On Mon, Apr 12, 2021 at 01:25:20PM -0600, Shuah Khan wrote:
> On 4/12/21 1:06 PM, Tom Seewald wrote:
> > On 4/12/21 1:59 PM, Tom Seewald wrote:
> >
> > > commit 9858af27e69247c5d04c3b093190a93ca365f33d upstream.
> > >
> > > Currently udc->ud.tcp_rx is being assigned twice, the second assignment
> > > is incorrect, it should be to udc->ud.tcp_tx instead of rx. Fix this.
> > >
> > > Fixes: 46613c9dfa96 ("usbip: fix vudc usbip_sockfd_store races leading to gpf")
> > > Acked-by: Shuah Khan <skhan@xxxxxxxxxxxxxxxxxxx>
> > > Signed-off-by: Colin Ian King <colin.king@xxxxxxxxxxxxx>
> > > Cc: stable <stable@xxxxxxxxxxxxxxx>
> > > Addresses-Coverity: ("Unused value")
> > > Link: https://lore.kernel.org/r/20210311104445.7811-1-colin.king@xxxxxxxxxxxxx
> > > Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
> > > Signed-off-by: Tom Seewald <tseewald@xxxxxxxxx>
> > > ---
> > > drivers/usb/usbip/vudc_sysfs.c | 2 +-
> > > 1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/drivers/usb/usbip/vudc_sysfs.c b/drivers/usb/usbip/vudc_sysfs.c
> > > index f44d98eeb36a..51cc5258b63e 100644
> > > --- a/drivers/usb/usbip/vudc_sysfs.c
> > > +++ b/drivers/usb/usbip/vudc_sysfs.c
> > > @@ -187,7 +187,7 @@ static ssize_t store_sockfd(struct device *dev,
> > > udc->ud.tcp_socket = socket;
> > > udc->ud.tcp_rx = tcp_rx;
> > > - udc->ud.tcp_rx = tcp_tx;
> > > + udc->ud.tcp_tx = tcp_tx;
> > > udc->ud.status = SDEV_ST_USED;
> > > spin_unlock_irq(&udc->ud.lock);
> > I sent this because I believe this patch needs to be backported to the
> > 4.9.y and 4.14.y stable trees.
> >
>
> Tom,
>
> Correct. This needs proting to 4.14 and 4.9. However, you have to also
> backport the patch it fixes to 4.14 and 4.9
>
> 46613c9dfa96 ("usbip: fix vudc usbip_sockfd_store races leading to gpf")
>
> You can combine the two patches when you backport to 4.14 and 4.9 and
> add both upstream commits in the change log.
Please do not ever combine patches for stable submissions, we want to
keep things as close as to what happened in Linus's tree as possible as
we track commit ids and putting 2 together is pretty impossible to
manage over time.
thanks,
greg k-h
