Message ID | 20180330014429.27824-1-jaegeuk@kernel.org |
---|---|
State | Superseded, archived |
Headers | show |
Series | ext4: handle errors on ext4_commit_super | expand |
On Thu 29-03-18 18:44:29, Jaegeuk Kim wrote: > From: Jaegeuk Kim <jaegeuk@google.com> > > When remounting ext4 from ro to rw, currently it allows its transition, > even if ext4_commit_super() returns EIO. Even worse thing is, after that, > fs/buffer complains buffer dirty bits like: > > Call trace: > [<ffffff9750c259dc>] mark_buffer_dirty+0x184/0x1a4 > [<ffffff9750cb398c>] __ext4_handle_dirty_super+0x4c/0xfc > [<ffffff9750c7a9fc>] ext4_file_open+0x154/0x1c0 > [<ffffff9750bea51c>] do_dentry_open+0x114/0x2d0 > [<ffffff9750bea75c>] vfs_open+0x5c/0x94 > [<ffffff9750bf879c>] path_openat+0x668/0xfe8 > [<ffffff9750bf8088>] do_filp_open+0x74/0x120 > [<ffffff9750beac98>] do_sys_open+0x148/0x254 > [<ffffff9750beade0>] SyS_openat+0x10/0x18 > [<ffffff9750a83ab0>] el0_svc_naked+0x24/0x28 > EXT4-fs (dm-1): previous I/O error to superblock detected > Buffer I/O error on dev dm-1, logical block 0, lost sync page write > EXT4-fs (dm-1): re-mounted. Opts: (null) > Buffer I/O error on dev dm-1, logical block 80, lost async page write > > Cc: "Theodore Ts'o" <tytso@mit.edu> > Cc: Andreas Dilger <adilger.kernel@dilger.ca> > Cc: linux-ext4@vger.kernel.org > Cc: Jaegeuk Kim <jaegeuk@kernel.org> > Signed-off-by: Jaegeuk Kim <jaegeuk@google.com> > --- > fs/ext4/super.c | 21 +++++++++++++-------- > 1 file changed, 13 insertions(+), 8 deletions(-) Thanks for the patch. Some comments below. > > diff --git a/fs/ext4/super.c b/fs/ext4/super.c > index 39bf464c35f1..b268bfea879c 100644 > --- a/fs/ext4/super.c > +++ b/fs/ext4/super.c > @@ -2151,7 +2151,7 @@ static int ext4_setup_super(struct super_block *sb, struct ext4_super_block *es, > if (sbi->s_journal) > ext4_set_feature_journal_needs_recovery(sb); > > - ext4_commit_super(sb, 1); > + res = ext4_commit_super(sb, 1); > done: > if (test_opt(sb, DEBUG)) > printk(KERN_INFO "[EXT4 FS bs=%lu, gc=%u, " So return value of ext4_setup_super() looks a bit strange and it seems it currently returns 0 or SB_RDONLY. So I think you first need to clean this up before making ext4_setup_super() return error code (as your patch could make ext4_fill_super() return SB_RDONLY under some circumstances). > @@ -4739,11 +4739,10 @@ static int ext4_commit_super(struct super_block *sb, int sync) > unlock_buffer(sbh); > error = __sync_dirty_buffer(sbh, > REQ_SYNC | (test_opt(sb, BARRIER) ? REQ_FUA : 0)); > - if (error) > + if (error && !buffer_write_io_error(sbh)) > return error; > > - error = buffer_write_io_error(sbh); > - if (error) { > + if (buffer_write_io_error(sbh)) { > ext4_msg(sb, KERN_ERR, "I/O error while writing " > "superblock"); > clear_buffer_write_io_error(sbh); I'd rather rewrite this as: if (buffer_write_io_error(sbh)) { do stuff as now } if (error) return error; Honza
