadow_pt = (pte64_t *)V3_VAddr((void *)BASE_TO_PAGE_ADDR(shadow_pde->pt_base_addr));
+    }
+
+    // Continue processing at the next level
+    if (guest_pde->large_page == 0) {
+       if (guest_pa_to_host_va(info, BASE_TO_PAGE_ADDR(guest_pde->pt_base_addr), (addr_t *)&guest_pt) == -1) {
+           // Machine check the guest
+           PrintError("Invalid Guest PTE Address: 0x%p\n", (void *)BASE_TO_PAGE_ADDR(guest_pde->pt_base_addr));
+           v3_raise_exception(info, MC_EXCEPTION);
+           return 0;
+       }
+    
+       if (handle_pte_shadow_pagefault_64(info, fault_addr, error_code, shadow_pt, guest_pt) == -1) {
+           PrintError("Error handling Page fault caused by PDE\n");
+           return -1;
+       }
+    } else {
+       if (handle_2MB_shadow_pagefault_64(info, fault_addr, error_code, shadow_pt, (pde64_2MB_t *)guest_pde) == -1) {
+           PrintError("Error handling large pagefault\n");
+           return -1;
+       } 
+    }
+
+    return 0;
+}
+
+
+static int handle_pte_shadow_pagefault_64(struct guest_info * info, addr_t fault_addr, pf_error_t error_code,
+                                         pte64_t * shadow_pt, pte64_t * guest_pt) {
+    pt_access_status_t guest_pte_access;
+    pt_access_status_t shadow_pte_access;
+    pte64_t * guest_pte = (pte64_t *)&(guest_pt[PTE64_INDEX(fault_addr)]);;
+    pte64_t * shadow_pte = (pte64_t *)&(shadow_pt[PTE64_INDEX(fault_addr)]);
+    addr_t guest_pa = BASE_TO_PAGE_ADDR((addr_t)(guest_pte->page_base_addr)) +  PAGE_OFFSET(fault_addr);
+    //  struct shadow_page_state * state = &(info->shdw_pg_state);
+
+    PrintDebug("Handling PTE fault\n");
+
+    struct v3_shadow_region * shdw_reg =  v3_get_shadow_region(info, guest_pa);
+
+
+
+    if (shdw_reg == NULL) {
+       // Inject a machine check in the guest
+       PrintError("Invalid Guest Address in page table (0x%p)\n", (void *)guest_pa);
+       v3_raise_exception(info, MC_EXCEPTION);
+       return 0;
+    }
+
+    // Check the guest page permissions
+    guest_pte_access = v3_can_access_pte64(guest_pt, fault_addr, error_code);
+
+    // Check the shadow page permissions
+    shadow_pte_access = v3_can_access_pte64(shadow_pt, fault_addr, error_code);
+
+    /* Was the page fault caused by the Guest's page tables? */
+    if (is_guest_pf(guest_pte_access, shadow_pte_access) == 1) {
+
+       PrintDebug("Access error injecting pf to guest (guest access error=%d) (pf error code=%d)\n", 
+                  guest_pte_access, *(uint_t*)&error_code);    
+
+       if (inject_guest_pf(info, fault_addr, error_code) == -1) {
+           PrintError("Could not inject guest page fault\n");
+           return -1;
+       }
+
+       return 0; 
+    }
+
+    if (shadow_pte_access == PT_ACCESS_OK) {
+       // Inconsistent state...
