[linux-2.6-block.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;
	sis->highest_bit = page_no - 1;
out:
	return ret;
bad_bmap:
	pr_err("swapon: swapfile has holes\n");
	ret = -EINVAL;
	goto out;
}

/*
 * We may have stale swap cache pages in memory: notice
 * them here and get rid of the unnecessary final write.
 */
int swap_writepage(struct page *page, struct writeback_control *wbc)
{
	struct folio *folio = page_folio(page);
	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;
	}
	if (zswap_store(folio)) {
		folio_start_writeback(folio);
		folio_unlock(folio);
		folio_end_writeback(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);
	}
	count_mthp_stat(folio_order(folio), MTHP_STAT_ANON_SWPOUT);
#endif
	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 but do not flag PageError like
		 * the normal direct-to-bio case as it could
		 * be temporary.
		 */
		pr_err_ratelimited("Write error %ld on dio swapfile (%llu)\n",
				   ret, page_file_offset(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 = folio_file_pos(folio);

	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);
	else if (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);

			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 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 = folio_file_pos(folio);

	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_vm_event(PSWPIN);
	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_vm_event(PSWPIN);
	submit_bio(bio);
}

void swap_read_folio(struct folio *folio, bool synchronous,
		struct swap_iocb **plug)
{
	struct swap_info_struct *sis = swp_swap_info(folio->swap);
	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 (zswap_load(folio)) {
		folio_mark_uptodate(folio);
		folio_unlock(folio);
	} else if (data_race(sis->flags & SWP_FS_OPS)) {
		swap_read_folio_fs(folio, plug);
	} else if (synchronous || (sis->flags & SWP_SYNCHRONOUS_IO)) {
		swap_read_folio_bdev_sync(folio, sis);
	} else {
		swap_read_folio_bdev_async(folio, sis);
	}

	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
b2441318	1	// SPDX-License-Identifier: GPL-2.0
1da177e4 LT	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>
5a0e3ad6	16	#include <linux/gfp.h>
1da177e4 LT	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>
b430e9d1	22	#include <linux/blkdev.h>
93779069	23	#include <linux/psi.h>
e2e40f2c	24	#include <linux/uio.h>
b0ba2d0f	25	#include <linux/sched/task.h>
a3d5dc90	26	#include <linux/delayacct.h>
42c06a0e	27	#include <linux/zswap.h>
014bb1de	28	#include "swap.h"
1da177e4	29
3222d8c2	30	static void __end_swap_bio_write(struct bio *bio)
1da177e4	31	{
a3ed1e9b	32	struct folio *folio = bio_first_folio_all(bio);
1da177e4	33
4e4cbee9	34	if (bio->bi_status) {
6ddab3b9 PZ	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	*
575ced1c	41	* Also clear PG_reclaim to avoid folio_rotate_reclaimable()
6ddab3b9	42	*/
a3ed1e9b	43	folio_mark_dirty(folio);
25eaab43 GD	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);
a3ed1e9b	47	folio_clear_reclaim(folio);
6ddab3b9	48	}
a3ed1e9b	49	folio_end_writeback(folio);
3222d8c2 CH	50	}
	51
	52	static void end_swap_bio_write(struct bio *bio)
	53	{
	54	__end_swap_bio_write(bio);
1da177e4	55	bio_put(bio);
1da177e4 LT	56	}
1da177e4 LT	57
9b4e30bd	58	static void __end_swap_bio_read(struct bio *bio)
1da177e4	59	{
bc74b53f	60	struct folio *folio = bio_first_folio_all(bio);
1da177e4	61
4e4cbee9	62	if (bio->bi_status) {
25eaab43 GD	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);
9b4e30bd	66	} else {
bc74b53f	67	folio_mark_uptodate(folio);
1da177e4	68	}
bc74b53f	69	folio_unlock(folio);
9b4e30bd CH	70	}
	71
	72	static void end_swap_bio_read(struct bio *bio)
	73	{
	74	__end_swap_bio_read(bio);
1da177e4	75	bio_put(bio);
1da177e4 LT	76	}
1da177e4 LT	77
a509bc1a MG	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	/*
4efaceb1	98	* Map all the blocks into the extent tree. This code doesn't try
a509bc1a MG	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
7e4411bf MP	109	cond_resched();
7e4411bf MP	110
30460e1e CM	111	first_block = probe_block;
	112	ret = bmap(inode, &first_block);
	113	if (ret \|\| !first_block)
a509bc1a MG	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
30460e1e CM	128	block = probe_block + block_in_page;
	129	ret = bmap(inode, &block);
	130	if (ret \|\| !block)
a509bc1a	131	goto bad_bmap;
30460e1e	132
a509bc1a MG	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	sis->highest_bit = page_no - 1;
	167	out:
	168	return ret;
	169	bad_bmap:
1170532b	170	pr_err("swapon: swapfile has holes\n");
a509bc1a MG	171	ret = -EINVAL;
	172	goto out;
	173	}
	174
1da177e4 LT	175	/*
	176	* We may have stale swap cache pages in memory: notice
	177	* them here and get rid of the unnecessary final write.
