Skip to contentSkip to conversation

btrfs: sysfs and unsupported temp-fsid features for clones

PATCH In review Anand Jain 3 participants 17 messages Oct 4, 2023 – Oct 19, 2023 Btrfs file system

lore
Share & tools
lore

Anand Jain

author
#0 lore raw

cover btrfs: sysfs and unsupported temp-fsid features for clones

Seed and device-add are the two features that must be unsupported if a cloned device is using temp-fsid to mount, as they conflict with multi-device functionality.

Additionally, add sysfs files for the temp-fsid feature.

Anand Jain (4):
btrfs: comment for temp-fsid, fsid, and metadata_uuid
btrfs: disable seed feature for temp-fsid
btrfs: disable the device add feature for temp-fsid
btrfs: show temp_fsid feature in sysfs

 fs/btrfs/ioctl.c   |  6 ++++++
 fs/btrfs/sysfs.c   | 20 ++++++++++++++++++++
 fs/btrfs/volumes.c |  8 ++++++++
 fs/btrfs/volumes.h |  4 ++++
 4 files changed, 38 insertions(+)
Signature · 1 line
2.38.1
author
#1 lore raw

PATCH 1/4 btrfs: comment for temp-fsid, fsid, and metadata_uuid

Add a comment to explain the relationship between temp-fsid, fsid, and metadata_uuid.

---
 fs/btrfs/volumes.h | 4 ++++
 1 file changed, 4 insertions(+)
fs/btrfs/volumes.h +4 -0
index fef46c0dbf19..5dd4ad775e5d 100644
@@ -299,6 +299,10 @@ struct btrfs_fs_devices {
 	 *   - When the BTRFS_FEATURE_INCOMPAT_METADATA_UUID flag is set:
 	 *       fs_devices->fsid == sb->fsid
 	 *       fs_devices->metadata_uuid == sb->metadata_uuid
+	 *
+	 *   - When in-memory fs_devices->temp_fsid is true
+	 *	 fs_devices->fsid = random
+	 *	 fs_devices->metadata_uuid == sb->fsid
 	 */
 	u8 metadata_uuid[BTRFS_FSID_SIZE];
 
Signature · 1 line
2.38.1
author
#2 lore raw

PATCH 2/4 btrfs: disable seed feature for temp-fsid

In reply to Anand Jain

A seed device is an integral component of the sprout device, which functions as a multi-device filesystem. Therefore, temp-fsid feature is not supported.

---
 fs/btrfs/volumes.c | 8 ++++++++
 1 file changed, 8 insertions(+)
fs/btrfs/volumes.c +8 -0
index 81b735f4efc1..1fdfa9153e30 100644
@@ -571,6 +571,14 @@ static struct btrfs_fs_devices *find_fsid_by_device(
 	if (btrfs_super_num_devices(disk_super) != 1)
 		return fsid_fs_devices;
 
+	/*
+	 * A seed device is an integral component of the sprout device, which
+	 * functions as a multi-device filesystem. So, temp-fsid feature is
+	 * not supported.
+	 */
+	if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_SEEDING)
+		return fsid_fs_devices;
+
 	/* Try to find a fs_devices by matching devt. */
 	list_for_each_entry(devt_fs_devices, &fs_uuids, fs_list) {
 		struct btrfs_device *device;
Signature · 1 line
2.38.1
author
#3 lore raw

PATCH 3/4 btrfs: disable the device add feature for temp-fsid

In reply to Anand Jain

The device addition operation will transform the cloned temp-fsid mounted device into a multi-device filesystem. Therefore, it is marked as unsupported.