Hi Jan, Thank you for the review. On 04/19, Jan Kara wrote: > On Thu 29-03-18 18:44:29, Jaegeuk Kim wrote: > > From: Jaegeuk Kim <jaegeuk@google.com> > > > > When remounting ext4 from ro to rw, currently it allows its transition, > > even if ext4_commit_super() returns EIO. Even worse thing is, after that, > > fs/buffer complains buffer dirty bits like: > > > > Call trace: > > [<ffffff9750c259dc>] mark_buffer_dirty+0x184/0x1a4 > > [<ffffff9750cb398c>] __ext4_handle_dirty_super+0x4c/0xfc > > [<ffffff9750c7a9fc>] ext4_file_open+0x154/0x1c0 > > [<ffffff9750bea51c>] do_dentry_open+0x114/0x2d0 > > [<ffffff9750bea75c>] vfs_open+0x5c/0x94 > > [<ffffff9750bf879c>] path_openat+0x668/0xfe8 > > [<ffffff9750bf8088>] do_filp_open+0x74/0x120 > > [<ffffff9750beac98>] do_sys_open+0x148/0x254 > > [<ffffff9750beade0>] SyS_openat+0x10/0x18 > > [<ffffff9750a83ab0>] el0_svc_naked+0x24/0x28 > > EXT4-fs (dm-1): previous I/O error to superblock detected > > Buffer I/O error on dev dm-1, logical block 0, lost sync page write > > EXT4-fs (dm-1): re-mounted. Opts: (null) > > Buffer I/O error on dev dm-1, logical block 80, lost async page write > > > > Cc: "Theodore Ts'o" <tytso@mit.edu> > > Cc: Andreas Dilger <adilger.kernel@dilger.ca> > > Cc: linux-ext4@vger.kernel.org > > Cc: Jaegeuk Kim <jaegeuk@kernel.org> > > Signed-off-by: Jaegeuk Kim <jaegeuk@google.com> > > --- > > fs/ext4/super.c | 21 +++++++++++++-------- > > 1 file changed, 13 insertions(+), 8 deletions(-) > > Thanks for the patch. Some comments below. > > > > > diff --git a/fs/ext4/super.c b/fs/ext4/super.c > > index 39bf464c35f1..b268bfea879c 100644 > > --- a/fs/ext4/super.c > > +++ b/fs/ext4/super.c > > @@ -2151,7 +2151,7 @@ static int ext4_setup_super(struct super_block *sb, struct ext4_super_block *es, > > if (sbi->s_journal) > > ext4_set_feature_journal_needs_recovery(sb); > > > > - ext4_commit_super(sb, 1); > > + res = ext4_commit_super(sb, 1); > > done: > > if (test_opt(sb, DEBUG)) > > printk(KERN_INFO "[EXT4 FS bs=%lu, gc=%u, " > > So return value of ext4_setup_super() looks a bit strange and it seems it > currently returns 0 or SB_RDONLY. So I think you first need to clean this Indeed. I'll update it in v2. > up before making ext4_setup_super() return error code (as your patch could > make ext4_fill_super() return SB_RDONLY under some circumstances). Interestingly, it only affects ext4_remount() only. :P > > > @@ -4739,11 +4739,10 @@ static int ext4_commit_super(struct super_block *sb, int sync) > > unlock_buffer(sbh); > > error = __sync_dirty_buffer(sbh, > > REQ_SYNC | (test_opt(sb, BARRIER) ? REQ_FUA : 0)); > > - if (error) > > + if (error && !buffer_write_io_error(sbh)) > > return error; > > > > - error = buffer_write_io_error(sbh); > > - if (error) { > > + if (buffer_write_io_error(sbh)) { > > ext4_msg(sb, KERN_ERR, "I/O error while writing " > > "superblock"); > > clear_buffer_write_io_error(sbh); > > I'd rather rewrite this as: > if (buffer_write_io_error(sbh)) { > do stuff as now > } > if (error) > return error; Got it. Thanks, > > Honza > -- > Jan Kara <jack@suse.com> > SUSE Labs, CR
When remounting ext4 from ro to rw, currently it allows its transition,
even if ext4_commit_super() returns EIO. Even worse thing is, after that,