	178	*/
	179	int swap_writepage(struct page page, struct writeback_control wbc)
	180	{
71fa1a53	181	struct folio *folio = page_folio(page);
e3e2762b	182	int ret;
1da177e4	183
71fa1a53 MWO	184	if (folio_free_swap(folio)) {
71fa1a53 MWO	185	folio_unlock(folio);
e3e2762b	186	return 0;
1da177e4	187	}
8a84802e SP	188	/*
	189	* Arch code may have to preserve more data than just the page
	190	* contents, e.g. memory tags.
	191	*/
f238b8c3	192	ret = arch_prepare_to_swap(folio);
8a84802e	193	if (ret) {
71fa1a53 MWO	194	folio_mark_dirty(folio);
71fa1a53 MWO	195	folio_unlock(folio);
e3e2762b	196	return ret;
8a84802e	197	}
34f4c198	198	if (zswap_store(folio)) {
71fa1a53 MWO	199	folio_start_writeback(folio);
	200	folio_unlock(folio);
	201	folio_end_writeback(folio);
e3e2762b	202	return 0;
38b5faf4	203	}
501a06fe NP	204	if (!mem_cgroup_zswap_writeback_enabled(folio_memcg(folio))) {
	205	folio_mark_dirty(folio);
	206	return AOP_WRITEPAGE_ACTIVATE;
	207	}
	208
b99b4e0d	209	__swap_writepage(folio, wbc);
e3e2762b	210	return 0;
2f772e6c SJ	211	}
2f772e6c SJ	212
9b72b134	213	static inline void count_swpout_vm_event(struct folio *folio)
225311a4 HY	214	{
225311a4 HY	215	#ifdef CONFIG_TRANSPARENT_HUGEPAGE
811244a5 XH	216	if (unlikely(folio_test_pmd_mappable(folio))) {
811244a5 XH	217	count_memcg_folio_events(folio, THP_SWPOUT, 1);
225311a4	218	count_vm_event(THP_SWPOUT);
811244a5	219	}
d0f048ac	220	count_mthp_stat(folio_order(folio), MTHP_STAT_ANON_SWPOUT);
225311a4	221	#endif
9b72b134	222	count_vm_events(PSWPOUT, folio_nr_pages(folio));
225311a4 HY	223	}
225311a4 HY	224
a18b9b15	225	#if defined(CONFIG_MEMCG) && defined(CONFIG_BLK_CGROUP)
98630cfd	226	static void bio_associate_blkg_from_page(struct bio bio, struct folio folio)
a18b9b15 CH	227	{
a18b9b15 CH	228	struct cgroup_subsys_state *css;
bcfe06bf	229	struct mem_cgroup *memcg;
a18b9b15	230
98630cfd	231	memcg = folio_memcg(folio);
bcfe06bf	232	if (!memcg)
a18b9b15 CH	233	return;
	234
	235	rcu_read_lock();
bcfe06bf	236	css = cgroup_e_css(memcg->css.cgroup, &io_cgrp_subsys);
a18b9b15 CH	237	bio_associate_blkg_from_css(bio, css);
	238	rcu_read_unlock();
	239	}
	240	#else
98630cfd	241	#define bio_associate_blkg_from_page(bio, folio) do { } while (0)
a18b9b15 CH	242	#endif /* CONFIG_MEMCG && CONFIG_BLK_CGROUP */
a18b9b15 CH	243
e1209d3a N	244	struct swap_iocb {
e1209d3a N	245	struct kiocb iocb;
5169b844 N	246	struct bio_vec bvec[SWAP_CLUSTER_MAX];
