Skip to contentSkip to conversation

ext4: save the error code which triggered an ext4_error() in the superblock

PATCH In review Theodore Ts'o 3 participants 8 messages Dec 4, 2019 – Dec 13, 2019 ext4 file system

lore
Share & tools
lore

Theodore Ts'o

author maintainer
#0 lore raw

PATCH 1/2 ext4: save the error code which triggered an ext4_error() in the superblock

This allows the cause of an ext4_error() report to be categorized based on whether it was triggered due to an I/O error, or an memory allocation error, or other possible causes. Most errors are caused by a detected file system inconsistency, so the default code stored in the superblock will be EXT4_ERR_EFSCORRUPTED.

Previous-Version-Link: https://lore.kernel.org/r/[email protected]

---

Notes:
v2: no changes from previous version

 fs/ext4/balloc.c    |  1 +
 fs/ext4/ext4.h      | 30 +++++++++++++++++++-
 fs/ext4/ext4_jbd2.c |  3 ++
 fs/ext4/extents.c   |  1 +
 fs/ext4/ialloc.c    |  2 ++
 fs/ext4/inline.c    |  2 ++
 fs/ext4/inode.c     |  8 +++++-
 fs/ext4/mballoc.c   |  4 +++
 fs/ext4/mmp.c       |  6 +++-
 fs/ext4/namei.c     |  4 +++
 fs/ext4/super.c     | 68 ++++++++++++++++++++++++++++++++++++++++++++-
 fs/ext4/xattr.c     |  4 ++-
 12 files changed, 128 insertions(+), 5 deletions(-)
fs/ext4/balloc.c +1 -0
index 0b202e00d93f..102c38527a10 100644
@@ -506,6 +506,7 @@ int ext4_wait_block_bitmap(struct super_block *sb, ext4_group_t block_group,
 		return -EFSCORRUPTED;
 	wait_on_buffer(bh);
 	if (!buffer_uptodate(bh)) {
+		ext4_set_errno(sb, EIO);
 		ext4_error(sb, "Cannot read block bitmap - "
 			   "block_group = %u, block_bitmap = %llu",
 			   block_group, (unsigned long long) bh->b_blocknr);
fs/ext4/ext4.h +29 -1
index 61987c106511..1c9ac0fc8715 100644
@@ -1343,7 +1343,8 @@ struct ext4_super_block {
 	__u8	s_lastcheck_hi;
 	__u8	s_first_error_time_hi;
 	__u8	s_last_error_time_hi;
-	__u8	s_pad[2];
+	__u8	s_first_error_errcode;
+	__u8    s_last_error_errcode;
 	__le16  s_encoding;		/* Filename charset encoding */
 	__le16  s_encoding_flags;	/* Filename charset encoding flags */
 	__le32	s_reserved[95];		/* Padding to the end of the block */
@@ -1574,6 +1575,32 @@ static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
 }
 
+/*
+ * Error number codes for s_{first,last}_error_errno
+ *
+ * Linux errno numbers are architecture specific, so we need to translate
+ * them into something which is architecture independent.   We don't define
+ * codes for all errno's; just the ones which are most likely to be the cause
+ * of an ext4_error() call.
+ */
+#define EXT4_ERR_UNKNOWN	 1
+#define EXT4_ERR_EIO		 2
+#define EXT4_ERR_ENOMEM		 3
+#define EXT4_ERR_EFSBADCRC	 4
+#define EXT4_ERR_EFSCORRUPTED	 5
+#define EXT4_ERR_ENOSPC		 6
+#define EXT4_ERR_ENOKEY		 7
+#define EXT4_ERR_EROFS		 8
+#define EXT4_ERR_EFBIG		 9
+#define EXT4_ERR_EEXIST		10
+#define EXT4_ERR_ERANGE		11
+#define EXT4_ERR_EOVERFLOW	12
+#define EXT4_ERR_EBUSY		13
+#define EXT4_ERR_ENOTDIR	14
+#define EXT4_ERR_ENOTEMPTY	15
+#define EXT4_ERR_ESHUTDOWN	16
+#define EXT4_ERR_EFAULT		17
+
 /*
  * Inode dynamic state flags
  */
@@ -2686,6 +2713,7 @@ extern const char *ext4_decode_error(struct super_block *sb, int errno,
 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
 					     ext4_group_t block_group,
 					     unsigned int flags);
+extern void ext4_set_errno(struct super_block *sb, int err);
 
 extern __printf(4, 5)
 void __ext4_error(struct super_block *, const char *, unsigned int,
fs/ext4/ext4_jbd2.c +3 -0
index d3b8cdea5df7..19217a3f1ae4 100644
@@ -58,6 +58,7 @@ static int ext4_journal_check_start(struct super_block *sb)
 	 * take the FS itself readonly cleanly.
 	 */
 	if (journal && is_journal_aborted(journal)) {
+		ext4_set_errno(sb, -journal->j_errno);
 		ext4_abort(sb, "Detected aborted journal");
 		return -EROFS;
 	}
@@ -249,6 +250,7 @@ int __ext4_forget(const char *where, unsigned int line, handle_t *handle,
 	if (err) {
 		ext4_journal_abort_handle(where, line, __func__,
 					  bh, handle, err);
+		ext4_set_errno(inode->i_sb, -err);
 		__ext4_abort(inode->i_sb, where, line,
 			   "error %d when attempting revoke", err);
 	}
@@ -320,6 +322,7 @@ int __ext4_handle_dirty_metadata(const char *where, unsigned int line,
 				es = EXT4_SB(inode->i_sb)->s_es;
 				es->s_last_error_block =
 					cpu_to_le64(bh->b_blocknr);
+				ext4_set_errno(inode->i_sb, EIO);
 				ext4_error_inode(inode, where, line,
 						 bh->b_blocknr,
 					"IO error syncing itable block");
fs/ext4/extents.c +1 -0
index 0e8708b77da6..ee83fe7c98aa 100644
@@ -492,6 +492,7 @@ static int __ext4_ext_check(const char *function, unsigned int line,
 	return 0;
 
 corrupted:
+	ext4_set_errno(inode->i_sb, -err);
 	ext4_error_inode(inode, function, line, 0,
 			 "pblk %llu bad header/extent: %s - magic %x, "
 			 "entries %u, max %u(%u), depth %u(%u)",
fs/ext4/ialloc.c +2 -0
index dc333e8e51e8..9979e1a55be5 100644
@@ -194,6 +194,7 @@ ext4_read_inode_bitmap(struct super_block *sb, ext4_group_t block_group)
 	wait_on_buffer(bh);
 	if (!buffer_uptodate(bh)) {
 		put_bh(bh);
+		ext4_set_errno(sb, EIO);
 		ext4_error(sb, "Cannot read inode bitmap - "
 			   "block_group = %u, inode_bitmap = %llu",
 			   block_group, bitmap_blk);
@@ -1223,6 +1224,7 @@ struct inode *ext4_orphan_get(struct super_block *sb, unsigned long ino)
 	inode = ext4_iget(sb, ino, EXT4_IGET_NORMAL);
 	if (IS_ERR(inode)) {
 		err = PTR_ERR(inode);
+		ext4_set_errno(sb, -err);
 		ext4_error(sb, "couldn't read orphan inode %lu (err %d)",
 			   ino, err);
 		return inode;
fs/ext4/inline.c +2 -0
index 2fec62d764fa..e61603f47035 100644
@@ -98,6 +98,7 @@ int ext4_get_max_inline_size(struct inode *inode)
 
 	error = ext4_get_inode_loc(inode, &iloc);
 	if (error) {
+		ext4_set_errno(inode->i_sb, -error);
 		ext4_error_inode(inode, __func__, __LINE__, 0,
 				 "can't get inode location %lu",
 				 inode->i_ino);
@@ -1761,6 +1762,7 @@ bool empty_inline_dir(struct inode *dir, int *has_inline_data)
 
