unsigned long l4x_set_pte(struct mm_struct *mm,
unsigned long addr,
pte_t old, pte_t pteval)
{
/*
* Check if any invalidation is necessary
*
* Invalidation (flush) necessary if:
* old page was present
* new page is not present OR
* new page has another physical address OR
* new page has another protection OR
* new page has other access attributes
*/
/* old was present && new not -> flush */
int flush_rights = L4_FPAGE_RWX;
#if 0
if ((pte_val(old) & PAGE_MASK) != (pte_val(pteval) & PAGE_MASK))
printk("spte %x->%x\n", pte_val(old), pte_val(pteval));
#endif
if (pte_present(pteval)) {
/* new page is present,
* now we have to find out what has changed */
if (((pte_val(old) ^ pte_val(pteval)) & PAGE_MASK)
|| (pte_young(old) && !pte_young(pteval))) {
/* physical page frame changed
* || access attribute changed -> flush */
/* flush is the default */
//pteval.pte_low &= ~_PAGE_MAPPED;
pteval = __pte(pte_val(pteval) & ~_PAGE_MAPPED);
} else if ((pte_write(old) && !pte_write(pteval))
|| (pte_dirty(old) && !pte_dirty(pteval))) {
/* Protection changed from r/w to ro
* or page now clean -> remap */
flush_rights = L4_FPAGE_W;
check_pte_mapped(old, pteval, "RW->RO");
} else {
/* nothing changed, simply return */
check_pte_mapped(old, pteval, "NoChg");
return pte_val(pteval);
}
}
/* Ok, now actually flush or remap the page */
L4XV_FN_v(l4x_flush_page(mm, pte_val(old), addr, PAGE_SHIFT, flush_rights));
return pte_val(pteval);
}
/*
* Changing some bits of contiguous entries requires us to follow a
* Break-Before-Make approach, breaking the whole contiguous set
* before we can change any entries. See ARM DDI 0487A.k_iss10775,
* "Misprogramming of the Contiguous bit", page D4-1762.
*
* This helper performs the break step.
*/
static pte_t get_clear_flush(struct mm_struct *mm,
unsigned long addr,
pte_t *ptep,
unsigned long pgsize,
unsigned long ncontig)
{
pte_t orig_pte = huge_ptep_get(ptep);
bool valid = pte_valid(orig_pte);
unsigned long i, saddr = addr;
for (i = 0; i < ncontig; i++, addr += pgsize, ptep++) {
pte_t pte = ptep_get_and_clear(mm, addr, ptep);
/*
* If HW_AFDBM is enabled, then the HW could turn on
* the dirty or accessed bit for any page in the set,
* so check them all.
*/
if (pte_dirty(pte))
orig_pte = pte_mkdirty(orig_pte);
if (pte_young(pte))
orig_pte = pte_mkyoung(orig_pte);
}
if (valid) {
struct vm_area_struct vma = TLB_FLUSH_VMA(mm, 0);
flush_tlb_range(&vma, saddr, addr);
}
return orig_pte;
}
static unsigned long clear_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
unsigned long addr, unsigned long end)
{
pte_t *pte;
pte_t ptecont;
do {
pte = pte_offset_map(pmd, addr);
ptecont = *pte;
if (pte_none(ptecont))
continue;
/*
* pte_young is a confusing name, though it AND _PAGE_ACCESSED
* Instead, I think we should call it pte_accessed
*/
if (pte_present(ptecont) && pte_young(ptecont)) {
/*
* The physical page, which this pte points to, has
* been read or written to during this time period.
