Merge tag 'f2fs-for-v5.2-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs

Pull f2fs updates from Jaegeuk Kim: "Another round of various bug fixes came in. Damien improved SMR drive support a bit, and Chao replaced BUG_ON() with reporting errors to user since we've not hit from users but did hit from crafted images. We've found a disk layout bug in large_nat_bits feature which supports very large NAT entries enabled at mkfs. If the feature is enabled, it will give a notice to run fsck to correct the on-disk layout. Enhancements: - reduce memory consumption for SMR drive - better discard handling for multiple partitions - tracepoints for f2fs_file_write_iter/f2fs_filemap_fault - allow to change CP_CHKSUM_OFFSET - detect wrong layout of large_nat_bitmap feature - enhance checking valid data indices Bug fixes: - Multiple partition support for SMR drive - deadlock problem in f2fs_balance_fs_bg - add boundary checks to fix abnormal behaviors on fuzzed images - inline_xattr space calculations - replace f2fs_bug_on with errors In addition, this series contains various memory boundary check and sanity check of on-disk consistency" * tag 'f2fs-for-v5.2-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs: (40 commits) f2fs: fix to avoid accessing xattr across the boundary f2fs: fix to avoid potential race on sbi->unusable_block_count access/update f2fs: add tracepoint for f2fs_filemap_fault() f2fs: introduce DATA_GENERIC_ENHANCE f2fs: fix to handle error in f2fs_disable_checkpoint() f2fs: remove redundant check in f2fs_file_write_iter() f2fs: fix to be aware of readonly device in write_checkpoint() f2fs: fix to skip recovery on readonly device f2fs: fix to consider multiple device for readonly check f2fs: relocate chksum_offset for large_nat_bitmap feature f2fs: allow unfixed f2fs_checkpoint.checksum_offset f2fs: Replace spaces with tab f2fs: insert space before the open parenthesis '(' f2fs: allow address pointer number of dnode aligning to specified size f2fs: introduce f2fs_read_single_page() for cleanup f2fs: mark is_extension_exist() inline f2fs: fix to set FI_UPDATE_WRITE correctly f2fs: fix to avoid panic in f2fs_inplace_write_data() f2fs: fix to do sanity check on valid block count of segment f2fs: fix to do sanity check on valid node/block count ...

Linus Torvalds committed 7 years ago

Commit 0d285442 parents fa4bff1 2777e6518 files · +688 −302Browse files

Unified · Split

18 files

fs/f2fs/acl.c +2 −2
fs/f2fs/checkpoint.c +86 −22
fs/f2fs/data.c +165 −120
fs/f2fs/f2fs.h +94 −33
fs/f2fs/file.c +47 −29
fs/f2fs/gc.c +13 −3
fs/f2fs/inline.c +17 −0
fs/f2fs/inode.c +7 −5
fs/f2fs/namei.c +1 −1
fs/f2fs/node.c +30 −13
fs/f2fs/recovery.c +32 −5
fs/f2fs/segment.c +38 −33
fs/f2fs/segment.h +8 −8
fs/f2fs/super.c +54 −16
fs/f2fs/xattr.c +27 −9
fs/f2fs/xattr.h +2 −0
include/linux/f2fs_fs.h +8 −3
include/trace/events/f2fs.h +57 −0

fs/f2fs/acl.c+2−2

@@ -285,7 +285,7 @@ static int f2fs_acl_create_masq(struct posix_acl acl, umode_t mode_p) · 284 unmodified lines
285	285	/* assert(atomic_read(acl->a_refcount) == 1); */
286	286
287	287	FOREACH_ACL_ENTRY(pa, acl, pe) {
288		− switch(pa->e_tag) {
	288	+ switch (pa->e_tag) {
289	289	case ACL_USER_OBJ:
290	290	pa->e_perm &= (mode >> 6) \| ~S_IRWXO;
291	291	mode &= (pa->e_perm << 6) \| ~S_IRWXU;
@@ -326,7 +326,7 @@ static int f2fs_acl_create_masq(struct posix_acl acl, umode_t mode_p) · 34 unmodified lines
326	326	}
327	327
328	328	mode_p = (mode_p & ~S_IRWXUGO) \| mode;
329		− return not_equiv;
	329	+ return not_equiv;
330	330	}
331	331
332	332	static int f2fs_acl_create(struct inode dir, umode_t mode,

fs/f2fs/checkpoint.c+86−22

@@ -66,7 +66,7 @@ static struct page __get_meta_page(struct f2fs_sb_info sbi, pgoff_t index, · 65 unmodified lines
66	66	.old_blkaddr = index,
67	67	.new_blkaddr = index,
68	68	.encrypted_page = NULL,
69		− .is_meta = is_meta,
	69	+ .is_por = !is_meta,
70	70	};
71	71	int err;
72	72
@@ -130,6 +130,30 @@ struct page f2fs_get_tmp_page(struct f2fs_sb_info sbi, pgoff_t index) · 57 unmodified lines
130	130	return __get_meta_page(sbi, index, false);
131	131	}
132	132
	133	+static bool __is_bitmap_valid(struct f2fs_sb_info *sbi, block_t blkaddr,
	134	+ int type)
	135	+{
	136	+ struct seg_entry *se;
	137	+ unsigned int segno, offset;
	138	+ bool exist;
	139	+
	140	+ if (type != DATA_GENERIC_ENHANCE && type != DATA_GENERIC_ENHANCE_READ)
	141	+ return true;
	142	+
	143	+ segno = GET_SEGNO(sbi, blkaddr);
	144	+ offset = GET_BLKOFF_FROM_SEG0(sbi, blkaddr);
	145	+ se = get_seg_entry(sbi, segno);
	146	+
	147	+ exist = f2fs_test_bit(offset, se->cur_valid_map);
	148	+ if (!exist && type == DATA_GENERIC_ENHANCE) {
	149	+ f2fs_msg(sbi->sb, KERN_ERR, "Inconsistent error "
	150	+ "blkaddr:%u, sit bitmap:%d", blkaddr, exist);
	151	+ set_sbi_flag(sbi, SBI_NEED_FSCK);
	152	+ WARN_ON(1);
	153	+ }
	154	+ return exist;
	155	+}
	156	+
133	157	bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
134	158	block_t blkaddr, int type)
135	159	{
@@ -151,15 +175,22 @@ bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, · 15 unmodified lines
151	175	return false;
152	176	break;
153	177	case META_POR:
	178	+ if (unlikely(blkaddr >= MAX_BLKADDR(sbi) \|\|
	179	+ blkaddr < MAIN_BLKADDR(sbi)))
	180	+ return false;
	181	+ break;
154	182	case DATA_GENERIC:
	183	+ case DATA_GENERIC_ENHANCE:
	184	+ case DATA_GENERIC_ENHANCE_READ:
155	185	if (unlikely(blkaddr >= MAX_BLKADDR(sbi) \|\|
156		− blkaddr < MAIN_BLKADDR(sbi))) {
157		− if (type == DATA_GENERIC) {
158		− f2fs_msg(sbi->sb, KERN_WARNING,
159		− "access invalid blkaddr:%u", blkaddr);
160		− WARN_ON(1);
161		− }
	186	+ blkaddr < MAIN_BLKADDR(sbi))) {
	187	+ f2fs_msg(sbi->sb, KERN_WARNING,
	188	+ "access invalid blkaddr:%u", blkaddr);
	189	+ set_sbi_flag(sbi, SBI_NEED_FSCK);
	190	+ WARN_ON(1);
162	191	return false;
	192	+ } else {
	193	+ return __is_bitmap_valid(sbi, blkaddr, type);
163	194	}
164	195	break;
165	196	case META_GENERIC:
@@ -189,7 +220,7 @@ int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, · 23 unmodified lines
189	220	.op_flags = sync ? (REQ_META \| REQ_PRIO) : REQ_RAHEAD,
190	221	.encrypted_page = NULL,
191	222	.in_list = false,
192		− .is_meta = (type != META_POR),
	223	+ .is_por = (type == META_POR),
193	224	};
194	225	struct blk_plug plug;
195	226
@@ -644,6 +675,12 @@ int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi) · 448 unmodified lines
644	675	if (!is_set_ckpt_flags(sbi, CP_ORPHAN_PRESENT_FLAG))
645	676	return 0;
646	677
	678	+ if (bdev_read_only(sbi->sb->s_bdev)) {
	679	+ f2fs_msg(sbi->sb, KERN_INFO, "write access "
	680	+ "unavailable, skipping orphan cleanup");
	681	+ return 0;
	682	+ }
	683	+
647	684	if (s_flags & SB_RDONLY) {
648	685	f2fs_msg(sbi->sb, KERN_INFO, "orphan cleanup on readonly fs");
649	686	sbi->sb->s_flags &= ~SB_RDONLY;
@@ -758,13 +795,27 @@ static void write_orphan_inodes(struct f2fs_sb_info *sbi, block_t start_blk) · 108 unmodified lines
758	795	}
759	796	}
760	797
	798	+static __u32 f2fs_checkpoint_chksum(struct f2fs_sb_info *sbi,
	799	+ struct f2fs_checkpoint *ckpt)
	800	+{
	801	+ unsigned int chksum_ofs = le32_to_cpu(ckpt->checksum_offset);
	802	+ __u32 chksum;
	803	+
	804	+ chksum = f2fs_crc32(sbi, ckpt, chksum_ofs);
	805	+ if (chksum_ofs < CP_CHKSUM_OFFSET) {
	806	+ chksum_ofs += sizeof(chksum);
	807	+ chksum = f2fs_chksum(sbi, chksum, (__u8 *)ckpt + chksum_ofs,
	808	+ F2FS_BLKSIZE - chksum_ofs);
	809	+ }
	810	+ return chksum;
	811	+}
	812	+
761	813	static int get_checkpoint_version(struct f2fs_sb_info *sbi, block_t cp_addr,
762	814	struct f2fs_checkpoint cp_block, struct page cp_page,
763	815	unsigned long long *version)
764	816	{
765		− unsigned long blk_size = sbi->blocksize;
766	817	size_t crc_offset = 0;
767		− __u32 crc = 0;
	818	+ __u32 crc;
768	819
769	820	*cp_page = f2fs_get_meta_page(sbi, cp_addr);
770	821	if (IS_ERR(*cp_page))
@@ -773,15 +824,27 @@ static int get_checkpoint_version(struct f2fs_sb_info *sbi, block_t cp_addr, · 2 unmodified lines
773	824	cp_block = (struct f2fs_checkpoint )page_address(*cp_page);
774	825
775	826	crc_offset = le32_to_cpu((*cp_block)->checksum_offset);
776		− if (crc_offset > (blk_size - sizeof(__le32))) {
	827	+ if (crc_offset < CP_MIN_CHKSUM_OFFSET \|\|
	828	+ crc_offset > CP_CHKSUM_OFFSET) {
777	829	f2fs_put_page(*cp_page, 1);
778	830	f2fs_msg(sbi->sb, KERN_WARNING,
779	831	"invalid crc_offset: %zu", crc_offset);
780	832	return -EINVAL;
781	833	}
782	834
783		− crc = cur_cp_crc(*cp_block);
784		− if (!f2fs_crc_valid(sbi, crc, *cp_block, crc_offset)) {
	835	+ if (__is_set_ckpt_flags(*cp_block, CP_LARGE_NAT_BITMAP_FLAG)) {
	836	+ if (crc_offset != CP_MIN_CHKSUM_OFFSET) {
	837	+ f2fs_put_page(*cp_page, 1);
	838	+ f2fs_msg(sbi->sb, KERN_WARNING,
	839	+ "layout of large_nat_bitmap is deprecated, "
	840	+ "run fsck to repair, chksum_offset: %zu",
	841	+ crc_offset);
	842	+ return -EINVAL;
	843	+ }
	844	+ }
	845	+
	846	+ crc = f2fs_checkpoint_chksum(sbi, *cp_block);
	847	+ if (crc != cur_cp_crc(*cp_block)) {
785	848	f2fs_put_page(*cp_page, 1);
786	849	f2fs_msg(sbi->sb, KERN_WARNING, "invalid crc value");
787	850	return -EINVAL;
@@ -1009,13 +1072,11 @@ retry: · 221 unmodified lines
1009	1072	if (inode) {
1010	1073	unsigned long cur_ino = inode->i_ino;
1011	1074
1012		− if (is_dir)
1013		− F2FS_I(inode)->cp_task = current;
	1075	+ F2FS_I(inode)->cp_task = current;
1014	1076
1015	1077	filemap_fdatawrite(inode->i_mapping);
1016	1078
1017		− if (is_dir)
1018		− F2FS_I(inode)->cp_task = NULL;
	1079	+ F2FS_I(inode)->cp_task = NULL;
1019	1080
1020	1081	iput(inode);
1021	1082	/* We need to give cpu to another writers. */
@@ -1391,7 +1452,7 @@ static int do_checkpoint(struct f2fs_sb_info sbi, struct cp_control cpc) · 369 unmodified lines
1391	1452	get_sit_bitmap(sbi, __bitmap_ptr(sbi, SIT_BITMAP));
1392	1453	get_nat_bitmap(sbi, __bitmap_ptr(sbi, NAT_BITMAP));
1393	1454
1394		− crc32 = f2fs_crc32(sbi, ckpt, le32_to_cpu(ckpt->checksum_offset));
	1455	+ crc32 = f2fs_checkpoint_chksum(sbi, ckpt);
1395	1456	((__le32 )((unsigned char *)ckpt +
1396	1457	le32_to_cpu(ckpt->checksum_offset)))
1397	1458	= cpu_to_le32(crc32);
@@ -1475,7 +1536,11 @@ static int do_checkpoint(struct f2fs_sb_info sbi, struct cp_control cpc) · 77 unmodified lines
1475	1536	clear_sbi_flag(sbi, SBI_IS_DIRTY);
1476	1537	clear_sbi_flag(sbi, SBI_NEED_CP);
1477	1538	clear_sbi_flag(sbi, SBI_QUOTA_SKIP_FLUSH);
	1539	+
	1540	+ spin_lock(&sbi->stat_lock);
1478	1541	sbi->unusable_block_count = 0;
	1542	+ spin_unlock(&sbi->stat_lock);
	1543	+
1479	1544	__set_cp_next_pack(sbi);
1480	1545
1481	1546	/*
@@ -1500,6 +1565,9 @@ int f2fs_write_checkpoint(struct f2fs_sb_info sbi, struct cp_control cpc) · 18 unmodified lines
1500	1565	unsigned long long ckpt_ver;
1501	1566	int err = 0;
1502	1567
	1568	+ if (f2fs_readonly(sbi->sb) \|\| f2fs_hw_is_readonly(sbi))
	1569	+ return -EROFS;
	1570	+
1503	1571	if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1504	1572	if (cpc->reason != CP_PAUSE)
1505	1573	return 0;
@@ -1516,10 +1584,6 @@ int f2fs_write_checkpoint(struct f2fs_sb_info sbi, struct cp_control cpc) · 10 unmodified lines
1516	1584	err = -EIO;
1517	1585	goto out;
1518	1586	}
1519		− if (f2fs_readonly(sbi->sb)) {
1520		− err = -EROFS;
1521		− goto out;
1522		− }
1523	1587
1524	1588	trace_f2fs_write_checkpoint(sbi->sb, cpc->reason, "start block_ops");
1525	1589