---
 fs/btrfs/ioctl.c | 6 ++++++
 1 file changed, 6 insertions(+)
fs/btrfs/ioctl.c +6 -0
index 848b7e6f6421..7d8d217cf5fd 100644
@@ -2635,6 +2635,12 @@ static long btrfs_ioctl_add_dev(struct btrfs_fs_info *fs_info, void __user *arg)
 		return -EINVAL;
 	}
 
+	if (fs_info->fs_devices->temp_fsid) {
+		btrfs_err(fs_info,
+			  "device add not supported on cloned temp-fsid mount");
+		return -EINVAL;
+	}
+
 	if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_ADD)) {
 		if (!btrfs_exclop_start_try_lock(fs_info, BTRFS_EXCLOP_DEV_ADD))
 			return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
Signature · 1 line
2.38.1
author
#4 lore raw

PATCH 4/4 btrfs: show temp_fsid feature in sysfs

In reply to Anand Jain

This adds sysfs objects to indicate temp_fsid feature support and its status.

/sys/fs/btrfs/features/temp_fsid
/sys/fs/btrfs/<UUID>/temp_fsid

For example:

     Consider two cloned and mounted devices.

	$ blkid /dev/sdc[1-2]
	/dev/sdc1: UUID="509ad44b-ad2a-4a8a-bc8d-fe69db7220d5" ..
	/dev/sdc2: UUID="509ad44b-ad2a-4a8a-bc8d-fe69db7220d5" ..

     One gets actual fsid, and the other gets the temp_fsid when
     mounted.

	$ btrfs filesystem show -m
	Label: none  uuid: 509ad44b-ad2a-4a8a-bc8d-fe69db7220d5
		Total devices 1 FS bytes used 54.14MiB
		devid    1 size 300.00MiB used 144.00MiB path /dev/sdc1

	Label: none  uuid: 33bad74e-c91b-43a5-aef8-b3cab97ae63a
		Total devices 1 FS bytes used 54.14MiB
		devid    1 size 300.00MiB used 144.00MiB path /dev/sdc2

     Their sysfs as below.

	$ cat /sys/fs/btrfs/features/temp_fsid
	0

	$ cat /sys/fs/btrfs/509ad44b-ad2a-4a8a-bc8d-fe69db7220d5/temp_fsid
	0

	$ cat /sys/fs/btrfs/33bad74e-c91b-43a5-aef8-b3cab97ae63a/temp_fsid
	1
---
 fs/btrfs/sysfs.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
fs/btrfs/sysfs.c +20 -0
index e07be193323a..7f9a4790e013 100644
@@ -425,6 +425,15 @@ static ssize_t acl_show(struct kobject *kobj, struct kobj_attribute *a, char *bu
 }
 BTRFS_ATTR(static_feature, acl, acl_show);
 
+static ssize_t temp_fsid_supported_show(struct kobject *kobj,
+					struct kobj_attribute *a, char *buf)
+{
+	int ret = 0;
+
+	return sysfs_emit(buf, "%d\n", ret);
+}
+BTRFS_ATTR(static_feature, temp_fsid, temp_fsid_supported_show);
+
 /*
  * Features which only depend on kernel version.
  *
@@ -438,6 +447,7 @@ static struct attribute *btrfs_supported_static_feature_attrs[] = {
 	BTRFS_ATTR_PTR(static_feature, send_stream_version),
 	BTRFS_ATTR_PTR(static_feature, supported_rescue_options),
 	BTRFS_ATTR_PTR(static_feature, supported_sectorsizes),
+	BTRFS_ATTR_PTR(static_feature, temp_fsid),
 	NULL
 };
 
@@ -1205,6 +1215,15 @@ static ssize_t btrfs_generation_show(struct kobject *kobj,
 }
 BTRFS_ATTR(, generation, btrfs_generation_show);
 
+static ssize_t btrfs_temp_fsid_show(struct kobject *kobj,
+				    struct kobj_attribute *a, char *buf)
+{
+	struct btrfs_fs_info *fs_info = to_fs_info(kobj);
+
+	return sysfs_emit(buf, "%d\n", fs_info->fs_devices->temp_fsid);
+}
+BTRFS_ATTR(, temp_fsid, btrfs_temp_fsid_show);
+
 static const char * const btrfs_read_policy_name[] = { "pid" };
 
 static ssize_t btrfs_read_policy_show(struct kobject *kobj,
@@ -1307,6 +1326,7 @@ static const struct attribute *btrfs_attrs[] = {
 	BTRFS_ATTR_PTR(, read_policy),
 	BTRFS_ATTR_PTR(, bg_reclaim_threshold),
 	BTRFS_ATTR_PTR(, commit_stats),
+	BTRFS_ATTR_PTR(, temp_fsid),
 	NULL,
 };
 
Signature · 1 line
2.38.1
#5 lore raw

Re: [PATCH 4/4] btrfs: show temp_fsid feature in sysfs

Quoted 28 lines · Anand Jain

This adds sysfs objects to indicate temp_fsid feature support and its status.