fs/buffer complains buffer dirty bits like:
Call trace:
[<ffffff9750c259dc>] mark_buffer_dirty+0x184/0x1a4
[<ffffff9750cb398c>] __ext4_handle_dirty_super+0x4c/0xfc
[<ffffff9750c7a9fc>] ext4_file_open+0x154/0x1c0
[<ffffff9750bea51c>] do_dentry_open+0x114/0x2d0
[<ffffff9750bea75c>] vfs_open+0x5c/0x94
[<ffffff9750bf879c>] path_openat+0x668/0xfe8
[<ffffff9750bf8088>] do_filp_open+0x74/0x120
[<ffffff9750beac98>] do_sys_open+0x148/0x254
[<ffffff9750beade0>] SyS_openat+0x10/0x18
[<ffffff9750a83ab0>] el0_svc_naked+0x24/0x28
EXT4-fs (dm-1): previous I/O error to superblock detected
Buffer I/O error on dev dm-1, logical block 0, lost sync page write
EXT4-fs (dm-1): re-mounted. Opts: (null)
Buffer I/O error on dev dm-1, logical block 80, lost async page write
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: linux-ext4@vger.kernel.org
Cc: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@google.com>
---
Change log from v1:
- modify return value of ext4_setup_super
- change the error handling order in ext4_commit_super
fs/ext4/super.c | 28 +++++++++++++++-------------
1 file changed, 15 insertions(+), 13 deletions(-)
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index 185f7e61f4cf..af4d373aa3ed 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -2116,12 +2116,12 @@ static int ext4_setup_super(struct super_block *sb, struct ext4_super_block *es,
int read_only)
{
struct ext4_sb_info *sbi = EXT4_SB(sb);
- int res = 0;
+ int err = 0;
if (le32_to_cpu(es->s_rev_level) > EXT4_MAX_SUPP_REV) {
ext4_msg(sb, KERN_ERR, "revision level too high, "
"forcing read-only mode");
- res = SB_RDONLY;
+ err = -EROFS;
}
if (read_only)
goto done;
@@ -2154,7 +2154,7 @@ static int ext4_setup_super(struct super_block *sb, struct ext4_super_block *es,
if (sbi->s_journal)
ext4_set_feature_journal_needs_recovery(sb);
- ext4_commit_super(sb, 1);
+ err = ext4_commit_super(sb, 1);
done:
if (test_opt(sb, DEBUG))
printk(KERN_INFO "[EXT4 FS bs=%lu, gc=%u, "
@@ -2166,7 +2166,7 @@ static int ext4_setup_super(struct super_block *sb, struct ext4_super_block *es,
sbi->s_mount_opt, sbi->s_mount_opt2);
cleancache_init_fs(sb);
- return res;
+ return err;
}
int ext4_alloc_flex_bg_array(struct super_block *sb, ext4_group_t ngroup)
@@ -4760,11 +4760,7 @@ static int ext4_commit_super(struct super_block *sb, int sync)
unlock_buffer(sbh);
error = __sync_dirty_buffer(sbh,
REQ_SYNC | (test_opt(sb, BARRIER) ? REQ_FUA : 0));
- if (error)
- return error;
-
- error = buffer_write_io_error(sbh);
- if (error) {
+ if (buffer_write_io_error(sbh)) {
ext4_msg(sb, KERN_ERR, "I/O error while writing "
"superblock");
clear_buffer_write_io_error(sbh);
@@ -5165,8 +5161,12 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data)
if (sbi->s_journal)
ext4_clear_journal_err(sb, es);
sbi->s_mount_state = le16_to_cpu(es->s_state);
- if (!ext4_setup_super(sb, es, 0))
- sb->s_flags &= ~SB_RDONLY;
+
+ err = ext4_setup_super(sb, es, 0);
+ if (err)
+ goto restore_opts;
+
+ sb->s_flags &= ~SB_RDONLY;
if (ext4_has_feature_mmp(sb))