 	err = ext4_get_inode_loc(dir, &iloc);
 	if (err) {
+		ext4_set_errno(dir->i_sb, -err);
 		EXT4_ERROR_INODE(dir, "error %d getting inode %lu block",
 				 err, dir->i_ino);
 		return true;
fs/ext4/inode.c +7 -1
index 310e4abd9aca..1aed7c653b42 100644
@@ -271,6 +271,7 @@ void ext4_evict_inode(struct inode *inode)
 	if (inode->i_blocks) {
 		err = ext4_truncate(inode);
 		if (err) {
+			ext4_set_errno(inode->i_sb, -err);
 			ext4_error(inode->i_sb,
 				   "couldn't truncate inode %lu (err %d)",
 				   inode->i_ino, err);
@@ -2478,10 +2479,12 @@ static int mpage_map_and_submit_extent(handle_t *handle,
 			EXT4_I(inode)->i_disksize = disksize;
 		up_write(&EXT4_I(inode)->i_data_sem);
 		err2 = ext4_mark_inode_dirty(handle, inode);
-		if (err2)
+		if (err2) {
+			ext4_set_errno(inode->i_sb, -err2);
 			ext4_error(inode->i_sb,
 				   "Failed to mark inode %lu dirty",
 				   inode->i_ino);
+		}
 		if (!err)
 			err = err2;
 	}
@@ -4338,6 +4341,7 @@ static int __ext4_get_inode_loc(struct inode *inode,
 		blk_finish_plug(&plug);
 		wait_on_buffer(bh);
 		if (!buffer_uptodate(bh)) {
+			ext4_set_errno(inode->i_sb, EIO);
 			EXT4_ERROR_INODE_BLOCK(inode, block,
 					       "unable to read itable block");
 			brelse(bh);
@@ -4552,6 +4556,7 @@ struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
 	}
 
 	if (!ext4_inode_csum_verify(inode, raw_inode, ei)) {
+		ext4_set_errno(inode->i_sb, EFSBADCRC);
 		ext4_error_inode(inode, function, line, 0,
 				 "iget: checksum invalid");
 		ret = -EFSBADCRC;
@@ -5090,6 +5095,7 @@ int ext4_write_inode(struct inode *inode, struct writeback_control *wbc)
 		if (wbc->sync_mode == WB_SYNC_ALL && !wbc->for_sync)
 			sync_dirty_buffer(iloc.bh);
 		if (buffer_req(iloc.bh) && !buffer_uptodate(iloc.bh)) {
+			ext4_set_errno(inode->i_sb, EIO);
 			EXT4_ERROR_INODE_BLOCK(inode, iloc.bh->b_blocknr,
 					 "IO error syncing inode");
 			err = -EIO;
fs/ext4/mballoc.c +4 -0
index a3e2767bdf2f..f64838187559 100644
@@ -3895,6 +3895,7 @@ ext4_mb_discard_group_preallocations(struct super_block *sb,
 	bitmap_bh = ext4_read_block_bitmap(sb, group);
 	if (IS_ERR(bitmap_bh)) {
 		err = PTR_ERR(bitmap_bh);
+		ext4_set_errno(sb, -err);
 		ext4_error(sb, "Error %d reading block bitmap for %u",
 			   err, group);
 		return 0;
@@ -4063,6 +4064,7 @@ void ext4_discard_preallocations(struct inode *inode)
 		err = ext4_mb_load_buddy_gfp(sb, group, &e4b,
 					     GFP_NOFS|__GFP_NOFAIL);
 		if (err) {
+			ext4_set_errno(sb, -err);
 			ext4_error(sb, "Error %d loading buddy information for %u",
 				   err, group);
 			continue;
@@ -4071,6 +4073,7 @@ void ext4_discard_preallocations(struct inode *inode)
 		bitmap_bh = ext4_read_block_bitmap(sb, group);
 		if (IS_ERR(bitmap_bh)) {
 			err = PTR_ERR(bitmap_bh);
+			ext4_set_errno(sb, -err);
 			ext4_error(sb, "Error %d reading block bitmap for %u",
 					err, group);
 			ext4_mb_unload_buddy(&e4b);
@@ -4325,6 +4328,7 @@ ext4_mb_discard_lg_preallocations(struct super_block *sb,
 		err = ext4_mb_load_buddy_gfp(sb, group, &e4b,
 					     GFP_NOFS|__GFP_NOFAIL);
 		if (err) {
+			ext4_set_errno(sb, -err);
 			ext4_error(sb, "Error %d loading buddy information for %u",
 				   err, group);
 			continue;
fs/ext4/mmp.c +5 -1
index 2305b4374fd3..1c44b1a32001 100644
@@ -173,8 +173,10 @@ static int kmmpd(void *data)
 		 * (s_mmp_update_interval * 60) seconds.
 		 */
 		if (retval) {
-			if ((failed_writes % 60) == 0)
+			if ((failed_writes % 60) == 0) {
+				ext4_set_errno(sb, -retval);
 				ext4_error(sb, "Error writing to MMP block");
+			}
 			failed_writes++;
 		}
 
@@ -205,6 +207,7 @@ static int kmmpd(void *data)
 
 			retval = read_mmp_block(sb, &bh_check, mmp_block);
 			if (retval) {
+				ext4_set_errno(sb, -retval);
 				ext4_error(sb, "error reading MMP data: %d",
 					   retval);
 				goto exit_thread;
@@ -218,6 +221,7 @@ static int kmmpd(void *data)
 					     "Error while updating MMP info. "
 					     "The filesystem seems to have been"
 					     " multiply mounted.");
+				ext4_set_errno(sb, EBUSY);
 				ext4_error(sb, "abort");
 				put_bh(bh_check);
 				retval = -EBUSY;
fs/ext4/namei.c +4 -0
index a856997d87b5..c74ed04d6781 100644
@@ -156,6 +156,7 @@ static struct buffer_head *__ext4_read_dirblock(struct inode *inode,
 		if (ext4_dx_csum_verify(inode, dirent))
 			set_buffer_verified(bh);
 		else {
+			ext4_set_errno(inode->i_sb, EFSBADCRC);
 			ext4_error_inode(inode, func, line, block,
 					 "Directory index failed checksum");
 			brelse(bh);
@@ -166,6 +167,7 @@ static struct buffer_head *__ext4_read_dirblock(struct inode *inode,
 		if (ext4_dirblock_csum_verify(inode, bh))
 			set_buffer_verified(bh);
 		else {
+			ext4_set_errno(inode->i_sb, EFSBADCRC);
 			ext4_error_inode(inode, func, line, block,
 					 "Directory block failed checksum");
 			brelse(bh);
@@ -1527,6 +1529,7 @@ static struct buffer_head *__ext4_find_entry(struct inode *dir,
 			goto next;
 		wait_on_buffer(bh);
 		if (!buffer_uptodate(bh)) {
+			ext4_set_errno(sb, EIO);
 			EXT4_ERROR_INODE(dir, "reading directory lblock %lu",
 					 (unsigned long) block);
 			brelse(bh);
@@ -1537,6 +1540,7 @@ static struct buffer_head *__ext4_find_entry(struct inode *dir,
 		    !is_dx_internal_node(dir, block,
 					 (struct ext4_dir_entry *)bh->b_data) &&
 		    !ext4_dirblock_csum_verify(dir, bh)) {
+			ext4_set_errno(sb, EFSBADCRC);
 			EXT4_ERROR_INODE(dir, "checksumming directory "
 					 "block %lu", (unsigned long)block);
 			brelse(bh);
fs/ext4/super.c +67 -1