/sys/fs/btrfs/features/temp_fsid
/sys/fs/btrfs/<UUID>/temp_fsid

For example:

     Consider two cloned and mounted devices.

	$ blkid /dev/sdc[1-2]
	/dev/sdc1: UUID="509ad44b-ad2a-4a8a-bc8d-fe69db7220d5" ..
	/dev/sdc2: UUID="509ad44b-ad2a-4a8a-bc8d-fe69db7220d5" ..

     One gets actual fsid, and the other gets the temp_fsid when
     mounted.

	$ btrfs filesystem show -m
	Label: none  uuid: 509ad44b-ad2a-4a8a-bc8d-fe69db7220d5
		Total devices 1 FS bytes used 54.14MiB
		devid    1 size 300.00MiB used 144.00MiB path /dev/sdc1

	Label: none  uuid: 33bad74e-c91b-43a5-aef8-b3cab97ae63a
		Total devices 1 FS bytes used 54.14MiB
		devid    1 size 300.00MiB used 144.00MiB path /dev/sdc2

     Their sysfs as below.
Quoted 8 lines
	$ cat /sys/fs/btrfs/features/temp_fsid
	0

	$ cat /sys/fs/btrfs/509ad44b-ad2a-4a8a-bc8d-fe69db7220d5/temp_fsid
	0

	$ cat /sys/fs/btrfs/33bad74e-c91b-43a5-aef8-b3cab97ae63a/temp_fsid
	1

So the fsid used for the directory is always the new one, is there a way to read which is the original filesystem's fsid? In this case it would be the 509ad44b-... We could print it in that file instead of '1', though it could be confusing that it's not the temp_fsid but the original one, file name mismatches the contents on first look.

Quoted 10 lines
---
 fs/btrfs/sysfs.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
fs/btrfs/sysfs.c +0 -0
index e07be193323a..7f9a4790e013 100644

fs/btrfs/sysfs.c static ssize_t acl_show(struct kobject *kobj, struct kobj_attribute *a, char *bu · line 433

Quoted 9 lines
fs/btrfs/sysfs.c +6 -0
@@ -425,6 +425,15 @@ static ssize_t acl_show(struct kobject *kobj, struct kobj_attribute *a, char *bu
 }
 BTRFS_ATTR(static_feature, acl, acl_show);
 
+static ssize_t temp_fsid_supported_show(struct kobject *kobj,
+					struct kobj_attribute *a, char *buf)
+{
+	int ret = 0;
+
+	return sysfs_emit(buf, "%d\n", ret);

This can be

	return sysfs_emit(buf, "0\n");
#8 lore raw
In reply to Anand Jain
Quoted 11 lines · Anand Jain

Seed and device-add are the two features that must be unsupported if a cloned device is using temp-fsid to mount, as they conflict with multi-device functionality.

Additionally, add sysfs files for the temp-fsid feature.

Anand Jain (4):
btrfs: comment for temp-fsid, fsid, and metadata_uuid
btrfs: disable seed feature for temp-fsid
btrfs: disable the device add feature for temp-fsid
btrfs: show temp_fsid feature in sysfs

Added to misc-next, thanks. The eventual change to the sysfs file can be done later.

How are we going to proceed with the patch from Guilherme?

author
#6 lore raw

Re: [PATCH 4/4] btrfs: show temp_fsid feature in sysfs

In reply to David Sterba
Quoted 35 lines · David Sterba
Quoted 34 lines · Anand Jain

This adds sysfs objects to indicate temp_fsid feature support and its status.

