[LU-9767] FS name checking got removed

• Type: Bug
• Resolution: Fixed
• Priority: Critical
• Fix Version/s: Lustre 2.10.1, Lustre 2.11.0
• Affects Version/s: Lustre 2.10.1, Lustre 2.11.0
• Severity: 3
• Assignee: James A Simmons
• Reporter: Ben Evans (Inactive)

Description

https://review.whamcloud.com/#/c/24325/ removed name length and content checking for the fsname.

You can now format a FS with up to 16 characters, but can only reference 8 of them in various places, 16 in others.

Issue Links

• is related to: LU-6401 Untangle lustre userland and kernel headers

Comments

[Gerrit Updater]

John L. Hammond ([e-mail]) merged in patch https://review.whamcloud.com/28996/
Subject: LU-9767 utils: validate filesystem name for mkfs.lustre
Project: fs/lustre-release
Branch: b2_10
Current Patch Set:
Commit: ba56efcfdd53f3066d95bb2274f6f9cca9e34c0b

[Gerrit Updater]

James Simmons ([e-mail]) uploaded a new patch: https://review.whamcloud.com/28996
Subject: LU-9767 utils: validate filesystem name for mkfs.lustre
Project: fs/lustre-release
Branch: b2_10
Current Patch Set: 1
Commit: 51deb7c46a0509732a9274bfc87a0d762f40114b

[Peter Jones]

Landed for 2.11

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/28070/
Subject: LU-9767 utils: validate filesystem name for mkfs.lustre
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 28216a73f4c0b19ef45e3ec2d6949801f6b9f828

[Gerrit Updater]

James Simmons ([e-mail]) uploaded a new patch: https://review.whamcloud.com/28070
Subject: LU-9767 utils: validate filesystem name for mkfs.lustre
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 654f635ed795d6eed28a37585dbfaec7f42b6342

[Andreas Dilger]

For mkfs.lustre, the problem is that there is no way for the kernel to verify the fsname because the kernel isn't even involved until the filesystem is mounted.

My recommendation would be make a quick patch to add checks into mount_utils_ldiskfs.c and mount_utils_zfs.c that the fsname is <= 8 characters, and do not include a '.' or other illegal character, and then separately we can examine whether it is safe to have a longer fsname where that works (e.g. ZFS where it isn't limited by the size of s_volume_name in the superblock) and make a patch to allow ZFS to have a longer fsname. This will need a code review of where fsname is used on disk and over the network, to verify there aren't explicit or implicit limitations based on the length.

[Ben Evans (Inactive)]

Well, you get some weird looking bugs from it:

tunefs.lustre:
Permanent disk data:
Target: asdfasdf-MDT0000
Index: 0
Lustre FS: asdfasdfasdfasd
Mount type: ldiskfs
Flags: 0x5
(MDT MGS )
Persistent mount opts: user_xattr,errors=remount-ro
Parameters:

so the fsname got accepted, but is truncated, except where it's not. I haven't fully sorted out what tools would refer to it by its full name, rather than the truncated one. mkfs.lustre happily accepts it, throwing no errors (up to 16 chars). I'd be curious what happens if you give it a raft of control characters. Regardless, this is a UI error, as the user should be informed of the problem in some manner.

[James A Simmons]

John Hammond asked for this. User land can scribble what ever it wants but kernel space should be rejecting the invalid file system name. The same goes for pool names. Is this not the case?