99 lines not on this page · raw message

fs/ext4/xattr.c +3 -1
index 8966a5439a22..246fbeeb6366 100644
@@ -2879,9 +2879,11 @@ int ext4_xattr_delete_inode(handle_t *handle, struct inode *inode,
 		bh = ext4_sb_bread(inode->i_sb, EXT4_I(inode)->i_file_acl, REQ_PRIO);
 		if (IS_ERR(bh)) {
 			error = PTR_ERR(bh);
-			if (error == -EIO)
+			if (error == -EIO) {
+				ext4_set_errno(inode->i_sb, EIO);
 				EXT4_ERROR_INODE(inode, "block %llu read error",
 						 EXT4_I(inode)->i_file_acl);
+			}
 			bh = NULL;
 			goto cleanup;
 		}
Signature · 1 line
2.23.0
author maintainer
#1 lore raw

PATCH 2/2 ext4: simulate various I/O and checksum errors when reading metadata

This allows us to test various error handling code paths

Previous-Version-Link: https://lore.kernel.org/r/[email protected]

---

Notes:
v2: change ext4_simulate_fail() to return a bool instead of an int

 fs/ext4/balloc.c |  4 +++-
 fs/ext4/ext4.h   | 42 ++++++++++++++++++++++++++++++++++++++++++
 fs/ext4/ialloc.c |  4 +++-
 fs/ext4/inode.c  |  6 +++++-
 fs/ext4/namei.c  | 11 ++++++++---
 fs/ext4/sysfs.c  | 23 +++++++++++++++++++++++
 6 files changed, 84 insertions(+), 6 deletions(-)
fs/ext4/balloc.c +3 -1
index 102c38527a10..5f993a411251 100644
@@ -371,7 +371,8 @@ static int ext4_validate_block_bitmap(struct super_block *sb,
 	if (buffer_verified(bh))
 		goto verified;
 	if (unlikely(!ext4_block_bitmap_csum_verify(sb, block_group,
-			desc, bh))) {
+						    desc, bh) ||
+		     ext4_simulate_fail(sb, EXT4_SIM_BBITMAP_CRC))) {
 		ext4_unlock_group(sb, block_group);
 		ext4_error(sb, "bg %u: bad block bitmap checksum", block_group);
 		ext4_mark_group_bitmap_corrupted(sb, block_group,
@@ -505,6 +506,7 @@ int ext4_wait_block_bitmap(struct super_block *sb, ext4_group_t block_group,
 	if (!desc)
 		return -EFSCORRUPTED;
 	wait_on_buffer(bh);
+	ext4_simulate_fail_bh(sb, bh, EXT4_SIM_BBITMAP_EIO);
 	if (!buffer_uptodate(bh)) {
 		ext4_set_errno(sb, EIO);
 		ext4_error(sb, "Cannot read block bitmap - "
fs/ext4/ext4.h +42 -0

54 lines not on this page · raw message

fs/ext4/ialloc.c +3 -1
index 9979e1a55be5..8e5bb8a3f151 100644
@@ -94,7 +94,8 @@ static int ext4_validate_inode_bitmap(struct super_block *sb,
 		goto verified;
 	blk = ext4_inode_bitmap(sb, desc);
 	if (!ext4_inode_bitmap_csum_verify(sb, block_group, desc, bh,
-					   EXT4_INODES_PER_GROUP(sb) / 8)) {
+					   EXT4_INODES_PER_GROUP(sb) / 8) ||
+	    ext4_simulate_fail(sb, EXT4_SIM_IBITMAP_CRC)) {
 		ext4_unlock_group(sb, block_group);
 		ext4_error(sb, "Corrupt inode bitmap - block_group = %u, "
 			   "inode_bitmap = %llu", block_group, blk);
@@ -192,6 +193,7 @@ ext4_read_inode_bitmap(struct super_block *sb, ext4_group_t block_group)
 	get_bh(bh);
 	submit_bh(REQ_OP_READ, REQ_META | REQ_PRIO, bh);
 	wait_on_buffer(bh);
+	ext4_simulate_fail_bh(sb, bh, EXT4_SIM_IBITMAP_EIO);
 	if (!buffer_uptodate(bh)) {
 		put_bh(bh);
 		ext4_set_errno(sb, EIO);
fs/ext4/inode.c +5 -1
index 1aed7c653b42..128e51c37b93 100644
@@ -4243,6 +4243,8 @@ static int __ext4_get_inode_loc(struct inode *inode,
 	bh = sb_getblk(sb, block);
 	if (unlikely(!bh))
 		return -ENOMEM;
+	if (ext4_simulate_fail(sb, EXT4_SIM_INODE_EIO))
+		goto simulate_eio;
 	if (!buffer_uptodate(bh)) {
 		lock_buffer(bh);
 
@@ -4341,6 +4343,7 @@ static int __ext4_get_inode_loc(struct inode *inode,
 		blk_finish_plug(&plug);
 		wait_on_buffer(bh);
 		if (!buffer_uptodate(bh)) {
+		simulate_eio:
 			ext4_set_errno(inode->i_sb, EIO);
 			EXT4_ERROR_INODE_BLOCK(inode, block,
 					       "unable to read itable block");
@@ -4555,7 +4558,8 @@ struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
 					      sizeof(gen));
 	}
 
-	if (!ext4_inode_csum_verify(inode, raw_inode, ei)) {
+	if (!ext4_inode_csum_verify(inode, raw_inode, ei) ||
+	    ext4_simulate_fail(sb, EXT4_SIM_INODE_CRC)) {
 		ext4_set_errno(inode->i_sb, EFSBADCRC);
 		ext4_error_inode(inode, function, line, 0,
 				 "iget: checksum invalid");
fs/ext4/namei.c +8 -3
index c74ed04d6781..3a0f65e21e8f 100644
@@ -109,7 +109,10 @@ static struct buffer_head *__ext4_read_dirblock(struct inode *inode,
 	struct ext4_dir_entry *dirent;
 	int is_dx_block = 0;
 
-	bh = ext4_bread(NULL, inode, block, 0);
+	if (ext4_simulate_fail(inode->i_sb, EXT4_SIM_DIRBLOCK_EIO))
+		bh = ERR_PTR(-EIO);
+	else
+		bh = ext4_bread(NULL, inode, block, 0);
 	if (IS_ERR(bh)) {
 		__ext4_warning(inode->i_sb, func, line,
 			       "inode #%lu: lblock %lu: comm %s: "
@@ -153,7 +156,8 @@ static struct buffer_head *__ext4_read_dirblock(struct inode *inode,
 	 * caller is sure it should be an index block.
 	 */
 	if (is_dx_block && type == INDEX) {
-		if (ext4_dx_csum_verify(inode, dirent))
+		if (ext4_dx_csum_verify(inode, dirent) &&
+		    !ext4_simulate_fail(inode->i_sb, EXT4_SIM_DIRBLOCK_CRC))
 			set_buffer_verified(bh);
 		else {
 			ext4_set_errno(inode->i_sb, EFSBADCRC);
@@ -164,7 +168,8 @@ static struct buffer_head *__ext4_read_dirblock(struct inode *inode,
 		}
 	}
 	if (!is_dx_block) {
-		if (ext4_dirblock_csum_verify(inode, bh))
+		if (ext4_dirblock_csum_verify(inode, bh) &&
+		    !ext4_simulate_fail(inode->i_sb, EXT4_SIM_DIRBLOCK_CRC))
 			set_buffer_verified(bh);
 		else {
 			ext4_set_errno(inode->i_sb, EFSBADCRC);
fs/ext4/sysfs.c +23 -0
index eb1efad0e20a..a990d28d191b 100644
@@ -29,6 +29,7 @@ typedef enum {
 	attr_last_error_time,
 	attr_feature,
 	attr_pointer_ui,
+	attr_pointer_ul,
 	attr_pointer_atomic,
 	attr_journal_task,
 } attr_id_t;
@@ -160,6 +161,9 @@ static struct ext4_attr ext4_attr_##_name = {			\
 #define EXT4_RW_ATTR_SBI_UI(_name,_elname)	\
 	EXT4_ATTR_OFFSET(_name, 0644, pointer_ui, ext4_sb_info, _elname)
 
+#define EXT4_RW_ATTR_SBI_UL(_name,_elname)	\
+	EXT4_ATTR_OFFSET(_name, 0644, pointer_ul, ext4_sb_info, _elname)
+
 #define EXT4_ATTR_PTR(_name,_mode,_id,_ptr) \
 static struct ext4_attr ext4_attr_##_name = {			\
 	.attr = {.name = __stringify(_name), .mode = _mode },	\
@@ -194,6 +198,9 @@ EXT4_RW_ATTR_SBI_UI(warning_ratelimit_interval_ms, s_warning_ratelimit_state.int
 EXT4_RW_ATTR_SBI_UI(warning_ratelimit_burst, s_warning_ratelimit_state.burst);
 EXT4_RW_ATTR_SBI_UI(msg_ratelimit_interval_ms, s_msg_ratelimit_state.interval);
 EXT4_RW_ATTR_SBI_UI(msg_ratelimit_burst, s_msg_ratelimit_state.burst);
+#ifdef CONFIG_EXT4_DEBUG
+EXT4_RW_ATTR_SBI_UL(simulate_fail, s_simulate_fail);
+#endif
 EXT4_RO_ATTR_ES_UI(errors_count, s_error_count);
 EXT4_ATTR(first_error_time, 0444, first_error_time);
 EXT4_ATTR(last_error_time, 0444, last_error_time);
@@ -228,6 +235,9 @@ static struct attribute *ext4_attrs[] = {
 	ATTR_LIST(first_error_time),
 	ATTR_LIST(last_error_time),
 	ATTR_LIST(journal_task),
+#ifdef CONFIG_EXT4_DEBUG
+	ATTR_LIST(simulate_fail),
+#endif
 	NULL,
 };
 ATTRIBUTE_GROUPS(ext4);
@@ -318,6 +328,11 @@ static ssize_t ext4_attr_show(struct kobject *kobj,
 		else
 			return snprintf(buf, PAGE_SIZE, "%u\n",
 					*((unsigned int *) ptr));
+	case attr_pointer_ul:
+		if (!ptr)
+			return 0;
+		return snprintf(buf, PAGE_SIZE, "%lu\n",
+				*((unsigned long *) ptr));
 	case attr_pointer_atomic:
 		if (!ptr)
 			return 0;
@@ -361,6 +376,14 @@ static ssize_t ext4_attr_store(struct kobject *kobj,
 		else
 			*((unsigned int *) ptr) = t;
 		return len;
+	case attr_pointer_ul:
+		if (!ptr)
+			return 0;
+		ret = kstrtoul(skip_spaces(buf), 0, &t);
+		if (ret)
+			return ret;
+		*((unsigned long *) ptr) = t;
+		return len;
 	case attr_inode_readahead:
 		return inode_readahead_blks_store(sbi, buf, len);
 	case attr_trigger_test_error:
Signature · 1 line
2.23.0
#2 lore raw

Re: [PATCH -v2 2/2] ext4: simulate various I/O and checksum errors when reading metadata

Quoted 8 lines · Theodore Ts'o

This allows us to test various error handling code paths

Previous-Version-Link: https://lore.kernel.org/r/[email protected]

---

fs/ext4/ext4.h +1 -0
+/*

fs/ext4/ext4.h static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) · line 1618

Quoted 10 lines
fs/ext4/ext4.h +10 -0
+ * Simulate_fail flags
+ */
+#define EXT4_SIM_BBITMAP_EIO	0x00000001
+#define EXT4_SIM_BBITMAP_CRC	0x00000002
+#define EXT4_SIM_IBITMAP_EIO	0x00000004
+#define EXT4_SIM_IBITMAP_CRC	0x00000008
+#define EXT4_SIM_INODE_EIO	0x00000010
+#define EXT4_SIM_INODE_CRC	0x00000020
+#define EXT4_SIM_DIRBLOCK_EIO	0x00000040
+#define EXT4_SIM_DIRBLOCK_CRC	0x00000080

Do we really need to have the ability to inject several different failures at the same time? This will only allow 32 or 64 different failure hooks.

IMHO it would be much more flexible to have these as an enum of independent values. That does somewhat limit the ability to inject multiple failures, but allows for many more different/specific fault injection points to be added in the future. We have hundreds of different fault points in Lustre to simulate hard-to-hit race conditions and trigger specific error paths.

If this patch has already been landed, it would still be possible to change ext4_simulate_fail() to check "if (likely(old != flag))" rather than just the bit. That allows making this into an enum that just happens to have an non-consecutive sequence of values assigned rather than a bitmask.

Cheers, Andreas

author maintainer
#3 lore raw

PATCH 1/1 ext4: simulate various I/O and checksum errors when reading metadata

This allows us to test various error handling code paths

Previous-Version-Link: https://lore.kernel.org/r/[email protected]

---

Andreas, thanks for the sugestion! I've made this change in the v3 patch.

Notes:
v2: change ext4_simulate_fail() to return a bool instead of an int
v3: use integer codes instead of bitflags to indicate which failure to simulate

 fs/ext4/balloc.c |  4 +++-
 fs/ext4/ext4.h   | 38 ++++++++++++++++++++++++++++++++++++++
 fs/ext4/ialloc.c |  4 +++-
 fs/ext4/inode.c  |  6 +++++-
 fs/ext4/namei.c  | 11 ++++++++---
 fs/ext4/sysfs.c  | 23 +++++++++++++++++++++++
 6 files changed, 80 insertions(+), 6 deletions(-)
fs/ext4/balloc.c +3 -1
index 102c38527a10..5f993a411251 100644
@@ -371,7 +371,8 @@ static int ext4_validate_block_bitmap(struct super_block *sb,
 	if (buffer_verified(bh))
 		goto verified;
 	if (unlikely(!ext4_block_bitmap_csum_verify(sb, block_group,
-			desc, bh))) {
+						    desc, bh) ||
+		     ext4_simulate_fail(sb, EXT4_SIM_BBITMAP_CRC))) {
 		ext4_unlock_group(sb, block_group);
 		ext4_error(sb, "bg %u: bad block bitmap checksum", block_group);
 		ext4_mark_group_bitmap_corrupted(sb, block_group,
@@ -505,6 +506,7 @@ int ext4_wait_block_bitmap(struct super_block *sb, ext4_group_t block_group,
 	if (!desc)
 		return -EFSCORRUPTED;
 	wait_on_buffer(bh);
+	ext4_simulate_fail_bh(sb, bh, EXT4_SIM_BBITMAP_EIO);
 	if (!buffer_uptodate(bh)) {
 		ext4_set_errno(sb, EIO);
 		ext4_error(sb, "Cannot read block bitmap - "
fs/ext4/ext4.h +38 -0
index 1c9ac0fc8715..689c8efc7dd1 100644
@@ -1557,6 +1557,9 @@ struct ext4_sb_info {
 	/* Barrier between changing inodes' journal flags and writepages ops. */
 	struct percpu_rw_semaphore s_journal_flag_rwsem;
 	struct dax_device *s_daxdev;
+#ifdef CONFIG_EXT4_DEBUG
+	unsigned long s_simulate_fail;
+#endif
 };
 
 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
@@ -1575,6 +1578,41 @@ static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
 }
 
+/*
+ * Simulate_fail codes
+ */
+#define EXT4_SIM_BBITMAP_EIO	1
+#define EXT4_SIM_BBITMAP_CRC	2
+#define EXT4_SIM_IBITMAP_EIO	3
+#define EXT4_SIM_IBITMAP_CRC	4
+#define EXT4_SIM_INODE_EIO	5
+#define EXT4_SIM_INODE_CRC	6
+#define EXT4_SIM_DIRBLOCK_EIO	7
+#define EXT4_SIM_DIRBLOCK_CRC	8
+
+static inline bool ext4_simulate_fail(struct super_block *sb,
+				     unsigned long code)
+{
+#ifdef CONFIG_EXT4_DEBUG
+	struct ext4_sb_info *sbi = EXT4_SB(sb);
+
+	if (sbi->s_simulate_fail != code)
+		return false;
+	sbi->s_simulate_fail = 0;
+	return true;
+#else
+	return false;
+#endif
+}
+
+static inline void ext4_simulate_fail_bh(struct super_block *sb,
+					 struct buffer_head *bh,
+					 unsigned long code)
+{
+	if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
+		clear_buffer_uptodate(bh);
+}
+
 /*
  * Error number codes for s_{first,last}_error_errno
  *
fs/ext4/ialloc.c +3 -1
index 9979e1a55be5..8e5bb8a3f151 100644
@@ -94,7 +94,8 @@ static int ext4_validate_inode_bitmap(struct super_block *sb,
 		goto verified;
 	blk = ext4_inode_bitmap(sb, desc);
 	if (!ext4_inode_bitmap_csum_verify(sb, block_group, desc, bh,
-					   EXT4_INODES_PER_GROUP(sb) / 8)) {
+					   EXT4_INODES_PER_GROUP(sb) / 8) ||
+	    ext4_simulate_fail(sb, EXT4_SIM_IBITMAP_CRC)) {
 		ext4_unlock_group(sb, block_group);
 		ext4_error(sb, "Corrupt inode bitmap - block_group = %u, "
 			   "inode_bitmap = %llu", block_group, blk);
@@ -192,6 +193,7 @@ ext4_read_inode_bitmap(struct super_block *sb, ext4_group_t block_group)
 	get_bh(bh);
 	submit_bh(REQ_OP_READ, REQ_META | REQ_PRIO, bh);
 	wait_on_buffer(bh);
+	ext4_simulate_fail_bh(sb, bh, EXT4_SIM_IBITMAP_EIO);
 	if (!buffer_uptodate(bh)) {
 		put_bh(bh);
 		ext4_set_errno(sb, EIO);
fs/ext4/inode.c +5 -1
index 1aed7c653b42..128e51c37b93 100644
@@ -4243,6 +4243,8 @@ static int __ext4_get_inode_loc(struct inode *inode,
 	bh = sb_getblk(sb, block);
 	if (unlikely(!bh))
 		return -ENOMEM;
+	if (ext4_simulate_fail(sb, EXT4_SIM_INODE_EIO))
+		goto simulate_eio;
 	if (!buffer_uptodate(bh)) {
 		lock_buffer(bh);
 
@@ -4341,6 +4343,7 @@ static int __ext4_get_inode_loc(struct inode *inode,
 		blk_finish_plug(&plug);
 		wait_on_buffer(bh);
 		if (!buffer_uptodate(bh)) {
+		simulate_eio:
 			ext4_set_errno(inode->i_sb, EIO);
 			EXT4_ERROR_INODE_BLOCK(inode, block,
 					       "unable to read itable block");
@@ -4555,7 +4558,8 @@ struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
 					      sizeof(gen));
 	}
 
-	if (!ext4_inode_csum_verify(inode, raw_inode, ei)) {
+	if (!ext4_inode_csum_verify(inode, raw_inode, ei) ||
+	    ext4_simulate_fail(sb, EXT4_SIM_INODE_CRC)) {
 		ext4_set_errno(inode->i_sb, EFSBADCRC);
 		ext4_error_inode(inode, function, line, 0,
 				 "iget: checksum invalid");
fs/ext4/namei.c +8 -3
index c74ed04d6781..3a0f65e21e8f 100644
@@ -109,7 +109,10 @@ static struct buffer_head *__ext4_read_dirblock(struct inode *inode,
 	struct ext4_dir_entry *dirent;
 	int is_dx_block = 0;
 
-	bh = ext4_bread(NULL, inode, block, 0);
+	if (ext4_simulate_fail(inode->i_sb, EXT4_SIM_DIRBLOCK_EIO))
+		bh = ERR_PTR(-EIO);
+	else
+		bh = ext4_bread(NULL, inode, block, 0);
 	if (IS_ERR(bh)) {
 		__ext4_warning(inode->i_sb, func, line,
 			       "inode #%lu: lblock %lu: comm %s: "
@@ -153,7 +156,8 @@ static struct buffer_head *__ext4_read_dirblock(struct inode *inode,
 	 * caller is sure it should be an index block.
 	 */
 	if (is_dx_block && type == INDEX) {
-		if (ext4_dx_csum_verify(inode, dirent))
+		if (ext4_dx_csum_verify(inode, dirent) &&
+		    !ext4_simulate_fail(inode->i_sb, EXT4_SIM_DIRBLOCK_CRC))
 			set_buffer_verified(bh);
 		else {
 			ext4_set_errno(inode->i_sb, EFSBADCRC);
@@ -164,7 +168,8 @@ static struct buffer_head *__ext4_read_dirblock(struct inode *inode,
 		}
 	}
 	if (!is_dx_block) {
-		if (ext4_dirblock_csum_verify(inode, bh))
+		if (ext4_dirblock_csum_verify(inode, bh) &&
+		    !ext4_simulate_fail(inode->i_sb, EXT4_SIM_DIRBLOCK_CRC))
 			set_buffer_verified(bh);
 		else {
 			ext4_set_errno(inode->i_sb, EFSBADCRC);
fs/ext4/sysfs.c +23 -0
index eb1efad0e20a..a990d28d191b 100644
@@ -29,6 +29,7 @@ typedef enum {
 	attr_last_error_time,
 	attr_feature,
 	attr_pointer_ui,
+	attr_pointer_ul,
 	attr_pointer_atomic,
 	attr_journal_task,
 } attr_id_t;
@@ -160,6 +161,9 @@ static struct ext4_attr ext4_attr_##_name = {			\
 #define EXT4_RW_ATTR_SBI_UI(_name,_elname)	\
 	EXT4_ATTR_OFFSET(_name, 0644, pointer_ui, ext4_sb_info, _elname)
 
+#define EXT4_RW_ATTR_SBI_UL(_name,_elname)	\
+	EXT4_ATTR_OFFSET(_name, 0644, pointer_ul, ext4_sb_info, _elname)
+
 #define EXT4_ATTR_PTR(_name,_mode,_id,_ptr) \
 static struct ext4_attr ext4_attr_##_name = {			\
 	.attr = {.name = __stringify(_name), .mode = _mode },	\
@@ -194,6 +198,9 @@ EXT4_RW_ATTR_SBI_UI(warning_ratelimit_interval_ms, s_warning_ratelimit_state.int
 EXT4_RW_ATTR_SBI_UI(warning_ratelimit_burst, s_warning_ratelimit_state.burst);
 EXT4_RW_ATTR_SBI_UI(msg_ratelimit_interval_ms, s_msg_ratelimit_state.interval);
 EXT4_RW_ATTR_SBI_UI(msg_ratelimit_burst, s_msg_ratelimit_state.burst);
+#ifdef CONFIG_EXT4_DEBUG
+EXT4_RW_ATTR_SBI_UL(simulate_fail, s_simulate_fail);
+#endif
 EXT4_RO_ATTR_ES_UI(errors_count, s_error_count);
 EXT4_ATTR(first_error_time, 0444, first_error_time);
 EXT4_ATTR(last_error_time, 0444, last_error_time);
@@ -228,6 +235,9 @@ static struct attribute *ext4_attrs[] = {
 	ATTR_LIST(first_error_time),
 	ATTR_LIST(last_error_time),
 	ATTR_LIST(journal_task),
+#ifdef CONFIG_EXT4_DEBUG
+	ATTR_LIST(simulate_fail),
+#endif
 	NULL,
 };
 ATTRIBUTE_GROUPS(ext4);
@@ -318,6 +328,11 @@ static ssize_t ext4_attr_show(struct kobject *kobj,
 		else
 			return snprintf(buf, PAGE_SIZE, "%u\n",
 					*((unsigned int *) ptr));
+	case attr_pointer_ul:
+		if (!ptr)
+			return 0;
+		return snprintf(buf, PAGE_SIZE, "%lu\n",
+				*((unsigned long *) ptr));
 	case attr_pointer_atomic:
 		if (!ptr)
 			return 0;
@@ -361,6 +376,14 @@ static ssize_t ext4_attr_store(struct kobject *kobj,
 		else
 			*((unsigned int *) ptr) = t;
 		return len;
+	case attr_pointer_ul:
+		if (!ptr)
+			return 0;
+		ret = kstrtoul(skip_spaces(buf), 0, &t);
+		if (ret)
+			return ret;
+		*((unsigned long *) ptr) = t;
+		return len;
 	case attr_inode_readahead:
 		return inode_readahead_blks_store(sbi, buf, len);
 	case attr_trigger_test_error:
Signature · 1 line
2.23.0
#4 lore raw

