Re: [PATCH net 1/3] ice: fix NULL pointer dereference in ice_update_vsi_tx_ring_stats()

[Tony Nguyen <anthony.l.nguyen@intel.com>]

On 3/16/2022 1:24 PM, Alexander Duyck wrote:
> On Wed, Mar 16, 2022 at 12:01 PM Nguyen, Anthony L
> <anthony.l.nguyen@intel.com> wrote:
>> On Tue, 2022-03-15 at 20:29 -0700, Jakub Kicinski wrote:
>>> On Tue, 15 Mar 2022 14:12:23 -0700 Tony Nguyen wrote:
>>>> From: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
> <snip>
>
>>>> pointer is valid, but later on ring is accessed to propagate
>>>> gathered Tx
>>>> stats onto VSI stats.
>>>>
>>>> Change the existing logic to move to next ring when ring is NULL.
>>>>
>>>> Fixes: e72bba21355d ("ice: split ice_ring onto Tx/Rx separate
>>>> structs")
>>>> Reported-by: kernel test robot <lkp@intel.com>
>>>> Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
>>>> Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
>>>> Acked-by: Alexander Lobakin <alexandr.lobakin@intel.com>
>>>> Tested-by: Gurucharan G <gurucharanx.g@intel.com> (A Contingent
>>>> worker at Intel)
>>>> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
>>>> ---
>>>>   drivers/net/ethernet/intel/ice/ice_main.c | 5 +++--
>>>>   1 file changed, 3 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/drivers/net/ethernet/intel/ice/ice_main.c
>>>> b/drivers/net/ethernet/intel/ice/ice_main.c
>>>> index 493942e910be..d4a7c39fd078 100644
>>>> --- a/drivers/net/ethernet/intel/ice/ice_main.c
>>>> +++ b/drivers/net/ethernet/intel/ice/ice_main.c
>>>> @@ -5962,8 +5962,9 @@ ice_update_vsi_tx_ring_stats(struct ice_vsi
>>>> *vsi,
>>>>                  u64 pkts = 0, bytes = 0;
>>>>
>>>>                  ring = READ_ONCE(rings[i]);
>>> Not really related to this patch but why is there a read_once() here?
>>> Aren't stats read under rtnl_lock? What is this protecting against?
>> It looks like it was based on a patch from i40e [1]. From the commit, I
>> gather this is the reason:
>>
>> "Previously the stats were 64 bit but highly racy due to the fact that
>> 64 bit transactions are not atomic on 32 bit systems."
>>
>> Thanks,
>>
>> Tony
>>
>> [1]
>> https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git/commit/?id=980e9b1186424fa3eb766d59fc91003d0ed1ed6a
>>
>>
>> (Resending as some non-text formatting snuck in to my reply. Sorry for
>> the spam)
> Actually the rcu locking and READ_ONCE has to do with the fact that
> the driver code for the igb/ixgbe/i40e driver had a bunch of code that
> could kick in from the sysfs and/or PCIe paths that would start
> tearing things down without necessarily holding the rtnl_lock as it
> should have.
>
> It might make sense to reevaluate the usage of the READ_ONCE and rcu
> locking to determine if it is actually needed since much of that is a
> hold over from when we were doing this in the watchdog and not while
> holding the rtnl_lock in the stats call.

Thanks for clarifying Alex. I'll ask the teams to look into this.