Messages in this thread | | | Subject | Re: [PATCH v3 1/2] ext4: commit super block if fs record error when journal record without error | From | "yebin (H)" <> | Date | Thu, 16 Feb 2023 17:29:57 +0800 |
| |
On 2023/2/16 17:17, Baokun Li wrote: > On 2023/2/16 15:44, yebin (H) wrote: >> >> >> On 2023/2/16 15:17, Baokun Li wrote: >>> On 2023/2/14 10:29, Ye Bin wrote: >>>> From: Ye Bin <yebin10@huawei.com> >>>> >>>> Now, 'es->s_state' maybe covered by recover journal. And journal errno >>>> maybe not recorded in journal sb as IO error. ext4_update_super() only >>>> update error information when 'sbi->s_add_error_count' large than >>>> zero. >>>> Then 'EXT4_ERROR_FS' flag maybe lost. >>>> To solve above issue commit error information after recover journal. >>>> >>>> Signed-off-by: Ye Bin <yebin10@huawei.com> >>>> --- >>>> fs/ext4/super.c | 12 ++++++++++++ >>>> 1 file changed, 12 insertions(+) >>>> >>>> diff --git a/fs/ext4/super.c b/fs/ext4/super.c >>>> index dc3907dff13a..b94754ba8556 100644 >>>> --- a/fs/ext4/super.c >>>> +++ b/fs/ext4/super.c >>>> @@ -5932,6 +5932,18 @@ static int ext4_load_journal(struct >>>> super_block *sb, >>>> goto err_out; >>>> } >>>> + if (unlikely(es->s_error_count && >>>> !jbd2_journal_errno(journal) && >>>> + !(le16_to_cpu(es->s_state) & EXT4_ERROR_FS))) { >>>> + EXT4_SB(sb)->s_mount_state |= EXT4_ERROR_FS; >>>> + es->s_state |= cpu_to_le16(EXT4_ERROR_FS); >>>> + err = ext4_commit_super(sb); >>>> + if (err) { >>>> + ext4_msg(sb, KERN_ERR, >>>> + "Failed to commit error information, please >>>> repair fs force!"); >>>> + goto err_out; >>>> + } >>>> + } >>>> + >>>> EXT4_SB(sb)->s_journal = journal; >>>> err = ext4_clear_journal_err(sb, es); >>>> if (err) { >>> I think we don't need such a complicated judgment, after the journal >>> replay and saving the error info, >>> if there is EXT4_ERROR_FS flag in ext4_sb_info->s_mount_state, just >>> add this flag directly to es->s_state. >>> This way the EXT4_ERROR_FS flag and the error message will be >>> written to disk the next time >> >> Thanks for your suggestion. There are two reasons for this: >> 1. We want to write the error mark to the disk as soon as possible. >> 2. Here we deal with the case where there is no error mark bit but >> there is an error record. >> In this case, the file system should be marked with an error and the >> user should be prompted. > The EXT4_ERROR_FS flag is always written to disk at the same time as > the error info, > except when the journal is replayed. During journal replay the error > info is additionally > copied to disk first, and the EXT4_ERROR_FS flag in the sbi is not > written to disk until > the ext4_put_super() is called. It is only when a failure occurs > during this time that > there is an error info but no EXT4_ERROR_FS flag. So we just need to > make sure that > the EXT4_ERROR_FS flag is also written to disk at the same time as the > error info > after the journal replay. The situation you said is based on the situation after the repair. What about the existing image with such inconsistency? >>> ext4_commit_super() is executed. The code change is as follows: >>> >>> diff --git a/fs/ext4/super.c b/fs/ext4/super.c >>> index 260c1b3e3ef2..341b11c589b3 100644 >>> --- a/fs/ext4/super.c >>> +++ b/fs/ext4/super.c >>> @@ -5935,6 +5935,7 @@ static int ext4_load_journal(struct >>> super_block *sb, >>> memcpy(((char *) es) + EXT4_S_ERR_START, >>> save, EXT4_S_ERR_LEN); >>> kfree(save); >>> + es->s_state |= >>> cpu_to_le16(EXT4_SB(sb)->s_mount_state & EXT4_ERROR_FS); >>> } >>> >>> if (err) { >>> >> >> >>
| |