struct page * read_swap_cache_async(unsigned long entry, int wait)
{
struct page *found_page = 0, *new_page;
unsigned long new_page_addr;
#ifdef DEBUG_SWAP
printk("DebugVM: read_swap_cache_async entry %08lx%s\n",
entry, wait ? ", wait" : "");
#endif
/*
* Make sure the swap entry is still in use.
*/
if (!swap_duplicate(entry)) /* Account for the swap cache */
goto out;
/*
* Look for the page in the swap cache.
*/
found_page = lookup_swap_cache(entry);
if (found_page)
goto out_free_swap;
new_page_addr = __get_free_page(GFP_USER);
if (!new_page_addr)
goto out_free_swap; /* Out of memory */
new_page = mem_map + MAP_NR(new_page_addr);
/*
* Check the swap cache again, in case we stalled above.
*/
found_page = lookup_swap_cache(entry);
if (found_page)
goto out_free_page;
/*
* Add it to the swap cache and read its contents.
*/
if (!add_to_swap_cache(new_page, entry))
goto out_free_page;
set_bit(PG_locked, &new_page->flags);
rw_swap_page(READ, entry, (char *) new_page_addr, wait);
#ifdef DEBUG_SWAP
printk("DebugVM: read_swap_cache_async created "
"entry %08lx at %p\n",
entry, (char *) page_address(new_page));
#endif
return new_page;
out_free_page:
__free_page(new_page);
out_free_swap:
swap_free(entry);
out:
return found_page;
}
/*
* We may have stale swap cache pages in memory: notice
* them here and get rid of the unnecessary final write.
*/
static int swap_writepage(struct page *page)
{
if (remove_exclusive_swap_page(page)) {
UnlockPage(page);
return 0;
}
TRACE_MEMORY(TRACE_EV_MEMORY_SWAP_OUT, (unsigned long) page);
rw_swap_page(WRITE, page);
return 0;
}
/*
* Locate a page of swap in physical memory, reserving swap cache space
* and reading the disk if it is not already cached.
* A failure return means that either the page allocation failed or that
* the swap entry is no longer in use.
*/
struct page * read_swap_cache_async(swp_entry_t entry)
{
struct page *found_page, *new_page = NULL;
int err;
do {
/*
* First check the swap cache. Since this is normally
* called after lookup_swap_cache() failed, re-calling
* that would confuse statistics: use find_get_page()
* directly.
*/
found_page = find_get_page(&swapper_space, entry.val);
if (found_page)
break;
/*
* Get a new page to read into from swap.
*/
if (!new_page) {
new_page = alloc_page(GFP_HIGHUSER);
if (!new_page)
break; /* Out of memory */
}
/*
* Associate the page with swap entry in the swap cache.
* May fail (-ENOENT) if swap entry has been freed since
* our caller observed it. May fail (-EEXIST) if there
* is already a page associated with this entry in the
* swap cache: added by a racing read_swap_cache_async,
* or by try_to_swap_out (or shmem_writepage) re-using
* the just freed swap entry for an existing page.
*/
err = add_to_swap_cache(new_page, entry);
if (!err) {
/*
* Initiate read into locked page and return.
*/
rw_swap_page(READ, new_page);
return new_page;
}
} while (err != -ENOENT);
if (new_page)
page_cache_release(new_page);
return found_page;
}
struct page * read_swap_cache_async(swp_entry_t entry, int wait)
{
struct page *found_page = 0, *new_page;
unsigned long new_page_addr;
/*
* Make sure the swap entry is still in use.
*/
if (!swap_duplicate(entry)) /* Account for the swap cache */
goto out;
/*
* Look for the page in the swap cache.
*/
found_page = lookup_swap_cache(entry);
if (found_page)
goto out_free_swap;
new_page_addr = __get_free_page(GFP_USER);
if (!new_page_addr)
goto out_free_swap; /* Out of memory */
new_page = virt_to_page(new_page_addr);
/*
* Check the swap cache again, in case we stalled above.
*/
found_page = lookup_swap_cache(entry);
if (found_page)
goto out_free_page;
/*
* Add it to the swap cache and read its contents.
*/
lock_page(new_page);
add_to_swap_cache(new_page, entry);
rw_swap_page(READ, new_page, wait);
return new_page;
out_free_page:
page_cache_release(new_page);
out_free_swap:
swap_free(entry);
out:
return found_page;
}
/*
* Setting up a new swap file needs a simple wrapper just to read the
* swap signature. SysV shared memory also needs a simple wrapper.
*/
void rw_swap_page_nocache(int rw, unsigned long entry, char *buffer)
{
struct page *page;
page = mem_map + MAP_NR((unsigned long) buffer);
wait_on_page(page);
set_bit(PG_locked, &page->flags);
if (test_and_set_bit(PG_swap_cache, &page->flags)) {
printk ("VM: read_swap_page: page already in swap cache!\n");
return;
}
if (page->inode) {
printk ("VM: read_swap_page: page already in page cache!\n");
return;
}
page->inode = &swapper_inode;
page->offset = entry;
atomic_inc(&page->count); /* Protect from shrink_mmap() */
rw_swap_page(rw, entry, buffer, 1);
atomic_dec(&page->count);
page->inode = 0;
clear_bit(PG_swap_cache, &page->flags);
}
/*
* We completely avoid races by reading each swap page in advance,
* and then search for the process using it. All the necessary
* page table adjustments can then be made atomically.
*/
static int try_to_unuse(unsigned int type)
{
struct swap_info_struct * si = &swap_info[type];
struct mm_struct *start_mm;
unsigned short *swap_map;
unsigned short swcount;
struct page *page;
swp_entry_t entry;
int i = 0;
int retval = 0;
int reset_overflow = 0;
/*
* When searching mms for an entry, a good strategy is to
* start at the first mm we freed the previous entry from
* (though actually we don't notice whether we or coincidence
* freed the entry). Initialize this start_mm with a hold.
