On Sun, May 30, 2004 at 11:21:58AM -0400, Alan Stern wrote:
> If the device _isn't_ messed up then the outside request and the reset
> request can't interfere with each other, because they both go to ep0 and
> hence are automatically serialized. The only problem we need to guard
> against is an outside request on ep0 arriving during a regular SCSI
> command -- so we only need to lock the device while sending regular
> commands.
> I guess there's no real harm in reusing ->serialize for this purpose.
> (After all, I'm going to do the same sort of thing with the hub driver,
> replacing its private semaphore with ->serialize.) So long as the lock is
> held only during regular commands and not during resets there shouldn't be
> a problem.
> P.S.: Vojtech, part of the patch you submitted is already out-of-date!
> devio.c has already been changed to lock the device during any nontrivial
> operation. So all that needs to be added are the locks in usb-storage.
Okay, so can we all agree on this one?
--
Vojtech Pavlik
SuSE Labs, SuSE CR
[EMAIL PROTECTED], 2004-05-30 18:12:53+02:00, [EMAIL PROTECTED]
usb: Based on a 2.4 patch from [EMAIL PROTECTED], this patch
serializes usb storage and usbfs operation, so that usbfs
cannot disturb storage by seemingly harmless control reads.
Signed-off-by: Vojtech Pavlik <[EMAIL PROTECTED]>
transport.c | 11 +++++++++++
1 files changed, 11 insertions(+)
diff -Nru a/drivers/usb/storage/transport.c b/drivers/usb/storage/transport.c
--- a/drivers/usb/storage/transport.c 2004-05-30 18:13:33 +02:00
+++ b/drivers/usb/storage/transport.c 2004-05-30 18:13:33 +02:00
@@ -527,9 +527,18 @@
int need_auto_sense;
int result;
+ /*
+ * Grab device's serialize mutex to prevent /usbfs and others from
+ * sending out a command in the middle of ours (if libusb sends a
+ * get_descriptor or something on pipe 0 after our CBW and before
+ * our CSW, and then we get a stall, we have trouble)
+ */
+
/* send the command to the transport layer */
+ down(&(us->pusb_dev->serialize));
srb->resid = 0;
result = us->transport(srb, us);
+ up(&(us->pusb_dev->serialize));
/* if the command gets aborted by the higher layers, we need to
* short-circuit all other processing
@@ -648,9 +657,11 @@
srb->serial_number ^= 0x80000000;
/* issue the auto-sense command */
+ down(&(us->pusb_dev->serialize));
old_resid = srb->resid;
srb->resid = 0;
temp_result = us->transport(us->srb, us);
+ up(&(us->pusb_dev->serialize));
/* let's clean up right away */
srb->resid = old_resid;
Signed-off-by: Vojtech Pavlik <[EMAIL PROTECTED]>
