[thirdparty/linux.git] / mm / page_io.c

// SPDX-License-Identifier: GPL-2.0
/*
 *  linux/mm/page_io.c
 *
 *  Copyright (C) 1991, 1992, 1993, 1994  Linus Torvalds
 *
 *  Swap reorganised 29.12.95, 
 *  Asynchronous swapping added 30.12.95. Stephen Tweedie
 *  Removed race in async swapping. 14.4.1996. Bruno Haible
 *  Add swap of shared pages through the page cache. 20.2.1998. Stephen Tweedie
 *  Always use brw_page, life becomes simpler. 12 May 1998 Eric Biederman
 */

#include <linux/mm.h>
#include <linux/kernel_stat.h>
#include <linux/gfp.h>
#include <linux/pagemap.h>
#include <linux/swap.h>
#include <linux/bio.h>
#include <linux/swapops.h>
#include <linux/writeback.h>
#include <linux/blkdev.h>
#include <linux/psi.h>
#include <linux/uio.h>
#include <linux/sched/task.h>
#include <linux/delayacct.h>
#include <linux/zswap.h>
#include "swap.h"

static void __end_swap_bio_write(struct bio *bio)
{
	struct folio *folio = bio_first_folio_all(bio);

	if (bio->bi_status) {
		/*
		 * We failed to write the page out to swap-space.
		 * Re-dirty the page in order to avoid it being reclaimed.
		 * Also print a dire warning that things will go BAD (tm)
		 * very quickly.
		 *
		 * Also clear PG_reclaim to avoid folio_rotate_reclaimable()
		 */
		folio_mark_dirty(folio);
		pr_alert_ratelimited("Write-error on swap-device (%u:%u:%llu)\n",
				     MAJOR(bio_dev(bio)), MINOR(bio_dev(bio)),
				     (unsigned long long)bio->bi_iter.bi_sector);
		folio_clear_reclaim(folio);
	}
	folio_end_writeback(folio);
}

static void end_swap_bio_write(struct bio *bio)
{
	__end_swap_bio_write(bio);
	bio_put(bio);
}

static void __end_swap_bio_read(struct bio *bio)
{
	struct folio *folio = bio_first_folio_all(bio);

	if (bio->bi_status) {
		pr_alert_ratelimited("Read-error on swap-device (%u:%u:%llu)\n",
				     MAJOR(bio_dev(bio)), MINOR(bio_dev(bio)),
				     (unsigned long long)bio->bi_iter.bi_sector);
	} else {
		folio_mark_uptodate(folio);
	}
	folio_unlock(folio);
}

static void end_swap_bio_read(struct bio *bio)
{
	__end_swap_bio_read(bio);
	bio_put(bio);
}

int generic_swapfile_activate(struct swap_info_struct *sis,
				struct file *swap_file,
				sector_t *span)
{
	struct address_space *mapping = swap_file->f_mapping;
	struct inode *inode = mapping->host;
	unsigned blocks_per_page;
	unsigned long page_no;
	unsigned blkbits;
	sector_t probe_block;
	sector_t last_block;
	sector_t lowest_block = -1;
	sector_t highest_block = 0;
	int nr_extents = 0;
	int ret;

	blkbits = inode->i_blkbits;
	blocks_per_page = PAGE_SIZE >> blkbits;

	/*
	 * Map all the blocks into the extent tree.  This code doesn't try
	 * to be very smart.
	 */
	probe_block = 0;
	page_no = 0;
	last_block = i_size_read(inode) >> blkbits;
	while ((probe_block + blocks_per_page) <= last_block &&
			page_no < sis->max) {
		unsigned block_in_page;
		sector_t first_block;

		cond_resched();

		first_block = probe_block;
		ret = bmap(inode, &first_block);
		if (ret || !first_block)
			goto bad_bmap;

		/*
		 * It must be PAGE_SIZE aligned on-disk
		 */
		if (first_block & (blocks_per_page - 1)) {
			probe_block++;
			goto reprobe;
		}

		for (block_in_page = 1; block_in_page < blocks_per_page;
					block_in_page++) {
			sector_t block;

			block = probe_block + block_in_page;
			ret = bmap(inode, &block);
			if (ret || !block)
				goto bad_bmap;

			if (block != first_block + block_in_page) {
				/* Discontiguity */
				probe_block++;
				goto reprobe;
			}
		}

		first_block >>= (PAGE_SHIFT - blkbits);
		if (page_no) {	/* exclude the header page */
			if (first_block < lowest_block)
				lowest_block = first_block;
			if (first_block > highest_block)
				highest_block = first_block;
		}

		/*
		 * We found a PAGE_SIZE-length, PAGE_SIZE-aligned run of blocks
		 */
		ret = add_swap_extent(sis, page_no, 1, first_block);
		if (ret < 0)
			goto out;
		nr_extents += ret;
		page_no++;
		probe_block += blocks_per_page;
reprobe:
		continue;
	}
	ret = nr_extents;
	*span = 1 + highest_block - lowest_block;
	if (page_no == 0)
		page_no = 1;	/* force Empty message */
	sis->max = page_no;
	sis->pages = page_no - 1;
out:
	return ret;
bad_bmap:
	pr_err("swapon: swapfile has holes\n");
	ret = -EINVAL;
	goto out;
}

static bool is_folio_zero_filled(struct folio *folio)
{
	unsigned int pos, last_pos;
	unsigned long *data;
	unsigned int i;

	last_pos = PAGE_SIZE / sizeof(*data) - 1;
	for (i = 0; i < folio_nr_pages(folio); i++) {
		data = kmap_local_folio(folio, i * PAGE_SIZE);
		/*
		 * Check last word first, incase the page is zero-filled at
		 * the start and has non-zero data at the end, which is common
		 * in real-world workloads.
		 */
		if (data[last_pos]) {
			kunmap_local(data);
			return false;
		}
		for (pos = 0; pos < last_pos; pos++) {
			if (data[pos]) {
				kunmap_local(data);
				return false;
			}
		}
		kunmap_local(data);
	}

