static void unmap_area_sections(unsigned long virt, unsigned long size)
{
unsigned long addr = virt, end = virt + (size & ~(SZ_1M - 1));
pgd_t *pgd;
pud_t *pud;
pmd_t *pmdp;
flush_cache_vunmap(addr, end);
pgd = pgd_offset_k(addr);
pud = pud_offset(pgd, addr);
pmdp = pmd_offset(pud, addr);
do {
pmd_t pmd = *pmdp;
if (!pmd_none(pmd)) {
pmd_clear(pmdp);
init_mm.context.kvm_seq++;
if ((pmd_val(pmd) & PMD_TYPE_MASK) == PMD_TYPE_TABLE)
pte_free_kernel(&init_mm, pmd_page_vaddr(pmd));
}
addr += PMD_SIZE;
pmdp += 2;
} while (addr < end);
if (current->active_mm->context.kvm_seq != init_mm.context.kvm_seq)
__check_kvm_seq(current->active_mm);
flush_tlb_kernel_range(virt, end);
}
/*
* Section support is unsafe on SMP - If you iounmap and ioremap a region,
* the other CPUs will not see this change until their next context switch.
* Meanwhile, (eg) if an interrupt comes in on one of those other CPUs
* which requires the new ioremap'd region to be referenced, the CPU will
* reference the _old_ region.
*
* Note that get_vm_area_caller() allocates a guard 4K page, so we need to
* mask the size back to 4MB aligned or we will overflow in the loop below.
*/
static void unmap_area_sections(unsigned long virt, unsigned long size)
{
unsigned long addr = virt, end = virt + (size & ~(SZ_4M - 1));
pgd_t *pgd;
flush_cache_vunmap(addr, end);
pgd = pgd_offset_k(addr);
do {
pmd_t pmd, *pmdp = pmd_offset((pud_t *)pgd, addr);
pmd = *pmdp;
if (!pmd_none(pmd)) {
/*
* Clear the PMD from the page table, and
* increment the kvm sequence so others
* notice this change.
*
* Note: this is still racy on SMP machines.
*/
pmd_clear(pmdp);
/*
* Free the page table, if there was one.
*/
if ((pmd_val(pmd) & PMD_TYPE_MASK) == PMD_TYPE_TABLE)
pte_free_kernel(&init_mm, pmd_page_vaddr(pmd));
}
addr += PGDIR_SIZE;
pgd++;
} while (addr < end);
flush_tlb_kernel_range(virt, end);
}
/*
* Place a pointer to an L2 page table in a middle page
* directory entry.
*/
static void __init assign_pte(pmd_t *pmd, pte_t *page_table)
{
phys_addr_t pa = __pa(page_table);
unsigned long l2_ptfn = pa >> HV_LOG2_PAGE_TABLE_ALIGN;
pte_t pteval = hv_pte_set_ptfn(__pgprot(_PAGE_TABLE), l2_ptfn);
BUG_ON((pa & (HV_PAGE_TABLE_ALIGN-1)) != 0);
pteval = pte_set_home(pteval, initial_heap_home());
*(pte_t *)pmd = pteval;
if (page_table != (pte_t *)pmd_page_vaddr(*pmd))
BUG();
}
static int init_stub_pte(struct mm_struct *mm, unsigned long proc,
unsigned long kernel)
{
pgd_t *pgd;
pud_t *pud;
pmd_t *pmd;
pte_t *pte;
pgd = pgd_offset(mm, proc);
pud = pud_alloc(mm, pgd, proc);
if (!pud)
goto out;
pmd = pmd_alloc(mm, pud, proc);
if (!pmd)
goto out_pmd;
pte = pte_alloc_map(mm, pmd, proc);
if (!pte)
goto out_pte;
/* There's an interaction between the skas0 stub pages, stack
* randomization, and the BUG at the end of exit_mmap. exit_mmap
* checks that the number of page tables freed is the same as had
* been allocated. If the stack is on the last page table page,
* then the stack pte page will be freed, and if not, it won't. To
* avoid having to know where the stack is, or if the process mapped
* something at the top of its address space for some other reason,
* we set TASK_SIZE to end at the start of the last page table.
* This keeps exit_mmap off the last page, but introduces a leak
* of that page. So, we hang onto it here and free it in
* destroy_context_skas.
*/
mm->context.skas.last_page_table = pmd_page_vaddr(*pmd);
#ifdef CONFIG_3_LEVEL_PGTABLES
mm->context.skas.last_pmd = (unsigned long) __va(pud_val(*pud));
#endif
*pte = mk_pte(virt_to_page(kernel), __pgprot(_PAGE_PRESENT));
*pte = pte_mkexec(*pte);
*pte = pte_wrprotect(*pte);
return(0);
out_pmd:
pud_free(pud);
out_pte:
pmd_free(pmd);
out:
return(-ENOMEM);
}
static void walk_pte_level(struct seq_file *m, struct pg_state *st, pmd_t addr,
unsigned long P)
{
int i;
pte_t *start;
start = (pte_t *) pmd_page_vaddr(addr);
for (i = 0; i < PTRS_PER_PTE; i++) {
pgprot_t prot = pte_pgprot(*start);
st->current_address = normalize_addr(P + i * PTE_LEVEL_MULT);
note_page(m, st, prot, 4);
start++;
}
}
pte_t *huge_pte_offset(struct mm_struct *mm, unsigned long addr)
{
pgd_t *pgd;
pud_t *pud;
pmd_t *pmd;
#ifdef CONFIG_HUGETLB_SUPER_PAGES
pte_t *pte;
#endif
/* Get the top-level page table entry. */
pgd = (pgd_t *)get_pte((pte_t *)mm->pgd, pgd_index(addr), 0);
/* We don't have four levels. */
pud = pud_offset(pgd, addr);
#ifndef __PAGETABLE_PUD_FOLDED
# error support fourth page table level
#endif
if (!pud_present(*pud))
return NULL;
/* Check for an L0 huge PTE, if we have three levels. */
#ifndef __PAGETABLE_PMD_FOLDED
if (pud_huge(*pud))
return (pte_t *)pud;
pmd = (pmd_t *)get_pte((pte_t *)pud_page_vaddr(*pud),
pmd_index(addr), 1);
if (!pmd_present(*pmd))
return NULL;
#else
pmd = pmd_offset(pud, addr);
#endif
/* Check for an L1 huge PTE. */
if (pmd_huge(*pmd))
return (pte_t *)pmd;
#ifdef CONFIG_HUGETLB_SUPER_PAGES
/* Check for an L2 huge PTE. */
pte = get_pte((pte_t *)pmd_page_vaddr(*pmd), pte_index(addr), 2);
if (!pte_present(*pte))
return NULL;
if (pte_super(*pte))
return pte;
#endif
return NULL;
}
/*
* Section support is unsafe on SMP - If you iounmap and ioremap a region,
* the other CPUs will not see this change until their next context switch.
