2.6.32-longterm review patch.  If anyone has any objections, please let me know.

------------------

From: Robert Lukassen <redacted>

commit 878b753e32ca765cd346a5d3038d630178ec78ff upstream.
[ changed s/usb_free_coherent/usb_buffer_free/ for 2.6.32.x]

In the WDM class driver a disconnect event leads to calls to
usb_free_coherent to put back two USB DMA buffers allocated earlier.
The call to usb_free_coherent uses a different size parameter
(desc->wMaxCommand) than the corresponding call to usb_alloc_coherent
(desc->bMaxPacketSize0).

When a disconnect event occurs, this leads to 'bad dma' complaints
from usb core because the USB DMA buffer is being pushed back to the
'buffer-2048' pool from which it has not been allocated.

This patch against the most recent linux-2.6 kernel ensures that the
parameters used by usb_alloc_coherent & usb_free_coherent calls in
cdc-wdm.c match.

Signed-off-by: Robert Lukassen <redacted>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Bj��rn Mork <bjorn@mork.no>

---
 drivers/usb/class/cdc-wdm.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

--- a/drivers/usb/class/cdc-wdm.c
+++ b/drivers/usb/class/cdc-wdm.c

@@ -277,7 +277,7 @@ static void cleanup(struct wdm_device *d
 			desc->sbuf,
 			desc->validity->transfer_dma);
 	usb_buffer_free(interface_to_usbdev(desc->intf),
-			desc->wMaxCommand,
+			desc->bMaxPacketSize0,
 			desc->inbuf,
 			desc->response->transfer_dma);
 	kfree(desc->orq);