Re: [PATCH iwl-next v2] ixgbe: Fix possible skb NULL pointer dereference

[Maciej Fijalkowski <maciej.fijalkowski@intel.com>]

On Thu, Jan 23, 2025 at 09:19:34AM +0100, Kwapulinski, Piotr wrote:
> >-----Original Message-----
> >From: Fijalkowski, Maciej <maciej.fijalkowski@intel.com> 
> >Sent: Monday, January 20, 2025 3:45 PM
> >To: Kwapulinski, Piotr <piotr.kwapulinski@intel.com>
> >Cc: intel-wired-lan@lists.osuosl.org; netdev@vger.kernel.org; dan.carpenter@linaro.org; yuehaibing@huawei.com; Kitszel, Przemyslaw <przemyslaw.kitszel@intel.com>
> >Subject: Re: [PATCH iwl-next v2] ixgbe: Fix possible skb NULL pointer dereference
> >
> >On Fri, Jan 17, 2025 at 04:49:35PM +0100, Piotr Kwapulinski wrote:
> >> The commit c824125cbb18 ("ixgbe: Fix passing 0 to ERR_PTR in
> >> ixgbe_run_xdp()") stopped utilizing the ERR-like macros for xdp status 
> >> encoding. Propagate this logic to the ixgbe_put_rx_buffer().
> >> 
> >> The commit also relaxed the skb NULL pointer check - caught by Smatch.
> >> Restore this check.
> >> 
> >> Fixes: c824125cbb18 ("ixgbe: Fix passing 0 to ERR_PTR in 
> >> ixgbe_run_xdp()")
> >> Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
> >> Signed-off-by: Piotr Kwapulinski <piotr.kwapulinski@intel.com>
> >> ---
> >>  drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 9 +++++----
> >>  1 file changed, 5 insertions(+), 4 deletions(-)
> >> 
> >> diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c 
> >> b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> >> index 7236f20..c682c3d 100644
> >> --- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> >> +++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> >> @@ -2098,14 +2098,14 @@ static struct ixgbe_rx_buffer 
> >> *ixgbe_get_rx_buffer(struct ixgbe_ring *rx_ring,
> >>  
> >>  static void ixgbe_put_rx_buffer(struct ixgbe_ring *rx_ring,
> >>  				struct ixgbe_rx_buffer *rx_buffer,
> >> -				struct sk_buff *skb,
> >> -				int rx_buffer_pgcnt)
> >> +				struct sk_buff *skb, int rx_buffer_pgcnt,
> >> +				int xdp_res)
> >>  {
> >>  	if (ixgbe_can_reuse_rx_page(rx_buffer, rx_buffer_pgcnt)) {
> >>  		/* hand second half of page back to the ring */
> >>  		ixgbe_reuse_rx_page(rx_ring, rx_buffer);
> >>  	} else {
> >> -		if (!IS_ERR(skb) && IXGBE_CB(skb)->dma == rx_buffer->dma) {
> >> +		if (skb && !xdp_res && IXGBE_CB(skb)->dma == rx_buffer->dma) {
> >
> >xdp_res check is redundant here. skb ptr will be non-null only for xdp_res == 0. so skb != NULL implies xdp_res == 0.
> That was tempting but eventually the ixgbe_run_xdp() handles the error exclusively. I suggest to leave it as it is.

ixgbe_run_xdp() on error paths will still return one of our homegrown XDP
statuses:

#define IXGBE_XDP_PASS		0
#define IXGBE_XDP_CONSUMED	BIT(0)
#define IXGBE_XDP_TX		BIT(1)
#define IXGBE_XDP_REDIR		BIT(2)
#define IXGBE_XDP_EXIT		BIT(3)

Plus the case "!xdp_res && !skb" is explicitly checked before calling
ixgbe_put_rx_buffer() and you're gonna break the rx loop altogether. This
happens when you failed to alloc skb for IXGBE_XDP_PASS.

> Thanks,
> Piotr
> 
> >
> >If I am not mistaken:D or ixgbe has some code path I missed.
> >
> >Besides this, thanks for improving commit message!
> >
> >>  			/* the page has been released from the ring */
> >>  			IXGBE_CB(skb)->page_released = true;
> >>  		} else {
> >> @@ -2415,7 +2415,8 @@ static int ixgbe_clean_rx_irq(struct ixgbe_q_vector *q_vector,
> >>  			break;
> >>  		}
> >>  
> >> -		ixgbe_put_rx_buffer(rx_ring, rx_buffer, skb, rx_buffer_pgcnt);
> >> +		ixgbe_put_rx_buffer(rx_ring, rx_buffer, skb, rx_buffer_pgcnt,
> >> +				    xdp_res);
> >>  		cleaned_count++;
> >>  
> >>  		/* place incomplete frames back on ring for completion */
> >> --
> >> v1 -> v2
> >>   Provide extra details in commit message for motivation of this patch.
> >> 
> >> 2.43.0
> >>