*/
DEBUG_INFO("[%#016lx - %#016lx], pfn = %#013lx", addr, end, pte_pfn(ptecont));
collect_statistics(pte_pfn(ptecont));
pte_clear_flags(ptecont, _PAGE_ACCESSED);
}
} while (pte++, addr += PAGE_SIZE, addr != end);
return addr;
}
int huge_ptep_set_access_flags(struct vm_area_struct *vma,
unsigned long addr, pte_t *ptep,
pte_t pte, int dirty)
{
int ncontig, i;
size_t pgsize = 0;
unsigned long pfn = pte_pfn(pte), dpfn;
pgprot_t hugeprot;
pte_t orig_pte;
if (!pte_cont(pte))
return ptep_set_access_flags(vma, addr, ptep, pte, dirty);
ncontig = find_num_contig(vma->vm_mm, addr, ptep, &pgsize);
dpfn = pgsize >> PAGE_SHIFT;
if (!__cont_access_flags_changed(ptep, pte, ncontig))
return 0;
orig_pte = get_clear_flush(vma->vm_mm, addr, ptep, pgsize, ncontig);
/* Make sure we don't lose the dirty or young state */
if (pte_dirty(orig_pte))
pte = pte_mkdirty(pte);
if (pte_young(orig_pte))
pte = pte_mkyoung(pte);
hugeprot = pte_pgprot(pte);
for (i = 0; i < ncontig; i++, ptep++, addr += pgsize, pfn += dpfn)
set_pte_at(vma->vm_mm, addr, ptep, pfn_pte(pfn, hugeprot));
return 1;
}
static int
pin_page_for_write(const void __user *_addr, pte_t **ptep, spinlock_t **ptlp)
{
unsigned long addr = (unsigned long)_addr;
pgd_t *pgd;
pmd_t *pmd;
pte_t *pte;
pud_t *pud;
spinlock_t *ptl;
pgd = pgd_offset(current->mm, addr);
if (unlikely(pgd_none(*pgd) || pgd_bad(*pgd)))
return 0;
pud = pud_offset(pgd, addr);
if (unlikely(pud_none(*pud) || pud_bad(*pud)))
return 0;
pmd = pmd_offset(pud, addr);
if (unlikely(pmd_none(*pmd) || pmd_bad(*pmd)))
return 0;
pte = pte_offset_map_lock(current->mm, pmd, addr, &ptl);
if (unlikely(!pte_present(*pte) || !pte_young(*pte) ||
!pte_write(*pte) || !pte_dirty(*pte))) {
pte_unmap_unlock(pte, ptl);
return 0;
}
*ptep = pte;
*ptlp = ptl;
return 1;
}
int kthread_wss(void *data)
{
unsigned long va;
int ret;
int wss;
pgd_t *pgd;
pmd_t *pmd;
pud_t *pud;
pte_t *ptep;
struct task_struct *task;
while(!kthread_should_stop())
{
printk(KERN_INFO "Checking process' WSS.\n");
for_each_process(task)
{
wss = 0;
if(task->mm != NULL)
{
struct vm_area_struct *temp = task->mm->mmap;
while(temp)
{
if(temp->vm_flags & VM_IO){}
else
{
for(va = temp->vm_start; va < temp->vm_end; va+=PAGE_SIZE)
{
pgd = pgd_offset(task->mm,va);
if(pgd_none(*pgd))
break;
pud = pud_offset(pgd,va);
if(pud_none(*pud))
break;
pmd = pmd_offset(pud,va);
if(pmd_none(*pmd))
break;
ptep = pte_offset_map(pmd,va);
ret = 0;
if(pte_young(*ptep))
{
ret = test_and_clear_bit(_PAGE_BIT_ACCESSED, (unsigned long *) &ptep->pte);
wss++;
}
if(ret)
{
pte_update(task->mm, va, ptep);
}
pte_unmap(ptep);
}
}
temp = temp->vm_next;
}
printk(KERN_INFO "%i: %i\n", task->pid, wss);
}
}
msleep(1000);
}
return 0;
}
/*
* huge_ptep_set_access_flags will update access flags (dirty, accesssed)
* and write permission.
*
* For a contiguous huge pte range we need to check whether or not write
* permission has to change only on the first pte in the set. Then for
* all the contiguous ptes we need to check whether or not there is a
* discrepancy between dirty or young.
*/
static int __cont_access_flags_changed(pte_t *ptep, pte_t pte, int ncontig)
{
int i;
if (pte_write(pte) != pte_write(huge_ptep_get(ptep)))
return 1;
for (i = 0; i < ncontig; i++) {
pte_t orig_pte = huge_ptep_get(ptep + i);
if (pte_dirty(pte) != pte_dirty(orig_pte))
return 1;
if (pte_young(pte) != pte_young(orig_pte))
return 1;
}
return 0;
}
/*
* This is called at the end of handling a user page fault, when the
* fault has been handled by updating a PTE in the linux page tables.
* We use it to preload an HPTE into the hash table corresponding to
* the updated linux PTE.
*
* This must always be called with the pte lock held.