*
* A simpler strategy would be to start at the last mm we
* freed the previous entry from; but that would take less
* advantage of mmlist ordering (now preserved by swap_out()),
* which clusters forked address spaces together, most recent
* child immediately after parent. If we race with dup_mmap(),
* we very much want to resolve parent before child, otherwise
* we may miss some entries: using last mm would invert that.
*/
start_mm = &init_mm;
atomic_inc(&init_mm.mm_users);
/*
* Keep on scanning until all entries have gone. Usually,
* one pass through swap_map is enough, but not necessarily:
* mmput() removes mm from mmlist before exit_mmap() and its
* zap_page_range(). That's not too bad, those entries are
* on their way out, and handled faster there than here.
* do_munmap() behaves similarly, taking the range out of mm's
* vma list before zap_page_range(). But unfortunately, when
* unmapping a part of a vma, it takes the whole out first,
* then reinserts what's left after (might even reschedule if
* open() method called) - so swap entries may be invisible
* to swapoff for a while, then reappear - but that is rare.
*/
while ((i = find_next_to_unuse(si, i))) {
/*
* Get a page for the entry, using the existing swap
* cache page if there is one. Otherwise, get a clean
* page and read the swap into it.
*/
swap_map = &si->swap_map[i];
entry = SWP_ENTRY(type, i);
page = read_swap_cache_async(entry);
if (!page) {
/*
* Either swap_duplicate() failed because entry
* has been freed independently, and will not be
* reused since sys_swapoff() already disabled
* allocation from here, or alloc_page() failed.
*/
if (!*swap_map)
continue;
retval = -ENOMEM;
break;
}
/*
* Don't hold on to start_mm if it looks like exiting.
*/
if (atomic_read(&start_mm->mm_users) == 1) {
mmput(start_mm);
start_mm = &init_mm;
atomic_inc(&init_mm.mm_users);
}
/*
* Wait for and lock page. When do_swap_page races with
* try_to_unuse, do_swap_page can handle the fault much
* faster than try_to_unuse can locate the entry. This
* apparently redundant "wait_on_page" lets try_to_unuse
* defer to do_swap_page in such a case - in some tests,
* do_swap_page and try_to_unuse repeatedly compete.
*/
wait_on_page(page);
lock_page(page);
/*
* Remove all references to entry, without blocking.
* Whenever we reach init_mm, there's no address space
* to search, but use it as a reminder to search shmem.
*/
swcount = *swap_map;
if (swcount > 1) {
flush_page_to_ram(page);
if (start_mm == &init_mm)
shmem_unuse(entry, page);
else
unuse_process(start_mm, entry, page);
}
if (*swap_map > 1) {
int set_start_mm = (*swap_map >= swcount);
struct list_head *p = &start_mm->mmlist;
struct mm_struct *new_start_mm = start_mm;
struct mm_struct *mm;
spin_lock(&mmlist_lock);
while (*swap_map > 1 &&
(p = p->next) != &start_mm->mmlist) {
mm = list_entry(p, struct mm_struct, mmlist);
swcount = *swap_map;
if (mm == &init_mm) {
set_start_mm = 1;
shmem_unuse(entry, page);
} else
unuse_process(mm, entry, page);
if (set_start_mm && *swap_map < swcount) {
new_start_mm = mm;
set_start_mm = 0;
}
}
atomic_inc(&new_start_mm->mm_users);
spin_unlock(&mmlist_lock);
mmput(start_mm);
start_mm = new_start_mm;
}
/*
* How could swap count reach 0x7fff when the maximum
* pid is 0x7fff, and there's no way to repeat a swap
* page within an mm (except in shmem, where it's the
* shared object which takes the reference count)?
* We believe SWAP_MAP_MAX cannot occur in Linux 2.4.
*
* If that's wrong, then we should worry more about
* exit_mmap() and do_munmap() cases described above:
* we might be resetting SWAP_MAP_MAX too early here.
* We know "Undead"s can happen, they're okay, so don't
* report them; but do report if we reset SWAP_MAP_MAX.
*/
if (*swap_map == SWAP_MAP_MAX) {
swap_list_lock();
swap_device_lock(si);
nr_swap_pages++;
*swap_map = 1;
swap_device_unlock(si);
swap_list_unlock();
reset_overflow = 1;
}
/*
* If a reference remains (rare), we would like to leave
* the page in the swap cache; but try_to_swap_out could
* then re-duplicate the entry once we drop page lock,
* so we might loop indefinitely; also, that page could
* not be swapped out to other storage meanwhile. So:
* delete from cache even if there's another reference,
* after ensuring that the data has been saved to disk -
* since if the reference remains (rarer), it will be
* read from disk into another page. Splitting into two
* pages would be incorrect if swap supported "shared
* private" pages, but they are handled by tmpfs files.
* Note shmem_unuse already deleted its from swap cache.
*/
if ((*swap_map > 1) && PageDirty(page) && PageSwapCache(page)) {
rw_swap_page(WRITE, page);
lock_page(page);
}
if (PageSwapCache(page))
delete_from_swap_cache(page);
/*
* So we could skip searching mms once swap count went
* to 1, we did not mark any present ptes as dirty: must
* mark page dirty so try_to_swap_out will preserve it.
*/
SetPageDirty(page);
UnlockPage(page);
page_cache_release(page);
/*
* Make sure that we aren't completely killing
* interactive performance. Interruptible check on
* signal_pending() would be nice, but changes the spec?
*/
if (current->need_resched)
schedule();
}
static int swap_writepage(struct page *page)
{
rw_swap_page(WRITE, page, 0);
return 0;
}