	return true;
}

static void swap_zeromap_folio_set(struct folio *folio)
{
	struct obj_cgroup *objcg = get_obj_cgroup_from_folio(folio);
	struct swap_info_struct *sis = swp_swap_info(folio->swap);
	int nr_pages = folio_nr_pages(folio);
	swp_entry_t entry;
	unsigned int i;

	for (i = 0; i < folio_nr_pages(folio); i++) {
		entry = page_swap_entry(folio_page(folio, i));
		set_bit(swp_offset(entry), sis->zeromap);
	}

	count_vm_events(SWPOUT_ZERO, nr_pages);
	if (objcg) {
		count_objcg_events(objcg, SWPOUT_ZERO, nr_pages);
		obj_cgroup_put(objcg);
	}
}

static void swap_zeromap_folio_clear(struct folio *folio)
{
	struct swap_info_struct *sis = swp_swap_info(folio->swap);
	swp_entry_t entry;
	unsigned int i;

	for (i = 0; i < folio_nr_pages(folio); i++) {
		entry = page_swap_entry(folio_page(folio, i));
		clear_bit(swp_offset(entry), sis->zeromap);
	}
}

/*
 * We may have stale swap cache pages in memory: notice
 * them here and get rid of the unnecessary final write.
 */
int swap_writeout(struct folio *folio, struct writeback_control *wbc)
{
	int ret;

	if (folio_free_swap(folio)) {
		folio_unlock(folio);
		return 0;
	}
	/*
	 * Arch code may have to preserve more data than just the page
	 * contents, e.g. memory tags.
	 */
	ret = arch_prepare_to_swap(folio);
	if (ret) {
		folio_mark_dirty(folio);
		folio_unlock(folio);
		return ret;
	}

	/*
	 * Use a bitmap (zeromap) to avoid doing IO for zero-filled pages.
	 * The bits in zeromap are protected by the locked swapcache folio
	 * and atomic updates are used to protect against read-modify-write
	 * corruption due to other zero swap entries seeing concurrent updates.
	 */
	if (is_folio_zero_filled(folio)) {
		swap_zeromap_folio_set(folio);
		folio_unlock(folio);
		return 0;
	} else {
		/*
		 * Clear bits this folio occupies in the zeromap to prevent
		 * zero data being read in from any previous zero writes that
		 * occupied the same swap entries.
		 */
		swap_zeromap_folio_clear(folio);
	}
	if (zswap_store(folio)) {
		count_mthp_stat(folio_order(folio), MTHP_STAT_ZSWPOUT);
		folio_unlock(folio);
		return 0;
	}
	if (!mem_cgroup_zswap_writeback_enabled(folio_memcg(folio))) {
		folio_mark_dirty(folio);
		return AOP_WRITEPAGE_ACTIVATE;
	}

	__swap_writepage(folio, wbc);
	return 0;
}

static inline void count_swpout_vm_event(struct folio *folio)
{
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	if (unlikely(folio_test_pmd_mappable(folio))) {
		count_memcg_folio_events(folio, THP_SWPOUT, 1);
		count_vm_event(THP_SWPOUT);
	}
#endif
	count_mthp_stat(folio_order(folio), MTHP_STAT_SWPOUT);
	count_memcg_folio_events(folio, PSWPOUT, folio_nr_pages(folio));
	count_vm_events(PSWPOUT, folio_nr_pages(folio));
}

#if defined(CONFIG_MEMCG) && defined(CONFIG_BLK_CGROUP)
static void bio_associate_blkg_from_page(struct bio *bio, struct folio *folio)
{
	struct cgroup_subsys_state *css;
	struct mem_cgroup *memcg;

	memcg = folio_memcg(folio);
	if (!memcg)
		return;

	rcu_read_lock();
	css = cgroup_e_css(memcg->css.cgroup, &io_cgrp_subsys);
	bio_associate_blkg_from_css(bio, css);
	rcu_read_unlock();
}
#else
#define bio_associate_blkg_from_page(bio, folio)		do { } while (0)
#endif /* CONFIG_MEMCG && CONFIG_BLK_CGROUP */

struct swap_iocb {
	struct kiocb		iocb;
	struct bio_vec		bvec[SWAP_CLUSTER_MAX];
	int			pages;
	int			len;
};
static mempool_t *sio_pool;

int sio_pool_init(void)
{
	if (!sio_pool) {
		mempool_t *pool = mempool_create_kmalloc_pool(
			SWAP_CLUSTER_MAX, sizeof(struct swap_iocb));
		if (cmpxchg(&sio_pool, NULL, pool))
			mempool_destroy(pool);
	}
	if (!sio_pool)
		return -ENOMEM;
	return 0;
}

static void sio_write_complete(struct kiocb *iocb, long ret)
{
	struct swap_iocb *sio = container_of(iocb, struct swap_iocb, iocb);
	struct page *page = sio->bvec[0].bv_page;
	int p;

	if (ret != sio->len) {
		/*
		 * In the case of swap-over-nfs, this can be a
		 * temporary failure if the system has limited
		 * memory for allocating transmit buffers.
		 * Mark the page dirty and avoid
		 * folio_rotate_reclaimable but rate-limit the
		 * messages.
		 */
		pr_err_ratelimited("Write error %ld on dio swapfile (%llu)\n",
				   ret, swap_dev_pos(page_swap_entry(page)));
		for (p = 0; p < sio->pages; p++) {
			page = sio->bvec[p].bv_page;
			set_page_dirty(page);
			ClearPageReclaim(page);
		}
	}

	for (p = 0; p < sio->pages; p++)
		end_page_writeback(sio->bvec[p].bv_page);

	mempool_free(sio, sio_pool);
}

static void swap_writepage_fs(struct folio *folio, struct writeback_control *wbc)
{
	struct swap_iocb *sio = NULL;
	struct swap_info_struct *sis = swp_swap_info(folio->swap);
	struct file *swap_file = sis->swap_file;
	loff_t pos = swap_dev_pos(folio->swap);

