strcpy() performs no bounds checking on the destination buffer. This
could result in linear overflows beyond the end of the buffer, leading
to all kinds of misbehaviors. The safe replacement is strscpy().
Signed-off-by: Len Baker <[email protected]>
---
drivers/staging/rtl8712/os_intfs.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/staging/rtl8712/os_intfs.c b/drivers/staging/rtl8712/os_intfs.c
index 2214aca09730..9502f6aa5306 100644
--- a/drivers/staging/rtl8712/os_intfs.c
+++ b/drivers/staging/rtl8712/os_intfs.c
@@ -203,7 +203,7 @@ struct net_device *r8712_init_netdev(void)
if (!pnetdev)
return NULL;
if (dev_alloc_name(pnetdev, ifname) < 0) {
- strcpy(ifname, "wlan%d");
+ strscpy(ifname, "wlan%d", sizeof(ifname));
dev_alloc_name(pnetdev, ifname);
}
padapter = netdev_priv(pnetdev);
--
2.25.1
On Sat, Jul 17, 2021 at 05:51:45PM +0200, Len Baker wrote:
> strcpy() performs no bounds checking on the destination buffer. This
> could result in linear overflows beyond the end of the buffer, leading
> to all kinds of misbehaviors. The safe replacement is strscpy().
>
> Signed-off-by: Len Baker <[email protected]>
> ---
> drivers/staging/rtl8712/os_intfs.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/staging/rtl8712/os_intfs.c b/drivers/staging/rtl8712/os_intfs.c
> index 2214aca09730..9502f6aa5306 100644
> --- a/drivers/staging/rtl8712/os_intfs.c
> +++ b/drivers/staging/rtl8712/os_intfs.c
> @@ -203,7 +203,7 @@ struct net_device *r8712_init_netdev(void)
> if (!pnetdev)
> return NULL;
> if (dev_alloc_name(pnetdev, ifname) < 0) {
> - strcpy(ifname, "wlan%d");
> + strscpy(ifname, "wlan%d", sizeof(ifname));
> dev_alloc_name(pnetdev, ifname);
Not related to your patch but this code is bad. What it does is the
"ifname" can be set as a module parameter. So instead of testing if it
has been set, it uses the checking inside dev_alloc_name() to see if we
can allocate what the user requested. If not then set it to "wlan%d".
If we cannot allocate what the user wants then we should return an
error.
It should do:
if (ifname[0] == '\0')
strscpy(ifname, "wlan%d", sizeof(ifname));
ret = dev_alloc_name(pnetdev, ifname);
if (ret < 0) {
dev_err(pnetdev, "allocating device name failed.\n");
return NULL;
}
regards,
dan carpenter
From: Dan Carpenter
> Sent: 19 July 2021 06:38
...
> Not related to your patch but this code is bad. What it does is the
> "ifname" can be set as a module parameter. So instead of testing if it
> has been set, it uses the checking inside dev_alloc_name() to see if we
> can allocate what the user requested. If not then set it to "wlan%d".
> If we cannot allocate what the user wants then we should return an
> error.
>
> It should do:
>
> if (ifname[0] == '\0')
> strscpy(ifname, "wlan%d", sizeof(ifname));
>
> ret = dev_alloc_name(pnetdev, ifname);
> if (ret < 0) {
> dev_err(pnetdev, "allocating device name failed.\n");
> return NULL;
> }
I know only root can set module parameters, but having one
that contains a string used as a printf format seems
dangerous at best.
Isn't it best to let userspace rename the interfaces later on?
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)
On Mon, Jul 19, 2021 at 03:24:38PM +0000, David Laight wrote:
> From: Dan Carpenter
> > Sent: 19 July 2021 06:38
> ...
> > Not related to your patch but this code is bad. What it does is the
> > "ifname" can be set as a module parameter. So instead of testing if it
> > has been set, it uses the checking inside dev_alloc_name() to see if we
> > can allocate what the user requested. If not then set it to "wlan%d".
> > If we cannot allocate what the user wants then we should return an
> > error.
> >
> > It should do:
> >
> > if (ifname[0] == '\0')
> > strscpy(ifname, "wlan%d", sizeof(ifname));
> >
> > ret = dev_alloc_name(pnetdev, ifname);
> > if (ret < 0) {
> > dev_err(pnetdev, "allocating device name failed.\n");
> > return NULL;
> > }
>
> I know only root can set module parameters, but having one
> that contains a string used as a printf format seems
> dangerous at best.