*/
void update_mmu_cache(struct vm_area_struct *vma, unsigned long address,
pte_t pte)
{
#ifdef CONFIG_PPC_STD_MMU
unsigned long access = 0, trap;
#endif
unsigned long pfn = pte_pfn(pte);
/* handle i-cache coherency */
if (!cpu_has_feature(CPU_FTR_COHERENT_ICACHE) &&
!cpu_has_feature(CPU_FTR_NOEXECUTE) &&
pfn_valid(pfn)) {
struct page *page = pfn_to_page(pfn);
#ifdef CONFIG_8xx
/* On 8xx, cache control instructions (particularly
* "dcbst" from flush_dcache_icache) fault as write
* operation if there is an unpopulated TLB entry
* for the address in question. To workaround that,
* we invalidate the TLB here, thus avoiding dcbst
* misbehaviour.
*/
_tlbie(address);
#endif
if (!PageReserved(page)
&& !test_bit(PG_arch_1, &page->flags)) {
if (vma->vm_mm == current->active_mm) {
__flush_dcache_icache((void *) address);
} else
flush_dcache_icache_page(page);
set_bit(PG_arch_1, &page->flags);
}
}
#ifdef CONFIG_PPC_STD_MMU
/* We only want HPTEs for linux PTEs that have _PAGE_ACCESSED set */
if (!pte_young(pte) || address >= TASK_SIZE)
return;
/* We try to figure out if we are coming from an instruction
* access fault and pass that down to __hash_page so we avoid
* double-faulting on execution of fresh text. We have to test
* for regs NULL since init will get here first thing at boot
*
* We also avoid filling the hash if not coming from a fault
*/
if (current->thread.regs == NULL)
return;
trap = TRAP(current->thread.regs);
if (trap == 0x400)
access |= _PAGE_EXEC;
else if (trap != 0x300)
return;
hash_preload(vma->vm_mm, address, access, trap);
#endif /* CONFIG_PPC_STD_MMU */
}
unsigned long l4x_set_pte(struct mm_struct *mm,
unsigned long addr,
pte_t old, pte_t pteval)
{
/*
* Check if any invalidation is necessary
*
* Invalidation (flush) necessary if:
* old page was present
* new page is not present OR
* new page has another physical address OR
* new page has another protection OR
* new page has other access attributes
*/
/* old was present && new not -> flush */
int flush_rights = L4_FPAGE_RWX;
if (pte_present(pteval)) {
/* new page is present,
* now we have to find out what has changed */
if (((pte_val(old) ^ pte_val(pteval)) & L4X_PHYSICAL_PAGE_MASK)
|| (pte_young(old) && !pte_young(pteval))) {
/* physical page frame changed
* || access attribute changed -> flush */
/* flush is the default */
} else if ((pte_write(old) && !pte_write(pteval))
|| (pte_dirty(old) && !pte_dirty(pteval))) {
/* Protection changed from r/w to ro
* or page now clean -> remap */
flush_rights = L4_FPAGE_W;
} else {
/* nothing changed, simply return */
return pte_val(pteval);
}
}
/* Ok, now actually flush or remap the page */
l4x_flush_page(mm, pte_val(old) & L4X_PHYSICAL_PAGE_MASK,
addr, PAGE_SHIFT, flush_rights, _RET_IP_);
return pte_val(pteval);
}
// Return true (!= 0) if any referenced bits are set.
static int ref_bits_set (int exclude_irqhandler) {
void *cur_addr;
pte_t *pte;
int i;
int ret_val = 0;
for (i = 0; i < cr_num_drivers; i++) {
if (exclude_irqhandler) uprintk ("i %d: ", i);
for (cur_addr = cr_base_address[i];
cur_addr < cr_base_address[i] + cr_module_size[i];
cur_addr += PAGE_SIZE) {
pte = virt_to_pte (cur_addr);
if (pte != NULL) {
// See if we're excluding the interrupt handler
// from this check.
if (exclude_irqhandler &&
addr_contains_irq_handler (cur_addr)) {
pte_unmap(pte);
if (exclude_irqhandler) uprintk ("X");
continue;
}
// See if the page was referenced lately.
