btrfs: Add "barrier" option to support "-o remount,barrier"
diff mbox

Message ID 1388040310-25092-1-git-send-email-quwenruo@cn.fujitsu.com
State New, archived
Headers show

Commit Message

Qu Wenruo Dec. 26, 2013, 6:45 a.m. UTC
Btrfs can be remounted without barrier, but there is no "barrier" option
so nobody can remount btrfs back with barrier on. Only umount and
mount again can re-enable barrier.(Quite awkward)

Reported-by: Daniel Blueman <daniel@quora.org>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
---
 Documentation/filesystems/btrfs.txt | 6 ++++++
 fs/btrfs/super.c                    | 8 +++++++-
 2 files changed, 13 insertions(+), 1 deletion(-)

Comments

David Sterba Jan. 2, 2014, 3:41 p.m. UTC | #1
On Thu, Dec 26, 2013 at 02:45:10PM +0800, Qu Wenruo wrote:
> Btrfs can be remounted without barrier, but there is no "barrier" option
> so nobody can remount btrfs back with barrier on. Only umount and
> mount again can re-enable barrier.(Quite awkward)

Ok for adding the mount option.

> Reported-by: Daniel Blueman <daniel@quora.org>
> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
> ---
>  Documentation/filesystems/btrfs.txt | 6 ++++++
>  fs/btrfs/super.c                    | 8 +++++++-
>  2 files changed, 13 insertions(+), 1 deletion(-)
> 
> diff --git a/Documentation/filesystems/btrfs.txt b/Documentation/filesystems/btrfs.txt
> index 5dd282d..f6f71d6 100644
> --- a/Documentation/filesystems/btrfs.txt
> +++ b/Documentation/filesystems/btrfs.txt
> @@ -51,6 +51,12 @@ Unless otherwise specified, all options default to off.
>  	defrag process.  Works best for small files; Not well suited for
>  	large database workloads.
>  
> +  barrier
> +	Enable the use of block layer write barriers.  Write barriers ensure
> +	that certain IOs make it through the device cache and are on persistent
> +	storage.
> +	Barriers are enabled by default.

There's the 'nobarrier' option already, I think it's better to keep the
pairing option documentation at one place and mark the default one with (*).

Related to that, there should imho be a pairing option for every mount
option where it makes sense, so all the combinations work through
remount.

The defaults do not need to be listed via btrfs_show_options, ie. keep
the output same as it is now.


david
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Qu Wenruo Jan. 3, 2014, 12:43 a.m. UTC | #2
On Thu, 2 Jan 2014 16:41:03 +0100, David Sterba wrote:
> On Thu, Dec 26, 2013 at 02:45:10PM +0800, Qu Wenruo wrote:
>> Btrfs can be remounted without barrier, but there is no "barrier" option
>> so nobody can remount btrfs back with barrier on. Only umount and
>> mount again can re-enable barrier.(Quite awkward)
> Ok for adding the mount option.
>
>> Reported-by: Daniel Blueman <daniel@quora.org>
>> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
>> ---
>>   Documentation/filesystems/btrfs.txt | 6 ++++++
>>   fs/btrfs/super.c                    | 8 +++++++-
>>   2 files changed, 13 insertions(+), 1 deletion(-)
>>
>> diff --git a/Documentation/filesystems/btrfs.txt b/Documentation/filesystems/btrfs.txt
>> index 5dd282d..f6f71d6 100644
>> --- a/Documentation/filesystems/btrfs.txt
>> +++ b/Documentation/filesystems/btrfs.txt
>> @@ -51,6 +51,12 @@ Unless otherwise specified, all options default to off.
>>   	defrag process.  Works best for small files; Not well suited for
>>   	large database workloads.
>>   
>> +  barrier
>> +	Enable the use of block layer write barriers.  Write barriers ensure
>> +	that certain IOs make it through the device cache and are on persistent
>> +	storage.
>> +	Barriers are enabled by default.
> There's the 'nobarrier' option already, I think it's better to keep the
> pairing option documentation at one place and mark the default one with (*).
>
> Related to that, there should imho be a pairing option for every mount
> option where it makes sense, so all the combinations work through
> remount.
>
> The defaults do not need to be listed via btrfs_show_options, ie. keep
> the output same as it is now.
>
>
> david
>
Thanks for pointing out these problem.

This makes sense.Like space_cache and nospace_cache follows the way you 
mentioned,
but most other options just followed the alphabet order, and a lot of
pairing is missing.

I'll try to add all the missingpairing and unify mount options in the 
document.

Qu
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch
diff mbox

diff --git a/Documentation/filesystems/btrfs.txt b/Documentation/filesystems/btrfs.txt
index 5dd282d..f6f71d6 100644
--- a/Documentation/filesystems/btrfs.txt
+++ b/Documentation/filesystems/btrfs.txt
@@ -51,6 +51,12 @@  Unless otherwise specified, all options default to off.
 	defrag process.  Works best for small files; Not well suited for
 	large database workloads.
 
+  barrier
+	Enable the use of block layer write barriers.  Write barriers ensure
+	that certain IOs make it through the device cache and are on persistent
+	storage.
+	Barriers are enabled by default.
+	
   check_int
   check_int_data
   check_int_print_mask=<value>
diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index e9c13fb..c8f4aca 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -323,7 +323,7 @@  enum {
 	Opt_no_space_cache, Opt_recovery, Opt_skip_balance,
 	Opt_check_integrity, Opt_check_integrity_including_extent_data,
 	Opt_check_integrity_print_mask, Opt_fatal_errors, Opt_rescan_uuid_tree,
-	Opt_commit_interval,
+	Opt_commit_interval, Opt_barrier,
 	Opt_err,
 };
 
@@ -335,6 +335,7 @@  static match_table_t tokens = {
 	{Opt_nodatasum, "nodatasum"},
 	{Opt_nodatacow, "nodatacow"},
 	{Opt_nobarrier, "nobarrier"},
+	{Opt_barrier, "barrier"},
 	{Opt_max_inline, "max_inline=%s"},
 	{Opt_alloc_start, "alloc_start=%s"},
 	{Opt_thread_pool, "thread_pool=%d"},
@@ -498,6 +499,11 @@  int btrfs_parse_options(struct btrfs_root *root, char *options)
 			btrfs_info(root->fs_info, "turning off barriers");
 			btrfs_set_opt(info->mount_opt, NOBARRIER);
 			break;
+		case Opt_barrier:
+			if (btrfs_test_opt(root, NOBARRIER))
+				btrfs_info(root->fs_info, "turning on barriers");
+			btrfs_clear_opt(info->mount_opt, NOBARRIER);
+			break;
 		case Opt_thread_pool:
 			ret = match_int(&args[0], &intarg);
 			if (ret) {