summaryrefslogtreecommitdiffstats
path: root/drivers/usb/class/cdc-wdm.c
diff options
context:
space:
mode:
authorBjørn Mork <bjorn@mork.no>2014-01-12 21:48:53 +0100
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>2014-01-12 20:13:28 -0800
commitf563926fed982f26b391ca42493f55f2447f1b0a (patch)
treef25186b2329acd71e206d510c01be31ece23aae9 /drivers/usb/class/cdc-wdm.c
parent52a6966c350624db89addc3e6a825f5e797a73e4 (diff)
downloadop-kernel-dev-f563926fed982f26b391ca42493f55f2447f1b0a.zip
op-kernel-dev-f563926fed982f26b391ca42493f55f2447f1b0a.tar.gz
usb: cdc-wdm: resp_count can be 0 even if WDM_READ is set
Do not decrement resp_count if it's already 0. We set resp_count to 0 when the device is closed. The next open and read will try to clear the WDM_READ flag if there was leftover data in the read buffer. This fix is necessary to prevent resubmitting the read URB in a tight loop because resp_count becomes negative. The bug can easily be triggered from userspace by not reading all data in the read buffer, and then closing and reopening the chardev. Fixes: 8dd5cd5395b9 ("usb: cdc-wdm: avoid hanging on zero length reads") Cc: <stable@vger.kernel.org> # 3.13 Signed-off-by: Bjørn Mork <bjorn@mork.no> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Diffstat (limited to 'drivers/usb/class/cdc-wdm.c')
-rw-r--r--drivers/usb/class/cdc-wdm.c2
1 files changed, 1 insertions, 1 deletions
diff --git a/drivers/usb/class/cdc-wdm.c b/drivers/usb/class/cdc-wdm.c
index 590ff8b..a051a7a 100644
--- a/drivers/usb/class/cdc-wdm.c
+++ b/drivers/usb/class/cdc-wdm.c
@@ -445,7 +445,7 @@ static int clear_wdm_read_flag(struct wdm_device *desc)
clear_bit(WDM_READ, &desc->flags);
/* submit read urb only if the device is waiting for it */
- if (!--desc->resp_count)
+ if (!desc->resp_count || !--desc->resp_count)
goto out;
set_bit(WDM_RESPONDING, &desc->flags);
OpenPOWER on IntegriCloud