5169b844 N	247	int pages;
a1a0dfd5	248	int len;
e1209d3a N	249	};
	250	static mempool_t *sio_pool;
	251
	252	int sio_pool_init(void)
2f772e6c	253	{
e1209d3a N	254	if (!sio_pool) {
	255	mempool_t *pool = mempool_create_kmalloc_pool(
	256	SWAP_CLUSTER_MAX, sizeof(struct swap_iocb));
	257	if (cmpxchg(&sio_pool, NULL, pool))
	258	mempool_destroy(pool);
	259	}
	260	if (!sio_pool)
	261	return -ENOMEM;
	262	return 0;
	263	}
62c230bc	264
7eadabc0 N	265	static void sio_write_complete(struct kiocb *iocb, long ret)
	266	{
	267	struct swap_iocb *sio = container_of(iocb, struct swap_iocb, iocb);
5169b844	268	struct page *page = sio->bvec[0].bv_page;
2282679f	269	int p;
62c230bc	270
a1a0dfd5	271	if (ret != sio->len) {
7eadabc0 N	272	/*
	273	* In the case of swap-over-nfs, this can be a
	274	* temporary failure if the system has limited
	275	* memory for allocating transmit buffers.
	276	* Mark the page dirty and avoid
	277	* folio_rotate_reclaimable but rate-limit the
	278	* messages but do not flag PageError like
	279	* the normal direct-to-bio case as it could
	280	* be temporary.
	281	*/
7eadabc0 N	282	pr_err_ratelimited("Write error %ld on dio swapfile (%llu)\n",
7eadabc0 N	283	ret, page_file_offset(page));
2282679f N	284	for (p = 0; p < sio->pages; p++) {
2282679f N	285	page = sio->bvec[p].bv_page;
2d30d31e	286	set_page_dirty(page);
0cdc444a	287	ClearPageReclaim(page);
62c230bc	288	}
62c230bc MG	289	}
62c230bc MG	290
2282679f N	291	for (p = 0; p < sio->pages; p++)
	292	end_page_writeback(sio->bvec[p].bv_page);
	293
7eadabc0 N	294	mempool_free(sio, sio_pool);
	295	}
	296
bfcd44d5	297	static void swap_writepage_fs(struct folio folio, struct writeback_control wbc)
7eadabc0	298	{
2282679f	299	struct swap_iocb *sio = NULL;
bfcd44d5	300	struct swap_info_struct *sis = swp_swap_info(folio->swap);
7eadabc0	301	struct file *swap_file = sis->swap_file;
bfcd44d5	302	loff_t pos = folio_file_pos(folio);
7eadabc0	303
bfcd44d5 MWO	304	count_swpout_vm_event(folio);
	305	folio_start_writeback(folio);
	306	folio_unlock(folio);
2282679f N	307	if (wbc->swap_plug)
	308	sio = *wbc->swap_plug;
	309	if (sio) {
	310	if (sio->iocb.ki_filp != swap_file \|\|
a1a0dfd5	311	sio->iocb.ki_pos + sio->len != pos) {
2282679f N	312	swap_write_unplug(sio);
	313	sio = NULL;
	314	}
	315	}
	316	if (!sio) {
	317	sio = mempool_alloc(sio_pool, GFP_NOIO);
	318	init_sync_kiocb(&sio->iocb, swap_file);
	319	sio->iocb.ki_complete = sio_write_complete;
