I feel silly that I'm getting my butt kicked by some auto-mounting USB dingus, but here I am.

UPDATE: Scroll down to [automount] if you don't care for the story of bungling which got me to this unfortunate pass. I actually do have two questions, at the bottom.

[background]
I'm running Suse 9.2 on a dual-booting hp pavilion ze4900 (aka el cheapo portable), with some kind of celeron and a busted touchpad. Suse runs happily, and until recently, so did Windows XP. The two co-existed on a 40GB hdd until some Bad Thing happened to the NTFS partition (hda1, 28GB).

[windowsXP]
I started getting lots of iPod error -69, which is connected to a full hdd. About the same time, I started getting XP's DISK CLEANER dialog boxes (the "let's delete some of your old cr4p" wizard), and soldiered along, figuring that I would really have to do something about that someday. Bam. Disk Full.
I still don't know what happened next, or perhaps more accurately, what had already happened and was just beginning to surface. Even after clearing some disk space, I had ridiculous slowdowns under XP and outright halts. Sometimes it had the grace to BSOD, and sometimes it just froze. HARD. Wouldn't even monitor the power button. Had to unplug & eject battery to cycle the power.
Somewhere along the way, I got a bunch of corruption, so there went a lot of files. Also, I think that the list of Bad Sectors got hashed, because only a chkdsk -r (after many chkdsk -f) started moving me in the right direction.
Eventually, I was able to SFC (or whatever the command is) which tells XP to repair or replace sytem files which are damaged or missing. Handy utility that one, as the slowdowns and halts finally stopped.

[SuSE]
Meanwhile, I had been trying in parallel to rescue files from the disk using the SuSE partition (root hda2, reiserfs, 10GB) to boot. Even SuSE had a hard time trying to access certain files on NTFS hda1. Sometimes SuSE would stall during transfers. I was really limited in what I could do because the asinine Windows name scheme meant that I could not use shell tools on the NTFS partition (due to gaping holes in my knowledge of how to do these things, but still!).
After repairing the batty operating system on hda1 and finally getting through a chkdsk -r (forces bad sector marking and replacement) with 0 items, then SuSE was able to pull files off of hda1, but still susceptible to stalling, just not so bad.

[automount]
So I went out and bought a usb hdd case to house a spare VFAT 20GB hdd I had just kicking around. Slapped it together, plugged it in, turned it on, and SuSE pops up a message asking if I would like KFMCLIENT to open this device as /media/usb-qwertyuiop-etc and I said YES. Then a window pops up from YAST (I guess) asking if I would like to configure this NEW HDD right now, and I said NO, because, WTH, it's already available as /media/asdfasdfasdf, and at any rate the less I configure myself, the better things are likely to turn out for all concerned.
I transferred a testfile to it, moved it to a different machine, modified the file there, then moved the drive back to the original machine, and read the changes. OPTEST SAT.
I deleted those files and started transferring from NTFS hda1 to /media/usb-dsfasdfasd and was pretty pleased with myself. Great, everything seems to go smoothly, and I found that if I am patient enough with a stalled transfer, KDE or Konqueror gives me a choice to CANCEL, SKIP, or AUTO-SKIP a file or files which cannot be transferred. Coolness. I said AUTO-SKIP, as I do believe that if Konqueror cannot get it, then there's probably not much left to get, a la HE'S DEAD, JIM.
So I shoveled a bunch of files onto /media/usb-sdfadfasdf, and was surprised when I got a DISK FULL. After some poking around, I was reminded of the df command, which revealed that I had only copied the files from hda1 (NTFS) to hda2 (root reiser)! So I deleted some picture folders to free up some room for a shutdown (in case it needed it), and shut down.
Moved the drive to the other machine. What do I see? The testfile from the OPTEST. No GB of anything.
It's as though the USB drive is treated like a CD to be burned, where everything exists on the local drive first, before being committed and burnt to disk. Except nothing ever commits. Fricking Write-Only Memory, that's what it is. It's "F-WOM".

[puzzlement]
Now I have mounted the usb drive by hand, at /mnt/xd (eXternal Drive), using the name given by cat /proc/partitions, which was /dev/sda1. It think it's a SCSI drive. Fine.
I'm okay with mounting by hand and all of that--in fact, I prefer it. I may be a newbie, but I have been a newbie since 1997, when I was struggling to install slackware on a hybrid 286/386/486 FrankenPuter I had cobbled together and heard that Diana had been killed in a Paris car crash--I know a good work-around when I see one.

[THE QUESTIONS]
But even though the /media/usb-asdfasdfa point reports its emptiness, df still tells me that hda2 is nearly full. Where is all of this cr4p hiding? How do I get rid of it?
Now, the transfer from NTFS hda1 to the hand-mounted /mnt/xd is going well, except for this: I get about five seconds of really active, high MB/s transfer, then twenty minutes of STALLED. Then another burst. It's not as though I'm actually running Mr. Gates' infernal code, I'm just transferring the sad remains of my files from his sulfurous partition. Anybody know why it does this?

UPDATE: My transfer rate has slowed to 52 B/s, at which rate SuSE calculates, er, politely averts its eyes declares that it will take 23:59:59 to complete. Ooh! back up to 32.0 KB/s. So I had another good burst, it seems. Very strange.

Of course, hda2 was nearly full (SuSE lives on hda2), so the process of transferring from NTFS hda1 to USB sda1 hung up not because of NTFS funkiness, but because the OS' own disk had no real room. I should make a separate swap partition after this. NOW I know what they meant!

Th USB issue remains frustrating. As my user-self, I deleted the files which I as a user had copied to /media/usb-asdfasdfa. Then, I emptied the trash. This desktop metaphor is not comfortable. When I went in as root, the files were all still in the same place! As root, I deleted the annoying /media/usb-sdfasdf and finally the damned thing stayed dead. At last I had freed up disk space on my root partition.

Deleted everything on the USB drive ITSELF (hand-mounted), because I cannot tell which of the 42,000 files may have been damaged, and re-started. This time, it chewed right through the first 8% (of 9.3GB, not bad at all), then seemed to hang. Every once in a while, there's a burst of successful activity, then 10-20 minutes of nearly frozen inactivity. I'm talking minutes from keyboard to screen.

Oh well.