if (ext4_multi_mount_protect(sb,
le64_to_cpu(es->s_mmp_block))) {
@@ -5190,8 +5190,10 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data)
}
ext4_setup_system_zone(sb);
- if (sbi->s_journal == NULL && !(old_sb_flags & SB_RDONLY))
- ext4_commit_super(sb, 1);
+ if (sbi->s_journal == NULL && !(old_sb_flags & SB_RDONLY)) {
+ err = ext4_commit_super(sb, 1);
+ goto restore_opts;
+ }
#ifdef CONFIG_QUOTA
/* Release old quota file names */
On Thu 19-04-18 08:59:57, Jaegeuk Kim wrote: > When remounting ext4 from ro to rw, currently it allows its transition, > even if ext4_commit_super() returns EIO. Even worse thing is, after that, > fs/buffer complains buffer dirty bits like: > > Call trace: > [<ffffff9750c259dc>] mark_buffer_dirty+0x184/0x1a4 > [<ffffff9750cb398c>] __ext4_handle_dirty_super+0x4c/0xfc > [<ffffff9750c7a9fc>] ext4_file_open+0x154/0x1c0 > [<ffffff9750bea51c>] do_dentry_open+0x114/0x2d0 > [<ffffff9750bea75c>] vfs_open+0x5c/0x94 > [<ffffff9750bf879c>] path_openat+0x668/0xfe8 > [<ffffff9750bf8088>] do_filp_open+0x74/0x120 > [<ffffff9750beac98>] do_sys_open+0x148/0x254 > [<ffffff9750beade0>] SyS_openat+0x10/0x18 > [<ffffff9750a83ab0>] el0_svc_naked+0x24/0x28 > EXT4-fs (dm-1): previous I/O error to superblock detected > Buffer I/O error on dev dm-1, logical block 0, lost sync page write > EXT4-fs (dm-1): re-mounted. Opts: (null) > Buffer I/O error on dev dm-1, logical block 80, lost async page write > > Cc: "Theodore Ts'o" <tytso@mit.edu> > Cc: Andreas Dilger <adilger.kernel@dilger.ca> > Cc: linux-ext4@vger.kernel.org > Cc: Jaegeuk Kim <jaegeuk@kernel.org> > Signed-off-by: Jaegeuk Kim <jaegeuk@google.com> Looks mostly good. Just: > @@ -5190,8 +5190,10 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data) > } > > ext4_setup_system_zone(sb); > - if (sbi->s_journal == NULL && !(old_sb_flags & SB_RDONLY)) > - ext4_commit_super(sb, 1); > + if (sbi->s_journal == NULL && !(old_sb_flags & SB_RDONLY)) { > + err = ext4_commit_super(sb, 1); > + goto restore_opts; > + } Why is goto restore_opts unconditional here? Honza
When remounting ext4 from ro to rw, currently it allows its transition,
even if ext4_commit_super() returns EIO. Even worse thing is, after that,
fs/buffer complains buffer dirty bits like:
Call trace:
[<ffffff9750c259dc>] mark_buffer_dirty+0x184/0x1a4
[<ffffff9750cb398c>] __ext4_handle_dirty_super+0x4c/0xfc
[<ffffff9750c7a9fc>] ext4_file_open+0x154/0x1c0
[<ffffff9750bea51c>] do_dentry_open+0x114/0x2d0
[<ffffff9750bea75c>] vfs_open+0x5c/0x94
[<ffffff9750bf879c>] path_openat+0x668/0xfe8
[<ffffff9750bf8088>] do_filp_open+0x74/0x120
[<ffffff9750beac98>] do_sys_open+0x148/0x254
[<ffffff9750beade0>] SyS_openat+0x10/0x18
[<ffffff9750a83ab0>] el0_svc_naked+0x24/0x28
EXT4-fs (dm-1): previous I/O error to superblock detected
Buffer I/O error on dev dm-1, logical block 0, lost sync page write
EXT4-fs (dm-1): re-mounted. Opts: (null)
Buffer I/O error on dev dm-1, logical block 80, lost async page write
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: linux-ext4@vger.kernel.org
