static void build_page_tables(pgentry_t *top_table,
unsigned long *pt_pfn, void *start_va, void *end_va)
{
//
// *pt_pfn - unused pages already mapped by domain builder (512K+)
// carve new page tables/directories from *pt_pfn as needed
// start_va - end_va - range of frames to map
//
struct mmu_update mmu_updates[L1_PAGETABLE_ENTRIES];
int count = 0;
while (start_va + PAGE_SIZE <= end_va)
{
pgentry_t *tab = top_table;
unsigned int offset;
pgentry_t pte;
#if defined(__x86_64__)
offset = l4_table_offset((unsigned long)start_va);
pte = tab[offset];
if ((pte & _PAGE_PRESENT) == 0)
pte = new_pt_page(pt_pfn, tab, offset, L3_FRAME);
tab = pte_to_virt(pte);
#endif
offset = l3_table_offset((unsigned long)start_va);
pte = tab[offset];
if ((pte & _PAGE_PRESENT) == 0)
pte = new_pt_page(pt_pfn, tab, offset, L2_FRAME);
tab = pte_to_virt(pte);
offset = l2_table_offset((unsigned long)start_va);
pte = tab[offset];
if ((pte & _PAGE_PRESENT) == 0)
pte = new_pt_page(pt_pfn, tab, offset, L1_FRAME);
tab = pte_to_virt(pte);
offset = l1_table_offset((unsigned long)start_va);
pte = tab[offset];
if ((pte & _PAGE_PRESENT) == 0)
{
unsigned long pt_mfn = virt_to_mfn(tab);
mmu_updates[count].ptr = ((pgentry_t)pt_mfn << PAGE_SHIFT) + sizeof(pgentry_t)*offset;
mmu_updates[count].val = virt_to_mfn(start_va) << PAGE_SHIFT | L1_PROT;
count++;
}
start_va += PAGE_SIZE;
if (count == L1_PAGETABLE_ENTRIES || start_va + PAGE_SIZE > end_va)
{
int rc = HYPERVISOR_mmu_update(mmu_updates, count, 0, DOMID_SELF);
if (rc < 0)
fatal_error("build_page_tables: mmu_update failed: %d", rc);
count = 0;
}
}
}
void page_walk(unsigned long virt_address)
{
pgentry_t *tab = (pgentry_t *)start_info.pt_base, page;
unsigned long addr = virt_address;
printk("Pagetable walk from virt %lx, base %lx:\n", virt_address, start_info.pt_base);
page = tab[l4_table_offset(addr)];
tab = pte_to_virt(page);
printk(" L4 = %"PRIpte" (%p) [offset = %lx]\n", page, tab, l4_table_offset(addr));
page = tab[l3_table_offset(addr)];
tab = pte_to_virt(page);
printk(" L3 = %"PRIpte" (%p) [offset = %lx]\n", page, tab, l3_table_offset(addr));
page = tab[l2_table_offset(addr)];
tab = pte_to_virt(page);
printk(" L2 = %"PRIpte" (%p) [offset = %lx]\n", page, tab, l2_table_offset(addr));
page = tab[l1_table_offset(addr)];
printk(" L1 = %"PRIpte" [offset = %lx]\n", page, l1_table_offset(addr));
}
static int handle_cow(unsigned long addr) {
pgentry_t *tab = (pgentry_t *)start_info.pt_base, page;
unsigned long new_page;
int rc;
page = tab[l4_table_offset(addr)];
if (!(page & _PAGE_PRESENT))
return 0;
tab = pte_to_virt(page);
page = tab[l3_table_offset(addr)];
if (!(page & _PAGE_PRESENT))
return 0;
tab = pte_to_virt(page);
page = tab[l2_table_offset(addr)];
if (!(page & _PAGE_PRESENT))
return 0;
tab = pte_to_virt(page);
page = tab[l1_table_offset(addr)];
if (!(page & _PAGE_PRESENT))
return 0;
/* Only support CoW for the zero page. */
if (PHYS_PFN(page) != mfn_zero)
return 0;
new_page = alloc_pages(0);
memset((void*) new_page, 0, PAGE_SIZE);
rc = HYPERVISOR_update_va_mapping(addr & PAGE_MASK, __pte(virt_to_mach(new_page) | L1_PROT), UVMF_INVLPG);
if (!rc)
return 1;
printk("Map zero page to %lx failed: %d.\n", addr, rc);
return 0;
}
/*
* Make pt_pfn a new 'level' page table frame and hook it into the page
* table at offset in previous level MFN (pref_l_mfn). pt_pfn is a guest
* PFN.
*/
static void new_pt_frame(unsigned long *pt_pfn, unsigned long prev_l_mfn,
unsigned long offset, unsigned long level)
{
pgentry_t *tab = (pgentry_t *)start_info.pt_base;
unsigned long pt_page = (unsigned long)pfn_to_virt(*pt_pfn);
pgentry_t prot_e, prot_t;
mmu_update_t mmu_updates[1];
int rc;
prot_e = prot_t = 0;
DEBUG("Allocating new L%d pt frame for pfn=%lx, "
"prev_l_mfn=%lx, offset=%lx",
level, *pt_pfn, prev_l_mfn, offset);
/* We need to clear the page, otherwise we might fail to map it
as a page table page */
memset((void*) pt_page, 0, PAGE_SIZE);
switch ( level )
{
case L1_FRAME:
prot_e = L1_PROT;
prot_t = L2_PROT;
break;
case L2_FRAME:
prot_e = L2_PROT;
prot_t = L3_PROT;
break;
#if defined(__x86_64__)
case L3_FRAME:
prot_e = L3_PROT;
prot_t = L4_PROT;
break;
#endif
default:
printk("new_pt_frame() called with invalid level number %d\n", level);
do_exit();
break;
}
/* Make PFN a page table page */
#if defined(__x86_64__)
tab = pte_to_virt(tab[l4_table_offset(pt_page)]);
#endif
tab = pte_to_virt(tab[l3_table_offset(pt_page)]);
mmu_updates[0].ptr = (tab[l2_table_offset(pt_page)] & PAGE_MASK) +
sizeof(pgentry_t) * l1_table_offset(pt_page);
mmu_updates[0].val = (pgentry_t)pfn_to_mfn(*pt_pfn) << PAGE_SHIFT |
(prot_e & ~_PAGE_RW);
if ( (rc = HYPERVISOR_mmu_update(mmu_updates, 1, NULL, DOMID_SELF)) < 0 )
{
printk("ERROR: PTE for new page table page could not be updated\n");
printk(" mmu_update failed with rc=%d\n", rc);
do_exit();
}
/* Hook the new page table page into the hierarchy */
mmu_updates[0].ptr =
((pgentry_t)prev_l_mfn << PAGE_SHIFT) + sizeof(pgentry_t) * offset;
mmu_updates[0].val = (pgentry_t)pfn_to_mfn(*pt_pfn) << PAGE_SHIFT | prot_t;
if ( (rc = HYPERVISOR_mmu_update(mmu_updates, 1, NULL, DOMID_SELF)) < 0 )
{
printk("ERROR: mmu_update failed with rc=%d\n", rc);
do_exit();
}
*pt_pfn += 1;
}