Re: [PATCH -v3] ext4: simulate various I/O and checksum errors when reading metadata

Quoted 7 lines · Theodore Ts'o

This allows us to test various error handling code paths

Previous-Version-Link: https://lore.kernel.org/r/[email protected]

---

fs/ext4/ext4.h static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) · line 1599

Quoted 8 lines
fs/ext4/ext4.h +8 -0
+
+static inline bool ext4_simulate_fail(struct super_block *sb,
+				     unsigned long code)
+{
+#ifdef CONFIG_EXT4_DEBUG
+	struct ext4_sb_info *sbi = EXT4_SB(sb);
+
+	if (sbi->s_simulate_fail != code)

This should be marked unlikely(), as it is definitely one of the places that is legitimately rarely true. Sorry for not pointing this out on the previous version of the patch.

Cheers, Andreas

author maintainer
#5 lore raw

Re: [PATCH -v3] ext4: simulate various I/O and checksum errors when reading metadata

Quoted 4 lines · Andreas Dilger

This should be marked unlikely(), as it is definitely one of the places that is legitimately rarely true. Sorry for not pointing this out on the previous version of the patch.

Good point; thanks, I've made that change.

- Ted

reviewer
#6 lore raw
In reply to Theodore Ts'o
Quoted 8 lines · Theodore Ts'o

This allows the cause of an ext4_error() report to be categorized based on whether it was triggered due to an I/O error, or an memory allocation error, or other possible causes. Most errors are caused by a detected file system inconsistency, so the default code stored in the superblock will be EXT4_ERR_EFSCORRUPTED.

Previous-Version-Link: https://lore.kernel.org/r/[email protected]

Somewhat late to the party but: Seeing that you basically call ext4_set_errno() before every ext4_error() call (or its variants), won't it be more concise and less error proned to add errno as an argument to ext4_error() and handle setting from there? Or are there times where you don't want error set?

Honza

Quoted 448 lines

---

Notes:
v2: no changes from previous version

 fs/ext4/balloc.c    |  1 +
 fs/ext4/ext4.h      | 30 +++++++++++++++++++-
 fs/ext4/ext4_jbd2.c |  3 ++
 fs/ext4/extents.c   |  1 +
 fs/ext4/ialloc.c    |  2 ++
 fs/ext4/inline.c    |  2 ++
 fs/ext4/inode.c     |  8 +++++-
 fs/ext4/mballoc.c   |  4 +++
 fs/ext4/mmp.c       |  6 +++-
 fs/ext4/namei.c     |  4 +++
 fs/ext4/super.c     | 68 ++++++++++++++++++++++++++++++++++++++++++++-
 fs/ext4/xattr.c     |  4 ++-
 12 files changed, 128 insertions(+), 5 deletions(-)
fs/ext4/balloc.c +1 -0
index 0b202e00d93f..102c38527a10 100644
@@ -506,6 +506,7 @@ int ext4_wait_block_bitmap(struct super_block *sb, ext4_group_t block_group,
 		return -EFSCORRUPTED;
 	wait_on_buffer(bh);
 	if (!buffer_uptodate(bh)) {
+		ext4_set_errno(sb, EIO);
 		ext4_error(sb, "Cannot read block bitmap - "
 			   "block_group = %u, block_bitmap = %llu",
 			   block_group, (unsigned long long) bh->b_blocknr);
fs/ext4/ext4.h +29 -1
index 61987c106511..1c9ac0fc8715 100644
@@ -1343,7 +1343,8 @@ struct ext4_super_block {
 	__u8	s_lastcheck_hi;
 	__u8	s_first_error_time_hi;
 	__u8	s_last_error_time_hi;
-	__u8	s_pad[2];
+	__u8	s_first_error_errcode;
+	__u8    s_last_error_errcode;
 	__le16  s_encoding;		/* Filename charset encoding */
 	__le16  s_encoding_flags;	/* Filename charset encoding flags */
 	__le32	s_reserved[95];		/* Padding to the end of the block */
@@ -1574,6 +1575,32 @@ static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
 }
 
+/*
+ * Error number codes for s_{first,last}_error_errno
+ *
+ * Linux errno numbers are architecture specific, so we need to translate
+ * them into something which is architecture independent.   We don't define
+ * codes for all errno's; just the ones which are most likely to be the cause
+ * of an ext4_error() call.
+ */
+#define EXT4_ERR_UNKNOWN	 1
+#define EXT4_ERR_EIO		 2
+#define EXT4_ERR_ENOMEM		 3
+#define EXT4_ERR_EFSBADCRC	 4
+#define EXT4_ERR_EFSCORRUPTED	 5
+#define EXT4_ERR_ENOSPC		 6
+#define EXT4_ERR_ENOKEY		 7
+#define EXT4_ERR_EROFS		 8
+#define EXT4_ERR_EFBIG		 9
+#define EXT4_ERR_EEXIST		10
+#define EXT4_ERR_ERANGE		11
+#define EXT4_ERR_EOVERFLOW	12
+#define EXT4_ERR_EBUSY		13
+#define EXT4_ERR_ENOTDIR	14
+#define EXT4_ERR_ENOTEMPTY	15
+#define EXT4_ERR_ESHUTDOWN	16
+#define EXT4_ERR_EFAULT		17
+
 /*
  * Inode dynamic state flags
  */
@@ -2686,6 +2713,7 @@ extern const char *ext4_decode_error(struct super_block *sb, int errno,
 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
 					     ext4_group_t block_group,
 					     unsigned int flags);
+extern void ext4_set_errno(struct super_block *sb, int err);
 
 extern __printf(4, 5)
 void __ext4_error(struct super_block *, const char *, unsigned int,
fs/ext4/ext4_jbd2.c +3 -0
index d3b8cdea5df7..19217a3f1ae4 100644
@@ -58,6 +58,7 @@ static int ext4_journal_check_start(struct super_block *sb)
 	 * take the FS itself readonly cleanly.
 	 */
 	if (journal && is_journal_aborted(journal)) {
+		ext4_set_errno(sb, -journal->j_errno);
 		ext4_abort(sb, "Detected aborted journal");
 		return -EROFS;
 	}
@@ -249,6 +250,7 @@ int __ext4_forget(const char *where, unsigned int line, handle_t *handle,
 	if (err) {
 		ext4_journal_abort_handle(where, line, __func__,
 					  bh, handle, err);
+		ext4_set_errno(inode->i_sb, -err);
 		__ext4_abort(inode->i_sb, where, line,
 			   "error %d when attempting revoke", err);
 	}
@@ -320,6 +322,7 @@ int __ext4_handle_dirty_metadata(const char *where, unsigned int line,
 				es = EXT4_SB(inode->i_sb)->s_es;
 				es->s_last_error_block =
 					cpu_to_le64(bh->b_blocknr);
+				ext4_set_errno(inode->i_sb, EIO);
 				ext4_error_inode(inode, where, line,
 						 bh->b_blocknr,
 					"IO error syncing itable block");
fs/ext4/extents.c +1 -0
index 0e8708b77da6..ee83fe7c98aa 100644
@@ -492,6 +492,7 @@ static int __ext4_ext_check(const char *function, unsigned int line,
 	return 0;
 