	320	sio->iocb.ki_pos = pos;
	321	sio->pages = 0;
a1a0dfd5	322	sio->len = 0;
2282679f	323	}
bfcd44d5 MWO	324	bvec_set_folio(&sio->bvec[sio->pages], folio, folio_size(folio), 0);
bfcd44d5 MWO	325	sio->len += folio_size(folio);
2282679f N	326	sio->pages += 1;
	327	if (sio->pages == ARRAY_SIZE(sio->bvec) \|\| !wbc->swap_plug) {
	328	swap_write_unplug(sio);
	329	sio = NULL;
	330	}
	331	if (wbc->swap_plug)
	332	*wbc->swap_plug = sio;
7eadabc0 N	333	}
7eadabc0 N	334
6de62c7b	335	static void swap_writepage_bdev_sync(struct folio *folio,
05cda97e	336	struct writeback_control wbc, struct swap_info_struct sis)
2f772e6c	337	{
3222d8c2 CH	338	struct bio_vec bv;
3222d8c2 CH	339	struct bio bio;
62c230bc	340
3222d8c2 CH	341	bio_init(&bio, sis->bdev, &bv, 1,
3222d8c2 CH	342	REQ_OP_WRITE \| REQ_SWAP \| wbc_to_write_flags(wbc));
3a61e6f6	343	bio.bi_iter.bi_sector = swap_folio_sector(folio);
6de62c7b	344	bio_add_folio_nofail(&bio, folio, folio_size(folio), 0);
62c230bc	345
98630cfd	346	bio_associate_blkg_from_page(&bio, folio);
9b72b134	347	count_swpout_vm_event(folio);
3222d8c2	348
f54fcaab Z	349	folio_start_writeback(folio);
f54fcaab Z	350	folio_unlock(folio);
3222d8c2 CH	351
	352	submit_bio_wait(&bio);
	353	__end_swap_bio_write(&bio);
	354	}
	355
ee1b1d9b	356	static void swap_writepage_bdev_async(struct folio *folio,
3222d8c2 CH	357	struct writeback_control wbc, struct swap_info_struct sis)
	358	{
	359	struct bio *bio;
dd6bd0d9	360
07888c66 CH	361	bio = bio_alloc(sis->bdev, 1,
	362	REQ_OP_WRITE \| REQ_SWAP \| wbc_to_write_flags(wbc),
	363	GFP_NOIO);
3a61e6f6	364	bio->bi_iter.bi_sector = swap_folio_sector(folio);
cf1e3fe4	365	bio->bi_end_io = end_swap_bio_write;
ee1b1d9b	366	bio_add_folio_nofail(bio, folio, folio_size(folio), 0);
48d15436	367
98630cfd	368	bio_associate_blkg_from_page(bio, folio);
9b72b134	369	count_swpout_vm_event(folio);
2675251d Z	370	folio_start_writeback(folio);
2675251d Z	371	folio_unlock(folio);
4e49ea4a	372	submit_bio(bio);
1da177e4	373	}
548d9782	374
b99b4e0d	375	void __swap_writepage(struct folio folio, struct writeback_control wbc)
05cda97e	376	{
b99b4e0d	377	struct swap_info_struct *sis = swp_swap_info(folio->swap);
05cda97e	378
b99b4e0d	379	VM_BUG_ON_FOLIO(!folio_test_swapcache(folio), folio);
05cda97e CH	380	/*
	381	* ->flags can be updated non-atomicially (scan_swap_map_slots),
	382	* but that will never affect SWP_FS_OPS, so the data_race
	383	* is safe.
