Re: [PATCH 2/2] ext4: Automatically enable journal_async_commit on ext4 file systems

[Ric Wheeler <ricwheeler@gmail.com>]

Eric Sandeen wrote:
> Theodore Ts'o wrote:
>> Now that we have cleaned up journal_async_commit, it's safe to enable
>> it all the time.  But we only want to do so if ext4-specific INCOMPAT
>> features are enabled, since otherwise we will prevent the filesystem
>> from being mounted using ext3.
>
> Has anyone done some serious power-fail/crash (real or simulated) 
> testing with this?  If not, I think we should before it's default...
>
> -Eric

I think that we also need to see some very compelling performance 
numbers to justify the complexity. Journaling is complex enough, so if 
this is not a win, we should (in my opinion) go with the simplest 
technique :-)

ric

>
>> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
>> ---
>>  Documentation/filesystems/ext4.txt |    8 ++++++--
>>  fs/ext4/super.c                    |   26 ++++++++++++++++++++++++--
>>  2 files changed, 30 insertions(+), 4 deletions(-)
>>
>> diff --git a/Documentation/filesystems/ext4.txt 
>> b/Documentation/filesystems/ext4.txt
>> index 3e329db..4f8d73a 100644
>> --- a/Documentation/filesystems/ext4.txt
>> +++ b/Documentation/filesystems/ext4.txt
>> @@ -135,8 +135,12 @@ ro                       Mount filesystem read 
>> only. Note that ext4 will
>>                   writes to the filesystem.
>>  
>>  journal_async_commit    Commit block can be written to disk without 
>> waiting
>> -            for descriptor blocks. If enabled older kernels cannot
>> -            mount the device.
>> +            for descriptor blocks.  This mount option will be 
>> +            automatically enabled if ext4-specific INCOMPAT
>> +            features are present in the file system.
>> +
>> +nojournal_async_commit    Disable the journal_async_commit option, even
>> +            for ext4 filesystems.
>>  
>>  journal=update        Update the ext4 file system's journal to the 
>> current
>>              format.
>> diff --git a/fs/ext4/super.c b/fs/ext4/super.c
>> index f1815d3..f644a5c 100644
>> --- a/fs/ext4/super.c
>> +++ b/fs/ext4/super.c
>> @@ -75,6 +75,15 @@ static int ext4_unfreeze(struct super_block *sb);
>>  static void ext4_write_super(struct super_block *sb);
>>  static int ext4_freeze(struct super_block *sb);
>>  
>> +/*
>> + * If ext4 filesystem features are enabled, then enable async_commits
>> + * by default.
>> + */
>> +#define ASYNC_COMMIT_DEFAULT(sb) (EXT4_HAS_INCOMPAT_FEATURE(sb, \
>> +                    (EXT4_FEATURE_INCOMPAT_EXTENTS| \
>> +                     EXT4_FEATURE_INCOMPAT_64BIT| \
>> +                     EXT4_FEATURE_INCOMPAT_FLEX_BG)))
>> +
>>  
>>  ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
>>                     struct ext4_group_desc *bg)
>> @@ -845,8 +854,13 @@ static int ext4_show_options(struct seq_file 
>> *seq, struct vfsmount *vfs)
>>       */
>>      seq_puts(seq, ",barrier=");
>>      seq_puts(seq, test_opt(sb, BARRIER) ? "1" : "0");
>> -    if (test_opt(sb, JOURNAL_ASYNC_COMMIT))
>> -        seq_puts(seq, ",journal_async_commit");
>> +    if (ASYNC_COMMIT_DEFAULT(sb)) {
>> +        if (!test_opt(sb, JOURNAL_ASYNC_COMMIT))
>> +            seq_puts(seq, ",nojournal_async_commit");
>> +    } else {
>> +        if (test_opt(sb, JOURNAL_ASYNC_COMMIT))
>> +            seq_puts(seq, ",journal_async_commit");
>> +    }
>>      if (test_opt(sb, NOBH))
>>          seq_puts(seq, ",nobh");
>>      if (test_opt(sb, I_VERSION))
>> @@ -1050,6 +1064,7 @@ enum {
>>      Opt_commit, Opt_min_batch_time, Opt_max_batch_time,
>>      Opt_journal_update, Opt_journal_dev,
>>      Opt_journal_checksum, Opt_journal_async_commit,
>> +    Opt_nojournal_async_commit,
>>      Opt_abort, Opt_data_journal, Opt_data_ordered, Opt_data_writeback,
>>      Opt_data_err_abort, Opt_data_err_ignore, Opt_mb_history_length,
>>      Opt_usrjquota, Opt_grpjquota, Opt_offusrjquota, Opt_offgrpjquota,
>> @@ -1092,6 +1107,7 @@ static const match_table_t tokens = {
>>      {Opt_journal_dev, "journal_dev=%u"},
>>      {Opt_journal_checksum, "journal_checksum"},
>>      {Opt_journal_async_commit, "journal_async_commit"},
>> +    {Opt_nojournal_async_commit, "nojournal_async_commit"},
>>      {Opt_abort, "abort"},
>>      {Opt_data_journal, "data=journal"},
>>      {Opt_data_ordered, "data=ordered"},
>> @@ -1284,6 +1300,9 @@ static int parse_options(char *options, struct 
>> super_block *sb,
>>          case Opt_journal_async_commit:
>>              set_opt(sbi->s_mount_opt, JOURNAL_ASYNC_COMMIT);
>>              break;
>> +        case Opt_nojournal_async_commit:
>> +            clear_opt(sbi->s_mount_opt, JOURNAL_ASYNC_COMMIT);
>> +            break;
>>          case Opt_noload:
>>              set_opt(sbi->s_mount_opt, NOLOAD);
>>              break;
>> @@ -2422,6 +2441,9 @@ static int ext4_fill_super(struct super_block 
>> *sb, void *data, int silent)
>>       */
>>      set_opt(sbi->s_mount_opt, DELALLOC);
>>  
>> +    if (ASYNC_COMMIT_DEFAULT(sb))
>> +        set_opt(sbi->s_mount_opt, JOURNAL_ASYNC_COMMIT);
>> +
>>      if (!parse_options((char *) data, sb, &journal_devnum,
>>                 &journal_ioprio, NULL, 0))
>>          goto failed_mount;
>
> -- 
> To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html