>Number:         4648
>Category:       port-mac68k
>Synopsis:       zip disk has bad data whith ncrsci
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    gnats-admin (GNATS administrator)
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Dec  7 08:20:01 1997
>Last-Modified:
>Originator:     Dan McMahill
>Organization:
	
>Release:        <NetBSD-current source date> 1.3 Beta 11/30 and 12/3/97
>Environment:
	
System: NetBSD dinah-moe.mit.edu 1.3_BETA NetBSD 1.3_BETA (SLOTMAN) #162: Wed Dec 3 18:36:18 PST 1997 paul@mac1.whooppee.com:/home/paul/src/sys/arch/mac68k/compile/SLOTMAN (current as of Tue Dec 2 06:35:07 1997) mac68k


>Description:
	
	After formatting, partitioning, and newfs'ing a zip disk,
	I see the 2 following disturbing things:

	1)  fsck -f /dev/sd2g   (sd2g is the NetBSD 4.2BSD filesystem)
	    dinah-moe-dan (103)-% su
	    Password:
	    dinah-moe# fsck -f /dev/sd2g
	    ** /dev/rsd2g
	    ** File system is already clean
	    ** Last Mounted on /zip
	    ** Phase 1 - Check Blocks and Sizes

	    CANNOT READ: BLK 33312
	    CONTINUE? [yn] y

	    THE FOLLOWING DISK SECTORS COULD NOT BE READ:

	    CANNOT READ: BLK 33424
	    CONTINUE? [yn] y

	    THE FOLLOWING DISK SECTORS COULD NOT BE READ:

	    CANNOT READ: BLK 33648
	    CONTINUE? [yn] y

	    THE FOLLOWING DISK SECTORS COULD NOT BE READ:

	    CANNOT READ: BLK 98800
	    CONTINUE? [yn] y

	    THE FOLLOWING DISK SECTORS COULD NOT BE READ:
	    ** Phase 2 - Check Pathnames
	    ** Phase 3 - Check Connectivity
	    ** Phase 4 - Check Reference Counts
	    ** Phase 5 - Check Cyl groups
	    1 files, 1 used, 95165 free (21 frags, 11893 blocks, 0.0% fragmentation)
	    dinah-moe# 
            
	    next, I repeat (without doing _anything_ else)

	    dinah-moe# !!
	    fsck -f /dev/sd2g
	    ** /dev/rsd2g
	    ** File system is already clean
	    ** Last Mounted on /zip
	    ** Phase 1 - Check Blocks and Sizes

	    CANNOT READ: BLK 131824
	    CONTINUE? [yn] y

	    THE FOLLOWING DISK SECTORS COULD NOT BE READ:
	    ** Phase 2 - Check Pathnames
	    ** Phase 3 - Check Connectivity
	    ** Phase 4 - Check Reference Counts
	    ** Phase 5 - Check Cyl groups
	    1 files, 1 used, 95165 free (21 frags, 11893 blocks, 0.0% fragmentation)

	    Note that I get different blocks which can't be read.

	    2)  The other readily observed problem which is more disturbing is:

	    dinah-moe# mount /dev/sd2g /zip
	    dinah-moe# cd /usr/dan
	    dinah-moe# cksum zip-HOWTO
	    2792940339 13249 zip-HOWTO
	    dinah-moe# cp zip-HOWTO /zip
	    dinah-moe# cd /zip
	    dinah-moe# cksum zip-HOWTO
	    2792940339 13249 zip-HOWTO
	    dinah-moe# 

	    note cksum matches (but zip-HOWTO is small)

	    now I try with a larger file:

	    dinah-moe# cd /usr/dan/src
	    dinah-moe# cksum scilab-2.2-src.tar.gz 
	    777269326 3981084 scilab-2.2-src.tar.gz
	    dinah-moe# cp scilab-2.2-src.tar.gz /zip
	    dinah-moe# cd /zip
	    dinah-moe# cksum scilab-2.2-src.tar.gz 
	    1818919436 3981084 scilab-2.2-src.tar.gz
	    dinah-moe# cksum scilab-2.2-src.tar.gz 
	    4012833890 3981084 scilab-2.2-src.tar.gz
	    dinah-moe# 

	    Now the cksum doesn't match and in fact I get a different answer if I
	    try it again on the zip.


>How-To-Repeat:
	
	This can be repeated by following the instructions in the problem 
	description section.
>Fix:
	
	The SBC based kernel from a few days ago didn't boot on my machine, but
	I've heard that a bug has been identified (and hopefully stamped out).
	It is rumored that SBC may work.

	My current "workaround" really isn't much of one.  I mostly want to use
	the zip for backups so I use the MacOs installer to copy the files out
	of the NetBSD partition and save on the zip as a HFS disk, not really
	a real workaround.

>Audit-Trail:
>Unformatted: