Re: [dpdk-dev] [PATCH v3] net/i40e: fix out of bounds read issue

["Di, ChenxuX" <chenxux.di@intel.com>]

Hi 

> -----Original Message-----
> From: Guo, Jia
> Sent: Thursday, May 14, 2020 2:17 PM
> To: Di, ChenxuX <chenxux.di@intel.com>; dev@dpdk.org
> Cc: Xing, Beilei <beilei.xing@intel.com>
> Subject: Re: [PATCH v3] net/i40e: fix out of bounds read issue
> 
> 
> On 5/14/2020 9:16 AM, Di, ChenxuX wrote:
> > Hi,
> >
> >
> >
> >> -----Original Message-----
> >> From: Guo, Jia
> >> Sent: Wednesday, May 13, 2020 2:51 PM
> >> To: Di, ChenxuX <chenxux.di@intel.com>; dev@dpdk.org
> >> Cc: Xing, Beilei <beilei.xing@intel.com>
> >> Subject: Re: [PATCH v3] net/i40e: fix out of bounds read issue
> >>
> >> hi, chenxu
> >>
> >> On 5/13/2020 10:26 AM, Chenxu Di wrote:
> >>> This patch fixes (out-of-bounds read) coverity issue.
> >>>
> >>> Coverity issue: 357699
> >>> Coverity issue: 357694
> >>> Fixes: feaae285b342 ("net/i40e: support hash configuration in RSS
> >>> flow")
> >>>
> >>> Signed-off-by: Chenxu Di <chenxux.di@intel.com>
> >>> ---
> >>>    drivers/net/i40e/i40e_ethdev.c | 6 ++++++
> >>>    1 file changed, 6 insertions(+)
> >>>
> >>> diff --git a/drivers/net/i40e/i40e_ethdev.c
> >>> b/drivers/net/i40e/i40e_ethdev.c index 749d85f54..c2d5c6835 100644
> >>> --- a/drivers/net/i40e/i40e_ethdev.c
> >>> +++ b/drivers/net/i40e/i40e_ethdev.c
> >>> @@ -13179,6 +13179,9 @@ i40e_rss_config_hash_function(struct i40e_pf
> *pf,
> >>>    				break;
> >>>    		}
> >>>
> >>> +		if (i == UINT64_BIT)
> >>> +			return 0;
> >>> +
> >>
> >> Should this function need to return none zero value and show error
> >> info out, or said should considerate "out of bound" as a configure failed.
> >>
> > It will just find the flow type and do the things. If not find the flow type, it just
> do nothing and return.
> 
> 
> i40e_rss_config_hash_function is return int value right, should it always return 0?
> Should the case of not finding be considerate as none-success config?
> 

The original code has too many tabs (for->if->for->if ), it is over 90 words per line.
So just break and move it out .

After break the loop,
if the first loop go to the end without break, the value i will be 64 while the length of pf->adapter->pctypes_tbl is 64.
That is the reason of the out of bound which the original code will not cause.
 
Actually it will not happen the case of not finding. It is just the code fix the  out of bound caused by breaking the loop .


> 
> >>>    		for (j = I40E_FILTER_PCTYPE_INVALID + 1;
> >>>    		     j < I40E_FILTER_PCTYPE_MAX; j++) {
> >>>    			if (pf->adapter->pctypes_tbl[i] & (1ULL << j)) @@ -
> 13311,6
> >>> +13314,9 @@ i40e_rss_clear_hash_function(struct i40e_pf *pf,
> >>>    				break;
> >>>    		}
> >>>
> >>> +		if (i == UINT64_BIT)
> >>> +			return 0;
> >>> +
> >>>    		for (j = I40E_FILTER_PCTYPE_INVALID + 1;
> >>>    		     j < I40E_FILTER_PCTYPE_MAX; j++) {
> >>>    			if (pf->adapter->pctypes_tbl[i] & (1ULL << j))