Subject: 1024 Byte Blocks
To: None <firstname.lastname@example.org>
From: Mark Gooderum <email@example.com>
Date: 09/12/1994 11:00:03
Well...I've gotten my disklabel patches for 1024 byte blocks working and
have tested that you can still label a 512 byte block disk...however,
FFS is *not* perfect on these disks...
If you write a file to a FFS filesystem on a 1024 byte sector disk,
the file gets the right size, and the filesystem structure is okay, but
the contents are off. The mismatch is usually at byte 8193, sometimes
elsewhere, but always starts at the beginning of a new cluster other
than the first in a file.
Micheal Hitch indicated that he had no FFS problems...and the code seems
to handle things okay for him, but this was the Amiga port. Endianess?
There are some comments in the fs.h header by
some of the block to cluster to fragment/etc macros that indicate that
they assume DEV_BSIZE blocks, but my best examination seems to indicate
that they behave correctly. I'll be adding some debugging/sanity check
code the FFS stuff and see what I get...
One suggestion was reformatting the disk(s) to 512 byte sectors, possibly using
the BIOS on the controller, which at this point I'm willing to try, but
it doesn't work...the disk still comes back 1024 byte sectors. This was
tried with both the Adaptec EZ-SCSI software and the 1542CF BIOS.