	count_swpout_vm_event(folio);
	folio_start_writeback(folio);
	folio_unlock(folio);
	if (wbc->swap_plug)
		sio = *wbc->swap_plug;
	if (sio) {
		if (sio->iocb.ki_filp != swap_file ||
		    sio->iocb.ki_pos + sio->len != pos) {
			swap_write_unplug(sio);
			sio = NULL;
		}
	}
	if (!sio) {
		sio = mempool_alloc(sio_pool, GFP_NOIO);
		init_sync_kiocb(&sio->iocb, swap_file);
		sio->iocb.ki_complete = sio_write_complete;
		sio->iocb.ki_pos = pos;
		sio->pages = 0;
		sio->len = 0;
	}
	bvec_set_folio(&sio->bvec[sio->pages], folio, folio_size(folio), 0);
	sio->len += folio_size(folio);
	sio->pages += 1;
	if (sio->pages == ARRAY_SIZE(sio->bvec) || !wbc->swap_plug) {
		swap_write_unplug(sio);
		sio = NULL;
	}
	if (wbc->swap_plug)
		*wbc->swap_plug = sio;
}

static void swap_writepage_bdev_sync(struct folio *folio,
		struct writeback_control *wbc, struct swap_info_struct *sis)
{
	struct bio_vec bv;
	struct bio bio;

	bio_init(&bio, sis->bdev, &bv, 1,
		 REQ_OP_WRITE | REQ_SWAP | wbc_to_write_flags(wbc));
	bio.bi_iter.bi_sector = swap_folio_sector(folio);
	bio_add_folio_nofail(&bio, folio, folio_size(folio), 0);

	bio_associate_blkg_from_page(&bio, folio);
	count_swpout_vm_event(folio);

	folio_start_writeback(folio);
	folio_unlock(folio);

	submit_bio_wait(&bio);
	__end_swap_bio_write(&bio);
}

static void swap_writepage_bdev_async(struct folio *folio,
		struct writeback_control *wbc, struct swap_info_struct *sis)
{
	struct bio *bio;

	bio = bio_alloc(sis->bdev, 1,
			REQ_OP_WRITE | REQ_SWAP | wbc_to_write_flags(wbc),
			GFP_NOIO);
	bio->bi_iter.bi_sector = swap_folio_sector(folio);
	bio->bi_end_io = end_swap_bio_write;
	bio_add_folio_nofail(bio, folio, folio_size(folio), 0);

	bio_associate_blkg_from_page(bio, folio);
	count_swpout_vm_event(folio);
	folio_start_writeback(folio);
	folio_unlock(folio);
	submit_bio(bio);
}

void __swap_writepage(struct folio *folio, struct writeback_control *wbc)
{
	struct swap_info_struct *sis = swp_swap_info(folio->swap);

	VM_BUG_ON_FOLIO(!folio_test_swapcache(folio), folio);
	/*
	 * ->flags can be updated non-atomicially (scan_swap_map_slots),
	 * but that will never affect SWP_FS_OPS, so the data_race
	 * is safe.
	 */
	if (data_race(sis->flags & SWP_FS_OPS))
		swap_writepage_fs(folio, wbc);
	/*
	 * ->flags can be updated non-atomicially (scan_swap_map_slots),
	 * but that will never affect SWP_SYNCHRONOUS_IO, so the data_race
	 * is safe.
	 */
	else if (data_race(sis->flags & SWP_SYNCHRONOUS_IO))
		swap_writepage_bdev_sync(folio, wbc, sis);
	else
		swap_writepage_bdev_async(folio, wbc, sis);
}

void swap_write_unplug(struct swap_iocb *sio)
{
	struct iov_iter from;
	struct address_space *mapping = sio->iocb.ki_filp->f_mapping;
	int ret;

	iov_iter_bvec(&from, ITER_SOURCE, sio->bvec, sio->pages, sio->len);
	ret = mapping->a_ops->swap_rw(&sio->iocb, &from);
	if (ret != -EIOCBQUEUED)
		sio_write_complete(&sio->iocb, ret);
}

static void sio_read_complete(struct kiocb *iocb, long ret)
{
	struct swap_iocb *sio = container_of(iocb, struct swap_iocb, iocb);
	int p;

	if (ret == sio->len) {
		for (p = 0; p < sio->pages; p++) {
			struct folio *folio = page_folio(sio->bvec[p].bv_page);

			count_mthp_stat(folio_order(folio), MTHP_STAT_SWPIN);
			count_memcg_folio_events(folio, PSWPIN, folio_nr_pages(folio));
			folio_mark_uptodate(folio);
			folio_unlock(folio);
		}
		count_vm_events(PSWPIN, sio->pages);
	} else {
		for (p = 0; p < sio->pages; p++) {
			struct folio *folio = page_folio(sio->bvec[p].bv_page);

			folio_unlock(folio);
		}
		pr_alert_ratelimited("Read-error on swap-device\n");
	}
	mempool_free(sio, sio_pool);
}

static bool swap_read_folio_zeromap(struct folio *folio)
{
	int nr_pages = folio_nr_pages(folio);
	struct obj_cgroup *objcg;
	bool is_zeromap;

	/*
	 * Swapping in a large folio that is partially in the zeromap is not
	 * currently handled. Return true without marking the folio uptodate so
	 * that an IO error is emitted (e.g. do_swap_page() will sigbus).
	 */
	if (WARN_ON_ONCE(swap_zeromap_batch(folio->swap, nr_pages,
			&is_zeromap) != nr_pages))
		return true;

	if (!is_zeromap)
		return false;

	objcg = get_obj_cgroup_from_folio(folio);
	count_vm_events(SWPIN_ZERO, nr_pages);
	if (objcg) {
		count_objcg_events(objcg, SWPIN_ZERO, nr_pages);
		obj_cgroup_put(objcg);
	}

	folio_zero_range(folio, 0, folio_size(folio));
	folio_mark_uptodate(folio);
	return true;
}

static void swap_read_folio_fs(struct folio *folio, struct swap_iocb **plug)
{
	struct swap_info_struct *sis = swp_swap_info(folio->swap);
	struct swap_iocb *sio = NULL;
	loff_t pos = swap_dev_pos(folio->swap);