fs/f2fs/data.c+165−120

@@ -218,12 +218,14 @@ struct block_device f2fs_target_device(struct f2fs_sb_info sbi, · 217 unmodified lines
218	218	struct block_device *bdev = sbi->sb->s_bdev;
219	219	int i;
220	220
221		− for (i = 0; i < sbi->s_ndevs; i++) {
222		− if (FDEV(i).start_blk <= blk_addr &&
223		− FDEV(i).end_blk >= blk_addr) {
224		− blk_addr -= FDEV(i).start_blk;
225		− bdev = FDEV(i).bdev;
226		− break;
	221	+ if (f2fs_is_multi_device(sbi)) {
	222	+ for (i = 0; i < sbi->s_ndevs; i++) {
	223	+ if (FDEV(i).start_blk <= blk_addr &&
	224	+ FDEV(i).end_blk >= blk_addr) {
	225	+ blk_addr -= FDEV(i).start_blk;
	226	+ bdev = FDEV(i).bdev;
	227	+ break;
	228	+ }
227	229	}
228	230	}
229	231	if (bio) {
@@ -237,6 +239,9 @@ int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr) · 7 unmodified lines
237	239	{
238	240	int i;
239	241
	242	+ if (!f2fs_is_multi_device(sbi))
	243	+ return 0;
	244	+
240	245	for (i = 0; i < sbi->s_ndevs; i++)
241	246	if (FDEV(i).start_blk <= blkaddr && FDEV(i).end_blk >= blkaddr)
242	247	return i;
@@ -420,7 +425,7 @@ static void __submit_merged_write_cond(struct f2fs_sb_info *sbi, · 177 unmodified lines
420	425
421	426	void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type)
422	427	{
423		− __submit_merged_write_cond(sbi, NULL, 0, 0, type, true);
	428	+ __submit_merged_write_cond(sbi, NULL, NULL, 0, type, true);
424	429	}
425	430
426	431	void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
@@ -448,7 +453,8 @@ int f2fs_submit_page_bio(struct f2fs_io_info *fio) · 21 unmodified lines
448	453	fio->encrypted_page : fio->page;
449	454
450	455	if (!f2fs_is_valid_blkaddr(fio->sbi, fio->new_blkaddr,
451		− __is_meta_io(fio) ? META_GENERIC : DATA_GENERIC))
	456	+ fio->is_por ? META_POR : (__is_meta_io(fio) ?
	457	+ META_GENERIC : DATA_GENERIC_ENHANCE)))
452	458	return -EFAULT;
453	459
454	460	trace_f2fs_submit_page_bio(page, fio);
@@ -498,9 +504,7 @@ next: · 43 unmodified lines
498	504	spin_unlock(&io->io_lock);
499	505	}
500	506
501		− if (__is_valid_data_blkaddr(fio->old_blkaddr))
502		− verify_block_addr(fio, fio->old_blkaddr);
503		− verify_block_addr(fio, fio->new_blkaddr);
	507	+ verify_fio_blkaddr(fio);
504	508
505	509	bio_page = fio->encrypted_page ? fio->encrypted_page : fio->page;
506	510
@@ -557,9 +561,6 @@ static struct bio f2fs_grab_read_bio(struct inode inode, block_t blkaddr, · 50 unmodified lines
557	561	struct bio_post_read_ctx *ctx;
558	562	unsigned int post_read_steps = 0;
559	563
560		− if (!f2fs_is_valid_blkaddr(sbi, blkaddr, DATA_GENERIC))
561		− return ERR_PTR(-EFAULT);
562		−
563	564	bio = f2fs_bio_alloc(sbi, min_t(int, nr_pages, BIO_MAX_PAGES), false);
564	565	if (!bio)
565	566	return ERR_PTR(-ENOMEM);
@@ -587,8 +588,10 @@ static struct bio f2fs_grab_read_bio(struct inode inode, block_t blkaddr, · 21 unmodified lines
587	588	static int f2fs_submit_page_read(struct inode inode, struct page page,
588	589	block_t blkaddr)
589	590	{
590		− struct bio *bio = f2fs_grab_read_bio(inode, blkaddr, 1, 0);
	591	+ struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	592	+ struct bio *bio;
591	593
	594	+ bio = f2fs_grab_read_bio(inode, blkaddr, 1, 0);
592	595	if (IS_ERR(bio))
593	596	return PTR_ERR(bio);
594	597
@@ -600,8 +603,8 @@ static int f2fs_submit_page_read(struct inode inode, struct page page, · 5 unmodified lines
600	603	return -EFAULT;
601	604	}
602	605	ClearPageError(page);
603		− inc_page_count(F2FS_I_SB(inode), F2FS_RD_DATA);
604		− __submit_bio(F2FS_I_SB(inode), bio, DATA);
	606	+ inc_page_count(sbi, F2FS_RD_DATA);
	607	+ __submit_bio(sbi, bio, DATA);
605	608	return 0;
606	609	}
607	610
@@ -729,6 +732,11 @@ struct page f2fs_get_read_data_page(struct inode inode, pgoff_t index, · 121 unmodified lines
729	732
730	733	if (f2fs_lookup_extent_cache(inode, index, &ei)) {
731	734	dn.data_blkaddr = ei.blk + index - ei.fofs;
	735	+ if (!f2fs_is_valid_blkaddr(F2FS_I_SB(inode), dn.data_blkaddr,
	736	+ DATA_GENERIC_ENHANCE_READ)) {
	737	+ err = -EFAULT;
	738	+ goto put_err;
	739	+ }
732	740	goto got_it;
733	741	}
734	742
@@ -742,6 +750,13 @@ struct page f2fs_get_read_data_page(struct inode inode, pgoff_t index, · 7 unmodified lines
742	750	err = -ENOENT;
743	751	goto put_err;
744	752	}
	753	+ if (dn.data_blkaddr != NEW_ADDR &&
	754	+ !f2fs_is_valid_blkaddr(F2FS_I_SB(inode),
	755	+ dn.data_blkaddr,
	756	+ DATA_GENERIC_ENHANCE)) {
	757	+ err = -EFAULT;
	758	+ goto put_err;
	759	+ }
745	760	got_it:
746	761	if (PageUptodate(page)) {
747	762	unlock_page(page);
@@ -1084,12 +1099,12 @@ next_block: · 336 unmodified lines
1084	1099	blkaddr = datablock_addr(dn.inode, dn.node_page, dn.ofs_in_node);
1085	1100
1086	1101	if (__is_valid_data_blkaddr(blkaddr) &&
1087		− !f2fs_is_valid_blkaddr(sbi, blkaddr, DATA_GENERIC)) {
	1102	+ !f2fs_is_valid_blkaddr(sbi, blkaddr, DATA_GENERIC_ENHANCE)) {
1088	1103	err = -EFAULT;
1089	1104	goto sync_out;
1090	1105	}
1091	1106
1092		− if (is_valid_data_blkaddr(sbi, blkaddr)) {
	1107	+ if (__is_valid_data_blkaddr(blkaddr)) {
1093	1108	/* use out-place-update for driect IO under LFS mode */
1094	1109	if (test_opt(sbi, LFS) && flag == F2FS_GET_BLOCK_DIO &&
1095	1110	map->m_may_create) {
@@ -1499,6 +1514,118 @@ out: · 403 unmodified lines
1499	1514	return ret;
1500	1515	}
1501	1516
	1517	+static int f2fs_read_single_page(struct inode inode, struct page page,
	1518	+ unsigned nr_pages,
	1519	+ struct f2fs_map_blocks *map,
	1520	+ struct bio **bio_ret,
	1521	+ sector_t *last_block_in_bio,
	1522	+ bool is_readahead)
	1523	+{
	1524	+ struct bio bio = bio_ret;
	1525	+ const unsigned blkbits = inode->i_blkbits;
	1526	+ const unsigned blocksize = 1 << blkbits;
	1527	+ sector_t block_in_file;
	1528	+ sector_t last_block;
	1529	+ sector_t last_block_in_file;
	1530	+ sector_t block_nr;
	1531	+ int ret = 0;
	1532	+
	1533	+ block_in_file = (sector_t)page->index;
	1534	+ last_block = block_in_file + nr_pages;
	1535	+ last_block_in_file = (i_size_read(inode) + blocksize - 1) >>
	1536	+ blkbits;
	1537	+ if (last_block > last_block_in_file)
	1538	+ last_block = last_block_in_file;
	1539	+
	1540	+ /* just zeroing out page which is beyond EOF */
	1541	+ if (block_in_file >= last_block)
	1542	+ goto zero_out;
	1543	+ /*
	1544	+ * Map blocks using the previous result first.
	1545	+ */
	1546	+ if ((map->m_flags & F2FS_MAP_MAPPED) &&
	1547	+ block_in_file > map->m_lblk &&
	1548	+ block_in_file < (map->m_lblk + map->m_len))
	1549	+ goto got_it;
	1550	+
	1551	+ /*
	1552	+ * Then do more f2fs_map_blocks() calls until we are
	1553	+ * done with this page.
	1554	+ */
	1555	+ map->m_lblk = block_in_file;
	1556	+ map->m_len = last_block - block_in_file;
	1557	+
	1558	+ ret = f2fs_map_blocks(inode, map, 0, F2FS_GET_BLOCK_DEFAULT);
	1559	+ if (ret)
	1560	+ goto out;
	1561	+got_it:
	1562	+ if ((map->m_flags & F2FS_MAP_MAPPED)) {
	1563	+ block_nr = map->m_pblk + block_in_file - map->m_lblk;
	1564	+ SetPageMappedToDisk(page);
	1565	+
	1566	+ if (!PageUptodate(page) && !cleancache_get_page(page)) {
	1567	+ SetPageUptodate(page);
	1568	+ goto confused;
	1569	+ }
	1570	+
	1571	+ if (!f2fs_is_valid_blkaddr(F2FS_I_SB(inode), block_nr,
	1572	+ DATA_GENERIC_ENHANCE_READ)) {
	1573	+ ret = -EFAULT;
	1574	+ goto out;
	1575	+ }
	1576	+ } else {
	1577	+zero_out:
	1578	+ zero_user_segment(page, 0, PAGE_SIZE);
	1579	+ if (!PageUptodate(page))
	1580	+ SetPageUptodate(page);
	1581	+ unlock_page(page);
	1582	+ goto out;
	1583	+ }
	1584	+
	1585	+ /*
	1586	+ * This page will go to BIO. Do we need to send this
	1587	+ * BIO off first?
	1588	+ */
	1589	+ if (bio && (*last_block_in_bio != block_nr - 1 \|\|
	1590	+ !__same_bdev(F2FS_I_SB(inode), block_nr, bio))) {
	1591	+submit_and_realloc:
	1592	+ __submit_bio(F2FS_I_SB(inode), bio, DATA);
	1593	+ bio = NULL;
	1594	+ }
	1595	+ if (bio == NULL) {
	1596	+ bio = f2fs_grab_read_bio(inode, block_nr, nr_pages,
	1597	+ is_readahead ? REQ_RAHEAD : 0);
	1598	+ if (IS_ERR(bio)) {
	1599	+ ret = PTR_ERR(bio);
	1600	+ bio = NULL;
	1601	+ goto out;
	1602	+ }
	1603	+ }
	1604	+
	1605	+ /*
	1606	+ * If the page is under writeback, we need to wait for
	1607	+ * its completion to see the correct decrypted data.
	1608	+ */
	1609	+ f2fs_wait_on_block_writeback(inode, block_nr);
	1610	+
	1611	+ if (bio_add_page(bio, page, blocksize, 0) < blocksize)
	1612	+ goto submit_and_realloc;
	1613	+
	1614	+ inc_page_count(F2FS_I_SB(inode), F2FS_RD_DATA);
	1615	+ ClearPageError(page);
	1616	+ *last_block_in_bio = block_nr;
	1617	+ goto out;
	1618	+confused:
	1619	+ if (bio) {
	1620	+ __submit_bio(F2FS_I_SB(inode), bio, DATA);
	1621	+ bio = NULL;
	1622	+ }
	1623	+ unlock_page(page);
	1624	+out:
	1625	+ *bio_ret = bio;
	1626	+ return ret;
	1627	+}
	1628	+
1502	1629	/*
1503	1630	* This function was originally taken from fs/mpage.c, and customized for f2fs.
1504	1631	* Major change was from block_size == page_size in f2fs by default.
@@ -1515,13 +1642,8 @@ static int f2fs_mpage_readpages(struct address_space *mapping, · 10 unmodified lines
1515	1642	struct bio *bio = NULL;
1516	1643	sector_t last_block_in_bio = 0;
1517	1644	struct inode *inode = mapping->host;
1518		− const unsigned blkbits = inode->i_blkbits;
1519		− const unsigned blocksize = 1 << blkbits;
1520		− sector_t block_in_file;
1521		− sector_t last_block;
1522		− sector_t last_block_in_file;
1523		− sector_t block_nr;
1524	1645	struct f2fs_map_blocks map;
	1646	+ int ret = 0;
1525	1647
1526	1648	map.m_pblk = 0;
1527	1649	map.m_lblk = 0;
@@ -1544,98 +1666,13 @@ static int f2fs_mpage_readpages(struct address_space *mapping, · 16 unmodified lines
1544	1666	goto next_page;
1545	1667	}
1546	1668
1547		− block_in_file = (sector_t)page->index;
1548		− last_block = block_in_file + nr_pages;
1549		− last_block_in_file = (i_size_read(inode) + blocksize - 1) >>
1550		− blkbits;
1551		− if (last_block > last_block_in_file)
1552		− last_block = last_block_in_file;
1553		−
1554		− /* just zeroing out page which is beyond EOF */
1555		− if (block_in_file >= last_block)
1556		− goto zero_out;
1557		− /*
1558		− * Map blocks using the previous result first.
1559		− */
1560		− if ((map.m_flags & F2FS_MAP_MAPPED) &&
1561		− block_in_file > map.m_lblk &&
1562		− block_in_file < (map.m_lblk + map.m_len))
1563		− goto got_it;
1564		−
1565		− /*
1566		− * Then do more f2fs_map_blocks() calls until we are
1567		− * done with this page.
1568		− */
1569		− map.m_lblk = block_in_file;
1570		− map.m_len = last_block - block_in_file;
1571		−
1572		− if (f2fs_map_blocks(inode, &map, 0, F2FS_GET_BLOCK_DEFAULT))
1573		− goto set_error_page;
1574		−got_it:
1575		− if ((map.m_flags & F2FS_MAP_MAPPED)) {
1576		− block_nr = map.m_pblk + block_in_file - map.m_lblk;
1577		− SetPageMappedToDisk(page);
1578		−
1579		− if (!PageUptodate(page) && !cleancache_get_page(page)) {
1580		− SetPageUptodate(page);
1581		− goto confused;
1582		− }
1583		−
1584		− if (!f2fs_is_valid_blkaddr(F2FS_I_SB(inode), block_nr,
1585		− DATA_GENERIC))
1586		− goto set_error_page;
1587		− } else {
1588		−zero_out:
	1669	+ ret = f2fs_read_single_page(inode, page, nr_pages, &map, &bio,
	1670	+ &last_block_in_bio, is_readahead);
	1671	+ if (ret) {
	1672	+ SetPageError(page);
1589	1673	zero_user_segment(page, 0, PAGE_SIZE);
1590		− if (!PageUptodate(page))
1591		− SetPageUptodate(page);
1592	1674	unlock_page(page);
1593		− goto next_page;
1594	1675	}
1595		−
1596		− /*
1597		− * This page will go to BIO. Do we need to send this
1598		− * BIO off first?
1599		− */
1600		− if (bio && (last_block_in_bio != block_nr - 1 \|\|
1601		− !__same_bdev(F2FS_I_SB(inode), block_nr, bio))) {
1602		−submit_and_realloc:
1603		− __submit_bio(F2FS_I_SB(inode), bio, DATA);
1604		− bio = NULL;
1605		− }
1606		− if (bio == NULL) {
1607		− bio = f2fs_grab_read_bio(inode, block_nr, nr_pages,
1608		− is_readahead ? REQ_RAHEAD : 0);
1609		− if (IS_ERR(bio)) {
1610		− bio = NULL;
1611		− goto set_error_page;
1612		− }
1613		− }
1614		−
1615		− /*
1616		− * If the page is under writeback, we need to wait for
1617		− * its completion to see the correct decrypted data.
1618		− */
1619		− f2fs_wait_on_block_writeback(inode, block_nr);
1620		−
1621		− if (bio_add_page(bio, page, blocksize, 0) < blocksize)
1622		− goto submit_and_realloc;
1623		−
1624		− inc_page_count(F2FS_I_SB(inode), F2FS_RD_DATA);
1625		− ClearPageError(page);
1626		− last_block_in_bio = block_nr;
1627		− goto next_page;
1628		−set_error_page:
1629		− SetPageError(page);
1630		− zero_user_segment(page, 0, PAGE_SIZE);
1631		− unlock_page(page);
1632		− goto next_page;
1633		−confused:
1634		− if (bio) {
1635		− __submit_bio(F2FS_I_SB(inode), bio, DATA);
1636		− bio = NULL;
1637		− }
1638		− unlock_page(page);
1639	1676	next_page:
1640	1677	if (pages)
1641	1678	put_page(page);
@@ -1643,7 +1680,7 @@ next_page: · 1 unmodified line
1643	1680	BUG_ON(pages && !list_empty(pages));
1644	1681	if (bio)
1645	1682	__submit_bio(F2FS_I_SB(inode), bio, DATA);
1646		− return 0;
	1683	+ return pages ? 0 : ret;
1647	1684	}
1648	1685
1649	1686	static int f2fs_read_data_page(struct file file, struct page page)
@@ -1813,7 +1850,7 @@ int f2fs_do_write_data_page(struct f2fs_io_info *fio) · 163 unmodified lines
1813	1850	fio->old_blkaddr = ei.blk + page->index - ei.fofs;
1814	1851
1815	1852	if (!f2fs_is_valid_blkaddr(fio->sbi, fio->old_blkaddr,
1816		− DATA_GENERIC))
	1853	+ DATA_GENERIC_ENHANCE))
1817	1854	return -EFAULT;
1818	1855
1819	1856	ipu_force = true;
@@ -1840,7 +1877,7 @@ int f2fs_do_write_data_page(struct f2fs_io_info *fio) · 20 unmodified lines
1840	1877	got_it:
1841	1878	if (__is_valid_data_blkaddr(fio->old_blkaddr) &&
1842	1879	!f2fs_is_valid_blkaddr(fio->sbi, fio->old_blkaddr,
1843		− DATA_GENERIC)) {
	1880	+ DATA_GENERIC_ENHANCE)) {
1844	1881	err = -EFAULT;
1845	1882	goto out_writepage;
1846	1883	}
@@ -1848,7 +1885,8 @@ got_it: · 1 unmodified line
1848	1885	* If current allocation needs SSR,
1849	1886	* it had better in-place writes for updated data.
1850	1887	*/
1851		− if (ipu_force \|\| (is_valid_data_blkaddr(fio->sbi, fio->old_blkaddr) &&
	1888	+ if (ipu_force \|\|
	1889	+ (__is_valid_data_blkaddr(fio->old_blkaddr) &&
1852	1890	need_inplace_update(fio))) {
1853	1891	err = encrypt_one_page(fio);
1854	1892	if (err)
@@ -1866,9 +1904,10 @@ got_it: · 11 unmodified lines
1866	1904	true);
1867	1905	if (PageWriteback(page))
1868	1906	end_page_writeback(page);
	1907	+ } else {
	1908	+ set_inode_flag(inode, FI_UPDATE_WRITE);
1869	1909	}
1870	1910	trace_f2fs_do_write_data_page(fio->page, IPU);
1871		− set_inode_flag(inode, FI_UPDATE_WRITE);
1872	1911	return err;
1873	1912	}
1874	1913
@@ -2030,7 +2069,8 @@ out: · 155 unmodified lines
2030	2069	}
2031	2070
2032	2071	unlock_page(page);
2033		− if (!S_ISDIR(inode->i_mode) && !IS_NOQUOTA(inode))
	2072	+ if (!S_ISDIR(inode->i_mode) && !IS_NOQUOTA(inode) &&
	2073	+ !F2FS_I(inode)->cp_task)
2034	2074	f2fs_balance_fs(sbi, need_balance_fs);
2035	2075
2036	2076	if (unlikely(f2fs_cp_error(sbi))) {
@@ -2491,6 +2531,11 @@ repeat: · 454 unmodified lines
2491	2531	zero_user_segment(page, 0, PAGE_SIZE);
2492	2532	SetPageUptodate(page);
2493	2533	} else {
	2534	+ if (!f2fs_is_valid_blkaddr(sbi, blkaddr,
	2535	+ DATA_GENERIC_ENHANCE_READ)) {
	2536	+ err = -EFAULT;
	2537	+ goto fail;
	2538	+ }
2494	2539	err = f2fs_submit_page_read(inode, page, blkaddr);
2495	2540	if (err)
2496	2541	goto fail;