* Meanwhile, (eg) if an interrupt comes in on one of those other CPUs
* which requires the new ioremap'd region to be referenced, the CPU will
* reference the _old_ region.
*
* Note that get_vm_area_caller() allocates a guard 4K page, so we need to
* mask the size back to 1MB aligned or we will overflow in the loop below.
*/
static void unmap_area_sections(unsigned long virt, unsigned long size)
{
unsigned long addr = virt, end = virt + (size & ~(SZ_1M - 1));
pgd_t *pgd;
pud_t *pud;
pmd_t *pmdp;
flush_cache_vunmap(addr, end);
pgd = pgd_offset_k(addr);
pud = pud_offset(pgd, addr);
pmdp = pmd_offset(pud, addr);
do {
pmd_t pmd = *pmdp;
if (!pmd_none(pmd)) {
/*
* Clear the PMD from the page table, and
* increment the kvm sequence so others
* notice this change.
*
* Note: this is still racy on SMP machines.
*/
pmd_clear(pmdp);
init_mm.context.kvm_seq++;
/*
* Free the page table, if there was one.
*/
if ((pmd_val(pmd) & PMD_TYPE_MASK) == PMD_TYPE_TABLE)
pte_free_kernel(&init_mm, pmd_page_vaddr(pmd));
}
addr += PMD_SIZE;
pmdp += 2;
} while (addr < end);
/*
* Ensure that the active_mm is up to date - we want to
* catch any use-after-iounmap cases.
*/
if (current->active_mm->context.kvm_seq != init_mm.context.kvm_seq)
__check_kvm_seq(current->active_mm);
flush_tlb_kernel_range(virt, end);
}
int exit_mm(struct mm_struct *mm)
{
pmd_t *pmd;
pgd_t *pgd;
uint32_t pgdno, pmdno;
physaddr_t pa;
struct vm_area_struct* vma = mm->mmap;
struct page *page;
if(!mm || !mm->mm_pgd)
return 0;
if(!atomic_dec_and_test(&mm->mm_count))
return 0;
delete_all_vma(mm);
for (pgdno = 0; pgdno < pgd_index(KERNEL_BASE_ADDR); pgdno++) {
pgd = mm->mm_pgd + pgdno;
if(!pgd_present(*pgd) || pgd_none(*pgd))
continue;
pmd_t* tmp = (pmd_t *)pgd_page_vaddr(*pgd);
for (pmdno = 0; pmdno < PTRS_PER_PMD; pmdno++) {
pmd = tmp + pmdno;
if(!pmd_present(*pmd) || pmd_none(*pmd))
continue;
struct page* p = virt2page(pmd_page_vaddr(*pmd));
page_decref(p);
pmd_set(pmd,0,0);
}
struct page* p = virt2page(pgd_page_vaddr(*pgd));
page_decref(p);
pgd_set(pgd,0,0);
}
page = virt2page((viraddr_t)mm->mm_pgd);
page_free(page);
kfree(mm);
return 0;
}
static void unmap_area_sections(unsigned long virt, unsigned long size)
{
unsigned long addr = virt, end = virt + (size & ~(SZ_4M - 1));
pgd_t *pgd;
flush_cache_vunmap(addr, end);
pgd = pgd_offset_k(addr);
do {
pmd_t pmd, *pmdp = pmd_offset((pud_t *)pgd, addr);
pmd = *pmdp;
if (!pmd_none(pmd)) {
pmd_clear(pmdp);
if ((pmd_val(pmd) & PMD_TYPE_MASK) == PMD_TYPE_TABLE)
pte_free_kernel(&init_mm, pmd_page_vaddr(pmd));
}
addr += PGDIR_SIZE;
pgd++;
} while (addr < end);
flush_tlb_kernel_range(virt, end);
}
/*
* For hugepage we have pfn in the pmd, we use PTE_RPN_SHIFT bits for flags
* For PTE page, we have a PTE_FRAG_SIZE (4K) aligned virtual address.
*/
struct page *pmd_page(pmd_t pmd)
{
if (pmd_trans_huge(pmd) || pmd_huge(pmd) || pmd_devmap(pmd))
return pte_page(pmd_pte(pmd));
return virt_to_page(pmd_page_vaddr(pmd));
}
pte_t *my_pte_offset_kernel(pmd_t *pmd, unsigned long address)
{
return (pte_t *)pmd_page_vaddr(*pmd) + my_pte_index(address);
}