	if (plug)
		sio = *plug;
	if (sio) {
		if (sio->iocb.ki_filp != sis->swap_file ||
		    sio->iocb.ki_pos + sio->len != pos) {
			swap_read_unplug(sio);
			sio = NULL;
		}
	}
	if (!sio) {
		sio = mempool_alloc(sio_pool, GFP_KERNEL);
		init_sync_kiocb(&sio->iocb, sis->swap_file);
		sio->iocb.ki_pos = pos;
		sio->iocb.ki_complete = sio_read_complete;
		sio->pages = 0;
		sio->len = 0;
	}
	bvec_set_folio(&sio->bvec[sio->pages], folio, folio_size(folio), 0);
	sio->len += folio_size(folio);
	sio->pages += 1;
	if (sio->pages == ARRAY_SIZE(sio->bvec) || !plug) {
		swap_read_unplug(sio);
		sio = NULL;
	}
	if (plug)
		*plug = sio;
}

static void swap_read_folio_bdev_sync(struct folio *folio,
		struct swap_info_struct *sis)
{
	struct bio_vec bv;
	struct bio bio;

	bio_init(&bio, sis->bdev, &bv, 1, REQ_OP_READ);
	bio.bi_iter.bi_sector = swap_folio_sector(folio);
	bio_add_folio_nofail(&bio, folio, folio_size(folio), 0);
	/*
	 * Keep this task valid during swap readpage because the oom killer may
	 * attempt to access it in the page fault retry time check.
	 */
	get_task_struct(current);
	count_mthp_stat(folio_order(folio), MTHP_STAT_SWPIN);
	count_memcg_folio_events(folio, PSWPIN, folio_nr_pages(folio));
	count_vm_events(PSWPIN, folio_nr_pages(folio));
	submit_bio_wait(&bio);
	__end_swap_bio_read(&bio);
	put_task_struct(current);
}

static void swap_read_folio_bdev_async(struct folio *folio,
		struct swap_info_struct *sis)
{
	struct bio *bio;

	bio = bio_alloc(sis->bdev, 1, REQ_OP_READ, GFP_KERNEL);
	bio->bi_iter.bi_sector = swap_folio_sector(folio);
	bio->bi_end_io = end_swap_bio_read;
	bio_add_folio_nofail(bio, folio, folio_size(folio), 0);
	count_mthp_stat(folio_order(folio), MTHP_STAT_SWPIN);
	count_memcg_folio_events(folio, PSWPIN, folio_nr_pages(folio));
	count_vm_events(PSWPIN, folio_nr_pages(folio));
	submit_bio(bio);
}

void swap_read_folio(struct folio *folio, struct swap_iocb **plug)
{
	struct swap_info_struct *sis = swp_swap_info(folio->swap);
	bool synchronous = sis->flags & SWP_SYNCHRONOUS_IO;
	bool workingset = folio_test_workingset(folio);
	unsigned long pflags;
	bool in_thrashing;

	VM_BUG_ON_FOLIO(!folio_test_swapcache(folio) && !synchronous, folio);
	VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio);
	VM_BUG_ON_FOLIO(folio_test_uptodate(folio), folio);

	/*
	 * Count submission time as memory stall and delay. When the device
	 * is congested, or the submitting cgroup IO-throttled, submission
	 * can be a significant part of overall IO time.
	 */
	if (workingset) {
		delayacct_thrashing_start(&in_thrashing);
		psi_memstall_enter(&pflags);
	}
	delayacct_swapin_start();

	if (swap_read_folio_zeromap(folio)) {
		folio_unlock(folio);
		goto finish;
	}

	if (zswap_load(folio) != -ENOENT)
		goto finish;

	/* We have to read from slower devices. Increase zswap protection. */
	zswap_folio_swapin(folio);

	if (data_race(sis->flags & SWP_FS_OPS)) {
		swap_read_folio_fs(folio, plug);
	} else if (synchronous) {
		swap_read_folio_bdev_sync(folio, sis);
	} else {
		swap_read_folio_bdev_async(folio, sis);
	}

finish:
	if (workingset) {
		delayacct_thrashing_end(&in_thrashing);
		psi_memstall_leave(&pflags);
	}
	delayacct_swapin_end();
}

void __swap_read_unplug(struct swap_iocb *sio)
{
	struct iov_iter from;
	struct address_space *mapping = sio->iocb.ki_filp->f_mapping;
	int ret;