fs/f2fs/f2fs.h+94−33

@@ -210,7 +210,14 @@ enum { · 209 unmodified lines
210	210	META_SSA,
211	211	META_MAX,
212	212	META_POR,
213		− DATA_GENERIC,
	213	+ DATA_GENERIC, /* check range only */
	214	+ DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
	215	+ DATA_GENERIC_ENHANCE_READ, /*
	216	+ * strong check on range and segment
	217	+ * bitmap but no warning due to race
	218	+ * condition of read on truncated area
	219	+ * by extent_cache
	220	+ */
214	221	META_GENERIC,
215	222	};
216	223
@@ -1041,7 +1048,7 @@ struct f2fs_io_info { · 824 unmodified lines
1041	1048	bool submitted; /* indicate IO submission */
1042	1049	int need_lock; /* indicate we need to lock cp_rwsem */
1043	1050	bool in_list; /* indicate fio is in io_list */
1044		− bool is_meta; /* indicate borrow meta inode mapping or not */
	1051	+ bool is_por; /* indicate IO is from recovery or not */
1045	1052	bool retry; /* need to reallocate block address */
1046	1053	enum iostat_type io_type; /* io type */
1047	1054	struct writeback_control io_wbc; / writeback control */
@@ -1068,8 +1075,8 @@ struct f2fs_dev_info { · 20 unmodified lines
1068	1075	block_t start_blk;
1069	1076	block_t end_blk;
1070	1077	#ifdef CONFIG_BLK_DEV_ZONED
1071		− unsigned int nr_blkz; /* Total number of zones */
1072		− u8 blkz_type; / Array of zones type */
	1078	+ unsigned int nr_blkz; /* Total number of zones */
	1079	+ unsigned long blkz_seq; / Bitmap indicating sequential zones */
1073	1080	#endif
1074	1081	};
1075	1082
@@ -1366,6 +1373,17 @@ static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) · 290 unmodified lines
1366	1373	}
1367	1374	#endif
1368	1375
	1376	+/*
	1377	+ * Test if the mounted volume is a multi-device volume.
	1378	+ * - For a single regular disk volume, sbi->s_ndevs is 0.
	1379	+ * - For a single zoned disk volume, sbi->s_ndevs is 1.
	1380	+ * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
	1381	+ */
	1382	+static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
	1383	+{
	1384	+ return sbi->s_ndevs > 1;
	1385	+}
	1386	+
1369	1387	/* For write statistics. Suppose sector size is 512 bytes,
1370	1388	* and the return value is in kbytes. s is of struct f2fs_sb_info.
1371	1389	*/
@@ -1777,6 +1795,7 @@ enospc: · 405 unmodified lines
1777	1795	return -ENOSPC;
1778	1796	}
1779	1797
	1798	+void f2fs_msg(struct super_block sb, const char level, const char *fmt, ...);
1780	1799	static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
1781	1800	struct inode *inode,
1782	1801	block_t count)
@@ -1785,13 +1804,21 @@ static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, · 2 unmodified lines
1785	1804
1786	1805	spin_lock(&sbi->stat_lock);
1787	1806	f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
1788		− f2fs_bug_on(sbi, inode->i_blocks < sectors);
1789	1807	sbi->total_valid_block_count -= (block_t)count;
1790	1808	if (sbi->reserved_blocks &&
1791	1809	sbi->current_reserved_blocks < sbi->reserved_blocks)
1792	1810	sbi->current_reserved_blocks = min(sbi->reserved_blocks,
1793	1811	sbi->current_reserved_blocks + count);
1794	1812	spin_unlock(&sbi->stat_lock);
	1813	+ if (unlikely(inode->i_blocks < sectors)) {
	1814	+ f2fs_msg(sbi->sb, KERN_WARNING,
	1815	+ "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
	1816	+ inode->i_ino,
	1817	+ (unsigned long long)inode->i_blocks,
	1818	+ (unsigned long long)sectors);
	1819	+ set_sbi_flag(sbi, SBI_NEED_FSCK);
	1820	+ return;
	1821	+ }
1795	1822	f2fs_i_blocks_write(inode, count, false, true);
1796	1823	}
1797	1824
@@ -1889,7 +1916,11 @@ static inline void __bitmap_ptr(struct f2fs_sb_info sbi, int flag) · 91 unmodified lines
1889	1916	if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
1890	1917	offset = (flag == SIT_BITMAP) ?
1891	1918	le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
1892		− return &ckpt->sit_nat_version_bitmap + offset;
	1919	+ /*
	1920	+ * if large_nat_bitmap feature is enabled, leave checksum
	1921	+ * protection for all nat/sit bitmaps.
	1922	+ */
	1923	+ return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32);
1893	1924	}
1894	1925
1895	1926	if (__cp_payload(sbi) > 0) {
@@ -2008,7 +2039,6 @@ static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, · 112 unmodified lines
2008	2039
2009	2040	f2fs_bug_on(sbi, !sbi->total_valid_block_count);
2010	2041	f2fs_bug_on(sbi, !sbi->total_valid_node_count);
2011		− f2fs_bug_on(sbi, !is_inode && !inode->i_blocks);
2012	2042
2013	2043	sbi->total_valid_node_count--;
2014	2044	sbi->total_valid_block_count--;
@@ -2018,10 +2048,19 @@ static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, · 3 unmodified lines
2018	2048
2019	2049	spin_unlock(&sbi->stat_lock);
2020	2050
2021		− if (is_inode)
	2051	+ if (is_inode) {
2022	2052	dquot_free_inode(inode);
2023		− else
	2053	+ } else {
	2054	+ if (unlikely(inode->i_blocks == 0)) {
	2055	+ f2fs_msg(sbi->sb, KERN_WARNING,
	2056	+ "Inconsistent i_blocks, ino:%lu, iblocks:%llu",
	2057	+ inode->i_ino,
	2058	+ (unsigned long long)inode->i_blocks);
	2059	+ set_sbi_flag(sbi, SBI_NEED_FSCK);
	2060	+ return;
	2061	+ }
2024	2062	f2fs_i_blocks_write(inode, 1, false, true);
	2063	+ }
2025	2064	}
2026	2065
2027	2066	static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
@@ -2545,7 +2584,14 @@ static inline int f2fs_has_inline_xattr(struct inode *inode) · 517 unmodified lines
2545	2584
2546	2585	static inline unsigned int addrs_per_inode(struct inode *inode)
2547	2586	{
2548		− return CUR_ADDRS_PER_INODE(inode) - get_inline_xattr_addrs(inode);
	2587	+ unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
	2588	+ get_inline_xattr_addrs(inode);
	2589	+ return ALIGN_DOWN(addrs, 1);
	2590	+}
	2591	+
	2592	+static inline unsigned int addrs_per_block(struct inode *inode)
	2593	+{
	2594	+ return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, 1);
2549	2595	}
2550	2596
2551	2597	static inline void inline_xattr_addr(struct inode inode, struct page *page)
@@ -2558,7 +2604,9 @@ static inline void inline_xattr_addr(struct inode inode, struct page *page) · 6 unmodified lines
2558	2604
2559	2605	static inline int inline_xattr_size(struct inode *inode)
2560	2606	{
2561		− return get_inline_xattr_addrs(inode) * sizeof(__le32);
	2607	+ if (f2fs_has_inline_xattr(inode))
	2608	+ return get_inline_xattr_addrs(inode) * sizeof(__le32);
	2609	+ return 0;
2562	2610	}
2563	2611
2564	2612	static inline int f2fs_has_inline_data(struct inode *inode)
@@ -2800,12 +2848,10 @@ static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, · 235 unmodified lines
2800	2848
2801	2849	#define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
2802	2850
2803		−#define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META && \
2804		− (!is_read_io((fio)->op) \|\| (fio)->is_meta))
	2851	+#define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
2805	2852
2806	2853	bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
2807	2854	block_t blkaddr, int type);
2808		−void f2fs_msg(struct super_block sb, const char level, const char *fmt, ...);
2809	2855	static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
2810	2856	block_t blkaddr, int type)
2811	2857	{
@@ -2824,15 +2870,6 @@ static inline bool __is_valid_data_blkaddr(block_t blkaddr) · 12 unmodified lines
2824	2870	return true;
2825	2871	}
2826	2872
2827		−static inline bool is_valid_data_blkaddr(struct f2fs_sb_info *sbi,
2828		− block_t blkaddr)
2829		−{
2830		− if (!__is_valid_data_blkaddr(blkaddr))
2831		− return false;
2832		− verify_blkaddr(sbi, blkaddr, DATA_GENERIC);
2833		− return true;
2834		−}
2835		−
2836	2873	static inline void f2fs_set_page_private(struct page *page,
2837	2874	unsigned long data)
2838	2875	{
@@ -3530,16 +3567,12 @@ F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); · 691 unmodified lines
3530	3567	F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
3531	3568
3532	3569	#ifdef CONFIG_BLK_DEV_ZONED
3533		−static inline int get_blkz_type(struct f2fs_sb_info *sbi,
3534		− struct block_device *bdev, block_t blkaddr)
	3570	+static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
	3571	+ block_t blkaddr)
3535	3572	{
3536	3573	unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
3537		− int i;
3538	3574
3539		− for (i = 0; i < sbi->s_ndevs; i++)
3540		− if (FDEV(i).bdev == bdev)
3541		− return FDEV(i).blkz_type[zno];
3542		− return -EINVAL;
	3575	+ return test_bit(zno, FDEV(devi).blkz_seq);
3543	3576	}
3544	3577	#endif
3545	3578
@@ -3548,9 +3581,23 @@ static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) · 2 unmodified lines
3548	3581	return f2fs_sb_has_blkzoned(sbi);
3549	3582	}
3550	3583
	3584	+static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
	3585	+{
	3586	+ return blk_queue_discard(bdev_get_queue(bdev)) \|\|
	3587	+ bdev_is_zoned(bdev);
	3588	+}
	3589	+
