The problem fixed in https://svnweb.freebsd.org/base?view=revision&revision=323157 was never merged to stable/11 and today I've faced with inability of newfs to work on GELI volumes with 8K sector size:
# diskinfo -v /dev/da1.eli.journal
8192 # sectorsize
1998251433984 # mediasize in bytes (1.8T)
243927177 # mediasize in sectors
0 # stripesize
0 # stripeoffset
No # TRIM/UNMAP support
Unknown # Rotation rate in RPM
# newfs -b 65536 -J /dev/da1.eli.journal
/dev/da1.eli.journal: 1905681.1MB (3902834832 sectors) block size 65536, fragment size 8192
using 759 cylinder groups of 2512.19MB, 40195 blks, 161024 inodes.
super-block backups (for fsck_ffs -b #) at:
256, 5145216, 10290176, 15435136, 20580096, 25725056, 30870016, 36014976, 41159936, 46304896,
3848430336, 3853575296, 3858720256, 3863865216, 3869010176, 3874155136, 3879300096, 3884445056,
3889590016, 3894734976, 3899879936
newfs: can't read recovery area: read error from block device: Invalid argument
Please consider merging r323157 to stable/11 before 11.4-RELEASE.
Forgot to note that I used command "svn diff -c323157" to prepare patch-file and it applied to stable/11 sources cleanly.
Then I've rebuilt /sbin/newfs binary and problem's gone.
I have no objection to your doing an MFC of 323157 to stable-11.
A commit references this bug:
Date: Mon Jan 20 08:28:55 UTC 2020
New revision: 356905
MFC r323157 by 323157: fix recovery information with sector sizes up to 64K.
Original commit log:
The new fsck recovery information to enable it to find backup
superblocks created in revision 322297 only works on disks
with sector sizes up to 4K. This update allows the recovery
information to be created by newfs and used by fsck on disks
with sector sizes up to 64K. Note that FFS currently limits
filesystem to be mounted from disks with up to 8K sectors.
Expanding this limitation will be the subject of another
For example, this allows newfs to work on GELI volumes with 8K sectors.
Approved by: mckusick