	384	*/
	385	if (data_race(sis->flags & SWP_FS_OPS))
bfcd44d5	386	swap_writepage_fs(folio, wbc);
3222d8c2	387	else if (sis->flags & SWP_SYNCHRONOUS_IO)
6de62c7b	388	swap_writepage_bdev_sync(folio, wbc, sis);
05cda97e	389	else
ee1b1d9b	390	swap_writepage_bdev_async(folio, wbc, sis);
1da177e4 LT	391	}
1da177e4 LT	392
2282679f N	393	void swap_write_unplug(struct swap_iocb *sio)
	394	{
	395	struct iov_iter from;
	396	struct address_space *mapping = sio->iocb.ki_filp->f_mapping;
	397	int ret;
	398
de4eda9d	399	iov_iter_bvec(&from, ITER_SOURCE, sio->bvec, sio->pages, sio->len);
2282679f N	400	ret = mapping->a_ops->swap_rw(&sio->iocb, &from);
	401	if (ret != -EIOCBQUEUED)
	402	sio_write_complete(&sio->iocb, ret);
	403	}
	404
e1209d3a N	405	static void sio_read_complete(struct kiocb *iocb, long ret)
	406	{
	407	struct swap_iocb *sio = container_of(iocb, struct swap_iocb, iocb);
5169b844	408	int p;
e1209d3a	409
a1a0dfd5	410	if (ret == sio->len) {
5169b844	411	for (p = 0; p < sio->pages; p++) {
6a8c0687	412	struct folio *folio = page_folio(sio->bvec[p].bv_page);
5169b844	413
6a8c0687 Z	414	folio_mark_uptodate(folio);
6a8c0687 Z	415	folio_unlock(folio);
5169b844 N	416	}
5169b844 N	417	count_vm_events(PSWPIN, sio->pages);
e1209d3a	418	} else {
5169b844	419	for (p = 0; p < sio->pages; p++) {
6a8c0687	420	struct folio *folio = page_folio(sio->bvec[p].bv_page);
5169b844	421
6a8c0687	422	folio_unlock(folio);
5169b844 N	423	}
5169b844 N	424	pr_alert_ratelimited("Read-error on swap-device\n");
e1209d3a	425	}
e1209d3a N	426	mempool_free(sio, sio_pool);
	427	}
	428
c9bdf768	429	static void swap_read_folio_fs(struct folio folio, struct swap_iocb *plug)
e1209d3a	430	{
64a24e55	431	struct swap_info_struct *sis = swp_swap_info(folio->swap);
5169b844	432	struct swap_iocb *sio = NULL;
64a24e55	433	loff_t pos = folio_file_pos(folio);
e1209d3a	434
5169b844 N	435	if (plug)
	436	sio = *plug;
	437	if (sio) {
	438	if (sio->iocb.ki_filp != sis->swap_file \|\|
a1a0dfd5	439	sio->iocb.ki_pos + sio->len != pos) {
5169b844 N	440	swap_read_unplug(sio);
	441	sio = NULL;
	442	}
	443	}
	444	if (!sio) {
	445	sio = mempool_alloc(sio_pool, GFP_KERNEL);
	446	init_sync_kiocb(&sio->iocb, sis->swap_file);
	447	sio->iocb.ki_pos = pos;
	448	sio->iocb.ki_complete = sio_read_complete;
	449	sio->pages = 0;
a1a0dfd5	450	sio->len = 0;
5169b844	451	}
64a24e55 MWO	452	bvec_set_folio(&sio->bvec[sio->pages], folio, folio_size(folio), 0);
64a24e55 MWO	453	sio->len += folio_size(folio);
5169b844 N	454	sio->pages += 1;
	455	if (sio->pages == ARRAY_SIZE(sio->bvec) \|\| !plug) {
	456	swap_read_unplug(sio);
	457	sio = NULL;
	458	}
	459	if (plug)
	460	*plug = sio;
e1209d3a N	461	}
e1209d3a N	462
c9bdf768	463	static void swap_read_folio_bdev_sync(struct folio *folio,
14bd75f5	464	struct swap_info_struct *sis)
1da177e4	465	{
9b4e30bd CH	466	struct bio_vec bv;
9b4e30bd CH	467	struct bio bio;
62c230bc	468
9b4e30bd	469	bio_init(&bio, sis->bdev, &bv, 1, REQ_OP_READ);
3a61e6f6	470	bio.bi_iter.bi_sector = swap_folio_sector(folio);
2c184d82	471	bio_add_folio_nofail(&bio, folio, folio_size(folio), 0);
b0ba2d0f TH	472	/*
	473	* Keep this task valid during swap readpage because the oom killer may
	474	* attempt to access it in the page fault retry time check.