3551	3590	static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
3552	3591	{
3553		− return blk_queue_discard(bdev_get_queue(sbi->sb->s_bdev));
	3592	+ int i;
	3593	+
	3594	+ if (!f2fs_is_multi_device(sbi))
	3595	+ return f2fs_bdev_support_discard(sbi->sb->s_bdev);
	3596	+
	3597	+ for (i = 0; i < sbi->s_ndevs; i++)
	3598	+ if (f2fs_bdev_support_discard(FDEV(i).bdev))
	3599	+ return true;
	3600	+ return false;
3554	3601	}
3555	3602
3556	3603	static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
@@ -3559,6 +3606,20 @@ static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) · 2 unmodified lines
3559	3606	f2fs_hw_should_discard(sbi);
3560	3607	}
3561	3608
	3609	+static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
	3610	+{
	3611	+ int i;
	3612	+
	3613	+ if (!f2fs_is_multi_device(sbi))
	3614	+ return bdev_read_only(sbi->sb->s_bdev);
	3615	+
	3616	+ for (i = 0; i < sbi->s_ndevs; i++)
	3617	+ if (bdev_read_only(FDEV(i).bdev))
	3618	+ return true;
	3619	+ return false;
	3620	+}
	3621	+
	3622	+
3562	3623	static inline void set_opt_mode(struct f2fs_sb_info *sbi, unsigned int mt)
3563	3624	{
3564	3625	clear_opt(sbi, ADAPTIVE);
@@ -3614,7 +3675,7 @@ static inline bool f2fs_force_buffered_io(struct inode *inode, · 49 unmodified lines
3614	3675
3615	3676	if (f2fs_post_read_required(inode))
3616	3677	return true;
3617		− if (sbi->s_ndevs)
	3678	+ if (f2fs_is_multi_device(sbi))
3618	3679	return true;
3619	3680	/*
3620	3681	* for blkzoned device, fallback direct IO to buffered IO, so
@@ -3651,4 +3712,4 @@ static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) · 30 unmodified lines
3651	3712	return false;
3652	3713	}
3653	3714
3654		−#endif
	3715	+#endif /* _LINUX_F2FS_H */

fs/f2fs/file.c+47−29

@@ -39,6 +39,8 @@ static vm_fault_t f2fs_filemap_fault(struct vm_fault *vmf) · 38 unmodified lines
39	39	ret = filemap_fault(vmf);
40	40	up_read(&F2FS_I(inode)->i_mmap_sem);
41	41
	42	+ trace_f2fs_filemap_fault(inode, vmf->pgoff, (unsigned long)ret);
	43	+
42	44	return ret;
43	45	}
44	46
@@ -356,7 +358,7 @@ static bool __found_offset(struct f2fs_sb_info *sbi, block_t blkaddr, · 311 unmodified lines
356	358	switch (whence) {
357	359	case SEEK_DATA:
358	360	if ((blkaddr == NEW_ADDR && dirty == pgofs) \|\|
359		− is_valid_data_blkaddr(sbi, blkaddr))
	361	+ __is_valid_data_blkaddr(blkaddr))
360	362	return true;
361	363	break;
362	364	case SEEK_HOLE:
@@ -422,7 +424,7 @@ static loff_t f2fs_seek_block(struct file *file, loff_t offset, int whence) · 59 unmodified lines
422	424
423	425	if (__is_valid_data_blkaddr(blkaddr) &&
424	426	!f2fs_is_valid_blkaddr(F2FS_I_SB(inode),
425		− blkaddr, DATA_GENERIC)) {
	427	+ blkaddr, DATA_GENERIC_ENHANCE)) {
426	428	f2fs_put_dnode(&dn);
427	429	goto fail;
428	430	}
@@ -523,7 +525,8 @@ void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count) · 94 unmodified lines
523	525	f2fs_set_data_blkaddr(dn);
524	526
525	527	if (__is_valid_data_blkaddr(blkaddr) &&
526		− !f2fs_is_valid_blkaddr(sbi, blkaddr, DATA_GENERIC))
	528	+ !f2fs_is_valid_blkaddr(sbi, blkaddr,
	529	+ DATA_GENERIC_ENHANCE))
527	530	continue;
528	531
529	532	f2fs_invalidate_blocks(sbi, blkaddr);
@@ -552,7 +555,7 @@ void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count) · 22 unmodified lines
552	555
553	556	void f2fs_truncate_data_blocks(struct dnode_of_data *dn)
554	557	{
555		− f2fs_truncate_data_blocks_range(dn, ADDRS_PER_BLOCK);
	558	+ f2fs_truncate_data_blocks_range(dn, ADDRS_PER_BLOCK(dn->inode));
556	559	}
557	560
558	561	static int truncate_partial_data_page(struct inode *inode, u64 from,
@@ -1006,7 +1009,8 @@ next_dnode: · 447 unmodified lines
1006	1009	} else if (ret == -ENOENT) {
1007	1010	if (dn.max_level == 0)
1008	1011	return -ENOENT;
1009		− done = min((pgoff_t)ADDRS_PER_BLOCK - dn.ofs_in_node, len);
	1012	+ done = min((pgoff_t)ADDRS_PER_BLOCK(inode) - dn.ofs_in_node,
	1013	+ len);
1010	1014	blkaddr += done;
1011	1015	do_replace += done;
1012	1016	goto next;
@@ -1017,6 +1021,14 @@ next_dnode: · 4 unmodified lines
1017	1021	for (i = 0; i < done; i++, blkaddr++, do_replace++, dn.ofs_in_node++) {
1018	1022	*blkaddr = datablock_addr(dn.inode,
1019	1023	dn.node_page, dn.ofs_in_node);
	1024	+
	1025	+ if (__is_valid_data_blkaddr(*blkaddr) &&
	1026	+ !f2fs_is_valid_blkaddr(sbi, *blkaddr,
	1027	+ DATA_GENERIC_ENHANCE)) {
	1028	+ f2fs_put_dnode(&dn);
	1029	+ return -EFAULT;
	1030	+ }
	1031	+
1020	1032	if (!f2fs_is_checkpointed_data(sbi, *blkaddr)) {
1021	1033
1022	1034	if (test_opt(sbi, LFS)) {
@@ -1157,7 +1169,7 @@ static int __exchange_data_block(struct inode *src_inode, · 134 unmodified lines
1157	1169	int ret;
1158	1170
1159	1171	while (len) {
1160		− olen = min((pgoff_t)4 * ADDRS_PER_BLOCK, len);
	1172	+ olen = min((pgoff_t)4 * ADDRS_PER_BLOCK(src_inode), len);
1161	1173
1162	1174	src_blkaddr = f2fs_kvzalloc(F2FS_I_SB(src_inode),
1163	1175	array_size(olen, sizeof(block_t)),
@@ -2573,10 +2585,10 @@ static int f2fs_ioc_flush_device(struct file *filp, unsigned long arg) · 1409 unmodified lines
2573	2585	sizeof(range)))
2574	2586	return -EFAULT;
2575	2587
2576		− if (sbi->s_ndevs <= 1 \|\| sbi->s_ndevs - 1 <= range.dev_num \|\|
	2588	+ if (!f2fs_is_multi_device(sbi) \|\| sbi->s_ndevs - 1 <= range.dev_num \|\|
2577	2589	__is_large_section(sbi)) {
2578	2590	f2fs_msg(sbi->sb, KERN_WARNING,
2579		− "Can't flush %u in %d for segs_per_sec %u != 1\n",
	2591	+ "Can't flush %u in %d for segs_per_sec %u != 1",
2580	2592	range.dev_num, sbi->s_ndevs,
2581	2593	sbi->segs_per_sec);
2582	2594	return -EINVAL;
@@ -2858,7 +2870,7 @@ int f2fs_pin_file_control(struct inode *inode, bool inc) · 275 unmodified lines
2858	2870
2859	2871	if (fi->i_gc_failures[GC_FAILURE_PIN] > sbi->gc_pin_file_threshold) {
2860	2872	f2fs_msg(sbi->sb, KERN_WARNING,
2861		− "%s: Enable GC = ino %lx after %x GC trials\n",
	2873	+ "%s: Enable GC = ino %lx after %x GC trials",
2862	2874	__func__, inode->i_ino,
2863	2875	fi->i_gc_failures[GC_FAILURE_PIN]);
2864	2876	clear_inode_flag(inode, FI_PIN_FILE);
@@ -3035,15 +3047,21 @@ static ssize_t f2fs_file_write_iter(struct kiocb iocb, struct iov_iter from) · 170 unmodified lines
3035	3047	struct inode *inode = file_inode(file);
3036	3048	ssize_t ret;
3037	3049
3038		− if (unlikely(f2fs_cp_error(F2FS_I_SB(inode))))
3039		− return -EIO;
	3050	+ if (unlikely(f2fs_cp_error(F2FS_I_SB(inode)))) {
	3051	+ ret = -EIO;
	3052	+ goto out;
	3053	+ }
3040	3054
3041		− if ((iocb->ki_flags & IOCB_NOWAIT) && !(iocb->ki_flags & IOCB_DIRECT))
3042		− return -EINVAL;
	3055	+ if ((iocb->ki_flags & IOCB_NOWAIT) && !(iocb->ki_flags & IOCB_DIRECT)) {
	3056	+ ret = -EINVAL;
	3057	+ goto out;
	3058	+ }
3043	3059
3044	3060	if (!inode_trylock(inode)) {
3045		− if (iocb->ki_flags & IOCB_NOWAIT)
3046		− return -EAGAIN;
	3061	+ if (iocb->ki_flags & IOCB_NOWAIT) {
	3062	+ ret = -EAGAIN;
	3063	+ goto out;
	3064	+ }
3047	3065	inode_lock(inode);
3048	3066	}
3049	3067
@@ -3056,19 +3074,16 @@ static ssize_t f2fs_file_write_iter(struct kiocb iocb, struct iov_iter from) · 6 unmodified lines
3056	3074	if (iov_iter_fault_in_readable(from, iov_iter_count(from)))
3057	3075	set_inode_flag(inode, FI_NO_PREALLOC);
3058	3076
3059		− if ((iocb->ki_flags & IOCB_NOWAIT) &&
3060		− (iocb->ki_flags & IOCB_DIRECT)) {
3061		− if (!f2fs_overwrite_io(inode, iocb->ki_pos,
	3077	+ if ((iocb->ki_flags & IOCB_NOWAIT)) {
	3078	+ if (!f2fs_overwrite_io(inode, iocb->ki_pos,
3062	3079	iov_iter_count(from)) \|\|
3063		− f2fs_has_inline_data(inode) \|\|
3064		− f2fs_force_buffered_io(inode,
3065		− iocb, from)) {
3066		− clear_inode_flag(inode,
3067		− FI_NO_PREALLOC);
3068		− inode_unlock(inode);
3069		− return -EAGAIN;
3070		− }
3071		−
	3080	+ f2fs_has_inline_data(inode) \|\|
	3081	+ f2fs_force_buffered_io(inode, iocb, from)) {
	3082	+ clear_inode_flag(inode, FI_NO_PREALLOC);
	3083	+ inode_unlock(inode);
	3084	+ ret = -EAGAIN;
	3085	+ goto out;
	3086	+ }
3072	3087	} else {
3073	3088	preallocated = true;
3074	3089	target_size = iocb->ki_pos + iov_iter_count(from);
@@ -3077,7 +3092,8 @@ static ssize_t f2fs_file_write_iter(struct kiocb iocb, struct iov_iter from) · 2 unmodified lines
3077	3092	if (err) {
3078	3093	clear_inode_flag(inode, FI_NO_PREALLOC);
3079	3094	inode_unlock(inode);
3080		− return err;
	3095	+ ret = err;
	3096	+ goto out;
3081	3097	}
3082	3098	}
3083	3099	ret = __generic_file_write_iter(iocb, from);
@@ -3091,7 +3107,9 @@ static ssize_t f2fs_file_write_iter(struct kiocb iocb, struct iov_iter from) · 7 unmodified lines
3091	3107	f2fs_update_iostat(F2FS_I_SB(inode), APP_WRITE_IO, ret);
3092	3108	}
3093	3109	inode_unlock(inode);
3094		−
	3110	+out:
	3111	+ trace_f2fs_file_write_iter(inode, iocb->ki_pos,
	3112	+ iov_iter_count(from), ret);
3095	3113	if (ret > 0)
3096	3114	ret = generic_write_sync(iocb, ret);
3097	3115	return ret;