	iov_iter_bvec(&from, ITER_DEST, sio->bvec, sio->pages, sio->len);
	ret = mapping->a_ops->swap_rw(&sio->iocb, &from);
	if (ret != -EIOCBQUEUED)
		sio_read_complete(&sio->iocb, ret);
}
Commit	Line	Data
	1	// SPDX-License-Identifier: GPL-2.0
	2	/*
	3	* linux/mm/page_io.c
	4	*
	5	* Copyright (C) 1991, 1992, 1993, 1994 Linus Torvalds
	6	*
	7	* Swap reorganised 29.12.95,
	8	* Asynchronous swapping added 30.12.95. Stephen Tweedie
	9	* Removed race in async swapping. 14.4.1996. Bruno Haible
	10	* Add swap of shared pages through the page cache. 20.2.1998. Stephen Tweedie
	11	* Always use brw_page, life becomes simpler. 12 May 1998 Eric Biederman
	12	*/
	13
	14	#include <linux/mm.h>
	15	#include <linux/kernel_stat.h>
	16	#include <linux/gfp.h>
	17	#include <linux/pagemap.h>
	18	#include <linux/swap.h>
	19	#include <linux/bio.h>
	20	#include <linux/swapops.h>
	21	#include <linux/writeback.h>
	22	#include <linux/blkdev.h>
	23	#include <linux/psi.h>
	24	#include <linux/uio.h>
	25	#include <linux/sched/task.h>
	26	#include <linux/delayacct.h>
	27	#include <linux/zswap.h>
	28	#include "swap.h"
	29
	30	static void __end_swap_bio_write(struct bio *bio)
	31	{
	32	struct folio *folio = bio_first_folio_all(bio);
	33
	34	if (bio->bi_status) {
	35	/*
	36	* We failed to write the page out to swap-space.
	37	* Re-dirty the page in order to avoid it being reclaimed.
	38	* Also print a dire warning that things will go BAD (tm)
	39	* very quickly.
	40	*
	41	* Also clear PG_reclaim to avoid folio_rotate_reclaimable()
	42	*/
	43	folio_mark_dirty(folio);
	44	pr_alert_ratelimited("Write-error on swap-device (%u:%u:%llu)\n",
	45	MAJOR(bio_dev(bio)), MINOR(bio_dev(bio)),
	46	(unsigned long long)bio->bi_iter.bi_sector);
	47	folio_clear_reclaim(folio);
	48	}
	49	folio_end_writeback(folio);
	50	}
	51
	52	static void end_swap_bio_write(struct bio *bio)
	53	{
	54	__end_swap_bio_write(bio);
	55	bio_put(bio);
	56	}
	57
	58	static void __end_swap_bio_read(struct bio *bio)
	59	{
	60	struct folio *folio = bio_first_folio_all(bio);
	61
	62	if (bio->bi_status) {
	63	pr_alert_ratelimited("Read-error on swap-device (%u:%u:%llu)\n",
	64	MAJOR(bio_dev(bio)), MINOR(bio_dev(bio)),
	65	(unsigned long long)bio->bi_iter.bi_sector);
	66	} else {
	67	folio_mark_uptodate(folio);
	68	}
	69	folio_unlock(folio);
	70	}
	71
	72	static void end_swap_bio_read(struct bio *bio)
	73	{
	74	__end_swap_bio_read(bio);
	75	bio_put(bio);
	76	}
	77
	78	int generic_swapfile_activate(struct swap_info_struct *sis,
	79	struct file *swap_file,
	80	sector_t *span)
	81	{
	82	struct address_space *mapping = swap_file->f_mapping;
	83	struct inode *inode = mapping->host;
	84	unsigned blocks_per_page;
	85	unsigned long page_no;
	86	unsigned blkbits;
	87	sector_t probe_block;
	88	sector_t last_block;
	89	sector_t lowest_block = -1;
	90	sector_t highest_block = 0;
	91	int nr_extents = 0;
	92	int ret;
	93
	94	blkbits = inode->i_blkbits;
	95	blocks_per_page = PAGE_SIZE >> blkbits;
	96
	97	/*
	98	* Map all the blocks into the extent tree. This code doesn't try
	99	* to be very smart.
	100	*/
	101	probe_block = 0;
	102	page_no = 0;
	103	last_block = i_size_read(inode) >> blkbits;
	104	while ((probe_block + blocks_per_page) <= last_block &&
	105	page_no < sis->max) {
	106	unsigned block_in_page;
	107	sector_t first_block;
	108
	109	cond_resched();
	110
	111	first_block = probe_block;
	112	ret = bmap(inode, &first_block);
	113	if (ret \|\| !first_block)
	114	goto bad_bmap;
	115
	116	/*
	117	* It must be PAGE_SIZE aligned on-disk
	118	*/
	119	if (first_block & (blocks_per_page - 1)) {
	120	probe_block++;
	121	goto reprobe;
	122	}
	123
	124	for (block_in_page = 1; block_in_page < blocks_per_page;
	125	block_in_page++) {
	126	sector_t block;
	127
	128	block = probe_block + block_in_page;
	129	ret = bmap(inode, &block);
	130	if (ret \|\| !block)
	131	goto bad_bmap;
	132
	133	if (block != first_block + block_in_page) {
	134	/* Discontiguity */
	135	probe_block++;
	136	goto reprobe;
	137	}
	138	}
	139
	140	first_block >>= (PAGE_SHIFT - blkbits);
	141	if (page_no) { /* exclude the header page */
	142	if (first_block < lowest_block)
	143	lowest_block = first_block;
	144	if (first_block > highest_block)
	145	highest_block = first_block;
	146	}
	147
	148	/*
	149	* We found a PAGE_SIZE-length, PAGE_SIZE-aligned run of blocks
	150	*/
	151	ret = add_swap_extent(sis, page_no, 1, first_block);
	152	if (ret < 0)
	153	goto out;
	154	nr_extents += ret;
	155	page_no++;
	156	probe_block += blocks_per_page;
	157	reprobe:
	158	continue;
	159	}
	160	ret = nr_extents;
	161	*span = 1 + highest_block - lowest_block;
	162	if (page_no == 0)
	163	page_no = 1; /* force Empty message */
	164	sis->max = page_no;
	165	sis->pages = page_no - 1;
	166	out:
	167	return ret;
	168	bad_bmap:
	169	pr_err("swapon: swapfile has holes\n");
	170	ret = -EINVAL;
	171	goto out;
	172	}
	173
	174	static bool is_folio_zero_filled(struct folio *folio)
	175	{
	176	unsigned int pos, last_pos;
	177	unsigned long *data;
	178	unsigned int i;
	179
	180	last_pos = PAGE_SIZE / sizeof(*data) - 1;