if (pte_young(*pte) != 0) {
// kunmap_atomic (page, KM_IRQ1);
pte_unmap(pte);
if (exclude_irqhandler) uprintk ("1");
ret_val = 1;
continue;
}
if (exclude_irqhandler) uprintk ("0");
// kunmap_atomic (page, KM_IRQ1);
pte_unmap(pte);
}
}
if (exclude_irqhandler) uprintk ("\n");
}
return ret_val;
}
static void smaps_pte_entry(pte_t *pte, unsigned long addr,
struct mm_walk *walk)
{
struct mem_size_stats *mss = walk->private;
struct vm_area_struct *vma = walk->vma;
struct page *page = NULL;
if (pte_present(*pte)) {
page = vm_normal_page(vma, addr, *pte);
} else if (is_swap_pte(*pte)) {
swp_entry_t swpent = pte_to_swp_entry(*pte);
if (!non_swap_entry(swpent))
mss->swap += PAGE_SIZE;
else if (is_migration_entry(swpent))
page = migration_entry_to_page(swpent);
}
if (!page)
return;
smaps_account(mss, page, PAGE_SIZE, pte_young(*pte), pte_dirty(*pte));
}
static int
pin_page_for_write(const void __user *_addr, pte_t **ptep, spinlock_t **ptlp)
{
unsigned long addr = (unsigned long)_addr;
pgd_t *pgd;
pmd_t *pmd;
pte_t *pte;
pud_t *pud;
spinlock_t *ptl;
pgd = pgd_offset(current->mm, addr);
if (unlikely(pgd_none(*pgd) || pgd_bad(*pgd)))
return 0;
pud = pud_offset(pgd, addr);
if (unlikely(pud_none(*pud) || pud_bad(*pud)))
return 0;
pmd = pmd_offset(pud, addr);
if (unlikely(pmd_none(*pmd)))
return 0;
/*
* A pmd can be bad if it refers to a HugeTLB or THP page.
*
* Both THP and HugeTLB pages have the same pmd layout
* and should not be manipulated by the pte functions.
*
* Lock the page table for the destination and check
* to see that it's still huge and whether or not we will
* need to fault on write, or if we have a splitting THP.
*/
if (unlikely(pmd_thp_or_huge(*pmd))) {
ptl = ¤t->mm->page_table_lock;
spin_lock(ptl);
if (unlikely(!pmd_thp_or_huge(*pmd)
|| pmd_hugewillfault(*pmd)
|| pmd_trans_splitting(*pmd))) {
spin_unlock(ptl);
return 0;
}
*ptep = NULL;
*ptlp = ptl;
return 1;
}
if (unlikely(pmd_bad(*pmd)))
return 0;
pte = pte_offset_map_lock(current->mm, pmd, addr, &ptl);
if (unlikely(!pte_present(*pte) || !pte_young(*pte) ||
!pte_write(*pte) || !pte_dirty(*pte))) {
pte_unmap_unlock(pte, ptl);
return 0;
}
*ptep = pte;
*ptlp = ptl;
return 1;
}
/*
* Note this is constrained to return 0, -EFAULT, -EACCESS, -ENOMEM by
* segv().
*/
int handle_page_fault(unsigned long address, unsigned long ip,
int is_write, int is_user, int *code_out)
{
struct mm_struct *mm = current->mm;
struct vm_area_struct *vma;
pgd_t *pgd;
pud_t *pud;
pmd_t *pmd;
pte_t *pte;
int err = -EFAULT;
*code_out = SEGV_MAPERR;
/*
* If the fault was during atomic operation, don't take the fault, just
* fail.
*/
if (in_atomic())
goto out_nosemaphore;
down_read(&mm->mmap_sem);
vma = find_vma(mm, address);
if (!vma)
goto out;
else if (vma->vm_start <= address)
goto good_area;
else if (!(vma->vm_flags & VM_GROWSDOWN))
goto out;
else if (is_user && !ARCH_IS_STACKGROW(address))
goto out;
else if (expand_stack(vma, address))
goto out;
good_area:
*code_out = SEGV_ACCERR;
if (is_write && !(vma->vm_flags & VM_WRITE))
goto out;
/* Don't require VM_READ|VM_EXEC for write faults! */
if (!is_write && !(vma->vm_flags & (VM_READ | VM_EXEC)))
goto out;
do {
int fault;
survive:
fault = handle_mm_fault(mm, vma, address, is_write);
if (unlikely(fault & VM_FAULT_ERROR)) {
if (fault & VM_FAULT_OOM) {
err = -ENOMEM;
goto out_of_memory;
} else if (fault & VM_FAULT_SIGBUS) {
err = -EACCES;
goto out;
}
BUG();
}
if (fault & VM_FAULT_MAJOR)
current->maj_flt++;
else
current->min_flt++;
pgd = pgd_offset(mm, address);
pud = pud_offset(pgd, address);
pmd = pmd_offset(pud, address);
pte = pte_offset_kernel(pmd, address);
} while (!pte_present(*pte));
err = 0;
/*
* The below warning was added in place of
* pte_mkyoung(); if (is_write) pte_mkdirty();
* If it's triggered, we'd see normally a hang here (a clean pte is
* marked read-only to emulate the dirty bit).