fs/f2fs/gc.c+13−3

@@ -591,7 +591,7 @@ block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode) · 590 unmodified lines
591	591	int dec = (node_ofs - indirect_blks - 3) / (NIDS_PER_BLOCK + 1);
592	592	bidx = node_ofs - 5 - dec;
593	593	}
594		− return bidx * ADDRS_PER_BLOCK + ADDRS_PER_INODE(inode);
	594	+ return bidx * ADDRS_PER_BLOCK(inode) + ADDRS_PER_INODE(inode);
595	595	}
596	596
597	597	static bool is_alive(struct f2fs_sb_info sbi, struct f2fs_summary sum,
@@ -656,6 +656,11 @@ static int ra_data_block(struct inode *inode, pgoff_t index) · 58 unmodified lines
656	656
657	657	if (f2fs_lookup_extent_cache(inode, index, &ei)) {
658	658	dn.data_blkaddr = ei.blk + index - ei.fofs;
	659	+ if (unlikely(!f2fs_is_valid_blkaddr(sbi, dn.data_blkaddr,
	660	+ DATA_GENERIC_ENHANCE_READ))) {
	661	+ err = -EFAULT;
	662	+ goto put_page;
	663	+ }
659	664	goto got_it;
660	665	}
661	666
@@ -665,8 +670,12 @@ static int ra_data_block(struct inode *inode, pgoff_t index) · 3 unmodified lines
665	670	goto put_page;
666	671	f2fs_put_dnode(&dn);
667	672
	673	+ if (!__is_valid_data_blkaddr(dn.data_blkaddr)) {
	674	+ err = -ENOENT;
	675	+ goto put_page;
	676	+ }
668	677	if (unlikely(!f2fs_is_valid_blkaddr(sbi, dn.data_blkaddr,
669		− DATA_GENERIC))) {
	678	+ DATA_GENERIC_ENHANCE))) {
670	679	err = -EFAULT;
671	680	goto put_page;
672	681	}
@@ -1175,6 +1184,7 @@ static int do_garbage_collect(struct f2fs_sb_info *sbi, · 502 unmodified lines
1175	1184	"type [%d, %d] in SSA and SIT",
1176	1185	segno, type, GET_SUM_TYPE((&sum->footer)));
1177	1186	set_sbi_flag(sbi, SBI_NEED_FSCK);
	1187	+ f2fs_stop_checkpoint(sbi, false);
1178	1188	goto skip;
1179	1189	}
1180	1190
@@ -1346,7 +1356,7 @@ void f2fs_build_gc_manager(struct f2fs_sb_info *sbi) · 165 unmodified lines
1346	1356	sbi->gc_pin_file_threshold = DEF_GC_FAILED_PINNED_FILES;
1347	1357
1348	1358	/* give warm/cold data area from slower device */
1349		− if (sbi->s_ndevs && !__is_large_section(sbi))
	1359	+ if (f2fs_is_multi_device(sbi) && !__is_large_section(sbi))
1350	1360	SIT_I(sbi)->last_victim[ALLOC_NEXT] =
1351	1361	GET_SEGNO(sbi, FDEV(0).end_blk) + 1;
1352	1362	}

fs/f2fs/inline.c+17−0

@@ -420,6 +420,14 @@ static int f2fs_move_inline_dirents(struct inode dir, struct page ipage, · 419 unmodified lines
420	420	stat_dec_inline_dir(dir);
421	421	clear_inode_flag(dir, FI_INLINE_DENTRY);
422	422
	423	+ /*
	424	+ * should retrieve reserved space which was used to keep
	425	+ * inline_dentry's structure for backward compatibility.
	426	+ */
	427	+ if (!f2fs_sb_has_flexible_inline_xattr(F2FS_I_SB(dir)) &&
	428	+ !f2fs_has_inline_xattr(dir))
	429	+ F2FS_I(dir)->i_inline_xattr_size = 0;
	430	+
423	431	f2fs_i_depth_write(dir, 1);
424	432	if (i_size_read(dir) < PAGE_SIZE)
425	433	f2fs_i_size_write(dir, PAGE_SIZE);
@@ -501,6 +509,15 @@ static int f2fs_move_rehashed_dirents(struct inode dir, struct page ipage, · 75 unmodified lines
501	509
502	510	stat_dec_inline_dir(dir);
503	511	clear_inode_flag(dir, FI_INLINE_DENTRY);
	512	+
	513	+ /*
	514	+ * should retrieve reserved space which was used to keep
	515	+ * inline_dentry's structure for backward compatibility.
	516	+ */
	517	+ if (!f2fs_sb_has_flexible_inline_xattr(F2FS_I_SB(dir)) &&
	518	+ !f2fs_has_inline_xattr(dir))
	519	+ F2FS_I(dir)->i_inline_xattr_size = 0;
	520	+
504	521	kvfree(backup_dentry);
505	522	return 0;
506	523	recover:

fs/f2fs/inode.c+7−5

@@ -73,7 +73,7 @@ static int __written_first_block(struct f2fs_sb_info *sbi, · 72 unmodified lines
73	73
74	74	if (!__is_valid_data_blkaddr(addr))
75	75	return 1;
76		− if (!f2fs_is_valid_blkaddr(sbi, addr, DATA_GENERIC))
	76	+ if (!f2fs_is_valid_blkaddr(sbi, addr, DATA_GENERIC_ENHANCE))
77	77	return -EFAULT;
78	78	return 0;
79	79	}
@@ -177,8 +177,8 @@ bool f2fs_inode_chksum_verify(struct f2fs_sb_info sbi, struct page page) · 97 unmodified lines
177	177
178	178	if (provided != calculated)
179	179	f2fs_msg(sbi->sb, KERN_WARNING,
180		− "checksum invalid, ino = %x, %x vs. %x",
181		− ino_of_node(page), provided, calculated);
	180	+ "checksum invalid, nid = %lu, ino_of_node = %x, %x vs. %x",
	181	+ page->index, ino_of_node(page), provided, calculated);
182	182
183	183	return provided == calculated;
184	184	}
@@ -267,9 +267,10 @@ static bool sanity_check_inode(struct inode inode, struct page node_page) · 82 unmodified lines
267	267	struct extent_info *ei = &F2FS_I(inode)->extent_tree->largest;
268	268
269	269	if (ei->len &&
270		− (!f2fs_is_valid_blkaddr(sbi, ei->blk, DATA_GENERIC) \|\|
	270	+ (!f2fs_is_valid_blkaddr(sbi, ei->blk,
	271	+ DATA_GENERIC_ENHANCE) \|\|
271	272	!f2fs_is_valid_blkaddr(sbi, ei->blk + ei->len - 1,
272		− DATA_GENERIC))) {
	273	+ DATA_GENERIC_ENHANCE))) {
273	274	set_sbi_flag(sbi, SBI_NEED_FSCK);
274	275	f2fs_msg(sbi->sb, KERN_WARNING,
275	276	"%s: inode (ino=%lx) extent info [%u, %u, %u] "
@@ -488,6 +489,7 @@ make_now: · 212 unmodified lines
488	489	return inode;
489	490
490	491	bad_inode:
	492	+ f2fs_inode_synced(inode);
491	493	iget_failed(inode);
492	494	trace_f2fs_iget_exit(inode, ret);
493	495	return ERR_PTR(ret);

fs/f2fs/namei.c+1−1

@@ -143,7 +143,7 @@ fail_drop: · 142 unmodified lines
143	143	return ERR_PTR(err);
144	144	}
145	145
146		−static int is_extension_exist(const unsigned char s, const char sub)
	146	+static inline int is_extension_exist(const unsigned char s, const char sub)
147	147	{
148	148	size_t slen = strlen(s);
149	149	size_t sublen = strlen(sub);

