If you are copying files to a freshly-formatted MS-DOS floppy (6.2.2)
and on a second process you do a ls -R of the same floppy, sometimes
the ls process will hang, and then the cp process may also hang.

Other types of filesystem reads will substitute for ls and probably
other programs can be used to put files on the floppy.

Fix: 

No idea.  However, the order of directories copied seems important.
When I used subdirectories named "one", "two" and "three" I could not
get a failure.   Another interesting thing is the order in which
the repeated ls commands show files being copied, almost a reverse
alphabetic order.

*END*
How-To-Repeat: 
To easily repeat the problem:
Take a freshly formatted MS-DOS floppy and mount it as /mnt:
mount -t msdos /dev/fd0a /mnt
cd /
mkdir test2
cd test2
create 26 files named a thru z of non-zero length.
I used   date > a
thru	date > z
(the script to automate this is left to the observer)
mkdir z1
mkdir z2
mkdir z3
cp * z1/.
cp * z2/.
cp * z3/.
sync

You should have 104 files and three directories in the test2 tree..
At this point, I suggest making a tar image of test2 so you can restore
it if you lose it in the next steps.

Make sure you are logged-in on at least three virtual terminals

cd to /test2 on all virtual terminals

On screen 1, issue the command:
	cp -R * /mnt/. 

On screen 2, issue the command
	ls -R /mnt
continue to repeat (!!) the ls command, waiting for the prompt each time.
Eventually the ls command will not terminate.  

At this point you may hear floppy I/O continue for another 15
seconds or so, and then the drive light will go off.  Switch to screen 1
and verify that the cp command is hung.

If the cp command finished normally and ls didn't hang up, 
rm -rf /mnt/. and repeat the above commands.  For me it always hung
during the first cp command.

On the third screen, you can do a ps -alxw and should see results like this

  UID   PID  PPID CPU PRI NI   VSZ  RSS WCHAN  STAT TT       TIME COMMAND
    0     0     0   0 -18  0     0    0 sched  DLs  ??    0:00.00  (swapper)
    0     1     0   2  10  0   412  176 wait   Is   ??    0:00.05 /sbin/init --
    0     2     0   0 -18  0     0   12 psleep DL   ??    0:00.00  (pagedaemon)
    0     3     0   0  28  0     0   12 psleep DL   ??    0:00.00  (vmdaemon)
    0     4     0   0  -6  0     0   12 update DL   ??    0:00.12  (update)
    0    19     1  23  18  0   212   88 pause  Is   ??    0:00.02 adjkerntz -i
    0    41     1   0   2  0   180  228 select Is   ??    0:00.03 routed -q
    0    57     1   0   2  0   180  296 select Ss   ??    0:00.43 syslogd
    1    63     1  75   2  0   176  220 select Is   ??    0:00.02 portmap
    0    72     1  18   2  0   208  312 select Is   ??    0:00.84 inetd
    0    79     1   0  18  0   264  332 pause  Is   ??    0:00.09 cron
    0    81     1  25   2  0   184  284 select Is   ??    0:00.13 lpd
    0    84     1  22   2  0   404  328 netcon Is   ??    0:00.04
	sendmail: accepting connections (sendmail)
    0   124     1   0   3  0   476  276 ttyin  Ss+  v0    0:01.08 -csh (csh)
    0   192   124   0  -6  0   252  132 getblk D    v0    0:00.30
	cp -R a b c d e f g h i j k l m n o p q r s t u v w x y z z1 z2 z
    0   125     1   0  18  0   472  336 pause  Is   v1    0:00.59 -csh (csh)
    0   212   125   0 -14  0   352  196 msdhgt D+   v1    0:00.07
	ls -g -k -R /mnt
    0   126     1   0  18  0   472  332 pause  Ss   v2    0:00.40 -csh (csh)
    0   216   126   6  29  0   440  180 -      R+   v2    0:00.04 ps -alxw

(The wait channel of the ls and cp is always the same as above.)

At this point, you can't kill the above processes, but you might be able
to do selected operations on the same floppy file system.  In one case I was
able to do a ls /mnt and got results and a prompt, but then I attempted a
ls -alR and that hung up.   Another time I was able to rm -rf /mnt/* 
all the files on the floppy without incident even though the other
processes were still hung.