Cc: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@google.com>
---
Change log from v2:
- fix missing goto condition
fs/ext4/super.c | 29 ++++++++++++++++-------------
1 file changed, 16 insertions(+), 13 deletions(-)
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index 185f7e61f4cf..32cf16c33cba 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -2116,12 +2116,12 @@ static int ext4_setup_super(struct super_block *sb, struct ext4_super_block *es,
int read_only)
{
struct ext4_sb_info *sbi = EXT4_SB(sb);
- int res = 0;
+ int err = 0;
if (le32_to_cpu(es->s_rev_level) > EXT4_MAX_SUPP_REV) {
ext4_msg(sb, KERN_ERR, "revision level too high, "
"forcing read-only mode");
- res = SB_RDONLY;
+ err = -EROFS;
}
if (read_only)
goto done;
@@ -2154,7 +2154,7 @@ static int ext4_setup_super(struct super_block *sb, struct ext4_super_block *es,
if (sbi->s_journal)
ext4_set_feature_journal_needs_recovery(sb);
- ext4_commit_super(sb, 1);
+ err = ext4_commit_super(sb, 1);
done:
if (test_opt(sb, DEBUG))
printk(KERN_INFO "[EXT4 FS bs=%lu, gc=%u, "
@@ -2166,7 +2166,7 @@ static int ext4_setup_super(struct super_block *sb, struct ext4_super_block *es,
sbi->s_mount_opt, sbi->s_mount_opt2);
cleancache_init_fs(sb);
- return res;
+ return err;
}
int ext4_alloc_flex_bg_array(struct super_block *sb, ext4_group_t ngroup)
@@ -4760,11 +4760,7 @@ static int ext4_commit_super(struct super_block *sb, int sync)
unlock_buffer(sbh);
error = __sync_dirty_buffer(sbh,
REQ_SYNC | (test_opt(sb, BARRIER) ? REQ_FUA : 0));
- if (error)
- return error;
-
- error = buffer_write_io_error(sbh);
- if (error) {
+ if (buffer_write_io_error(sbh)) {
ext4_msg(sb, KERN_ERR, "I/O error while writing "
"superblock");
clear_buffer_write_io_error(sbh);
@@ -5165,8 +5161,12 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data)
if (sbi->s_journal)
ext4_clear_journal_err(sb, es);
sbi->s_mount_state = le16_to_cpu(es->s_state);
- if (!ext4_setup_super(sb, es, 0))
- sb->s_flags &= ~SB_RDONLY;
+
+ err = ext4_setup_super(sb, es, 0);
+ if (err)
+ goto restore_opts;
+
+ sb->s_flags &= ~SB_RDONLY;
if (ext4_has_feature_mmp(sb))
if (ext4_multi_mount_protect(sb,
le64_to_cpu(es->s_mmp_block))) {
@@ -5190,8 +5190,11 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data)
}
ext4_setup_system_zone(sb);
- if (sbi->s_journal == NULL && !(old_sb_flags & SB_RDONLY))
- ext4_commit_super(sb, 1);
+ if (sbi->s_journal == NULL && !(old_sb_flags & SB_RDONLY)) {
+ err = ext4_commit_super(sb, 1);
+ if (err)
+ goto restore_opts;
+ }
#ifdef CONFIG_QUOTA
/* Release old quota file names */
On Mon, Apr 23, 2018 at 08:46:26AM -0600, Jaegeuk Kim wrote: > When remounting ext4 from ro to rw, currently it allows its transition, > even if ext4_commit_super() returns EIO. Even worse thing is, after that, > fs/buffer complains buffer dirty bits like: > > Call trace: > [<ffffff9750c259dc>] mark_buffer_dirty+0x184/0x1a4 > [<ffffff9750cb398c>] __ext4_handle_dirty_super+0x4c/0xfc > [<ffffff9750c7a9fc>] ext4_file_open+0x154/0x1c0 > [<ffffff9750bea51c>] do_dentry_open+0x114/0x2d0 > [<ffffff9750bea75c>] vfs_open+0x5c/0x94 > [<ffffff9750bf879c>] path_openat+0x668/0xfe8 > [<ffffff9750bf8088>] do_filp_open+0x74/0x120 > [<ffffff9750beac98>] do_sys_open+0x148/0x254 > [<ffffff9750beade0>] SyS_openat+0x10/0x18 > [<ffffff9750a83ab0>] el0_svc_naked+0x24/0x28 > EXT4-fs (dm-1): previous I/O error to superblock detected > Buffer I/O error on dev dm-1, logical block 0, lost sync page write > EXT4-fs (dm-1): re-mounted. Opts: (null) > Buffer I/O error on dev dm-1, logical block 80, lost async page write > > Cc: "Theodore Ts'o" <tytso@mit.edu> > Cc: Andreas Dilger <adilger.kernel@dilger.ca> > Cc: linux-ext4@vger.kernel.org > Cc: Jaegeuk Kim <jaegeuk@kernel.org> > Signed-off-by: Jaegeuk Kim <jaegeuk@google.com> Applied with a fix up to ext4_fill_super() when it calls ext4_setup_super() with the changed error return semantics. - Ted diff --git a/fs/ext4/super.c b/fs/ext4/super.c index aac33c155363..1388e56bb3f5 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -4254,8 +4254,12 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) goto failed_mount4; } - if (ext4_setup_super(sb, es, sb_rdonly(sb))) + ret = ext4_setup_super(sb, es, sb_rdonly(sb)); + if (ret == -EROFS) { sb->s_flags |= SB_RDONLY; + ret = 0; + } else if (ret) + goto failed_mount4a; /* determine the minimum size of new large inodes, if present */ if (sbi->s_inode_size > EXT4_GOOD_OLD_INODE_SIZE &&
diff --git a/fs/ext4/super.c b/fs/ext4/super.c index 39bf464c35f1..b268bfea879c 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -2151,7 +2151,7 @@ static int ext4_setup_super(struct super_block *sb, struct ext4_super_block *es, if (sbi->s_journal) ext4_set_feature_journal_needs_recovery(sb); - ext4_commit_super(sb, 1); + res = ext4_commit_super(sb, 1); done: if (test_opt(sb, DEBUG)) printk(KERN_INFO "[EXT4 FS bs=%lu, gc=%u, " @@ -4739,11 +4739,10 @@ static int ext4_commit_super(struct super_block *sb, int sync) unlock_buffer(sbh); error = __sync_dirty_buffer(sbh, REQ_SYNC | (test_opt(sb, BARRIER) ? REQ_FUA : 0)); - if (error) + if (error && !buffer_write_io_error(sbh)) return error; - error = buffer_write_io_error(sbh); - if (error) { + if (buffer_write_io_error(sbh)) { ext4_msg(sb, KERN_ERR, "I/O error while writing " "superblock"); clear_buffer_write_io_error(sbh); @@ -5144,8 +5143,12 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data) if (sbi->s_journal) ext4_clear_journal_err(sb, es); sbi->s_mount_state = le16_to_cpu(es->s_state); - if (!ext4_setup_super(sb, es, 0)) - sb->s_flags &= ~SB_RDONLY; + + err = ext4_setup_super(sb, es, 0); + if (err) + goto restore_opts; + + sb->s_flags &= ~SB_RDONLY; if (ext4_has_feature_mmp(sb)) if (ext4_multi_mount_protect(sb, le64_to_cpu(es->s_mmp_block))) { @@ -5169,8 +5172,10 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data) } ext4_setup_system_zone(sb); - if (sbi->s_journal == NULL && !(old_sb_flags & SB_RDONLY)) - ext4_commit_super(sb, 1); + if (sbi->s_journal == NULL && !(old_sb_flags & SB_RDONLY)) { + err = ext4_commit_super(sb, 1); + goto restore_opts; + } #ifdef CONFIG_QUOTA /* Release old quota file names */