 corrupted:
+	ext4_set_errno(inode->i_sb, -err);
 	ext4_error_inode(inode, function, line, 0,
 			 "pblk %llu bad header/extent: %s - magic %x, "
 			 "entries %u, max %u(%u), depth %u(%u)",
fs/ext4/ialloc.c +2 -0
index dc333e8e51e8..9979e1a55be5 100644
@@ -194,6 +194,7 @@ ext4_read_inode_bitmap(struct super_block *sb, ext4_group_t block_group)
 	wait_on_buffer(bh);
 	if (!buffer_uptodate(bh)) {
 		put_bh(bh);
+		ext4_set_errno(sb, EIO);
 		ext4_error(sb, "Cannot read inode bitmap - "
 			   "block_group = %u, inode_bitmap = %llu",
 			   block_group, bitmap_blk);
@@ -1223,6 +1224,7 @@ struct inode *ext4_orphan_get(struct super_block *sb, unsigned long ino)
 	inode = ext4_iget(sb, ino, EXT4_IGET_NORMAL);
 	if (IS_ERR(inode)) {
 		err = PTR_ERR(inode);
+		ext4_set_errno(sb, -err);
 		ext4_error(sb, "couldn't read orphan inode %lu (err %d)",
 			   ino, err);
 		return inode;
fs/ext4/inline.c +2 -0
index 2fec62d764fa..e61603f47035 100644
@@ -98,6 +98,7 @@ int ext4_get_max_inline_size(struct inode *inode)
 
 	error = ext4_get_inode_loc(inode, &iloc);
 	if (error) {
+		ext4_set_errno(inode->i_sb, -error);
 		ext4_error_inode(inode, __func__, __LINE__, 0,
 				 "can't get inode location %lu",
 				 inode->i_ino);
@@ -1761,6 +1762,7 @@ bool empty_inline_dir(struct inode *dir, int *has_inline_data)
 
 	err = ext4_get_inode_loc(dir, &iloc);
 	if (err) {
+		ext4_set_errno(dir->i_sb, -err);
 		EXT4_ERROR_INODE(dir, "error %d getting inode %lu block",
 				 err, dir->i_ino);
 		return true;
fs/ext4/inode.c +7 -1
index 310e4abd9aca..1aed7c653b42 100644
@@ -271,6 +271,7 @@ void ext4_evict_inode(struct inode *inode)
 	if (inode->i_blocks) {
 		err = ext4_truncate(inode);
 		if (err) {
+			ext4_set_errno(inode->i_sb, -err);
 			ext4_error(inode->i_sb,
 				   "couldn't truncate inode %lu (err %d)",
 				   inode->i_ino, err);
@@ -2478,10 +2479,12 @@ static int mpage_map_and_submit_extent(handle_t *handle,
 			EXT4_I(inode)->i_disksize = disksize;
 		up_write(&EXT4_I(inode)->i_data_sem);
 		err2 = ext4_mark_inode_dirty(handle, inode);
-		if (err2)
+		if (err2) {
+			ext4_set_errno(inode->i_sb, -err2);
 			ext4_error(inode->i_sb,
 				   "Failed to mark inode %lu dirty",
 				   inode->i_ino);
+		}
 		if (!err)
 			err = err2;
 	}
@@ -4338,6 +4341,7 @@ static int __ext4_get_inode_loc(struct inode *inode,
 		blk_finish_plug(&plug);
 		wait_on_buffer(bh);
 		if (!buffer_uptodate(bh)) {
+			ext4_set_errno(inode->i_sb, EIO);
 			EXT4_ERROR_INODE_BLOCK(inode, block,
 					       "unable to read itable block");
 			brelse(bh);
@@ -4552,6 +4556,7 @@ struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
 	}
 
 	if (!ext4_inode_csum_verify(inode, raw_inode, ei)) {
+		ext4_set_errno(inode->i_sb, EFSBADCRC);
 		ext4_error_inode(inode, function, line, 0,
 				 "iget: checksum invalid");
 		ret = -EFSBADCRC;
@@ -5090,6 +5095,7 @@ int ext4_write_inode(struct inode *inode, struct writeback_control *wbc)
 		if (wbc->sync_mode == WB_SYNC_ALL && !wbc->for_sync)
 			sync_dirty_buffer(iloc.bh);
 		if (buffer_req(iloc.bh) && !buffer_uptodate(iloc.bh)) {
+			ext4_set_errno(inode->i_sb, EIO);
 			EXT4_ERROR_INODE_BLOCK(inode, iloc.bh->b_blocknr,
 					 "IO error syncing inode");
 			err = -EIO;
fs/ext4/mballoc.c +4 -0
index a3e2767bdf2f..f64838187559 100644
@@ -3895,6 +3895,7 @@ ext4_mb_discard_group_preallocations(struct super_block *sb,
 	bitmap_bh = ext4_read_block_bitmap(sb, group);
 	if (IS_ERR(bitmap_bh)) {
 		err = PTR_ERR(bitmap_bh);
+		ext4_set_errno(sb, -err);
 		ext4_error(sb, "Error %d reading block bitmap for %u",
 			   err, group);
 		return 0;
@@ -4063,6 +4064,7 @@ void ext4_discard_preallocations(struct inode *inode)
 		err = ext4_mb_load_buddy_gfp(sb, group, &e4b,
 					     GFP_NOFS|__GFP_NOFAIL);
 		if (err) {
+			ext4_set_errno(sb, -err);
 			ext4_error(sb, "Error %d loading buddy information for %u",
 				   err, group);
 			continue;
@@ -4071,6 +4073,7 @@ void ext4_discard_preallocations(struct inode *inode)
 		bitmap_bh = ext4_read_block_bitmap(sb, group);
 		if (IS_ERR(bitmap_bh)) {
 			err = PTR_ERR(bitmap_bh);
+			ext4_set_errno(sb, -err);
 			ext4_error(sb, "Error %d reading block bitmap for %u",
 					err, group);
 			ext4_mb_unload_buddy(&e4b);
@@ -4325,6 +4328,7 @@ ext4_mb_discard_lg_preallocations(struct super_block *sb,
 		err = ext4_mb_load_buddy_gfp(sb, group, &e4b,
 					     GFP_NOFS|__GFP_NOFAIL);
 		if (err) {
+			ext4_set_errno(sb, -err);
 			ext4_error(sb, "Error %d loading buddy information for %u",
 				   err, group);
 			continue;
fs/ext4/mmp.c +5 -1
index 2305b4374fd3..1c44b1a32001 100644
@@ -173,8 +173,10 @@ static int kmmpd(void *data)
 		 * (s_mmp_update_interval * 60) seconds.
 		 */
 		if (retval) {
-			if ((failed_writes % 60) == 0)
+			if ((failed_writes % 60) == 0) {
+				ext4_set_errno(sb, -retval);
 				ext4_error(sb, "Error writing to MMP block");
+			}
 			failed_writes++;
 		}
 
