Description of problem:
I bought a USB 2.0 card (Adaptec AUA-2000 -- this is a PCI card w/ two USB
ports) and media reader/writer (SanDisk 5-in-1 Imagemate) so I could fill my SD
cards in a reasonable amount of time.  I am copying files from my external USB
hard drive (it's an IDE drive in a USB enclosure).  About the fourth time I
copied a directory of .oggs the write hung.  Killing the process that is doing
the copy has no effect -- kill -9 does not remove it, it is just completely
hung.  System can't even be rebooted to get rid of it, as drives will not
umount; I have to power cycle (would appreciate any advice on avoiding this).

The problem does not seem to manifest when copying data to an intermediary
directory, nor if the write device is on a slower USB port; owing to the
annoying nature of reproducing this (i.e. requiring reboot) I haven't had
abundant opportunities to explore this, but it seems to manifest ONLY when both
devices are on the fast USB ports.

Version-Release number of selected component (if applicable):
# uname -a
Linux desktop.sosiouxme.net 2.6.10-1.760_FC3 #1 Wed Feb 2 00:14:23 EST 2005 i686
athlon i386 GNU/Linux


How reproducible:
Perhaps 1 in 3 times I try to copy a directory of ~50 MB.

Steps to Reproduce:
1.  Hook up two high-speed USB 2.0 devices to card ports as described.
2.  Copy ~50MB data from one to the other.  Try a few times.
  
Actual results:
Write hangs, both devices "busy", write process(es) cannot be halted.

Expected results:
Data copied, devices free to unmount.

Additional info:
# usbtree
/:  Bus 05.Port 1: Dev 1, Class=root_hub, Driver=ohci_hcd/2p, 12M
/:  Bus 04.Port 1: Dev 1, Class=root_hub, Driver=ohci_hcd/3p, 12M
/:  Bus 03.Port 1: Dev 1, Class=root_hub, Driver=ehci_hcd/5p, 480M
    |__ Port 3: Dev 5, If 0, Class=stor., Driver=usb-storage, 480M
    |__ Port 4: Dev 6, If 0, Class=stor., Driver=usb-storage, 480M
/:  Bus 02.Port 1: Dev 1, Class=root_hub, Driver=uhci_hcd/2p, 12M
/:  Bus 01.Port 1: Dev 1, Class=root_hub, Driver=uhci_hcd/2p, 12M
    |__ Port 1: Dev 2, If 0, Class=hub, Driver=hub/4p, 12M


strace cp -R
/media/usbdisk/music/stevie_ray_vaughan__double_trouble/greatest_hits/
/media/usbdisk1/audio/srv >& /tmp/cp.strace.log
[...]
read(3, "\275\207\245\327\212\331A\310\2$\30\353=\\\304p@ \1^\313"..., 16384) =
16384
write(4, "\275\207\245\327\212\331A\310\2$\30\353=\\\304p@ \1^\313"..., 16384) =
16384
read(3, "\366\35\263\353\366\333\213\220/\'/\205\327\346\367F\36"..., 16384) = 16384
write(4, "\366\35\263\353\366\333\213\220/\'/\205\327\346\367F\36"..., 16384) =
16384
read(3, "\20\216\324N\351\250L\231\4\360\332\362\316\204\371Y\327"..., 16384) =
16384
write(4, "\20\216\324N\351\250L\231\4\360\332\362\316\204\371Y\327"..., 16384

(looks like the relevant bit, if any, may have been buffered... doh)

I'd be happy to perform any further diagnostics if someone can tell me which
ones would be relevant!