f2fs: fix to do sanity check on total_data_blocks

PATCH Superseded Chao Yu 2 participants 3 messages Apr 29, 2022 – May 4, 2022

Subsystems (2)

Stable branch F2FS file system

lore

Share & tools

lore

Chao Yu

author maintainer

2022-04-29 14:45 UTC #0 lore raw

PATCH 1/1 f2fs: fix to do sanity check on total_data_blocks

As Yanming reported in bugzilla:

https://bugzilla.kernel.org/show_bug.cgi?id=215916

The kernel message is shown below:

kernel BUG at fs/f2fs/segment.c:2560!

Call Trace:
 allocate_segment_by_default+0x228/0x440
 f2fs_allocate_data_block+0x13d1/0x31f0
 do_write_page+0x18d/0x710
 f2fs_outplace_write_data+0x151/0x250
 f2fs_do_write_data_page+0xef9/0x1980
 move_data_page+0x6af/0xbc0
 do_garbage_collect+0x312f/0x46f0
 f2fs_gc+0x6b0/0x3bc0
 f2fs_balance_fs+0x921/0x2260
 f2fs_write_single_data_page+0x16be/0x2370
 f2fs_write_cache_pages+0x428/0xd00
 f2fs_write_data_pages+0x96e/0xd50
 do_writepages+0x168/0x550
 __writeback_single_inode+0x9f/0x870
 writeback_sb_inodes+0x47d/0xb20
 __writeback_inodes_wb+0xb2/0x200
 wb_writeback+0x4bd/0x660
 wb_workfn+0x5f3/0xab0
 process_one_work+0x79f/0x13e0
 worker_thread+0x89/0xf60
 kthread+0x26a/0x300
 ret_from_fork+0x22/0x30
RIP: 0010:new_curseg+0xe8d/0x15f0

The root cause is: ckpt.valid_block_count is inconsistent with SIT table, stat info indicates filesystem has free blocks, but SIT table indicates filesystem has no free segment.

So that during garbage colloection, it triggers panic when LFS allocator fails to find free segment.

This patch tries to fix this issue by checking consistency in between ckpt.valid_block_count and block accounted from SIT.

Cc [email protected]

Reported-by Ming Yan <[email protected]>

Signed-off-by Chao Yu <[email protected]>

---
 fs/f2fs/segment.c | 24 +++++++++++++++++++++---
 1 file changed, 21 insertions(+), 3 deletions(-)

fs/f2fs/segment.c +21 -3

index 8c17fed8987e..eddaf3b45b25 100644
@@ -4462,6 +4462,7 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
 	unsigned int readed, start_blk = 0;
 	int err = 0;
 	block_t total_node_blocks = 0;
+	block_t total_data_blocks = 0;
 
 	do {
 		readed = f2fs_ra_meta_pages(sbi, start_blk, BIO_MAX_VECS,

@@ -4488,6 +4489,8 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
 			seg_info_from_raw_sit(se, &sit);
 			if (IS_NODESEG(se->type))
 				total_node_blocks += se->valid_blocks;
+			else
+				total_data_blocks += se->valid_blocks;
 
 			if (f2fs_block_unit_discard(sbi)) {
 				/* build discard map only one time */

@@ -4529,6 +4532,8 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
 		old_valid_blocks = se->valid_blocks;
 		if (IS_NODESEG(se->type))
 			total_node_blocks -= old_valid_blocks;
+		else
+			total_data_blocks -= old_valid_blocks;
 
 		err = check_block_count(sbi, start, &sit);
 		if (err)

@@ -4536,6 +4541,8 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
 		seg_info_from_raw_sit(se, &sit);
 		if (IS_NODESEG(se->type))
 			total_node_blocks += se->valid_blocks;
+		else
+			total_data_blocks += se->valid_blocks;
 
 		if (f2fs_block_unit_discard(sbi)) {
 			if (is_set_ckpt_flags(sbi, CP_TRIMMED_FLAG)) {

@@ -4557,13 +4564,24 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
 	}
 	up_read(&curseg->journal_rwsem);
 
-	if (!err && total_node_blocks != valid_node_count(sbi)) {
+	if (err)
+		return err;
+
+	if (total_node_blocks != valid_node_count(sbi)) {
 		f2fs_err(sbi, "SIT is corrupted node# %u vs %u",
 			 total_node_blocks, valid_node_count(sbi));
-		err = -EFSCORRUPTED;
+		return -EFSCORRUPTED;
 	}
 
-	return err;
+	if (total_data_blocks + total_node_blocks !=
+				valid_user_blocks(sbi)) {
+		f2fs_err(sbi, "SIT is corrupted data# %u vs %u",
+			 total_data_blocks,
+			 valid_user_blocks(sbi) - total_node_blocks);
+		return -EFSCORRUPTED;
+	}
+
+	return 0;
 }
 
 static void init_free_segmap(struct f2fs_sb_info *sbi)

Signature · 1 line

2.25.1

Jaegeuk Kim

maintainer

2022-05-03 21:47 UTC #1 lore raw

Quoted 99 lines · Chao Yu

As Yanming reported in bugzilla:

https://bugzilla.kernel.org/show_bug.cgi?id=215916

The kernel message is shown below:

kernel BUG at fs/f2fs/segment.c:2560!

Call Trace:
 allocate_segment_by_default+0x228/0x440
 f2fs_allocate_data_block+0x13d1/0x31f0
 do_write_page+0x18d/0x710
 f2fs_outplace_write_data+0x151/0x250
 f2fs_do_write_data_page+0xef9/0x1980
 move_data_page+0x6af/0xbc0
 do_garbage_collect+0x312f/0x46f0
 f2fs_gc+0x6b0/0x3bc0
 f2fs_balance_fs+0x921/0x2260
 f2fs_write_single_data_page+0x16be/0x2370
 f2fs_write_cache_pages+0x428/0xd00
 f2fs_write_data_pages+0x96e/0xd50
 do_writepages+0x168/0x550
 __writeback_single_inode+0x9f/0x870
 writeback_sb_inodes+0x47d/0xb20
 __writeback_inodes_wb+0xb2/0x200
 wb_writeback+0x4bd/0x660
 wb_workfn+0x5f3/0xab0
 process_one_work+0x79f/0x13e0
 worker_thread+0x89/0xf60
 kthread+0x26a/0x300
 ret_from_fork+0x22/0x30
RIP: 0010:new_curseg+0xe8d/0x15f0

The root cause is: ckpt.valid_block_count is inconsistent with SIT table, stat info indicates filesystem has free blocks, but SIT table indicates filesystem has no free segment.

So that during garbage colloection, it triggers panic when LFS allocator fails to find free segment.

This patch tries to fix this issue by checking consistency in between ckpt.valid_block_count and block accounted from SIT.

Cc [email protected]

Reported-by Ming Yan <[email protected]>

Signed-off-by Chao Yu <[email protected]>

---
 fs/f2fs/segment.c | 24 +++++++++++++++++++++---
 1 file changed, 21 insertions(+), 3 deletions(-)

fs/f2fs/segment.c +11 -1

index 8c17fed8987e..eddaf3b45b25 100644
@@ -4462,6 +4462,7 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
 	unsigned int readed, start_blk = 0;
 	int err = 0;
 	block_t total_node_blocks = 0;
+	block_t total_data_blocks = 0;
 
 	do {
 		readed = f2fs_ra_meta_pages(sbi, start_blk, BIO_MAX_VECS,

@@ -4488,6 +4489,8 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
 			seg_info_from_raw_sit(se, &sit);
 			if (IS_NODESEG(se->type))
 				total_node_blocks += se->valid_blocks;
+			else
+				total_data_blocks += se->valid_blocks;
 
 			if (f2fs_block_unit_discard(sbi)) {
 				/* build discard map only one time */

@@ -4529,6 +4532,8 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
 		old_valid_blocks = se->valid_blocks;
 		if (IS_NODESEG(se->type))
 			total_node_blocks -= old_valid_blocks;
+		else
+			total_data_blocks -= old_valid_blocks;
 
 		err = check_block_count(sbi, start, &sit);
 		if (err)

@@ -4536,6 +4541,8 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
 		seg_info_from_raw_sit(se, &sit);
 		if (IS_NODESEG(se->type))
 			total_node_blocks += se->valid_blocks;
+		else
+			total_data_blocks += se->valid_blocks;
 
 		if (f2fs_block_unit_discard(sbi)) {
 			if (is_set_ckpt_flags(sbi, CP_TRIMMED_FLAG)) {

@@ -4557,13 +4564,24 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
 	}
 	up_read(&curseg->journal_rwsem);
 
-	if (!err && total_node_blocks != valid_node_count(sbi)) {
+	if (err)
+		return err;
+
+	if (total_node_blocks != valid_node_count(sbi)) {
 		f2fs_err(sbi, "SIT is corrupted node# %u vs %u",
 			 total_node_blocks, valid_node_count(sbi));

fs/f2fs/segment.c static int build_sit_entries(struct f2fs_sb_info *sbi) · line 4580

Quoted 10 lines

fs/f2fs/segment.c +6 -2

-		err = -EFSCORRUPTED;
+		return -EFSCORRUPTED;
 	}
 
-	return err;
+	if (total_data_blocks + total_node_blocks !=
+				valid_user_blocks(sbi)) {
+		f2fs_err(sbi, "SIT is corrupted data# %u vs %u",
+			 total_data_blocks,
+			 valid_user_blocks(sbi) - total_node_blocks);

This doesn't work, since some NEW_ADDR is not counted from SIT.

Quoted 9 lines

fs/f2fs/segment.c +4 -0

+		return -EFSCORRUPTED;
+	}
+
+	return 0;
 }
 
 static void init_free_segmap(struct f2fs_sb_info *sbi)

Signature · 1 line

2.25.1

_______________________________________________
Linux-f2fs-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

Chao Yu

author maintainer

2022-05-04 00:42 UTC #2 lore raw

Quoted 104 lines · Jaegeuk Kim

Quoted 103 lines · Chao Yu

As Yanming reported in bugzilla:

https://bugzilla.kernel.org/show_bug.cgi?id=215916

The kernel message is shown below:

kernel BUG at fs/f2fs/segment.c:2560!

