Am 03.12.24 um 22:58 schrieb Theodore Ts'o:
On Tue, Dec 03, 2024 at 08:22:49PM +0100, Michael Biebl wrote:

Just to clarify: The failing test uses -r 1, not 0:

fs_ext2.c#383,extfs_mkfs(): command [mke2fs  -q  -F  -b 1024  -L 'Test' -U
31ec6c69-80a1-4ed6-9642-f2b78f798bff  -I 256  -r 1  -O 
has_journal,ext_attr,resize_inode,dir_index,^sparse_super2,^fast_commit,orphan_file,filetype,extent,^journal_dev,flex_bg,^meta_bg,^mmp,64bit,^inline_data,^ea_inode,^large_dir,large_file,huge_file,sparse_super,^uninit_bg,dir_nlink,extra_isize,^bigalloc,metadata_csum,^project
/dev/loop0] failed with return status=1

Yes, that's why I proposed option (C).  This would allow mke2fs -r 1
to succeed, but -r 0 would issue some kind of "this has been changed
to -E revision=0 but you almost certainly didn't mean to use -r 0".

This would not require any changes to fsarchiver, *except* in the case
where some fsarchiver user was trying to archive a file system which
should only be used on a 30-year old Linux kernel.  So that's why I
was asking whether fsarchiver would care about that use case --- how
do people use fsarchiver, in practice?

fsarchiver records the file system properties when you create an archive and creates the file system with the same properties as stored in the archive when you run a restore.

Assume a user has a file system that was created a long time ago using rev=0. This doesn't necessarily mean that he is using a 30-year old Linux kernel! They have made continuous backups of that file system. With this change in e2fsprogs, creating new backups of that file system will now fail and even worse, they can no longer restore any existing archives, which I would consider pretty bad.

Now, I can't tell you how many users have such file systems with rev=0 (my assumption would be that those are not widespread). That said, I would prefer not breaking this use case knowingly i.e. be conservative especially when it concerns backups.

Patching fsarchiver to use -E revision=0 (or rather try both as I can't assume I have a new e2fsprogs) could be an option if someone provided a patch. Are you willing to work on such a patch?

Michael

Attachment: OpenPGP_signature.asc
Description: OpenPGP digital signature

Reply via email to