Skip to contentSkip to conversation

btrfs-progs: btrfstune -M|m drop test_uuid_unique

PATCH RESEND In review Anand Jain 2 participants 3 messages Oct 10, 2019 – Oct 17, 2019 Btrfs file system

lore
Share & tools
lore

Anand Jain

author
#0 lore raw

PATCH RESEND 1/1 btrfs-progs: btrfstune -M|m drop test_uuid_unique

It's common to copy/snapshot an OS image to run another instance of the OS. A duplicate fsid can't be mounted on the same system unless the fsid is changed by using btrfstune -m.

However in some circumstances the image needs to go back to the original fsid /metadata_uuid.

As of now btrfstune -M fails if the specified uuid isn't unique, as show below.

btrfstune -M $(btrfs in dump-super ./2-2g.img | grep metadata_uuid | \
					awk '{print $2}') ./2-2g.img

ERROR: fsid 87f8d9c5-a8b7-438e-a890-17bbe11c95e5 is not unique

But as we are changing the fsid of an unmounted image, so its ok to leave it to the users choice if the fsid is not unique, so that the image can be sent back the system where it was used with that fsid.

So this patch drops the check test_uuid_unique() in btrfstune -M|m. Thanks.

---
Previous sent patch and its discussions:
https://patchwork.kernel.org/patch/11134157/

 btrfstune.c | 4 ----
 1 file changed, 4 deletions(-)
btrfstune.c +0 -4
index afa3aae35412..4befcadef8b1 100644
@@ -570,10 +570,6 @@ int BOX_MAIN(btrfstune)(int argc, char *argv[])
 			error("could not parse UUID: %s", new_fsid_str);
 			return 1;
 		}
-		if (!test_uuid_unique(new_fsid_str)) {
-			error("fsid %s is not unique", new_fsid_str);
-			return 1;
-		}
 	}
 
 	fd = open(device, O_RDWR);
Signature · 1 line
2.21.0 (Apple Git-120)
author
#1 lore raw

David,

Ping (4th time).

Per general rule of thumb tools like btrfs-progs must be flexible and kernel should be harden for the duplicate fsid.

This patch relaxes the unnecessary check in btrfstune -M.

Without this patch there is no way to undo a changed fsid on the system.

Per original author there is no reasonable explanation why this check is there in the first place until now.

IMO this patch is good integrate.

Please let me know if otherwise.

Thanks, Anand

Quoted 45 lines · Anand Jain

It's common to copy/snapshot an OS image to run another instance of the OS. A duplicate fsid can't be mounted on the same system unless the fsid is changed by using btrfstune -m.

However in some circumstances the image needs to go back to the original fsid /metadata_uuid.

As of now btrfstune -M fails if the specified uuid isn't unique, as show below.

btrfstune -M $(btrfs in dump-super ./2-2g.img | grep metadata_uuid | \
					awk '{print $2}') ./2-2g.img

ERROR: fsid 87f8d9c5-a8b7-438e-a890-17bbe11c95e5 is not unique

But as we are changing the fsid of an unmounted image, so its ok to leave it to the users choice if the fsid is not unique, so that the image can be sent back the system where it was used with that fsid.

So this patch drops the check test_uuid_unique() in btrfstune -M|m. Thanks.

---
Previous sent patch and its discussions:
https://patchwork.kernel.org/patch/11134157/

  btrfstune.c | 4 ----
  1 file changed, 4 deletions(-)
btrfstune.c +0 -4
index afa3aae35412..4befcadef8b1 100644
@@ -570,10 +570,6 @@ int BOX_MAIN(btrfstune)(int argc, char *argv[])
  			error("could not parse UUID: %s", new_fsid_str);
  			return 1;
  		}
-		if (!test_uuid_unique(new_fsid_str)) {
-			error("fsid %s is not unique", new_fsid_str);
-			return 1;
-		}
  	}
 
  	fd = open(device, O_RDWR);
#2 lore raw
Quoted 1 line · Anand Jain

Ping (4th time).

Anand, the patch queue is big and ever growing, pinging can help to bring back the attention but for patches that have disgreements or unclear reasoning the turnaround time can be longer.

Specifically, patches that have impact on usability extra care must be taken not to cause chaos by seemingly "simple" change.

Quoted 1 line

IMO this patch is good integrate.

I'll reply under the previous submission as there is the discussion in progress.