Call Trace:
  allocate_segment_by_default+0x228/0x440
  f2fs_allocate_data_block+0x13d1/0x31f0
  do_write_page+0x18d/0x710
  f2fs_outplace_write_data+0x151/0x250
  f2fs_do_write_data_page+0xef9/0x1980
  move_data_page+0x6af/0xbc0
  do_garbage_collect+0x312f/0x46f0
  f2fs_gc+0x6b0/0x3bc0
  f2fs_balance_fs+0x921/0x2260
  f2fs_write_single_data_page+0x16be/0x2370
  f2fs_write_cache_pages+0x428/0xd00
  f2fs_write_data_pages+0x96e/0xd50
  do_writepages+0x168/0x550
  __writeback_single_inode+0x9f/0x870
  writeback_sb_inodes+0x47d/0xb20
  __writeback_inodes_wb+0xb2/0x200
  wb_writeback+0x4bd/0x660
  wb_workfn+0x5f3/0xab0
  process_one_work+0x79f/0x13e0
  worker_thread+0x89/0xf60
  kthread+0x26a/0x300
  ret_from_fork+0x22/0x30
RIP: 0010:new_curseg+0xe8d/0x15f0

The root cause is: ckpt.valid_block_count is inconsistent with SIT table, stat info indicates filesystem has free blocks, but SIT table indicates filesystem has no free segment.

So that during garbage colloection, it triggers panic when LFS allocator fails to find free segment.

This patch tries to fix this issue by checking consistency in between ckpt.valid_block_count and block accounted from SIT.

Cc [email protected]

Reported-by Ming Yan <[email protected]>

Signed-off-by Chao Yu <[email protected]>

---
  fs/f2fs/segment.c | 24 +++++++++++++++++++++---
  1 file changed, 21 insertions(+), 3 deletions(-)

fs/f2fs/segment.c +12 -2

index 8c17fed8987e..eddaf3b45b25 100644
@@ -4462,6 +4462,7 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
  	unsigned int readed, start_blk = 0;
  	int err = 0;
  	block_t total_node_blocks = 0;
+	block_t total_data_blocks = 0;
 
  	do {
  		readed = f2fs_ra_meta_pages(sbi, start_blk, BIO_MAX_VECS,

@@ -4488,6 +4489,8 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
  			seg_info_from_raw_sit(se, &sit);
  			if (IS_NODESEG(se->type))
  				total_node_blocks += se->valid_blocks;
+			else
+				total_data_blocks += se->valid_blocks;
 
  			if (f2fs_block_unit_discard(sbi)) {
  				/* build discard map only one time */

@@ -4529,6 +4532,8 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
  		old_valid_blocks = se->valid_blocks;
  		if (IS_NODESEG(se->type))
  			total_node_blocks -= old_valid_blocks;
+		else
+			total_data_blocks -= old_valid_blocks;
 
  		err = check_block_count(sbi, start, &sit);
  		if (err)

@@ -4536,6 +4541,8 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
  		seg_info_from_raw_sit(se, &sit);
  		if (IS_NODESEG(se->type))
  			total_node_blocks += se->valid_blocks;
+		else
+			total_data_blocks += se->valid_blocks;
 
  		if (f2fs_block_unit_discard(sbi)) {
  			if (is_set_ckpt_flags(sbi, CP_TRIMMED_FLAG)) {

@@ -4557,13 +4564,24 @@ static int build_sit_entries(struct f2fs_sb_info *sbi)
  	}
  	up_read(&curseg->journal_rwsem);
 
-	if (!err && total_node_blocks != valid_node_count(sbi)) {
+	if (err)
+		return err;
+
+	if (total_node_blocks != valid_node_count(sbi)) {
  		f2fs_err(sbi, "SIT is corrupted node# %u vs %u",
  			 total_node_blocks, valid_node_count(sbi));
-		err = -EFSCORRUPTED;
+		return -EFSCORRUPTED;
  	}

Quoted 8 lines

fs/f2fs/segment.c static int build_sit_entries(struct f2fs_sb_info *sbi) · line 4580
Quoted 6 lines
fs/f2fs/segment.c +5 -1
-	return err;
+	if (total_data_blocks + total_node_blocks !=
+				valid_user_blocks(sbi)) {
+		f2fs_err(sbi, "SIT is corrupted data# %u vs %u",
+			 total_data_blocks,
+			 valid_user_blocks(sbi) - total_node_blocks);
This doesn't work, since some NEW_ADDR is not counted from SIT.

Yup, so how about updating check condition as below? it means image may be corrupted if blocks counted from SIT is more than the one counted from CKPT.

if (total_data_blocks + total_node_blocks >
valid_user_blocks(sbi))

Thanks,

Quoted 10 lines

fs/f2fs/segment.c +4 -0

+		return -EFSCORRUPTED;
+	}
+
+	return 0;
  }
 
  static void init_free_segmap(struct f2fs_sb_info *sbi)

Signature · 1 line

2.25.1