* However, the generic code can mark a PTE writable but clean on a
* concurrent read fault, triggering this harmlessly. So comment it out.
*/
#if 0
WARN_ON(!pte_young(*pte) || (is_write && !pte_dirty(*pte)));
#endif
flush_tlb_page(vma, address);
out:
up_read(&mm->mmap_sem);
out_nosemaphore:
return err;
/*
* We ran out of memory, or some other thing happened to us that made
* us unable to handle the page fault gracefully.
*/
out_of_memory:
if (is_global_init(current)) {
up_read(&mm->mmap_sem);
yield();
down_read(&mm->mmap_sem);
goto survive;
}
goto out;
}
/*
* Note this is constrained to return 0, -EFAULT, -EACCESS, -ENOMEM by
* segv().
*/
int handle_page_fault(unsigned long address, unsigned long ip,
int is_write, int is_user, int *code_out)
{
struct mm_struct *mm = current->mm;
struct vm_area_struct *vma;
pgd_t *pgd;
pud_t *pud;
pmd_t *pmd;
pte_t *pte;
int err = -EFAULT;
unsigned int flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE |
(is_write ? FAULT_FLAG_WRITE : 0);
*code_out = SEGV_MAPERR;
/*
* If the fault was during atomic operation, don't take the fault, just
* fail.
*/
if (in_atomic())
goto out_nosemaphore;
retry:
down_read(&mm->mmap_sem);
vma = find_vma(mm, address);
if (!vma)
goto out;
else if (vma->vm_start <= address)
goto good_area;
else if (!(vma->vm_flags & VM_GROWSDOWN))
goto out;
else if (is_user && !ARCH_IS_STACKGROW(address))
goto out;
else if (expand_stack(vma, address))
goto out;
good_area:
*code_out = SEGV_ACCERR;
if (is_write && !(vma->vm_flags & VM_WRITE))
goto out;
/* Don't require VM_READ|VM_EXEC for write faults! */
if (!is_write && !(vma->vm_flags & (VM_READ | VM_EXEC)))
goto out;
do {
int fault;
fault = handle_mm_fault(mm, vma, address, flags);
if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current))
goto out_nosemaphore;
if (unlikely(fault & VM_FAULT_ERROR)) {
if (fault & VM_FAULT_OOM) {
goto out_of_memory;
} else if (fault & VM_FAULT_SIGBUS) {
err = -EACCES;
goto out;
}
BUG();
}
if (flags & FAULT_FLAG_ALLOW_RETRY) {
if (fault & VM_FAULT_MAJOR)
current->maj_flt++;
else
current->min_flt++;
if (fault & VM_FAULT_RETRY) {
flags &= ~FAULT_FLAG_ALLOW_RETRY;
flags |= FAULT_FLAG_TRIED;
goto retry;
}
}
pgd = pgd_offset(mm, address);
pud = pud_offset(pgd, address);
pmd = pmd_offset(pud, address);
pte = pte_offset_kernel(pmd, address);
} while (!pte_present(*pte));
err = 0;
/*
* The below warning was added in place of
* pte_mkyoung(); if (is_write) pte_mkdirty();
* If it's triggered, we'd see normally a hang here (a clean pte is
* marked read-only to emulate the dirty bit).
* However, the generic code can mark a PTE writable but clean on a
* concurrent read fault, triggering this harmlessly. So comment it out.
*/
#if 0
WARN_ON(!pte_young(*pte) || (is_write && !pte_dirty(*pte)));
#endif
flush_tlb_page(vma, address);
out:
up_read(&mm->mmap_sem);
out_nosemaphore:
return err;
out_of_memory:
/*
* We ran out of memory, call the OOM killer, and return the userspace
* (which will retry the fault, or kill us if we got oom-killed).