	181	for (i = 0; i < folio_nr_pages(folio); i++) {
	182	data = kmap_local_folio(folio, i * PAGE_SIZE);
	183	/*
	184	* Check last word first, incase the page is zero-filled at
	185	* the start and has non-zero data at the end, which is common
	186	* in real-world workloads.
	187	*/
	188	if (data[last_pos]) {
	189	kunmap_local(data);
	190	return false;
	191	}
	192	for (pos = 0; pos < last_pos; pos++) {
	193	if (data[pos]) {
	194	kunmap_local(data);
	195	return false;
	196	}
	197	}
	198	kunmap_local(data);
	199	}
	200
	201	return true;
	202	}
	203
	204	static void swap_zeromap_folio_set(struct folio *folio)
	205	{
	206	struct obj_cgroup *objcg = get_obj_cgroup_from_folio(folio);
	207	struct swap_info_struct *sis = swp_swap_info(folio->swap);
	208	int nr_pages = folio_nr_pages(folio);
	209	swp_entry_t entry;
	210	unsigned int i;
	211
	212	for (i = 0; i < folio_nr_pages(folio); i++) {
	213	entry = page_swap_entry(folio_page(folio, i));
	214	set_bit(swp_offset(entry), sis->zeromap);
	215	}
	216
	217	count_vm_events(SWPOUT_ZERO, nr_pages);
	218	if (objcg) {
	219	count_objcg_events(objcg, SWPOUT_ZERO, nr_pages);
	220	obj_cgroup_put(objcg);
	221	}
	222	}
	223
	224	static void swap_zeromap_folio_clear(struct folio *folio)
	225	{
	226	struct swap_info_struct *sis = swp_swap_info(folio->swap);
	227	swp_entry_t entry;
	228	unsigned int i;
	229
	230	for (i = 0; i < folio_nr_pages(folio); i++) {
	231	entry = page_swap_entry(folio_page(folio, i));
	232	clear_bit(swp_offset(entry), sis->zeromap);
	233	}
	234	}
	235
	236	/*
	237	* We may have stale swap cache pages in memory: notice
	238	* them here and get rid of the unnecessary final write.
	239	*/
	240	int swap_writeout(struct folio folio, struct writeback_control wbc)
	241	{
	242	int ret;
	243
	244	if (folio_free_swap(folio)) {
	245	folio_unlock(folio);
	246	return 0;
	247	}
	248	/*
	249	* Arch code may have to preserve more data than just the page
	250	* contents, e.g. memory tags.
	251	*/
	252	ret = arch_prepare_to_swap(folio);
	253	if (ret) {
	254	folio_mark_dirty(folio);
	255	folio_unlock(folio);
	256	return ret;
	257	}
	258
	259	/*
	260	* Use a bitmap (zeromap) to avoid doing IO for zero-filled pages.
	261	* The bits in zeromap are protected by the locked swapcache folio
	262	* and atomic updates are used to protect against read-modify-write
	263	* corruption due to other zero swap entries seeing concurrent updates.
	264	*/
	265	if (is_folio_zero_filled(folio)) {
	266	swap_zeromap_folio_set(folio);
	267	folio_unlock(folio);
	268	return 0;
	269	} else {
	270	/*
	271	* Clear bits this folio occupies in the zeromap to prevent
	272	* zero data being read in from any previous zero writes that
	273	* occupied the same swap entries.
	274	*/
	275	swap_zeromap_folio_clear(folio);
	276	}
	277	if (zswap_store(folio)) {
	278	count_mthp_stat(folio_order(folio), MTHP_STAT_ZSWPOUT);
	279	folio_unlock(folio);
	280	return 0;
	281	}
	282	if (!mem_cgroup_zswap_writeback_enabled(folio_memcg(folio))) {
	283	folio_mark_dirty(folio);
	284	return AOP_WRITEPAGE_ACTIVATE;
	285	}
	286
	287	__swap_writepage(folio, wbc);
	288	return 0;
	289	}
	290
	291	static inline void count_swpout_vm_event(struct folio *folio)
	292	{
	293	#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	294	if (unlikely(folio_test_pmd_mappable(folio))) {
	295	count_memcg_folio_events(folio, THP_SWPOUT, 1);
	296	count_vm_event(THP_SWPOUT);
	297	}
	298	#endif
	299	count_mthp_stat(folio_order(folio), MTHP_STAT_SWPOUT);
	300	count_memcg_folio_events(folio, PSWPOUT, folio_nr_pages(folio));
	301	count_vm_events(PSWPOUT, folio_nr_pages(folio));
	302	}
	303
	304	#if defined(CONFIG_MEMCG) && defined(CONFIG_BLK_CGROUP)
	305	static void bio_associate_blkg_from_page(struct bio bio, struct folio folio)
	306	{
	307	struct cgroup_subsys_state *css;
	308	struct mem_cgroup *memcg;
	309
	310	memcg = folio_memcg(folio);
	311	if (!memcg)
	312	return;
	313
	314	rcu_read_lock();
	315	css = cgroup_e_css(memcg->css.cgroup, &io_cgrp_subsys);
	316	bio_associate_blkg_from_css(bio, css);
	317	rcu_read_unlock();
	318	}
	319	#else
	320	#define bio_associate_blkg_from_page(bio, folio) do { } while (0)
	321	#endif /* CONFIG_MEMCG && CONFIG_BLK_CGROUP */
	322
	323	struct swap_iocb {
	324	struct kiocb iocb;
	325	struct bio_vec bvec[SWAP_CLUSTER_MAX];
	326	int pages;
	327	int len;
	328	};
	329	static mempool_t *sio_pool;
	330
	331	int sio_pool_init(void)
	332	{
	333	if (!sio_pool) {
	334	mempool_t *pool = mempool_create_kmalloc_pool(
	335	SWAP_CLUSTER_MAX, sizeof(struct swap_iocb));
	336	if (cmpxchg(&sio_pool, NULL, pool))
	337	mempool_destroy(pool);
	338	}
	339	if (!sio_pool)
	340	return -ENOMEM;
	341	return 0;
	342	}
	343
	344	static void sio_write_complete(struct kiocb *iocb, long ret)
	345	{
	346	struct swap_iocb *sio = container_of(iocb, struct swap_iocb, iocb);
	347	struct page *page = sio->bvec[0].bv_page;
	348	int p;
	349
	350	if (ret != sio->len) {
	351	/*
	352	* In the case of swap-over-nfs, this can be a
	353	* temporary failure if the system has limited
