Someone explain badblocks discrepancy?

[Mike Sabin]

I have an ordinary 1.44 MB floppy that I am trying to check with badblocks (stock Redhat 7.1 w/ all updates on i586; docs for e2fsprogs in /usr/share/doc indicate e2fsprogs-1.19).  I am getting inconsistent results when I run badblocks with the -w flag (shows no bad blocks), and without (shows plenty of bad blocks).  Can someone explain this to me?  I would have thought that -w would catch the same or more bad blocks than the read-only version, but never identify _fewer_ bad blocks.  Here is the output of my bash session:

[mike@localhost]$ su
Password:
[root@localhost]# /sbin/badblocks /dev/fd0
2
11
20
29
38
47
128
137
146
164
[root@localhost]# /sbin/badblocks -w /dev/fd0
[root@localhost]# /sbin/badblocks /dev/fd0
2
11
20
29
47
128
137
146
164
[root@localhost]# /sbin/badblocks -svw /dev/fd0
Checking for bad blocks in read-write mode
From block 0 to 1440
Writing pattern 0xaaaaaaaa: done                       
Reading and comparing: done                       
Writing pattern 0x55555555: done                       
Reading and comparing: done                       
Writing pattern 0xffffffff: done                       
Reading and comparing: done                       
Writing pattern 0x00000000: done                       
Reading and comparing: done                       
Pass completed, 0 bad blocks found.
[root@localhost]# exit

[cira]

i know this i old, but we might as well record the answer if anyone wants to know.

To the best of my understanding, floppy disks could sometimes be recovered using a write test as the data was refreshed. In this case, the data was difficult to read due to something such as bit drift, or perhaps the data was written using a different drive & the tracks were not aligned exactly the same. In your case the medium was still good, only the strength of the magnetic field in those blocks was weak. Rewriting those blocks restored the magnetic field.