fs/f2fs/node.c+30−13

@@ -454,7 +454,7 @@ static void set_node_addr(struct f2fs_sb_info sbi, struct node_info ni, · 453 unmodified lines
454	454	new_blkaddr == NULL_ADDR);
455	455	f2fs_bug_on(sbi, nat_get_blkaddr(e) == NEW_ADDR &&
456	456	new_blkaddr == NEW_ADDR);
457		− f2fs_bug_on(sbi, is_valid_data_blkaddr(sbi, nat_get_blkaddr(e)) &&
	457	+ f2fs_bug_on(sbi, __is_valid_data_blkaddr(nat_get_blkaddr(e)) &&
458	458	new_blkaddr == NEW_ADDR);
459	459
460	460	/* increment version no as node is removed */
@@ -465,7 +465,7 @@ static void set_node_addr(struct f2fs_sb_info sbi, struct node_info ni, · 4 unmodified lines
465	465
466	466	/* change address */
467	467	nat_set_blkaddr(e, new_blkaddr);
468		− if (!is_valid_data_blkaddr(sbi, new_blkaddr))
	468	+ if (!__is_valid_data_blkaddr(new_blkaddr))
469	469	set_nat_flag(e, IS_CHECKPOINTED, false);
470	470	__set_nat_cache_dirty(nm_i, e);
471	471
@@ -526,6 +526,7 @@ int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, · 54 unmodified lines
526	526	struct f2fs_nat_entry ne;
527	527	struct nat_entry *e;
528	528	pgoff_t index;
	529	+ block_t blkaddr;
529	530	int i;
530	531
531	532	ni->nid = nid;
@@ -569,6 +570,11 @@ int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, · 37 unmodified lines
569	570	node_info_from_raw_nat(ni, &ne);
570	571	f2fs_put_page(page, 1);
571	572	cache:
	573	+ blkaddr = le32_to_cpu(ne.block_addr);
	574	+ if (__is_valid_data_blkaddr(blkaddr) &&
	575	+ !f2fs_is_valid_blkaddr(sbi, blkaddr, DATA_GENERIC_ENHANCE))
	576	+ return -EFAULT;
	577	+
572	578	/* cache nat entry */
573	579	cache_nat_entry(sbi, nid, &ne);
574	580	return 0;
@@ -600,9 +606,9 @@ static void f2fs_ra_node_pages(struct page *parent, int start, int n) · 25 unmodified lines
600	606	pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs)
601	607	{
602	608	const long direct_index = ADDRS_PER_INODE(dn->inode);
603		− const long direct_blks = ADDRS_PER_BLOCK;
604		− const long indirect_blks = ADDRS_PER_BLOCK * NIDS_PER_BLOCK;
605		− unsigned int skipped_unit = ADDRS_PER_BLOCK;
	609	+ const long direct_blks = ADDRS_PER_BLOCK(dn->inode);
	610	+ const long indirect_blks = ADDRS_PER_BLOCK(dn->inode) * NIDS_PER_BLOCK;
	611	+ unsigned int skipped_unit = ADDRS_PER_BLOCK(dn->inode);
606	612	int cur_level = dn->cur_level;
607	613	int max_level = dn->max_level;
608	614	pgoff_t base = 0;
@@ -638,9 +644,9 @@ static int get_node_path(struct inode *inode, long block, · 29 unmodified lines
638	644	int offset[4], unsigned int noffset[4])
639	645	{
640	646	const long direct_index = ADDRS_PER_INODE(inode);
641		− const long direct_blks = ADDRS_PER_BLOCK;
	647	+ const long direct_blks = ADDRS_PER_BLOCK(inode);
642	648	const long dptrs_per_blk = NIDS_PER_BLOCK;
643		− const long indirect_blks = ADDRS_PER_BLOCK * NIDS_PER_BLOCK;
	649	+ const long indirect_blks = ADDRS_PER_BLOCK(inode) * NIDS_PER_BLOCK;
644	650	const long dindirect_blks = indirect_blks * NIDS_PER_BLOCK;
645	651	int n = 0;
646	652	int level = 0;
@@ -1181,8 +1187,14 @@ int f2fs_remove_inode_page(struct inode *inode) · 534 unmodified lines
1181	1187	f2fs_put_dnode(&dn);
1182	1188	return -EIO;
1183	1189	}
1184		− f2fs_bug_on(F2FS_I_SB(inode),
1185		− inode->i_blocks != 0 && inode->i_blocks != 8);
	1190	+
	1191	+ if (unlikely(inode->i_blocks != 0 && inode->i_blocks != 8)) {
	1192	+ f2fs_msg(F2FS_I_SB(inode)->sb, KERN_WARNING,
	1193	+ "Inconsistent i_blocks, ino:%lu, iblocks:%llu",
	1194	+ inode->i_ino,
	1195	+ (unsigned long long)inode->i_blocks);
	1196	+ set_sbi_flag(F2FS_I_SB(inode), SBI_NEED_FSCK);
	1197	+ }
1186	1198
1187	1199	/* will put inode & node pages */
1188	1200	err = truncate_node(&dn);
@@ -1277,9 +1289,10 @@ static int read_node_page(struct page *page, int op_flags) · 88 unmodified lines
1277	1289	int err;
1278	1290
1279	1291	if (PageUptodate(page)) {
1280		−#ifdef CONFIG_F2FS_CHECK_FS
1281		− f2fs_bug_on(sbi, !f2fs_inode_chksum_verify(sbi, page));
1282		−#endif
	1292	+ if (!f2fs_inode_chksum_verify(sbi, page)) {
	1293	+ ClearPageUptodate(page);
	1294	+ return -EBADMSG;
	1295	+ }
1283	1296	return LOCKED_PAGE;
1284	1297	}
1285	1298
@@ -1543,7 +1556,8 @@ static int __write_node_page(struct page page, bool atomic, bool submitted, · 257 unmodified lines
1543	1556	}
1544	1557
1545	1558	if (__is_valid_data_blkaddr(ni.blk_addr) &&
1546		− !f2fs_is_valid_blkaddr(sbi, ni.blk_addr, DATA_GENERIC)) {
	1559	+ !f2fs_is_valid_blkaddr(sbi, ni.blk_addr,
	1560	+ DATA_GENERIC_ENHANCE)) {
1547	1561	up_read(&sbi->node_write);
1548	1562	goto redirty_out;
1549	1563	}
@@ -2078,6 +2092,9 @@ static bool add_free_nid(struct f2fs_sb_info *sbi, · 528 unmodified lines
2078	2092	if (unlikely(nid == 0))
2079	2093	return false;
2080	2094
	2095	+ if (unlikely(f2fs_check_nid_range(sbi, nid)))
	2096	+ return false;
	2097	+
2081	2098	i = f2fs_kmem_cache_alloc(free_nid_slab, GFP_NOFS);
2082	2099	i->nid = nid;
2083	2100	i->state = FREE_NID;

fs/f2fs/recovery.c+32−5

@@ -325,8 +325,10 @@ static int find_fsync_dnodes(struct f2fs_sb_info sbi, struct list_head head, · 324 unmodified lines
325	325	break;
326	326	}
327	327
328		− if (!is_recoverable_dnode(page))
	328	+ if (!is_recoverable_dnode(page)) {
	329	+ f2fs_put_page(page, 1);
329	330	break;
	331	+ }
330	332
331	333	if (!is_fsync_dnode(page))
332	334	goto next;
@@ -338,8 +340,10 @@ static int find_fsync_dnodes(struct f2fs_sb_info sbi, struct list_head head, · 5 unmodified lines
338	340	if (!check_only &&
339	341	IS_INODE(page) && is_dent_dnode(page)) {
340	342	err = f2fs_recover_inode_page(sbi, page);
341		− if (err)
	343	+ if (err) {
	344	+ f2fs_put_page(page, 1);
342	345	break;
	346	+ }
343	347	quota_inode = true;
344	348	}
345	349
@@ -355,6 +359,7 @@ static int find_fsync_dnodes(struct f2fs_sb_info sbi, struct list_head head, · 9 unmodified lines
355	359	err = 0;
356	360	goto next;
357	361	}
	362	+ f2fs_put_page(page, 1);
358	363	break;
359	364	}
360	365	}
@@ -370,6 +375,7 @@ next: · 9 unmodified lines
370	375	"%s: detect looped node chain, "
371	376	"blkaddr:%u, next:%u",
372	377	__func__, blkaddr, next_blkaddr_of_node(page));
	378	+ f2fs_put_page(page, 1);
373	379	err = -EINVAL;
374	380	break;
375	381	}
@@ -380,7 +386,6 @@ next: · 4 unmodified lines
380	386
381	387	f2fs_ra_meta_pages_cond(sbi, blkaddr);
382	388	}
383		− f2fs_put_page(page, 1);
384	389	return err;
385	390	}
386	391
@@ -546,7 +551,15 @@ retry_dn: · 159 unmodified lines
546	551	goto err;
547	552
548	553	f2fs_bug_on(sbi, ni.ino != ino_of_node(page));
549		− f2fs_bug_on(sbi, ofs_of_node(dn.node_page) != ofs_of_node(page));
	554	+
	555	+ if (ofs_of_node(dn.node_page) != ofs_of_node(page)) {
	556	+ f2fs_msg(sbi->sb, KERN_WARNING,
	557	+ "Inconsistent ofs_of_node, ino:%lu, ofs:%u, %u",
	558	+ inode->i_ino, ofs_of_node(dn.node_page),
	559	+ ofs_of_node(page));
	560	+ err = -EFAULT;
	561	+ goto err;
	562	+ }
550	563
551	564	for (; start < end; start++, dn.ofs_in_node++) {
552	565	block_t src, dest;
@@ -554,6 +567,18 @@ retry_dn: · 1 unmodified line
554	567	src = datablock_addr(dn.inode, dn.node_page, dn.ofs_in_node);
555	568	dest = datablock_addr(dn.inode, page, dn.ofs_in_node);
556	569
	570	+ if (__is_valid_data_blkaddr(src) &&
	571	+ !f2fs_is_valid_blkaddr(sbi, src, META_POR)) {
	572	+ err = -EFAULT;
	573	+ goto err;
	574	+ }
	575	+
	576	+ if (__is_valid_data_blkaddr(dest) &&
	577	+ !f2fs_is_valid_blkaddr(sbi, dest, META_POR)) {
	578	+ err = -EFAULT;
	579	+ goto err;
	580	+ }
	581	+
557	582	/* skip recovering if dest is the same as src */
558	583	if (src == dest)
559	584	continue;
@@ -666,8 +691,10 @@ static int recover_data(struct f2fs_sb_info sbi, struct list_head inode_list, · 106 unmodified lines
666	691	*/
667	692	if (IS_INODE(page)) {
668	693	err = recover_inode(entry->inode, page);
669		− if (err)
	694	+ if (err) {
	695	+ f2fs_put_page(page, 1);
670	696	break;
	697	+ }
671	698	}
672	699	if (entry->last_dentry == blkaddr) {
673	700	err = recover_dentry(entry->inode, page, dir_list);

