Re: [PATCH net-next v2 1/2] usbnet: ipheth: fix risk of NULL pointer deallocation

[Simon Horman <simon.horman@corigine.com>]

On Fri, May 26, 2023 at 10:33:21AM +0200, George Valkov wrote:
> 
> > On 26 May 2023, at 10:52 AM, Simon Horman <simon.horman@corigine.com> wrote:
> >
> > On Thu, May 25, 2023 at 09:42:54PM +0200, Foster Snowhill wrote:
> >> From: Georgi Valkov <gvalkov@gmail.com>
> >>
> >> The cleanup precedure in ipheth_probe will attempt to free a
> >> NULL pointer in dev->ctrl_buf if the memory allocation for
> >> this buffer is not successful. Rearrange the goto labels to
> >> avoid this risk.
> >
> > Hi Georgi and Foster,
> >
> > kfree will ignore a NULL argument, so I think the existing code is safe.
> > But given the name of the label I do agree there is scope for a cleanup
> > here.
> 
> It’s good to know that precaution has been taken in kfree to avoid this, yet at
> my opinion knowingly attempting to free a NULL pointer is a red flag and bad
> design. Likely a misplaced label.
> 
> > Could you consider rewording the patch description accordingly?
> 
> What would you like me to use as title and description? Can I use this?
> 
> usbnet: ipheth: avoid kfree with a NULL pointer
> 
> The cleanup precedure in ipheth_probe will attempt to free a
> NULL pointer in dev->ctrl_buf if the memory allocation for
> this buffer is not successful. While kfree ignores NULL pointers,
> and the existing code is safe, it is a better design to rearrange
> the goto labels and avoid this.

Thanks, that looks good to me.

> >> Signed-off-by: Georgi Valkov <gvalkov@gmail.com>
> >
> > If Georgi is the author of the patch, which seems to be the case,
> > then the above is correct. But as the patch is being posted by Foster
> > I think it should be followed by a Signed-off-by line for Foster.
> 
> Yes, I discovered the potential issue and authored the patch to help. We’ll
> append Signed-off-by Foster as you suggested. Thanks Simon!
> 
> Something like that?

Yes, I think that sounds good.

Please wait 24h before the posting of v2 before posting v3,
to allow time for more review of v3 (from others).

...