@@ -205,6 +207,7 @@ static int kmmpd(void *data)
 
 			retval = read_mmp_block(sb, &bh_check, mmp_block);
 			if (retval) {
+				ext4_set_errno(sb, -retval);
 				ext4_error(sb, "error reading MMP data: %d",
 					   retval);
 				goto exit_thread;
@@ -218,6 +221,7 @@ static int kmmpd(void *data)
 					     "Error while updating MMP info. "
 					     "The filesystem seems to have been"
 					     " multiply mounted.");
+				ext4_set_errno(sb, EBUSY);
 				ext4_error(sb, "abort");
 				put_bh(bh_check);
 				retval = -EBUSY;
fs/ext4/namei.c +4 -0
index a856997d87b5..c74ed04d6781 100644
@@ -156,6 +156,7 @@ static struct buffer_head *__ext4_read_dirblock(struct inode *inode,
 		if (ext4_dx_csum_verify(inode, dirent))
 			set_buffer_verified(bh);
 		else {
+			ext4_set_errno(inode->i_sb, EFSBADCRC);
 			ext4_error_inode(inode, func, line, block,
 					 "Directory index failed checksum");
 			brelse(bh);
@@ -166,6 +167,7 @@ static struct buffer_head *__ext4_read_dirblock(struct inode *inode,
 		if (ext4_dirblock_csum_verify(inode, bh))
 			set_buffer_verified(bh);
 		else {
+			ext4_set_errno(inode->i_sb, EFSBADCRC);
 			ext4_error_inode(inode, func, line, block,
 					 "Directory block failed checksum");
 			brelse(bh);
@@ -1527,6 +1529,7 @@ static struct buffer_head *__ext4_find_entry(struct inode *dir,
 			goto next;
 		wait_on_buffer(bh);
 		if (!buffer_uptodate(bh)) {
+			ext4_set_errno(sb, EIO);
 			EXT4_ERROR_INODE(dir, "reading directory lblock %lu",
 					 (unsigned long) block);
 			brelse(bh);
@@ -1537,6 +1540,7 @@ static struct buffer_head *__ext4_find_entry(struct inode *dir,
 		    !is_dx_internal_node(dir, block,
 					 (struct ext4_dir_entry *)bh->b_data) &&
 		    !ext4_dirblock_csum_verify(dir, bh)) {
+			ext4_set_errno(sb, EFSBADCRC);
 			EXT4_ERROR_INODE(dir, "checksumming directory "
 					 "block %lu", (unsigned long)block);
 			brelse(bh);
fs/ext4/super.c +67 -1
index b205112ca051..c1b143cc1ab1 100644
@@ -367,6 +367,8 @@ static void __save_error_info(struct super_block *sb, const char *func,
 	ext4_update_tstamp(es, s_last_error_time);
 	strncpy(es->s_last_error_func, func, sizeof(es->s_last_error_func));
 	es->s_last_error_line = cpu_to_le32(line);
+	if (es->s_last_error_errcode == 0)
+		es->s_last_error_errcode = EXT4_ERR_EFSCORRUPTED;
 	if (!es->s_first_error_time) {
 		es->s_first_error_time = es->s_last_error_time;
 		es->s_first_error_time_hi = es->s_last_error_time_hi;
@@ -375,6 +377,7 @@ static void __save_error_info(struct super_block *sb, const char *func,
 		es->s_first_error_line = cpu_to_le32(line);
 		es->s_first_error_ino = es->s_last_error_ino;
 		es->s_first_error_block = es->s_last_error_block;
+		es->s_first_error_errcode = es->s_last_error_errcode;
 	}
 	/*
 	 * Start the daily error reporting function if it hasn't been
@@ -631,6 +634,66 @@ const char *ext4_decode_error(struct super_block *sb, int errno,
 	return errstr;
 }
 
+void ext4_set_errno(struct super_block *sb, int err)
+{
+	if (err < 0)
+		err = -err;
+
+	switch (err) {
+	case EIO:
+		err = EXT4_ERR_EIO;
+		break;
+	case ENOMEM:
+		err = EXT4_ERR_ENOMEM;
+		break;
+	case EFSBADCRC:
+		err = EXT4_ERR_EFSBADCRC;
+		break;
+	case EFSCORRUPTED:
+		err = EXT4_ERR_EFSCORRUPTED;
+		break;
+	case ENOSPC:
+		err = EXT4_ERR_ENOSPC;
+		break;
+	case ENOKEY:
+		err = EXT4_ERR_ENOKEY;
+		break;
+	case EROFS:
+		err = EXT4_ERR_EROFS;
+		break;
+	case EFBIG:
+		err = EXT4_ERR_EFBIG;
+		break;
+	case EEXIST:
+		err = EXT4_ERR_EEXIST;
+		break;
+	case ERANGE:
+		err = EXT4_ERR_ERANGE;
+		break;
+	case EOVERFLOW:
+		err = EXT4_ERR_EOVERFLOW;
+		break;
+	case EBUSY:
+		err = EXT4_ERR_EBUSY;
+		break;
+	case ENOTDIR:
+		err = EXT4_ERR_ENOTDIR;
+		break;
+	case ENOTEMPTY:
+		err = EXT4_ERR_ENOTEMPTY;
+		break;
+	case ESHUTDOWN:
+		err = EXT4_ERR_ESHUTDOWN;
+		break;
+	case EFAULT:
+		err = EXT4_ERR_EFAULT;
+		break;
+	default:
+		err = EXT4_ERR_UNKNOWN;
+	}
+	EXT4_SB(sb)->s_es->s_last_error_errcode = err;
+}
+
 /* __ext4_std_error decodes expected errors from journaling functions
  * automatically and invokes the appropriate error response.  */
 
@@ -655,6 +718,7 @@ void __ext4_std_error(struct super_block *sb, const char *function,
 		       sb->s_id, function, line, errstr);
 	}
 
+	ext4_set_errno(sb, -errno);
 	save_error_info(sb, function, line);
 	ext4_handle_error(sb);
 }
@@ -982,8 +1046,10 @@ static void ext4_put_super(struct super_block *sb)
 		aborted = is_journal_aborted(sbi->s_journal);
 		err = jbd2_journal_destroy(sbi->s_journal);
 		sbi->s_journal = NULL;
-		if ((err < 0) && !aborted)
+		if ((err < 0) && !aborted) {
+			ext4_set_errno(sb, -err);
 			ext4_abort(sb, "Couldn't clean up the journal");
+		}
 	}
 
 	ext4_unregister_sysfs(sb);
fs/ext4/xattr.c +3 -1
index 8966a5439a22..246fbeeb6366 100644
@@ -2879,9 +2879,11 @@ int ext4_xattr_delete_inode(handle_t *handle, struct inode *inode,
 		bh = ext4_sb_bread(inode->i_sb, EXT4_I(inode)->i_file_acl, REQ_PRIO);
 		if (IS_ERR(bh)) {
 			error = PTR_ERR(bh);
-			if (error == -EIO)
+			if (error == -EIO) {
+				ext4_set_errno(inode->i_sb, EIO);
 				EXT4_ERROR_INODE(inode, "block %llu read error",
 						 EXT4_I(inode)->i_file_acl);
+			}
 			bh = NULL;
 			goto cleanup;
 		}
Signature · 1 line
2.23.0
Signature · 2 lines
Jan Kara <[email protected]>
SUSE Labs, CR
author maintainer
#7 lore raw
Quoted 7 lines · Jan Kara
Quoted 6 lines · Theodore Ts'o

This allows the cause of an ext4_error() report to be categorized based on whether it was triggered due to an I/O error, or an memory allocation error, or other possible causes. Most errors are caused by a detected file system inconsistency, so the default code stored in the superblock will be EXT4_ERR_EFSCORRUPTED.

Quoted 8 lines
Quoted 2 lines

Previous-Version-Link: https://lore.kernel.org/r/[email protected]

Somewhat late to the party but: Seeing that you basically call ext4_set_errno() before every ext4_error() call (or its variants), won't it be more concise and less error proned to add errno as an argument to ext4_error() and handle setting from there? Or are there times where you don't want error set?

There are about 100 calls to ext4_error or its variants in fs/ext4. This patch inserts ext4_set_errno() before roughly 20 of them. Most of the time, we call ext4_error() because we've found a file system inconsistency. So the default if ext4_set_errno() hasn't been called is EXT4_ERR_EFSCORRUPTED.

I thought about adding errno as an argument to ext4_error(), but it would have substantially increased the size of the patch.

Cheers,

- Ted