+       // Guest Re-Entry will flush page tables and everything should now work
+       PrintDebug("Inconsistent state... Guest re-entry should flush tlb\n");
+       return 0;
+    }
+
+
+    if (shadow_pte_access == PT_ACCESS_NOT_PRESENT) {
+       // Page Table Entry Not Present
+       PrintDebug("guest_pa =%p\n", (void *)guest_pa);
+
+       if ((shdw_reg->host_type == SHDW_REGION_ALLOCATED) ||
+           (shdw_reg->host_type == SHDW_REGION_WRITE_HOOK)) {
+           addr_t shadow_pa = v3_get_shadow_addr(shdw_reg, guest_pa);
+      
+           shadow_pte->page_base_addr = PAGE_BASE_ADDR(shadow_pa);
+      
+           shadow_pte->present = guest_pte->present;
+           shadow_pte->user_page = guest_pte->user_page;
+      
+           //set according to VMM policy
+           shadow_pte->write_through = guest_pte->write_through;
+           shadow_pte->cache_disable = guest_pte->cache_disable;
+           shadow_pte->global_page = guest_pte->global_page;
+           //
+      
+           guest_pte->accessed = 1;
+      
+           if (guest_pte->dirty == 1) {
+               shadow_pte->writable = guest_pte->writable;
+           } else if ((guest_pte->dirty == 0) && (error_code.write == 1)) {
+               shadow_pte->writable = guest_pte->writable;
+               guest_pte->dirty = 1;
+           } else if ((guest_pte->dirty == 0) && (error_code.write == 0)) {
+               shadow_pte->writable = 0;
+           }
+
+
+           // Write hooks trump all, and are set Read Only
+           if (shdw_reg->host_type == SHDW_REGION_WRITE_HOOK) {
+               shadow_pte->writable = 0;
+           }
+
+       } else {
+           // Page fault handled by hook functions
+
+           if (v3_handle_mem_full_hook(info, fault_addr, guest_pa, shdw_reg, error_code) == -1) {
+               PrintError("Special Page fault handler returned error for address: %p\n",  (void *)fault_addr);
+               return -1;
+           }
+       }
+    } else if (shadow_pte_access == PT_ACCESS_WRITE_ERROR) {
+       guest_pte->dirty = 1;
+
+       if (shdw_reg->host_type == SHDW_REGION_WRITE_HOOK) {
+           if (v3_handle_mem_wr_hook(info, fault_addr, guest_pa, shdw_reg, error_code) == -1) {
+               PrintError("Special Page fault handler returned error for address: %p\n",  (void *)fault_addr);
+               return -1;
+           }
+       } else {
+           PrintDebug("Shadow PTE Write Error\n");
+           shadow_pte->writable = guest_pte->writable;
+       }
+
+
+       return 0;
+
+    } else {
+       // Inject page fault into the guest     
+       if (inject_guest_pf(info, fault_addr, error_code) == -1) {
+           PrintError("Could not inject guest page fault\n");
+           return -1;
+       }
+       PrintError("PTE Page fault fell through... Not sure if this should ever happen\n");
+       PrintError("Manual Says to inject page fault into guest\n");
+       return -1;
+    }
+
+    return 0;
+}
+
+
+
+static int handle_2MB_shadow_pagefault_64(struct guest_info * info, 
+                                         addr_t fault_addr, pf_error_t error_code, 
+                                         pte64_t * shadow_pt, pde64_2MB_t * large_guest_pde) 
+{
+    pt_access_status_t shadow_pte_access = v3_can_access_pte64(shadow_pt, fault_addr, error_code);
+    pte64_t * shadow_pte = (pte64_t *)&(shadow_pt[PTE64_INDEX(fault_addr)]);
+    addr_t guest_fault_pa = BASE_TO_PAGE_ADDR_2MB(large_guest_pde->page_base_addr) + PAGE_OFFSET_2MB(fault_addr);
+    //  struct shadow_page_state * state = &(info->shdw_pg_state);
+
+    PrintDebug("Handling 2MB fault (guest_fault_pa=%p) (error_code=%x)\n", (void *)guest_fault_pa, *(uint_t*)&error_code);
+    PrintDebug("ShadowPT=%p, LargeGuestPDE=%p\n", shadow_pt, large_guest_pde);
+
+    struct v3_shadow_region * shdw_reg = v3_get_shadow_region(info, guest_fault_pa);
+
+    if (shdw_reg == NULL) {
+       // Inject a machine check in the guest
+       PrintError("Invalid Guest Address in page table (0x%p)\n", (void *)guest_fault_pa);
+       v3_raise_exception(info, MC_EXCEPTION);
+       return 0;
+    }
+
+    if (shadow_pte_access == PT_ACCESS_OK) {
+       // Inconsistent state...