	354	* memory for allocating transmit buffers.
	355	* Mark the page dirty and avoid
	356	* folio_rotate_reclaimable but rate-limit the
	357	* messages.
	358	*/
	359	pr_err_ratelimited("Write error %ld on dio swapfile (%llu)\n",
	360	ret, swap_dev_pos(page_swap_entry(page)));
	361	for (p = 0; p < sio->pages; p++) {
	362	page = sio->bvec[p].bv_page;
	363	set_page_dirty(page);
	364	ClearPageReclaim(page);
	365	}
	366	}
	367
	368	for (p = 0; p < sio->pages; p++)
	369	end_page_writeback(sio->bvec[p].bv_page);
	370
	371	mempool_free(sio, sio_pool);
	372	}
	373
	374	static void swap_writepage_fs(struct folio folio, struct writeback_control wbc)
	375	{
	376	struct swap_iocb *sio = NULL;
	377	struct swap_info_struct *sis = swp_swap_info(folio->swap);
	378	struct file *swap_file = sis->swap_file;
	379	loff_t pos = swap_dev_pos(folio->swap);
	380
	381	count_swpout_vm_event(folio);
	382	folio_start_writeback(folio);
	383	folio_unlock(folio);
	384	if (wbc->swap_plug)
	385	sio = *wbc->swap_plug;
	386	if (sio) {
	387	if (sio->iocb.ki_filp != swap_file \|\|
	388	sio->iocb.ki_pos + sio->len != pos) {
	389	swap_write_unplug(sio);
	390	sio = NULL;
	391	}
	392	}
	393	if (!sio) {
	394	sio = mempool_alloc(sio_pool, GFP_NOIO);
	395	init_sync_kiocb(&sio->iocb, swap_file);
	396	sio->iocb.ki_complete = sio_write_complete;
	397	sio->iocb.ki_pos = pos;
	398	sio->pages = 0;
	399	sio->len = 0;
	400	}
	401	bvec_set_folio(&sio->bvec[sio->pages], folio, folio_size(folio), 0);
	402	sio->len += folio_size(folio);
	403	sio->pages += 1;
	404	if (sio->pages == ARRAY_SIZE(sio->bvec) \|\| !wbc->swap_plug) {
	405	swap_write_unplug(sio);
	406	sio = NULL;
	407	}
	408	if (wbc->swap_plug)
	409	*wbc->swap_plug = sio;
	410	}
	411
	412	static void swap_writepage_bdev_sync(struct folio *folio,
	413	struct writeback_control wbc, struct swap_info_struct sis)
	414	{
	415	struct bio_vec bv;
	416	struct bio bio;
	417
	418	bio_init(&bio, sis->bdev, &bv, 1,
	419	REQ_OP_WRITE \| REQ_SWAP \| wbc_to_write_flags(wbc));
	420	bio.bi_iter.bi_sector = swap_folio_sector(folio);
	421	bio_add_folio_nofail(&bio, folio, folio_size(folio), 0);
	422
	423	bio_associate_blkg_from_page(&bio, folio);
	424	count_swpout_vm_event(folio);
	425
	426	folio_start_writeback(folio);
	427	folio_unlock(folio);
	428
	429	submit_bio_wait(&bio);
	430	__end_swap_bio_write(&bio);
	431	}
	432
	433	static void swap_writepage_bdev_async(struct folio *folio,
	434	struct writeback_control wbc, struct swap_info_struct sis)
	435	{
	436	struct bio *bio;
	437
	438	bio = bio_alloc(sis->bdev, 1,
	439	REQ_OP_WRITE \| REQ_SWAP \| wbc_to_write_flags(wbc),
	440	GFP_NOIO);
	441	bio->bi_iter.bi_sector = swap_folio_sector(folio);
	442	bio->bi_end_io = end_swap_bio_write;
	443	bio_add_folio_nofail(bio, folio, folio_size(folio), 0);
	444
	445	bio_associate_blkg_from_page(bio, folio);
	446	count_swpout_vm_event(folio);
	447	folio_start_writeback(folio);
	448	folio_unlock(folio);
	449	submit_bio(bio);
	450	}
	451
	452	void __swap_writepage(struct folio folio, struct writeback_control wbc)
	453	{
	454	struct swap_info_struct *sis = swp_swap_info(folio->swap);
	455
	456	VM_BUG_ON_FOLIO(!folio_test_swapcache(folio), folio);
	457	/*
	458	* ->flags can be updated non-atomicially (scan_swap_map_slots),
	459	* but that will never affect SWP_FS_OPS, so the data_race
	460	* is safe.
	461	*/
	462	if (data_race(sis->flags & SWP_FS_OPS))
	463	swap_writepage_fs(folio, wbc);
	464	/*
	465	* ->flags can be updated non-atomicially (scan_swap_map_slots),
	466	* but that will never affect SWP_SYNCHRONOUS_IO, so the data_race
	467	* is safe.
	468	*/
	469	else if (data_race(sis->flags & SWP_SYNCHRONOUS_IO))
	470	swap_writepage_bdev_sync(folio, wbc, sis);
	471	else
	472	swap_writepage_bdev_async(folio, wbc, sis);
	473	}
	474
	475	void swap_write_unplug(struct swap_iocb *sio)
	476	{
	477	struct iov_iter from;
	478	struct address_space *mapping = sio->iocb.ki_filp->f_mapping;
	479	int ret;
	480
	481	iov_iter_bvec(&from, ITER_SOURCE, sio->bvec, sio->pages, sio->len);
	482	ret = mapping->a_ops->swap_rw(&sio->iocb, &from);
	483	if (ret != -EIOCBQUEUED)
	484	sio_write_complete(&sio->iocb, ret);
	485	}
	486
	487	static void sio_read_complete(struct kiocb *iocb, long ret)
	488	{
	489	struct swap_iocb *sio = container_of(iocb, struct swap_iocb, iocb);
	490	int p;
	491
	492	if (ret == sio->len) {
	493	for (p = 0; p < sio->pages; p++) {
	494	struct folio *folio = page_folio(sio->bvec[p].bv_page);
	495
	496	count_mthp_stat(folio_order(folio), MTHP_STAT_SWPIN);
	497	count_memcg_folio_events(folio, PSWPIN, folio_nr_pages(folio));
	498	folio_mark_uptodate(folio);
	499	folio_unlock(folio);
	500	}
	501	count_vm_events(PSWPIN, sio->pages);
	502	} else {
	503	for (p = 0; p < sio->pages; p++) {
	504	struct folio *folio = page_folio(sio->bvec[p].bv_page);
	505
	506	folio_unlock(folio);
	507	}
	508	pr_alert_ratelimited("Read-error on swap-device\n");
	509	}
	510	mempool_free(sio, sio_pool);
	511	}
	512
	513	static bool swap_read_folio_zeromap(struct folio *folio)
	514	{
	515	int nr_pages = folio_nr_pages(folio);
