I tried to see why the error occurred with modifying fsck/setup.c.

Following data were obtained. Numbers are all hexadecimal.

# ./fsck -d
** /dev/rsd1a (NO WRITE)
&sblock 2e000, &altsblock 31000, sblock.fs_sbsize, 800
superblock mismatches
offset 0, original 0, alternate 0
offset 1, original 0, alternate 0
offset 2, original 10, alternate 5b5b5b5b
offset 3, original 18, alternate 5b5b5b5b
offset 4, original 20, alternate 5b5b5b5b
offset 5, original 1d0, alternate 5b5b5b5b
offset 6, original 30, alternate 5b5b5b5b
offset 7, original ffffffe0, alternate 5b5b5b5b
offset 8, original 2f5e14db, alternate 2f5e14db
offset 9, original 1d9a20, alternate 5b5b5b5b
offset a, original 1cb30d, alternate 5b5b5b5b
offset b, original 84, alternate 5b5b5b5b
offset c, original 2000, alternate 5b5b5b5b
offset d, original 400, alternate 5b5b5b5b
offset e, original 8, alternate 5b5b5b5b
offset f, original a, alternate a
offset 10, original 4, alternate 4
offset 11, original 3c, alternate 5b5b5b5b
offset 12, original ffffe000, alternate 5b5b5b5b
offset 13, original fffffc00, alternate 5b5b5b5b
offset 14, original d, alternate 5b5b5b5b
offset 15, original a, alternate 5b5b5b5b
offset 16, original 1, alternate 1
offset 17, original 800, alternate 800
offset 18, original 3, alternate 5b5b5b5b
offset 19, original 1, alternate 1
offset 1a, original 800, alternate 5b5b5b5b
offset 1b, original fffffe00, alternate 5b5b5b5b

Thus the original superblocks are different from the alternate ones.
Does it help finding the true root of the problem?

Yoshihisa Sugimoto / sugimoto@sums.shiga-med.ac.jp
The First Department of Medicine  / The Medical Information Center             
Shiga University of Medical Science, Ohtsu, Shiga, Japan