*/
up_read(&mm->mmap_sem);
pagefault_out_of_memory();
return 0;
}
static void fix_range(struct mm_struct *mm, unsigned long start_addr,
unsigned long end_addr, int force)
{
pgd_t *npgd;
pud_t *npud;
pmd_t *npmd;
pte_t *npte;
unsigned long addr, end;
int r, w, x, err, fd;
if(mm == NULL) return;
fd = mm->context.skas.mm_fd;
for(addr = start_addr; addr < end_addr;){
npgd = pgd_offset(mm, addr);
if(!pgd_present(*npgd)){
if(force || pgd_newpage(*npgd)){
end = addr + PGDIR_SIZE;
if(end > end_addr)
end = end_addr;
err = unmap(fd, (void *) addr, end - addr);
if(err < 0)
panic("munmap failed, errno = %d\n",
-err);
pgd_mkuptodate(*npgd);
}
addr += PGDIR_SIZE;
continue;
}
npud = pud_offset(npgd, addr);
if(!pud_present(*npud)){
if(force || pud_newpage(*npud)){
end = addr + PUD_SIZE;
if(end > end_addr)
end = end_addr;
err = unmap(fd, (void *) addr, end - addr);
if(err < 0)
panic("munmap failed, errno = %d\n",
-err);
pud_mkuptodate(*npud);
}
addr += PUD_SIZE;
continue;
}
npmd = pmd_offset(npud, addr);
if(!pmd_present(*npmd)){
if(force || pmd_newpage(*npmd)){
end = addr + PMD_SIZE;
if(end > end_addr)
end = end_addr;
err = unmap(fd, (void *) addr, end - addr);
if(err < 0)
panic("munmap failed, errno = %d\n",
-err);
pmd_mkuptodate(*npmd);
}
addr += PMD_SIZE;
continue;
}
npte = pte_offset_kernel(npmd, addr);
r = pte_read(*npte);
w = pte_write(*npte);
x = pte_exec(*npte);
if(!pte_dirty(*npte))
w = 0;
if(!pte_young(*npte)){
r = 0;
w = 0;
}
if(force || pte_newpage(*npte)){
err = unmap(fd, (void *) addr, PAGE_SIZE);
if(err < 0)
panic("munmap failed, errno = %d\n", -err);
if(pte_present(*npte))
map(fd, addr, pte_val(*npte) & PAGE_MASK,
PAGE_SIZE, r, w, x);
}
else if(pte_newprot(*npte))
protect(fd, addr, PAGE_SIZE, r, w, x, 1);
*npte = pte_mkuptodate(*npte);
addr += PAGE_SIZE;
}
}
/*pgtable sequential scan and count for __access_bits.*/
static int scan_pgtable(void)
{
pgd_t *pgd = NULL;
pud_t *pud = NULL;
pmd_t *pmd = NULL;
pte_t *ptep, pte;
spinlock_t *ptl;
struct mm_struct *mm;
struct vm_area_struct *vma;
unsigned long start = 0; /*the start of address.*/
unsigned long end = 0; /*the end of address.*/
unsigned long address = 0; /* the address of vma.*/
int number_hotpages = 0; /* the number of hot pages */
int number_vpages = 0;
int cycle_index = 0; /* the loop counter, which denotes ITERATIONS. */
/* the array that records the number of hot page in every cycle */
int hot_page[ITERATIONS];
int number_current_pg = 0;
int pg_count = 0;
int j = 0;
int times = 0; /* records reuse time*/
/* some variables that describe page "heat" */
int hig = 0;
int mid = 0;
int low = 0;
int llow = 0;
int lllow = 0;
int llllow = 0;
int all_pages = 0;/* the total number of pages */
/*the average number of hot pages in each iteration.*/
long avg_hotpage=0;
/*the total number of memory accesses across all pages*/
long num_access=0;
/* avg utilization of each page */
int avg_page_utilization = 0;
/*get the handle of current running benchmark.*/
struct task_struct *bench_process = get_current_process();
if(bench_process == NULL)
{
printk("sysmon: get no process handle in scan_pgtable function...exit&trying again...\n");
return 0;
}
else /* get the process*/
mm = bench_process->mm;
if(mm == NULL)
{
printk("sysmon: error mm is NULL, return back & trying...\n");
return 0;
}
for(j = 0; j < PAGE_ALL; j++)
page_heat[j] = -1;