   /sys/fs/btrfs/features/temp_fsid
   /sys/fs/btrfs/<UUID>/temp_fsid

   For example:

      Consider two cloned and mounted devices.

	$ blkid /dev/sdc[1-2]
	/dev/sdc1: UUID="509ad44b-ad2a-4a8a-bc8d-fe69db7220d5" ..
	/dev/sdc2: UUID="509ad44b-ad2a-4a8a-bc8d-fe69db7220d5" ..

      One gets actual fsid, and the other gets the temp_fsid when
      mounted.

	$ btrfs filesystem show -m
	Label: none  uuid: 509ad44b-ad2a-4a8a-bc8d-fe69db7220d5
		Total devices 1 FS bytes used 54.14MiB
		devid    1 size 300.00MiB used 144.00MiB path /dev/sdc1

	Label: none  uuid: 33bad74e-c91b-43a5-aef8-b3cab97ae63a
		Total devices 1 FS bytes used 54.14MiB
		devid    1 size 300.00MiB used 144.00MiB path /dev/sdc2

      Their sysfs as below.

	$ cat /sys/fs/btrfs/features/temp_fsid
	0

	$ cat /sys/fs/btrfs/509ad44b-ad2a-4a8a-bc8d-fe69db7220d5/temp_fsid
	0
Quoted 8 lines
Quoted 2 lines
	$ cat /sys/fs/btrfs/33bad74e-c91b-43a5-aef8-b3cab97ae63a/temp_fsid
	1

So the fsid used for the directory is always the new one, is there a way to read which is the original filesystem's fsid? In this case it would be the 509ad44b-... We could print it in that file instead of '1', though it could be confusing that it's not the temp_fsid but the original one, file name mismatches the contents on first look.

Instead, can we emit 'fsid' in another kobject altogether? Furthermore, we also have a 'metadata_uuid' kobject. Here is how they relate.

1. normally:

 $ cat /sys/fs/btrfs/<meta-fsid>/fsid
 <meta-fsid>
 $ cat /sys/fs/btrfs/<meta-fsid>/metadata_uuid
 <meta-fsid>

2. metadata-uuid flag is set:

 $ cat /sys/fs/btrfs/<sb-fsid>/fsid
 <sb-fsid>
 $ cat /sys/fs/btrfs/<sb-fsid>/metadata_uuid
 <meta-fsid>

3. normal + temp-fsid:

 $ cat /sys/fs/btrfs/<temp-fsid>/fsid
 <meta-fsid>
 $ cat /sys/fs/btrfs/<temp-fsid>/metadata_uuid
 <meta-fsid>

4. metadata-uuid flag is set + temp-fsid:

 $ cat /sys/fs/btrfs/<temp-fsid>/fsid
 <sb-fsid>
 $ cat /sys/fs/btrfs/<temp-fsid>/metadata_uuid
 <meta-fsid>
Quoted 23 lines
Quoted 11 lines
---
  fs/btrfs/sysfs.c | 20 ++++++++++++++++++++
  1 file changed, 20 insertions(+)
fs/btrfs/sysfs.c +0 -0
index e07be193323a..7f9a4790e013 100644

fs/btrfs/sysfs.c static ssize_t acl_show(struct kobject *kobj, struct kobj_attribute *a, char *bu · line 433

Quoted 9 lines
fs/btrfs/sysfs.c +6 -0
@@ -425,6 +425,15 @@ static ssize_t acl_show(struct kobject *kobj, struct kobj_attribute *a, char *bu
  }
  BTRFS_ATTR(static_feature, acl, acl_show);
 
+static ssize_t temp_fsid_supported_show(struct kobject *kobj,
+					struct kobj_attribute *a, char *buf)
+{
+	int ret = 0;
+
+	return sysfs_emit(buf, "%d\n", ret);

This can be

Quoted 1 line
	return sysfs_emit(buf, "0\n");

Changes in the misc-next looks better.

