3.16.62-rc1 review patch. If anyone has any objections, please let me know.
------------------
From: Alan Stern <stern@xxxxxxxxxxxxxxxxxxx>
commit 665c365a77fbfeabe52694aedf3446d5f2f1ce42 upstream.
Commit 7a68d9fb8510 ("USB: usbdevfs: sanitize flags more") checks the
transfer flags for URBs submitted from userspace via usbfs. However,
the check for whether the USBDEVFS_URB_SHORT_NOT_OK flag should be
allowed for a control transfer was added in the wrong place, before
the code has properly determined the direction of the control
transfer. (Control transfers are special because for them, the
direction is set by the bRequestType byte of the Setup packet rather
than direction bit of the endpoint address.)
This patch moves code which sets up the allow_short flag for control
transfers down after is_in has been set to the correct value.
Signed-off-by: Alan Stern <stern@xxxxxxxxxxxxxxxxxxx>
Reported-and-tested-by: syzbot+24a30223a4b609bb802e@xxxxxxxxxxxxxxxxxxxxxxxxx
Fixes: 7a68d9fb8510 ("USB: usbdevfs: sanitize flags more")
CC: Oliver Neukum <oneukum@xxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
Signed-off-by: Ben Hutchings <ben@xxxxxxxxxxxxxxx>
---
drivers/usb/core/devio.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
--- a/drivers/usb/core/devio.c
+++ b/drivers/usb/core/devio.c
@@ -1329,8 +1329,6 @@ static int proc_do_submiturb(struct usb_
u = 0;
switch(uurb->type) {
case USBDEVFS_URB_TYPE_CONTROL:
- if (is_in)
- allow_short = true;
if (!usb_endpoint_xfer_control(&ep->desc))
return -EINVAL;
/* min 8 byte setup packet */
@@ -1360,6 +1358,8 @@ static int proc_do_submiturb(struct usb_
is_in = 0;
uurb->endpoint &= ~USB_DIR_IN;
}
+ if (is_in)
+ allow_short = true;
snoop(&ps->dev->dev, "control urb: bRequestType=%02x "
"bRequest=%02x wValue=%04x "
"wIndex=%04x wLength=%04x\n",