fs/f2fs/segment.c+38−33

@@ -580,7 +580,7 @@ static int submit_flush_wait(struct f2fs_sb_info *sbi, nid_t ino) · 579 unmodified lines
580	580	int ret = 0;
581	581	int i;
582	582
583		− if (!sbi->s_ndevs)
	583	+ if (!f2fs_is_multi_device(sbi))
584	584	return __submit_flush_wait(sbi, sbi->sb->s_bdev);
585	585
586	586	for (i = 0; i < sbi->s_ndevs; i++) {
@@ -648,7 +648,8 @@ int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino) · 61 unmodified lines
648	648	return ret;
649	649	}
650	650
651		− if (atomic_inc_return(&fcc->queued_flush) == 1 \|\| sbi->s_ndevs > 1) {
	651	+ if (atomic_inc_return(&fcc->queued_flush) == 1 \|\|
	652	+ f2fs_is_multi_device(sbi)) {
652	653	ret = submit_flush_wait(sbi, ino);
653	654	atomic_dec(&fcc->queued_flush);
654	655
@@ -754,7 +755,7 @@ int f2fs_flush_device_cache(struct f2fs_sb_info *sbi) · 99 unmodified lines
754	755	{
755	756	int ret = 0, i;
756	757
757		− if (!sbi->s_ndevs)
	758	+ if (!f2fs_is_multi_device(sbi))
758	759	return 0;
759	760
760	761	for (i = 1; i < sbi->s_ndevs; i++) {
@@ -1367,9 +1368,12 @@ static int __queue_discard_cmd(struct f2fs_sb_info *sbi, · 606 unmodified lines
1367	1368	{
1368	1369	block_t lblkstart = blkstart;
1369	1370
	1371	+ if (!f2fs_bdev_support_discard(bdev))
	1372	+ return 0;
	1373	+
1370	1374	trace_f2fs_queue_discard(bdev, blkstart, blklen);
1371	1375
1372		− if (sbi->s_ndevs) {
	1376	+ if (f2fs_is_multi_device(sbi)) {
1373	1377	int devi = f2fs_target_device_index(sbi, blkstart);
1374	1378
1375	1379	blkstart -= FDEV(devi).start_blk;
@@ -1732,42 +1736,36 @@ static int __f2fs_issue_discard_zone(struct f2fs_sb_info *sbi, · 356 unmodified lines
1732	1736	block_t lblkstart = blkstart;
1733	1737	int devi = 0;
1734	1738
1735		− if (sbi->s_ndevs) {
	1739	+ if (f2fs_is_multi_device(sbi)) {
1736	1740	devi = f2fs_target_device_index(sbi, blkstart);
	1741	+ if (blkstart < FDEV(devi).start_blk \|\|
	1742	+ blkstart > FDEV(devi).end_blk) {
	1743	+ f2fs_msg(sbi->sb, KERN_ERR, "Invalid block %x",
	1744	+ blkstart);
	1745	+ return -EIO;
	1746	+ }
1737	1747	blkstart -= FDEV(devi).start_blk;
1738	1748	}
1739	1749
1740		− /*
1741		− * We need to know the type of the zone: for conventional zones,
1742		− * use regular discard if the drive supports it. For sequential
1743		− * zones, reset the zone write pointer.
1744		− */
1745		− switch (get_blkz_type(sbi, bdev, blkstart)) {
1746		−
1747		− case BLK_ZONE_TYPE_CONVENTIONAL:
1748		− if (!blk_queue_discard(bdev_get_queue(bdev)))
1749		− return 0;
1750		− return __queue_discard_cmd(sbi, bdev, lblkstart, blklen);
1751		− case BLK_ZONE_TYPE_SEQWRITE_REQ:
1752		− case BLK_ZONE_TYPE_SEQWRITE_PREF:
	1750	+ /* For sequential zones, reset the zone write pointer */
	1751	+ if (f2fs_blkz_is_seq(sbi, devi, blkstart)) {
1753	1752	sector = SECTOR_FROM_BLOCK(blkstart);
1754	1753	nr_sects = SECTOR_FROM_BLOCK(blklen);
1755	1754
1756	1755	if (sector & (bdev_zone_sectors(bdev) - 1) \|\|
1757	1756	nr_sects != bdev_zone_sectors(bdev)) {
1758		− f2fs_msg(sbi->sb, KERN_INFO,
1759		− "(%d) %s: Unaligned discard attempted (block %x + %x)",
	1757	+ f2fs_msg(sbi->sb, KERN_ERR,
	1758	+ "(%d) %s: Unaligned zone reset attempted (block %x + %x)",
1760	1759	devi, sbi->s_ndevs ? FDEV(devi).path: "",
1761	1760	blkstart, blklen);
1762	1761	return -EIO;
1763	1762	}
1764	1763	trace_f2fs_issue_reset_zone(bdev, blkstart);
1765		− return blkdev_reset_zones(bdev, sector,
1766		− nr_sects, GFP_NOFS);
1767		− default:
1768		− /* Unknown zone type: broken device ? */
1769		− return -EIO;
	1764	+ return blkdev_reset_zones(bdev, sector, nr_sects, GFP_NOFS);
1770	1765	}
	1766	+
	1767	+ /* For conventional zones, use regular discard if supported */
	1768	+ return __queue_discard_cmd(sbi, bdev, lblkstart, blklen);
1771	1769	}
1772	1770	#endif
1773	1771
@@ -1775,8 +1773,7 @@ static int __issue_discard_async(struct f2fs_sb_info *sbi, · 1 unmodified line
1775	1773	struct block_device *bdev, block_t blkstart, block_t blklen)
1776	1774	{
1777	1775	#ifdef CONFIG_BLK_DEV_ZONED
1778		− if (f2fs_sb_has_blkzoned(sbi) &&
1779		− bdev_zoned_model(bdev) != BLK_ZONED_NONE)
	1776	+ if (f2fs_sb_has_blkzoned(sbi) && bdev_is_zoned(bdev))
1780	1777	return __f2fs_issue_discard_zone(sbi, bdev, blkstart, blklen);
1781	1778	#endif
1782	1779	return __queue_discard_cmd(sbi, bdev, blkstart, blklen);
@@ -2172,8 +2169,11 @@ static void update_sit_entry(struct f2fs_sb_info *sbi, block_t blkaddr, int del) · 389 unmodified lines
2172	2169	* before, we must track that to know how much space we
2173	2170	* really have.
2174	2171	*/
2175		− if (f2fs_test_bit(offset, se->ckpt_valid_map))
	2172	+ if (f2fs_test_bit(offset, se->ckpt_valid_map)) {
	2173	+ spin_lock(&sbi->stat_lock);
2176	2174	sbi->unusable_block_count++;
	2175	+ spin_unlock(&sbi->stat_lock);
	2176	+ }
2177	2177	}
2178	2178
2179	2179	if (f2fs_test_and_clear_bit(offset, se->discard_map))
@@ -2220,7 +2220,7 @@ bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr) · 40 unmodified lines
2220	2220	struct seg_entry *se;
2221	2221	bool is_cp = false;
2222	2222
2223		− if (!is_valid_data_blkaddr(sbi, blkaddr))
	2223	+ if (!__is_valid_data_blkaddr(blkaddr))
2224	2224	return true;
2225	2225
2226	2226	down_read(&sit_i->sentry_lock);
@@ -3089,7 +3089,7 @@ static void update_device_state(struct f2fs_io_info *fio) · 862 unmodified lines
3089	3089	struct f2fs_sb_info *sbi = fio->sbi;
3090	3090	unsigned int devidx;
3091	3091
3092		− if (!sbi->s_ndevs)
	3092	+ if (!f2fs_is_multi_device(sbi))
3093	3093	return;
3094	3094
3095	3095	devidx = f2fs_target_device_index(sbi, fio->new_blkaddr);
@@ -3187,13 +3187,18 @@ int f2fs_inplace_write_data(struct f2fs_io_info *fio) · 91 unmodified lines
3187	3187	{
3188	3188	int err;
3189	3189	struct f2fs_sb_info *sbi = fio->sbi;
	3190	+ unsigned int segno;
3190	3191
3191	3192	fio->new_blkaddr = fio->old_blkaddr;
3192	3193	/* i/o temperature is needed for passing down write hints */
3193	3194	__get_segment_type(fio);
3194	3195
3195		− f2fs_bug_on(sbi, !IS_DATASEG(get_seg_entry(sbi,
3196		− GET_SEGNO(sbi, fio->new_blkaddr))->type));
	3196	+ segno = GET_SEGNO(sbi, fio->new_blkaddr);
	3197	+
	3198	+ if (!IS_DATASEG(get_seg_entry(sbi, segno)->type)) {
	3199	+ set_sbi_flag(sbi, SBI_NEED_FSCK);
	3200	+ return -EFAULT;
	3201	+ }
3197	3202
3198	3203	stat_inc_inplace_blocks(fio->sbi);
3199	3204
@@ -3336,7 +3341,7 @@ void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr) · 136 unmodified lines
3336	3341	if (!f2fs_post_read_required(inode))
3337	3342	return;
3338	3343
3339		− if (!is_valid_data_blkaddr(sbi, blkaddr))
	3344	+ if (!__is_valid_data_blkaddr(blkaddr))
3340	3345	return;
3341	3346
3342	3347	cpage = find_lock_page(META_MAPPING(sbi), blkaddr);

fs/f2fs/segment.h+8−8

@@ -82,7 +82,7 @@ · 81 unmodified lines
82	82	(GET_SEGOFF_FROM_SEG0(sbi, blk_addr) & ((sbi)->blocks_per_seg - 1))
83	83
84	84	#define GET_SEGNO(sbi, blk_addr) \
85		− ((!is_valid_data_blkaddr(sbi, blk_addr)) ? \
	85	+ ((!__is_valid_data_blkaddr(blk_addr)) ? \
86	86	NULL_SEGNO : GET_L2R_SEGNO(FREE_I(sbi), \
87	87	GET_SEGNO_FROM_SEG0(sbi, blk_addr)))
88	88	#define BLKS_PER_SEC(sbi) \
@@ -656,14 +656,15 @@ static inline void check_seg_range(struct f2fs_sb_info *sbi, unsigned int segno) · 567 unmodified lines
656	656	f2fs_bug_on(sbi, segno > TOTAL_SEGS(sbi) - 1);
657	657	}
658	658
659		−static inline void verify_block_addr(struct f2fs_io_info *fio, block_t blk_addr)
	659	+static inline void verify_fio_blkaddr(struct f2fs_io_info *fio)
660	660	{
661	661	struct f2fs_sb_info *sbi = fio->sbi;
662	662
663		− if (__is_meta_io(fio))
664		− verify_blkaddr(sbi, blk_addr, META_GENERIC);
665		− else
666		− verify_blkaddr(sbi, blk_addr, DATA_GENERIC);
	663	+ if (__is_valid_data_blkaddr(fio->old_blkaddr))
	664	+ verify_blkaddr(sbi, fio->old_blkaddr, __is_meta_io(fio) ?
	665	+ META_GENERIC : DATA_GENERIC);
	666	+ verify_blkaddr(sbi, fio->new_blkaddr, __is_meta_io(fio) ?
	667	+ META_GENERIC : DATA_GENERIC_ENHANCE);
667	668	}
668	669
669	670	/*
@@ -672,7 +673,6 @@ static inline void verify_block_addr(struct f2fs_io_info *fio, block_t blk_addr) · 2 unmodified lines
672	673	static inline int check_block_count(struct f2fs_sb_info *sbi,
673	674	int segno, struct f2fs_sit_entry *raw_sit)
674	675	{
675		−#ifdef CONFIG_F2FS_CHECK_FS
676	676	bool is_valid = test_bit_le(0, raw_sit->valid_map) ? true : false;
677	677	int valid_blocks = 0;
678	678	int cur_pos = 0, next_pos;
@@ -699,7 +699,7 @@ static inline int check_block_count(struct f2fs_sb_info *sbi, · 20 unmodified lines
699	699	set_sbi_flag(sbi, SBI_NEED_FSCK);
700	700	return -EINVAL;
701	701	}
702		−#endif
	702	+
703	703	/* check segment usage, and check boundary of a given segment number */
704	704	if (unlikely(GET_SIT_VBLOCKS(raw_sit) > sbi->blocks_per_seg
705	705	\|\| segno > TOTAL_SEGS(sbi) - 1)) {

fs/f2fs/super.c+54−16

@@ -1019,7 +1019,7 @@ static void destroy_device_list(struct f2fs_sb_info *sbi) · 1018 unmodified lines
1019	1019	for (i = 0; i < sbi->s_ndevs; i++) {
1020	1020	blkdev_put(FDEV(i).bdev, FMODE_EXCL);
1021	1021	#ifdef CONFIG_BLK_DEV_ZONED
1022		− kvfree(FDEV(i).blkz_type);
	1022	+ kvfree(FDEV(i).blkz_seq);
1023	1023	#endif
1024	1024	}
1025	1025	kvfree(sbi->devs);
@@ -1221,10 +1221,13 @@ static int f2fs_statfs(struct dentry dentry, struct kstatfs buf) · 195 unmodified lines
1221	1221	buf->f_blocks = total_count - start_count;
1222	1222	buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
1223	1223	sbi->current_reserved_blocks;
	1224	+
	1225	+ spin_lock(&sbi->stat_lock);
1224	1226	if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1225	1227	buf->f_bfree = 0;
1226	1228	else
1227	1229	buf->f_bfree -= sbi->unusable_block_count;
	1230	+ spin_unlock(&sbi->stat_lock);
1228	1231
1229	1232	if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1230	1233	buf->f_bavail = buf->f_bfree -
@@ -1499,9 +1502,15 @@ static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi) · 268 unmodified lines
1499	1502	mutex_lock(&sbi->gc_mutex);
1500	1503	cpc.reason = CP_PAUSE;
1501	1504	set_sbi_flag(sbi, SBI_CP_DISABLED);
1502		− f2fs_write_checkpoint(sbi, &cpc);
	1505	+ err = f2fs_write_checkpoint(sbi, &cpc);
	1506	+ if (err)
	1507	+ goto out_unlock;
1503	1508
	1509	+ spin_lock(&sbi->stat_lock);
1504	1510	sbi->unusable_block_count = 0;
	1511	+ spin_unlock(&sbi->stat_lock);
	1512	+
	1513	+out_unlock:
1505	1514	mutex_unlock(&sbi->gc_mutex);
1506	1515	restore_flag:
1507	1516	sbi->sb->s_flags = s_flags; /* Restore MS_RDONLY status */
@@ -2271,7 +2280,7 @@ static const struct export_operations f2fs_export_ops = { · 763 unmodified lines
2271	2280	static loff_t max_file_blocks(void)
2272	2281	{
2273	2282	loff_t result = 0;
2274		− loff_t leaf_count = ADDRS_PER_BLOCK;
	2283	+ loff_t leaf_count = DEF_ADDRS_PER_BLOCK;
2275	2284
2276	2285	/*
2277	2286	* note: previously, result is equal to (DEF_ADDRS_PER_INODE -
@@ -2449,7 +2458,7 @@ static int sanity_check_raw_super(struct f2fs_sb_info *sbi, · 171 unmodified lines
2449	2458	/* Currently, support only 4KB page cache size */
2450	2459	if (F2FS_BLKSIZE != PAGE_SIZE) {
2451	2460	f2fs_msg(sb, KERN_INFO,
2452		− "Invalid page_cache_size (%lu), supports only 4KB\n",
	2461	+ "Invalid page_cache_size (%lu), supports only 4KB",
2453	2462	PAGE_SIZE);
2454	2463	return 1;
2455	2464	}
@@ -2458,7 +2467,7 @@ static int sanity_check_raw_super(struct f2fs_sb_info *sbi, · 2 unmodified lines
2458	2467	blocksize = 1 << le32_to_cpu(raw_super->log_blocksize);
2459	2468	if (blocksize != F2FS_BLKSIZE) {
2460	2469	f2fs_msg(sb, KERN_INFO,
2461		− "Invalid blocksize (%u), supports only 4KB\n",
	2470	+ "Invalid blocksize (%u), supports only 4KB",
2462	2471	blocksize);
2463	2472	return 1;
2464	2473	}
@@ -2466,7 +2475,7 @@ static int sanity_check_raw_super(struct f2fs_sb_info *sbi, · 1 unmodified line
2466	2475	/* check log blocks per segment */
2467	2476	if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
2468	2477	f2fs_msg(sb, KERN_INFO,
2469		− "Invalid log blocks per segment (%u)\n",
	2478	+ "Invalid log blocks per segment (%u)",
2470	2479	le32_to_cpu(raw_super->log_blocks_per_seg));
2471	2480	return 1;
2472	2481	}
@@ -2587,7 +2596,8 @@ int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi) · 114 unmodified lines
2587	2596	unsigned int log_blocks_per_seg;
2588	2597	unsigned int segment_count_main;
2589	2598	unsigned int cp_pack_start_sum, cp_payload;
2590		− block_t user_block_count;
	2599	+ block_t user_block_count, valid_user_blocks;
	2600	+ block_t avail_node_count, valid_node_count;
2591	2601	int i, j;
2592	2602
2593	2603	total = le32_to_cpu(raw_super->segment_count);
@@ -2622,6 +2632,24 @@ int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi) · 28 unmodified lines
2622	2632	return 1;
2623	2633	}
2624	2634
	2635	+ valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
	2636	+ if (valid_user_blocks > user_block_count) {
	2637	+ f2fs_msg(sbi->sb, KERN_ERR,
	2638	+ "Wrong valid_user_blocks: %u, user_block_count: %u",
	2639	+ valid_user_blocks, user_block_count);
	2640	+ return 1;
	2641	+ }
	2642	+
	2643	+ valid_node_count = le32_to_cpu(ckpt->valid_node_count);
	2644	+ avail_node_count = sbi->total_node_count - sbi->nquota_files -
	2645	+ F2FS_RESERVED_NODE_NUM;
	2646	+ if (valid_node_count > avail_node_count) {
	2647	+ f2fs_msg(sbi->sb, KERN_ERR,
	2648	+ "Wrong valid_node_count: %u, avail_node_count: %u",
	2649	+ valid_node_count, avail_node_count);
	2650	+ return 1;
	2651	+ }
	2652	+
2625	2653	main_segs = le32_to_cpu(raw_super->segment_count_main);
2626	2654	blocks_per_seg = sbi->blocks_per_seg;
2627	2655
@@ -2793,9 +2821,11 @@ static int init_blkz_info(struct f2fs_sb_info *sbi, int devi) · 165 unmodified lines
2793	2821	if (nr_sectors & (bdev_zone_sectors(bdev) - 1))
2794	2822	FDEV(devi).nr_blkz++;
2795	2823
2796		− FDEV(devi).blkz_type = f2fs_kmalloc(sbi, FDEV(devi).nr_blkz,
2797		− GFP_KERNEL);
2798		− if (!FDEV(devi).blkz_type)
	2824	+ FDEV(devi).blkz_seq = f2fs_kzalloc(sbi,
	2825	+ BITS_TO_LONGS(FDEV(devi).nr_blkz)
	2826	+ * sizeof(unsigned long),
	2827	+ GFP_KERNEL);
	2828	+ if (!FDEV(devi).blkz_seq)
2799	2829	return -ENOMEM;
2800	2830
2801	2831	#define F2FS_REPORT_NR_ZONES 4096
@@ -2822,7 +2852,8 @@ static int init_blkz_info(struct f2fs_sb_info *sbi, int devi) · 20 unmodified lines
2822	2852	}
2823	2853
2824	2854	for (i = 0; i < nr_zones; i++) {
2825		− FDEV(devi).blkz_type[n] = zones[i].type;
	2855	+ if (zones[i].type != BLK_ZONE_TYPE_CONVENTIONAL)
	2856	+ set_bit(n, FDEV(devi).blkz_seq);
2826	2857	sector += zones[i].len;
2827	2858	n++;
2828	2859	}
@@ -3105,7 +3136,7 @@ try_onemore: · 276 unmodified lines
3105	3136	#ifndef CONFIG_BLK_DEV_ZONED
3106	3137	if (f2fs_sb_has_blkzoned(sbi)) {
3107	3138	f2fs_msg(sb, KERN_ERR,
3108		− "Zoned block device support is not enabled\n");
	3139	+ "Zoned block device support is not enabled");
3109	3140	err = -EOPNOTSUPP;
3110	3141	goto free_sb_buf;
3111	3142	}
@@ -3350,10 +3381,17 @@ try_onemore: · 238 unmodified lines
3350	3381	* mount should be failed, when device has readonly mode, and
3351	3382	* previous checkpoint was not done by clean system shutdown.
3352	3383	*/
3353		− if (bdev_read_only(sb->s_bdev) &&
3354		− !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
3355		− err = -EROFS;
3356		− goto free_meta;
	3384	+ if (f2fs_hw_is_readonly(sbi)) {
	3385	+ if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
	3386	+ err = -EROFS;
	3387	+ f2fs_msg(sb, KERN_ERR,
	3388	+ "Need to recover fsync data, but "
	3389	+ "write access unavailable");
	3390	+ goto free_meta;
	3391	+ }
	3392	+ f2fs_msg(sbi->sb, KERN_INFO, "write access "
	3393	+ "unavailable, skipping recovery");
	3394	+ goto reset_checkpoint;
3357	3395	}
3358	3396
3359	3397	if (need_fsck)