Thanks, Anand

author
#7 lore raw

Re: [PATCH 4/4] btrfs: show temp_fsid feature in sysfs

Quoted 76 lines · Anand Jain
Quoted 35 lines · David Sterba
Quoted 34 lines · Anand Jain

This adds sysfs objects to indicate temp_fsid feature support and its status.

   /sys/fs/btrfs/features/temp_fsid
   /sys/fs/btrfs/<UUID>/temp_fsid

   For example:

      Consider two cloned and mounted devices.

    $ blkid /dev/sdc[1-2]
    /dev/sdc1: UUID="509ad44b-ad2a-4a8a-bc8d-fe69db7220d5" ..
    /dev/sdc2: UUID="509ad44b-ad2a-4a8a-bc8d-fe69db7220d5" ..

      One gets actual fsid, and the other gets the temp_fsid when mounted.

    $ btrfs filesystem show -m
    Label: none  uuid: 509ad44b-ad2a-4a8a-bc8d-fe69db7220d5
        Total devices 1 FS bytes used 54.14MiB
        devid    1 size 300.00MiB used 144.00MiB path /dev/sdc1

    Label: none  uuid: 33bad74e-c91b-43a5-aef8-b3cab97ae63a Total devices 1 FS bytes used 54.14MiB
        devid    1 size 300.00MiB used 144.00MiB path /dev/sdc2

      Their sysfs as below.

    $ cat /sys/fs/btrfs/features/temp_fsid
    0
    $ cat /sys/fs/btrfs/509ad44b-ad2a-4a8a-bc8d-fe69db7220d5/temp_fsid
    0
Quoted 8 lines
Quoted 2 lines
    $ cat /sys/fs/btrfs/33bad74e-c91b-43a5-aef8-b3cab97ae63a/temp_fsid
    1

So the fsid used for the directory is always the new one, is there a way to read which is the original filesystem's fsid? In this case it would be the 509ad44b-... We could print it in that file instead of '1', though it could be confusing that it's not the temp_fsid but the original one, file name mismatches the contents on first look.

Instead, can we emit 'fsid' in another kobject altogether? Furthermore, we also have a 'metadata_uuid' kobject. Here is how they relate.

1. normally:

  $ cat /sys/fs/btrfs/<meta-fsid>/fsid
  <meta-fsid>
  $ cat /sys/fs/btrfs/<meta-fsid>/metadata_uuid
  <meta-fsid>

2. metadata-uuid flag is set:

  $ cat /sys/fs/btrfs/<sb-fsid>/fsid
  <sb-fsid>
  $ cat /sys/fs/btrfs/<sb-fsid>/metadata_uuid
  <meta-fsid>

3. normal + temp-fsid:

  $ cat /sys/fs/btrfs/<temp-fsid>/fsid
  <meta-fsid>
  $ cat /sys/fs/btrfs/<temp-fsid>/metadata_uuid
  <meta-fsid>
Quoted 6 lines

4. metadata-uuid flag is set + temp-fsid:

  $ cat /sys/fs/btrfs/<temp-fsid>/fsid
  <sb-fsid>
  $ cat /sys/fs/btrfs/<temp-fsid>/metadata_uuid
  <meta-fsid>

Let's consider this thought:

The proposal above implies that /sys/fs/btrfs/<temp-fsid>/temp_fsid will continue to function as described in this patch. Otherwise, determining whether temp_fsid is enabled becomes challenging.

As part of the temp-fsid feature, we plan to introduce two additional kobjects. However, if this results in an excessive number of kobjects, then...

We have the option to rename /sys/fs/btrfs/<temp-fsid>/temp_fsid to /sys/fs/btrfs/<temp-fsid>/temp-fsid_enabled_super_fsid to display the actual sb::fsid.

Thanks, Anand

author
#9 lore raw
In reply to David Sterba
Quoted 13 lines · David Sterba
Quoted 11 lines · Anand Jain

Seed and device-add are the two features that must be unsupported if a cloned device is using temp-fsid to mount, as they conflict with multi-device functionality.

Additionally, add sysfs files for the temp-fsid feature.