	475	*/
9b4e30bd	476	get_task_struct(current);
f8891e5e	477	count_vm_event(PSWPIN);
9b4e30bd CH	478	submit_bio_wait(&bio);
	479	__end_swap_bio_read(&bio);
	480	put_task_struct(current);
	481	}
	482
c9bdf768	483	static void swap_read_folio_bdev_async(struct folio *folio,
9b4e30bd	484	struct swap_info_struct *sis)
1da177e4 LT	485	{
1da177e4 LT	486	struct bio *bio;
23955622	487
9b4e30bd	488	bio = bio_alloc(sis->bdev, 1, REQ_OP_READ, GFP_KERNEL);
3a61e6f6	489	bio->bi_iter.bi_sector = swap_folio_sector(folio);
9b4e30bd	490	bio->bi_end_io = end_swap_bio_read;
3c3ebd82	491	bio_add_folio_nofail(bio, folio, folio_size(folio), 0);
9b4e30bd CH	492	count_vm_event(PSWPIN);
9b4e30bd CH	493	submit_bio(bio);
14bd75f5 CH	494	}
14bd75f5 CH	495
c9bdf768 MWO	496	void swap_read_folio(struct folio *folio, bool synchronous,
c9bdf768 MWO	497	struct swap_iocb **plug)
14bd75f5	498	{
c9bdf768	499	struct swap_info_struct *sis = swp_swap_info(folio->swap);
fbcec6a3	500	bool workingset = folio_test_workingset(folio);
93779069	501	unsigned long pflags;
3a9bb7b1	502	bool in_thrashing;
1da177e4	503
fbcec6a3 MWO	504	VM_BUG_ON_FOLIO(!folio_test_swapcache(folio) && !synchronous, folio);
	505	VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio);
	506	VM_BUG_ON_FOLIO(folio_test_uptodate(folio), folio);
93779069 MK	507
93779069 MK	508	/*
3a9bb7b1 YY	509	* Count submission time as memory stall and delay. When the device
	510	* is congested, or the submitting cgroup IO-throttled, submission
	511	* can be a significant part of overall IO time.
93779069	512	*/
3a9bb7b1 YY	513	if (workingset) {
3a9bb7b1 YY	514	delayacct_thrashing_start(&in_thrashing);
d8c47cc7	515	psi_memstall_enter(&pflags);
3a9bb7b1	516	}
a3d5dc90	517	delayacct_swapin_start();
93779069	518
ca54f6d8	519	if (zswap_load(folio)) {
fbcec6a3 MWO	520	folio_mark_uptodate(folio);
fbcec6a3 MWO	521	folio_unlock(folio);
14bd75f5	522	} else if (data_race(sis->flags & SWP_FS_OPS)) {
c9bdf768	523	swap_read_folio_fs(folio, plug);
3222d8c2	524	} else if (synchronous \|\| (sis->flags & SWP_SYNCHRONOUS_IO)) {
c9bdf768	525	swap_read_folio_bdev_sync(folio, sis);
14bd75f5	526	} else {
c9bdf768	527	swap_read_folio_bdev_async(folio, sis);
23955622	528	}
23955622	529
3a9bb7b1 YY	530	if (workingset) {
3a9bb7b1 YY	531	delayacct_thrashing_end(&in_thrashing);
d8c47cc7	532	psi_memstall_leave(&pflags);
3a9bb7b1	533	}
a3d5dc90	534	delayacct_swapin_end();
1da177e4	535	}
62c230bc	536
5169b844	537	void __swap_read_unplug(struct swap_iocb *sio)
62c230bc	538	{
5169b844 N	539	struct iov_iter from;
	540	struct address_space *mapping = sio->iocb.ki_filp->f_mapping;
	541	int ret;
cc30c5d6	542
de4eda9d	543	iov_iter_bvec(&from, ITER_DEST, sio->bvec, sio->pages, sio->len);
5169b844 N	544	ret = mapping->a_ops->swap_rw(&sio->iocb, &from);
	545	if (ret != -EIOCBQUEUED)
	546	sio_read_complete(&sio->iocb, ret);
62c230bc	547	}