fs/f2fs/xattr.c+27−9

@@ -202,12 +202,17 @@ static inline const struct xattr_handler *f2fs_xattr_handler(int index) · 201 unmodified lines
202	202	return handler;
203	203	}
204	204
205		−static struct f2fs_xattr_entry __find_xattr(void base_addr, int index,
206		− size_t len, const char *name)
	205	+static struct f2fs_xattr_entry __find_xattr(void base_addr,
	206	+ void *last_base_addr, int index,
	207	+ size_t len, const char *name)
207	208	{
208	209	struct f2fs_xattr_entry *entry;
209	210
210	211	list_for_each_xattr(entry, base_addr) {
	212	+ if ((void *)(entry) + sizeof(__u32) > last_base_addr \|\|
	213	+ (void *)XATTR_NEXT_ENTRY(entry) > last_base_addr)
	214	+ return NULL;
	215	+
211	216	if (entry->e_name_index != index)
212	217	continue;
213	218	if (entry->e_name_len != len)
@@ -297,20 +302,22 @@ static int lookup_all_xattrs(struct inode inode, struct page ipage, · 83 unmodified lines
297	302	const char name, struct f2fs_xattr_entry *xe,
298	303	void *base_addr, int base_size)
299	304	{
300		− void cur_addr, txattr_addr, *last_addr = NULL;
	305	+ void cur_addr, txattr_addr, *last_txattr_addr;
	306	+ void *last_addr = NULL;
301	307	nid_t xnid = F2FS_I(inode)->i_xattr_nid;
302		− unsigned int size = xnid ? VALID_XATTR_BLOCK_SIZE : 0;
303	308	unsigned int inline_size = inline_xattr_size(inode);
304	309	int err = 0;
305	310
306		− if (!size && !inline_size)
	311	+ if (!xnid && !inline_size)
307	312	return -ENODATA;
308	313
309		− *base_size = inline_size + size + XATTR_PADDING_SIZE;
	314	+ *base_size = XATTR_SIZE(xnid, inode) + XATTR_PADDING_SIZE;
310	315	txattr_addr = f2fs_kzalloc(F2FS_I_SB(inode), *base_size, GFP_NOFS);
311	316	if (!txattr_addr)
312	317	return -ENOMEM;
313	318
	319	+ last_txattr_addr = (void *)txattr_addr + XATTR_SIZE(xnid, inode);
	320	+
314	321	/* read from inline xattr */
315	322	if (inline_size) {
316	323	err = read_inline_xattr(inode, ipage, txattr_addr);
@@ -337,7 +344,11 @@ static int lookup_all_xattrs(struct inode inode, struct page ipage, · 20 unmodified lines
337	344	else
338	345	cur_addr = txattr_addr;
339	346
340		− *xe = __find_xattr(cur_addr, index, len, name);
	347	+ *xe = __find_xattr(cur_addr, last_txattr_addr, index, len, name);
	348	+ if (!*xe) {
	349	+ err = -EFAULT;
	350	+ goto out;
	351	+ }
341	352	check:
342	353	if (IS_XATTR_LAST_ENTRY(*xe)) {
343	354	err = -ENODATA;
@@ -581,7 +592,8 @@ static int __f2fs_setxattr(struct inode *inode, int index, · 237 unmodified lines
581	592	struct page *ipage, int flags)
582	593	{
583	594	struct f2fs_xattr_entry here, last;
584		− void *base_addr;
	595	+ void base_addr, last_base_addr;
	596	+ nid_t xnid = F2FS_I(inode)->i_xattr_nid;
585	597	int found, newsize;
586	598	size_t len;
587	599	__u32 new_hsize;
@@ -605,8 +617,14 @@ static int __f2fs_setxattr(struct inode *inode, int index, · 17 unmodified lines
605	617	if (error)
606	618	return error;
607	619
	620	+ last_base_addr = (void *)base_addr + XATTR_SIZE(xnid, inode);
	621	+
608	622	/* find entry with wanted name. */
609		− here = __find_xattr(base_addr, index, len, name);
	623	+ here = __find_xattr(base_addr, last_base_addr, index, len, name);
	624	+ if (!here) {
	625	+ error = -EFAULT;
	626	+ goto exit;
	627	+ }
610	628
611	629	found = IS_XATTR_LAST_ENTRY(here) ? 0 : 1;
612	630

fs/f2fs/xattr.h+2−0

@@ -71,6 +71,8 @@ struct f2fs_xattr_entry { · 70 unmodified lines
71	71	entry = XATTR_NEXT_ENTRY(entry))
72	72	#define VALID_XATTR_BLOCK_SIZE (PAGE_SIZE - sizeof(struct node_footer))
73	73	#define XATTR_PADDING_SIZE (sizeof(__u32))
	74	+#define XATTR_SIZE(x,i) (((x) ? VALID_XATTR_BLOCK_SIZE : 0) + \
	75	+ (inline_xattr_size(i)))
74	76	#define MIN_OFFSET(i) XATTR_ALIGN(inline_xattr_size(i) + \
75	77	VALID_XATTR_BLOCK_SIZE)
76	78

include/linux/f2fs_fs.h+8−3

@@ -164,6 +164,10 @@ struct f2fs_checkpoint { · 163 unmodified lines
164	164	unsigned char sit_nat_version_bitmap[1];
165	165	} __packed;
166	166
	167	+#define CP_CHKSUM_OFFSET 4092 /* default chksum offset in checkpoint */
	168	+#define CP_MIN_CHKSUM_OFFSET \
	169	+ (offsetof(struct f2fs_checkpoint, sit_nat_version_bitmap))
	170	+
167	171	/*
168	172	* For orphan inode management
169	173	*/
@@ -198,11 +202,12 @@ struct f2fs_extent { · 28 unmodified lines
198	202	get_extra_isize(inode))
199	203	#define DEF_NIDS_PER_INODE 5 /* Node IDs in an Inode */
200	204	#define ADDRS_PER_INODE(inode) addrs_per_inode(inode)
201		−#define ADDRS_PER_BLOCK 1018 /* Address Pointers in a Direct Block */
	205	+#define DEF_ADDRS_PER_BLOCK 1018 /* Address Pointers in a Direct Block */
	206	+#define ADDRS_PER_BLOCK(inode) addrs_per_block(inode)
202	207	#define NIDS_PER_BLOCK 1018 /* Node IDs in an Indirect Block */
203	208
204	209	#define ADDRS_PER_PAGE(page, inode) \
205		− (IS_INODE(page) ? ADDRS_PER_INODE(inode) : ADDRS_PER_BLOCK)
	210	+ (IS_INODE(page) ? ADDRS_PER_INODE(inode) : ADDRS_PER_BLOCK(inode))
206	211
207	212	#define NODE_DIR1_BLOCK (DEF_ADDRS_PER_INODE + 1)
208	213	#define NODE_DIR2_BLOCK (DEF_ADDRS_PER_INODE + 2)
@@ -267,7 +272,7 @@ struct f2fs_inode { · 58 unmodified lines
267	272	} __packed;
268	273
269	274	struct direct_node {
270		− __le32 addr[ADDRS_PER_BLOCK]; /* array of data block address */
	275	+ __le32 addr[DEF_ADDRS_PER_BLOCK]; /* array of data block address */
271	276	} __packed;
272	277
273	278	struct indirect_node {

include/trace/events/f2fs.h+57−0

@@ -533,6 +533,37 @@ TRACE_EVENT(f2fs_truncate_partial_nodes, · 532 unmodified lines
533	533	__entry->err)
534	534	);
535	535
	536	+TRACE_EVENT(f2fs_file_write_iter,
	537	+
	538	+ TP_PROTO(struct inode *inode, unsigned long offset,
	539	+ unsigned long length, int ret),
	540	+
	541	+ TP_ARGS(inode, offset, length, ret),
	542	+
	543	+ TP_STRUCT__entry(
	544	+ __field(dev_t, dev)
	545	+ __field(ino_t, ino)
	546	+ __field(unsigned long, offset)
	547	+ __field(unsigned long, length)
	548	+ __field(int, ret)
	549	+ ),
	550	+
	551	+ TP_fast_assign(
	552	+ __entry->dev = inode->i_sb->s_dev;
	553	+ __entry->ino = inode->i_ino;
	554	+ __entry->offset = offset;
	555	+ __entry->length = length;
	556	+ __entry->ret = ret;
	557	+ ),
	558	+
	559	+ TP_printk("dev = (%d,%d), ino = %lu, "
	560	+ "offset = %lu, length = %lu, written(err) = %d",
	561	+ show_dev_ino(__entry),
	562	+ __entry->offset,
	563	+ __entry->length,
	564	+ __entry->ret)
	565	+);
	566	+
536	567	TRACE_EVENT(f2fs_map_blocks,
537	568	TP_PROTO(struct inode inode, struct f2fs_map_blocks map, int ret),
538	569
@@ -1253,6 +1284,32 @@ DEFINE_EVENT(f2fs__page, f2fs_commit_inmem_page, · 714 unmodified lines
1253	1284	TP_ARGS(page, type)
1254	1285	);
1255	1286
	1287	+TRACE_EVENT(f2fs_filemap_fault,
	1288	+
	1289	+ TP_PROTO(struct inode *inode, pgoff_t index, unsigned long ret),
	1290	+
	1291	+ TP_ARGS(inode, index, ret),
	1292	+
	1293	+ TP_STRUCT__entry(
	1294	+ __field(dev_t, dev)
	1295	+ __field(ino_t, ino)
	1296	+ __field(pgoff_t, index)
	1297	+ __field(unsigned long, ret)
	1298	+ ),
	1299	+
	1300	+ TP_fast_assign(
	1301	+ __entry->dev = inode->i_sb->s_dev;
	1302	+ __entry->ino = inode->i_ino;
	1303	+ __entry->index = index;
	1304	+ __entry->ret = ret;
	1305	+ ),
	1306	+
	1307	+ TP_printk("dev = (%d,%d), ino = %lu, index = %lu, ret = %lx",
	1308	+ show_dev_ino(__entry),
	1309	+ (unsigned long)__entry->index,
	1310	+ __entry->ret)
	1311	+);
	1312	+
1256	1313	TRACE_EVENT(f2fs_writepages,
1257	1314
1258	1315	TP_PROTO(struct inode inode, struct writeback_control wbc, int type),

Merge tag 'f2fs-for-v5.2-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs

Keyboard shortcuts

General

Issues and pull requests

Issues, pull requests and Inbox

Repository code