	516	struct obj_cgroup *objcg;
	517	bool is_zeromap;
	518
	519	/*
	520	* Swapping in a large folio that is partially in the zeromap is not
	521	* currently handled. Return true without marking the folio uptodate so
	522	* that an IO error is emitted (e.g. do_swap_page() will sigbus).
	523	*/
	524	if (WARN_ON_ONCE(swap_zeromap_batch(folio->swap, nr_pages,
	525	&is_zeromap) != nr_pages))
	526	return true;
	527
	528	if (!is_zeromap)
	529	return false;
	530
	531	objcg = get_obj_cgroup_from_folio(folio);
	532	count_vm_events(SWPIN_ZERO, nr_pages);
	533	if (objcg) {
	534	count_objcg_events(objcg, SWPIN_ZERO, nr_pages);
	535	obj_cgroup_put(objcg);
	536	}
	537
	538	folio_zero_range(folio, 0, folio_size(folio));
	539	folio_mark_uptodate(folio);
	540	return true;
	541	}
	542
	543	static void swap_read_folio_fs(struct folio folio, struct swap_iocb *plug)
	544	{
	545	struct swap_info_struct *sis = swp_swap_info(folio->swap);
	546	struct swap_iocb *sio = NULL;
	547	loff_t pos = swap_dev_pos(folio->swap);
	548
	549	if (plug)
	550	sio = *plug;
	551	if (sio) {
	552	if (sio->iocb.ki_filp != sis->swap_file \|\|
	553	sio->iocb.ki_pos + sio->len != pos) {
	554	swap_read_unplug(sio);
	555	sio = NULL;
	556	}
	557	}
	558	if (!sio) {
	559	sio = mempool_alloc(sio_pool, GFP_KERNEL);
	560	init_sync_kiocb(&sio->iocb, sis->swap_file);
	561	sio->iocb.ki_pos = pos;
	562	sio->iocb.ki_complete = sio_read_complete;
	563	sio->pages = 0;
	564	sio->len = 0;
	565	}
	566	bvec_set_folio(&sio->bvec[sio->pages], folio, folio_size(folio), 0);
	567	sio->len += folio_size(folio);
	568	sio->pages += 1;
	569	if (sio->pages == ARRAY_SIZE(sio->bvec) \|\| !plug) {
	570	swap_read_unplug(sio);
	571	sio = NULL;
	572	}
	573	if (plug)
	574	*plug = sio;
	575	}
	576
	577	static void swap_read_folio_bdev_sync(struct folio *folio,
	578	struct swap_info_struct *sis)
	579	{
	580	struct bio_vec bv;
	581	struct bio bio;
	582
	583	bio_init(&bio, sis->bdev, &bv, 1, REQ_OP_READ);
	584	bio.bi_iter.bi_sector = swap_folio_sector(folio);
	585	bio_add_folio_nofail(&bio, folio, folio_size(folio), 0);
	586	/*
	587	* Keep this task valid during swap readpage because the oom killer may
	588	* attempt to access it in the page fault retry time check.
	589	*/
	590	get_task_struct(current);
	591	count_mthp_stat(folio_order(folio), MTHP_STAT_SWPIN);
	592	count_memcg_folio_events(folio, PSWPIN, folio_nr_pages(folio));
	593	count_vm_events(PSWPIN, folio_nr_pages(folio));
	594	submit_bio_wait(&bio);
	595	__end_swap_bio_read(&bio);
	596	put_task_struct(current);
	597	}
	598
	599	static void swap_read_folio_bdev_async(struct folio *folio,
	600	struct swap_info_struct *sis)
	601	{
	602	struct bio *bio;
	603
	604	bio = bio_alloc(sis->bdev, 1, REQ_OP_READ, GFP_KERNEL);
	605	bio->bi_iter.bi_sector = swap_folio_sector(folio);
	606	bio->bi_end_io = end_swap_bio_read;
	607	bio_add_folio_nofail(bio, folio, folio_size(folio), 0);
	608	count_mthp_stat(folio_order(folio), MTHP_STAT_SWPIN);
	609	count_memcg_folio_events(folio, PSWPIN, folio_nr_pages(folio));
	610	count_vm_events(PSWPIN, folio_nr_pages(folio));
	611	submit_bio(bio);
	612	}
	613
	614	void swap_read_folio(struct folio folio, struct swap_iocb *plug)
	615	{
	616	struct swap_info_struct *sis = swp_swap_info(folio->swap);
	617	bool synchronous = sis->flags & SWP_SYNCHRONOUS_IO;
	618	bool workingset = folio_test_workingset(folio);
	619	unsigned long pflags;
	620	bool in_thrashing;
	621
	622	VM_BUG_ON_FOLIO(!folio_test_swapcache(folio) && !synchronous, folio);
	623	VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio);
	624	VM_BUG_ON_FOLIO(folio_test_uptodate(folio), folio);
	625
	626	/*
	627	* Count submission time as memory stall and delay. When the device
	628	* is congested, or the submitting cgroup IO-throttled, submission
	629	* can be a significant part of overall IO time.
	630	*/
	631	if (workingset) {
	632	delayacct_thrashing_start(&in_thrashing);
	633	psi_memstall_enter(&pflags);
	634	}
	635	delayacct_swapin_start();
	636
	637	if (swap_read_folio_zeromap(folio)) {
	638	folio_unlock(folio);
	639	goto finish;
	640	}
	641
	642	if (zswap_load(folio) != -ENOENT)
	643	goto finish;
	644
	645	/* We have to read from slower devices. Increase zswap protection. */
	646	zswap_folio_swapin(folio);
	647
	648	if (data_race(sis->flags & SWP_FS_OPS)) {
	649	swap_read_folio_fs(folio, plug);
	650	} else if (synchronous) {
	651	swap_read_folio_bdev_sync(folio, sis);
	652	} else {
	653	swap_read_folio_bdev_async(folio, sis);
	654	}
	655
	656	finish:
	657	if (workingset) {
	658	delayacct_thrashing_end(&in_thrashing);
	659	psi_memstall_leave(&pflags);
	660	}
	661	delayacct_swapin_end();
	662	}
	663
	664	void __swap_read_unplug(struct swap_iocb *sio)
	665	{
	666	struct iov_iter from;
	667	struct address_space *mapping = sio->iocb.ki_filp->f_mapping;
	668	int ret;
	669
	670	iov_iter_bvec(&from, ITER_DEST, sio->bvec, sio->pages, sio->len);
	671	ret = mapping->a_ops->swap_rw(&sio->iocb, &from);
	672	if (ret != -EIOCBQUEUED)
	673	sio_read_complete(&sio->iocb, ret);
	674	}