Anand Jain (4):
btrfs: comment for temp-fsid, fsid, and metadata_uuid
btrfs: disable seed feature for temp-fsid
btrfs: disable the device add feature for temp-fsid
btrfs: show temp_fsid feature in sysfs

Quoted 4 lines

Added to misc-next, thanks. The eventual change to the sysfs file can be done later.

How are we going to proceed with the patch from Guilherme?

The last step is to ensure that the temp-fsid feature is restricted with the temp-fsid superblock flag. Guilherme's patches (kernel, mkfs, and tune) already handle it but need a rebase. Can Guilherme send an RFC patch for feedback from others and copy suggested-by. Because, I haven't found a compelling reason for the restriction, except to improve the user experience.

His fstests patch will be accepted. And progressively we should add more coverage when the fstests configuration does not include the temp-fsid environment.

Lastly, in fact even for this patchset:

Thanks, Anand

#10 lore raw
Quoted 14 lines · Anand Jain

[...]

Quoted 1 line

How are we going to proceed with the patch from Guilherme?

The last step is to ensure that the temp-fsid feature is restricted with the temp-fsid superblock flag. Guilherme's patches (kernel, mkfs, and tune) already handle it but need a rebase. Can Guilherme send an RFC patch for feedback from others and copy suggested-by. Because, I haven't found a compelling reason for the restriction, except to improve the user experience.

His fstests patch will be accepted. And progressively we should add more coverage when the fstests configuration does not include the temp-fsid environment.

Quoted 6 lines

Lastly, in fact even for this patchset:

Thanks, Anand

Thanks Anand and David. The first thing I need to do, is to build misc-next and test it if if works with my setup, to double-check Anand's approach fits the use-case (it seems to, so far). Then, I guess we'll need to see if there's a missing piece on that - in the other thread Anand mentioned maybe the superblock flag would be useful after all, so if that's necessary, I can send it of course.

Finally, this week I'm away from my regular system and cannot provide the test results, next week I can do that for sure.

Thank you both for looping me in and for the details and clarifications! Cheers,

Guilherme

author
#11 lore raw
Quoted 7 lines
Quoted 7 lines

Can Guilherme send an RFC patch for feedback from others and copy suggested-by. Because, I haven't found a compelling reason for the restriction, except to improve the user experience.

My comments about the superblock flag are above.

User experiences are subjective, so we need others to comment; an RFC will help.

Quoted 11 lines

Thanks Anand and David. The first thing I need to do, is to build misc-next and test it if if works with my setup, to double-check Anand's approach fits the use-case (it seems to, so far). Then, I guess we'll need to see if there's a missing piece on that - in the other thread Anand mentioned maybe the superblock flag would be useful after all, so if that's necessary, I can send it of course.

Quoted 1 line

Finally, this week I'm away from my regular system and cannot provide

the test results,

next week I can do that for sure.

That's great; much appreciated.

Thanks, Anand

#12 lore raw
Quoted 8 lines · Anand Jain
Quoted 3 lines
Quoted 3 lines

Can Guilherme send an RFC patch for feedback from others and copy suggested-by. Because, I haven't found a compelling reason for the restriction, except to improve the user experience.

My comments about the superblock flag are above.

User experiences are subjective, so we need others to comment; an RFC will help.

A few things changed, the incompat bit was supposed to prevent accidentally duplicated fsids but with your recent changes this is safe. This would need to let Guilherme check if the A/B use case still works but this seems to be so as I'm reading the changelog.

In a controlled environment the incompat bit will not bring much value other than yet another sanity check preventing some user error, but related only to the multiple devices.

author
#13 lore raw
Quoted 17 lines · David Sterba
Quoted 8 lines · Anand Jain
Quoted 3 lines
Quoted 3 lines

Can Guilherme send an RFC patch for feedback from others and copy suggested-by. Because, I haven't found a compelling reason for the restriction, except to improve the user experience.

My comments about the superblock flag are above.

User experiences are subjective, so we need others to comment; an RFC will help.

A few things changed, the incompat bit was supposed to prevent accidentally duplicated fsids but with your recent changes this is safe. This would need to let Guilherme check if the A/B use case still works but this seems to be so as I'm reading the changelog.