for(j = 0; j < ITERATIONS; j++)
{
hot_page[j] = 0;
reuse_time[j] = 0;
dirty_page[j] = 0;
}
/*yanghao*/
times = 0;
for(cycle_index = 0; cycle_index < ITERATIONS; cycle_index++)
{
number_hotpages = 0;
/*scan each vma*/
for(vma = mm->mmap; vma; vma = vma->vm_next)
{
start = vma->vm_start;
end = vma->vm_end;
mm = vma->vm_mm;
/*in each vma, we check all pages*/
for(address = start; address < end; address += PAGE_SIZE)
{
/*scan page table for each page in this VMA*/
pgd = pgd_offset(mm, address);
if (pgd_none(*pgd) || unlikely(pgd_bad(*pgd)))
continue;
pud = pud_offset(pgd, address);
if (pud_none(*pud) || unlikely(pud_bad(*pud)))
continue;
pmd = pmd_offset(pud, address);
if (pmd_none(*pmd) || unlikely(pmd_bad(*pmd)))
continue;
ptep = pte_offset_map_lock(mm, pmd, address, &ptl);
pte = *ptep;
if(pte_present(pte))
{
if(pte_young(pte)) /*hot page*/
{
/*re-set and clear _access_bits to 0*/
pte = pte_mkold(pte);
set_pte_at(mm, address, ptep, pte);
/*yanghao:re-set and clear _dirty_bits to 0*/
pte = pte_mkclean(pte);
set_pte_at(mm, address, ptep, pte);
}
}
else /*no page pte_none*/
{
pte_unmap_unlock(ptep, ptl);
continue;
}
pte_unmap_unlock(ptep, ptl);
page_counts++;
}
}
/*count the number of hot pages*/
if(bench_process == NULL)
{
printk("sysmon: get no process handle in scan_pgtable function...exit&trying again...\n");
return 0;
}
else /*get the process*/
mm = bench_process->mm;
if(mm == NULL)
{
printk("sysmon: error mm is NULL, return back & trying...\n");
return 0;
}
number_vpages = 0;
sampling_interval = page_counts / 250; /*yanghao:*/
page_counts = 0;
for(vma = mm->mmap; vma; vma = vma->vm_next)
{
start = vma->vm_start;
end = vma->vm_end;
/*scan each page in this VMA*/
mm = vma->vm_mm;
pg_count = 0;
for(address = start; address < end; address += PAGE_SIZE)
{
/*scan page table for each page in this VMA*/
pgd = pgd_offset(mm, address);
if (pgd_none(*pgd) || unlikely(pgd_bad(*pgd)))
continue;
pud = pud_offset(pgd, address);
if (pud_none(*pud) || unlikely(pud_bad(*pud)))
continue;
pmd = pmd_offset(pud, address);
if (pmd_none(*pmd) || unlikely(pmd_bad(*pmd)))
continue;
ptep = pte_offset_map_lock(mm, pmd, address, &ptl);
pte = *ptep;
if(pte_present(pte))
{
if(pte_young(pte)) /* hot pages*/
{
number_current_pg = pg_count + number_vpages;
page_heat[number_current_pg]++;
hot_page[cycle_index]++;
/*yanghao:*/
if (page_counts == random_page)
{
times++;
if (pte_dirty(pte))
dirty_page[cycle_index] = 1;
}
}
else
{
if (page_counts == random_page)
reuse_time[times]++;
}
}
pg_count++;
pte_unmap_unlock(ptep, ptl);
page_counts++;
}
number_vpages += (int)(end - start)/PAGE_SIZE;
}
}
/*yanghao:cal. the No. of random_page*/
random_page += sampling_interval;
if(random_page >= page_counts)
random_page=page_counts / 300;
/*****************************OUTPUT************************************/
for(j = 0; j < PAGE_ALL; j++)
{
if(page_heat[j] < VH && page_heat[j] > H)
hig++;
if(page_heat[j] > M && page_heat[j] <= H)
mid++;
if(page_heat[j] <= M && page_heat[j] > L)
low++;
if(page_heat[j] > VL_MAX && page_heat[j] <= L)
llow++;
if(page_heat[j] > VL_MIN && page_heat[j] <= VL_MAX)
lllow++;
if(page_heat[j] >= 0 && page_heat[j] <= VL_MIN)
llllow++;
if(page_heat[j] > -1)
all_pages++;
}
/*the values reflect the accessing frequency of each physical page.*/
printk("[LOG: after sampling (%d loops) ...] ",ITERATIONS);