+       // Guest Re-Entry will flush tables and everything should now workd
+       PrintDebug("Inconsistent state... Guest re-entry should flush tlb\n");
+       //PrintHostPageTree(info, fault_addr, info->ctrl_regs.cr3);
+       return 0;
+    }
+
+  
+    if (shadow_pte_access == PT_ACCESS_NOT_PRESENT) {
+       // Get the guest physical address of the fault
+
+       if ((shdw_reg->host_type == SHDW_REGION_ALLOCATED) || 
+           (shdw_reg->host_type == SHDW_REGION_WRITE_HOOK)) {
+           addr_t shadow_pa = v3_get_shadow_addr(shdw_reg, guest_fault_pa);
+
+           shadow_pte->page_base_addr = PAGE_BASE_ADDR(shadow_pa);
+
+           shadow_pte->present = 1;
+
+           /* We are assuming that the PDE entry has precedence
+            * so the Shadow PDE will mirror the guest PDE settings, 
+            * and we don't have to worry about them here
+            * Allow everything
+            */
+           shadow_pte->user_page = 1;
+
+           if (shdw_reg->host_type == SHDW_REGION_WRITE_HOOK) {
+               shadow_pte->writable = 0;
+           } else {
+               shadow_pte->writable = 1;
+           }
+
+           //set according to VMM policy
+           shadow_pte->write_through = large_guest_pde->write_through;
+           shadow_pte->cache_disable = large_guest_pde->cache_disable;
+           shadow_pte->global_page = large_guest_pde->global_page;
+           //
+      
+       } else {
+           if (v3_handle_mem_full_hook(info, fault_addr, guest_fault_pa, shdw_reg, error_code) == -1) {
+               PrintError("Special Page Fault handler returned error for address: %p\n", (void *)fault_addr);
+               return -1;
+           }
+       }
+    } else if (shadow_pte_access == PT_ACCESS_WRITE_ERROR) {
+
+       if (shdw_reg->host_type == SHDW_REGION_WRITE_HOOK) {
+
+           if (v3_handle_mem_wr_hook(info, fault_addr, guest_fault_pa, shdw_reg, error_code) == -1) {
+               PrintError("Special Page Fault handler returned error for address: %p\n", (void *)fault_addr);
+               return -1;
+           }
+       }
+
+    } else {
+       PrintError("Error in large page fault handler...\n");
+       PrintError("This case should have been handled at the top level handler\n");
+       return -1;
+    }
+
+    //  PrintHostPageTree(info, fault_addr, info->ctrl_regs.cr3);
+    PrintDebug("Returning from large page fault handler\n");
+    return 0;
+}
+
+
+
+
+static int invalidation_cb_64(struct guest_info * info, page_type_t type, 
+                             addr_t vaddr, addr_t page_ptr, addr_t page_pa, 
+                             void * private_data) {
+
+    switch (type) {
+       case PAGE_PML464:
+           {    
+               pml4e64_t * pml = (pml4e64_t *)page_ptr;
+
+               if (pml[PML4E64_INDEX(vaddr)].present == 0) {
+                   return 1;
+               }
+               return 0;
+           }
+       case PAGE_PDP64:
+           {
+               pdpe64_t * pdp = (pdpe64_t *)page_ptr;
+               pdpe64_t * pdpe = &(pdp[PDPE64_INDEX(vaddr)]);
+
+               if (pdpe->present == 0) {
+                   return 1;
+               }
+     
+               if (pdpe->vmm_info == V3_LARGE_PG) {
+                   PrintError("1 Gigabyte pages not supported\n");
+                   return -1;
+
+                   pdpe->present = 0;
+                   return 1;
+               }
+
+               return 0;
+           }
+       case PAGE_PD64:
+           {
+               pde64_t * pd = (pde64_t *)page_ptr;
+               pde64_t * pde = &(pd[PDE64_INDEX(vaddr)]);
+
+               if (pde->present == 0) {
+                   return 1;
+               }
+      
+               if (pde->vmm_info == V3_LARGE_PG) {
+                   pde->present = 0;
+                   return 1;
+               }
+
+               return 0;
+           }
+       case PAGE_PT64:
+           {
+               pte64_t * pt = (pte64_t *)page_ptr;
+
+               pt[PTE64_INDEX(vaddr)].present = 0;
+
+               return 1;
+           }
+       default:
+           PrintError("Invalid Page Type\n");
+           return -1;
+
+    }
+
+    // should not get here
+    PrintError("Should not get here....\n");
+    return -1;
 }
 
 
 static inline int handle_shadow_invlpg_64(struct guest_info * info, addr_t vaddr) {
-  PrintError("64 bit shadow paging not implemented\n");
-  return -1;
+    PrintDebug("INVLPG64 - %p\n",(void*)vaddr);
+
+    int ret =  v3_drill_host_pt_64(info, info->ctrl_regs.cr3, vaddr, invalidation_cb_64, NULL);
+    if (ret == -1) {
+       PrintError("Page table drill returned error.... \n");
+       PrintHostPageTree(info, vaddr, info->ctrl_regs.cr3);
+    }
+
+    return (ret == -1) ? -1 : 0; 
 }