Quoted 3 lines

In a controlled environment the incompat bit will not bring much value other than yet another sanity check preventing some user error, > but related only to the multiple devices.

Agreed. Anyway, the kernel would continue to fail the mount of the duplicate-fsid for a multi-device filesystem.

And

   $ mkfs.btrfs -U <duplicate-fsid> /dev/sda1 /dev/sda2 ..

for a multi-device filesystem will also fail.

Therefore, the only avenue for the user to make a mistake is by using dd to copy for a multi-device setup.

Thanks, Anand

#14 lore raw
In reply to David Sterba
Quoted 10 lines · David Sterba
Quoted 8 lines · Anand Jain
Quoted 3 lines
Quoted 3 lines

Can Guilherme send an RFC patch for feedback from others and copy suggested-by. Because, I haven't found a compelling reason for the restriction, except to improve the user experience.

My comments about the superblock flag are above.

User experiences are subjective, so we need others to comment; an RFC will help.

Quoted 8 lines

A few things changed, the incompat bit was supposed to prevent accidentally duplicated fsids but with your recent changes this is safe. This would need to let Guilherme check if the A/B use case still works but this seems to be so as I'm reading the changelog.

In a controlled environment the incompat bit will not bring much value other than yet another sanity check preventing some user error, but related only to the multiple devices.

Hi David and Anand, I've manage to test misc-next of today, that includes both this patchset as well as the "support cloned-device mount capability​" one.

It seems to be working fine for our use case, though I'll test a bit more on Deck. I was able to mount the same filesystem (spread in 2 nvme devices) at the same time, in any order...the second one always get the temp-fsid. Tested also re-mounting the devices on other locations, and it seems all consistent, with no error observed.

I also question the value of the incompat flag, not seeing much use for that..looping Qu Wenruo as they first suggested this flag-based approach, in case there is some more feedback...

Anyway, thanks for your improved approach Anand and to David: is it expected to land on 6.7?
Cheers,

Guilherme

#15 lore raw
Quoted 24 lines · Guilherme G. Piccoli
Quoted 10 lines · David Sterba
Quoted 8 lines · Anand Jain
Quoted 3 lines
Quoted 3 lines

Can Guilherme send an RFC patch for feedback from others and copy suggested-by. Because, I haven't found a compelling reason for the restriction, except to improve the user experience.

My comments about the superblock flag are above.

User experiences are subjective, so we need others to comment; an RFC will help.

Quoted 8 lines

A few things changed, the incompat bit was supposed to prevent accidentally duplicated fsids but with your recent changes this is safe. This would need to let Guilherme check if the A/B use case still works but this seems to be so as I'm reading the changelog.

In a controlled environment the incompat bit will not bring much value other than yet another sanity check preventing some user error, but related only to the multiple devices.

Hi David and Anand, I've manage to test misc-next of today, that includes both this patchset as well as the "support cloned-device mount capability​" one.

Quoted 5 lines

It seems to be working fine for our use case, though I'll test a bit more on Deck. I was able to mount the same filesystem (spread in 2 nvme devices) at the same time, in any order...the second one always get the temp-fsid. Tested also re-mounting the devices on other locations, and it seems all consistent, with no error observed.

Great, thanks.

Quoted 3 lines

I also question the value of the incompat flag, not seeing much use for that..looping Qu Wenruo as they first suggested this flag-based approach, in case there is some more feedback...

Yeah at this point I don't see the need for the incompat bit, which is the better outcome.

Quoted 2 lines

Anyway, thanks for your improved approach Anand and to David: is it expected to land on 6.7?

Yes, what's in misc-next is queued for 6.7, also we have the whole development cycle to fix remaining bugs.

#16 lore raw
Quoted 6 lines · David Sterba

[...]

Quoted 2 lines

Anyway, thanks for your improved approach Anand and to David: is it expected to land on 6.7?

Yes, what's in misc-next is queued for 6.7, also we have the whole development cycle to fix remaining bugs.

Thanks a bunch for confirming and all guidance through this feature process.