printk("the values denote the physical page accessing frequence.\n");
printk("-->hig (150,200) is %d. Indicating the number of re-used pages is high.\n",hig);
printk("-->mid (100,150] is %d.\n",mid);
printk("-->low (64,100] is %d.\n",low);
printk("-->llow (10,64] is %d. In locality,no too many re-used pages.\n",llow);
printk("-->lllow (5,10] is %d.\n",lllow);
printk("-->llllow [1,5] is %d.\n",llllow);
for(j = 0;j < ITERATIONS; j++)
avg_hotpage += hot_page[j];
avg_hotpage /= (j+1);
/*
* new step@20140704
* (1)the different phases of memory utilization
* (2)the avg. page accessing utilization
* (3)memory pages layout and spectrum
*/
for(j = 0; j < PAGE_ALL; j++)
if(page_heat[j] > -1) /*the page that is accessed at least once.*/
num_access += (page_heat[j] + 1);
printk("the total number of memory accesses is %ld, the average is %ld\n",
num_access, num_access / ITERATIONS);
avg_page_utilization = num_access / all_pages;
printk("Avg hot pages num is %ld, all used pages num is %d, avg utilization of each page is %d\n",
avg_hotpage, all_pages, avg_page_utilization);
/*yanghao:print the information about reuse-distance*/
if ((times == 0) && (reuse_time[0] ==0))
printk("the page No.%d is not available.",random_page);
else
{
if ((times == 0) && (reuse_time[0] == 0))
printk("the page No.%d was not used in this 200 loops.",random_page);
else
{
if (times < ITERATIONS)
times++;
printk("the reusetime of page No.%d is:",random_page);
for (j = 0; j < times; j++)
printk("%d ",reuse_time[j]);
printk("\n");
printk("the total number of the digit above denotes the sum that page NO.%d be accessd in %d loops.\n",
random_page,ITERATIONS);
printk("each digit means the sum loops that between current loop and the last loop.\n");
}
}
printk("\n\n");
return 1;
}
static void fix_range(struct mm_struct *mm, unsigned long start_addr,
unsigned long end_addr, int force)
{
pgd_t *npgd;
pmd_t *npmd;
pte_t *npte;
unsigned long addr;
int r, w, x, err;
if((current->thread.mode.tt.extern_pid != -1) &&
(current->thread.mode.tt.extern_pid != os_getpid()))
panic("fix_range fixing wrong address space, current = 0x%p",
current);
if(mm == NULL) return;
for(addr=start_addr;addr<end_addr;){
if(addr == TASK_SIZE){
/* Skip over kernel text, kernel data, and physical
* memory, which don't have ptes, plus kernel virtual
* memory, which is flushed separately, and remap
* the process stack. The only way to get here is
* if (end_addr == STACK_TOP) > TASK_SIZE, which is
* only true in the honeypot case.
*/
addr = STACK_TOP - ABOVE_KMEM;
continue;
}
npgd = pgd_offset(mm, addr);
npmd = pmd_offset(npgd, addr);
if(pmd_present(*npmd)){
npte = pte_offset_kernel(npmd, addr);
r = pte_read(*npte);
w = pte_write(*npte);
x = pte_exec(*npte);
if(!pte_dirty(*npte)) w = 0;
if(!pte_young(*npte)){
r = 0;
w = 0;
}
if(force || pte_newpage(*npte)){
err = os_unmap_memory((void *) addr,
PAGE_SIZE);
if(err < 0)
panic("munmap failed, errno = %d\n",
-err);
if(pte_present(*npte))
map_memory(addr,
pte_val(*npte) & PAGE_MASK,
PAGE_SIZE, r, w, x);
}
else if(pte_newprot(*npte)){
protect_memory(addr, PAGE_SIZE, r, w, x, 1);
}
*npte = pte_mkuptodate(*npte);
addr += PAGE_SIZE;
}
else {
if(force || pmd_newpage(*npmd)){
err = os_unmap_memory((void *) addr, PMD_SIZE);
if(err < 0)
panic("munmap failed, errno = %d\n",
-err);
pmd_mkuptodate(*npmd);
}
addr += PMD_SIZE;
}
}
}