>
> Isn't it best to let userspace rename the interfaces later on?
Yeah. I think you're right.
regards,
dan carpenter
Hi,
On Wed, Jul 21, 2021 at 11:06:24AM +0300, Dan Carpenter wrote:
> On Mon, Jul 19, 2021 at 03:24:38PM +0000, David Laight wrote:
> > From: Dan Carpenter
> > > Sent: 19 July 2021 06:38
> > ...
> > > Not related to your patch but this code is bad. What it does is the
> > > "ifname" can be set as a module parameter. So instead of testing if it
> > > has been set, it uses the checking inside dev_alloc_name() to see if we
> > > can allocate what the user requested. If not then set it to "wlan%d".
> > > If we cannot allocate what the user wants then we should return an
> > > error.
> > >
> > > It should do:
> > >
> > > if (ifname[0] == '\0')
> > > strscpy(ifname, "wlan%d", sizeof(ifname));
> > >
> > > ret = dev_alloc_name(pnetdev, ifname);
> > > if (ret < 0) {
> > > dev_err(pnetdev, "allocating device name failed.\n");
> > > return NULL;
> > > }
> >
> > I know only root can set module parameters, but having one
> > that contains a string used as a printf format seems
> > dangerous at best.
> >
> > Isn't it best to let userspace rename the interfaces later on?
>
> Yeah. I think you're right.
Sorry, but I don't understand if the code needs to be improved or not.
Is the code shown by Dan not correct?
>
> regards,
> dan carpenter
>
Thanks for your time,
Len
On Fri, Jul 23, 2021 at 05:15:10PM +0200, Len Baker wrote:
> Hi,
>
> On Wed, Jul 21, 2021 at 11:06:24AM +0300, Dan Carpenter wrote:
> > On Mon, Jul 19, 2021 at 03:24:38PM +0000, David Laight wrote:
> > > From: Dan Carpenter
> > > > Sent: 19 July 2021 06:38
> > > ...
> > > > Not related to your patch but this code is bad. What it does is the
> > > > "ifname" can be set as a module parameter. So instead of testing if it
> > > > has been set, it uses the checking inside dev_alloc_name() to see if we
> > > > can allocate what the user requested. If not then set it to "wlan%d".
> > > > If we cannot allocate what the user wants then we should return an
> > > > error.
> > > >
> > > > It should do:
> > > >
> > > > if (ifname[0] == '\0')
> > > > strscpy(ifname, "wlan%d", sizeof(ifname));
> > > >
> > > > ret = dev_alloc_name(pnetdev, ifname);
> > > > if (ret < 0) {
> > > > dev_err(pnetdev, "allocating device name failed.\n");
> > > > return NULL;
> > > > }
> > >
> > > I know only root can set module parameters, but having one
> > > that contains a string used as a printf format seems
> > > dangerous at best.
> > >
> > > Isn't it best to let userspace rename the interfaces later on?
> >
> > Yeah. I think you're right.
>
> Sorry, but I don't understand if the code needs to be improved or not.
> Is the code shown by Dan not correct?
We should remove the ifname[] array and the module option completely.
regards,
dan carpenter
On Mon, Jul 26, 2021 at 11:11:48AM +0300, Dan Carpenter wrote:
> On Fri, Jul 23, 2021 at 05:15:10PM +0200, Len Baker wrote:
> > Hi,
> >
> > On Wed, Jul 21, 2021 at 11:06:24AM +0300, Dan Carpenter wrote:
> > > On Mon, Jul 19, 2021 at 03:24:38PM +0000, David Laight wrote:
> > > >
> > > > I know only root can set module parameters, but having one
> > > > that contains a string used as a printf format seems
> > > > dangerous at best.
> > > >
> > > > Isn't it best to let userspace rename the interfaces later on?
> > >
> > > Yeah. I think you're right.
> >
> > Sorry, but I don't understand if the code needs to be improved or not.
> > Is the code shown by Dan not correct?
>
> We should remove the ifname[] array and the module option completely.
Thanks for the clarification.
Regards,
Len