[dpdk-dev] [PATCH] lpm: fix unchecked return value

[Ruifeng Wang]

> -----Original Message-----
> From: Medvedkin, Vladimir <vladimir.medvedkin at intel.com>
> Sent: Thursday, July 16, 2020 7:00 PM
> To: Ruifeng Wang <Ruifeng.Wang at arm.com>; Bruce Richardson
> <bruce.richardson at intel.com>
> Cc: dev at dpdk.org; nd <nd at arm.com>; Honnappa Nagarahalli
> <Honnappa.Nagarahalli at arm.com>; Phil Yang <Phil.Yang at arm.com>
> Subject: Re: [PATCH] lpm: fix unchecked return value
> 
> Hi Ruifeng,
> 
> On 16/07/2020 06:19, Ruifeng Wang wrote:
> > Coverity complains about unchecked return value of
> rte_rcu_qsbr_dq_enqueue.
> > By default, defer queue size is big enough to hold all tbl8 groups.
> > When enqueue fails, return error to the user to indicate system issue.
> >
> > Coverity issue: 360832
> > Fixes: 8a9f8564e9f9 ("lpm: implement RCU rule reclamation")
> >
> > Signed-off-by: Ruifeng Wang <ruifeng.wang at arm.com>
> > ---
> >   lib/librte_lpm/rte_lpm.c | 16 +++++++++++-----
> >   1 file changed, 11 insertions(+), 5 deletions(-)
> >
> > diff --git a/lib/librte_lpm/rte_lpm.c b/lib/librte_lpm/rte_lpm.c index
> > 2db9e16a2..a6d3a7894 100644
> > --- a/lib/librte_lpm/rte_lpm.c
> > +++ b/lib/librte_lpm/rte_lpm.c
> > @@ -532,11 +532,12 @@ tbl8_alloc(struct rte_lpm *lpm)
> >   	return group_idx;
> >   }
> >
> > -static void
> > +static int
> >   tbl8_free(struct rte_lpm *lpm, uint32_t tbl8_group_start)
> >   {
> >   	struct rte_lpm_tbl_entry zero_tbl8_entry = {0};
> >   	struct __rte_lpm *internal_lpm;
> > +	int rc = 0;
> >
> >   	internal_lpm = container_of(lpm, struct __rte_lpm, lpm);
> >   	if (internal_lpm->v == NULL) {
> > @@ -552,9 +553,13 @@ tbl8_free(struct rte_lpm *lpm, uint32_t
> tbl8_group_start)
> >   				__ATOMIC_RELAXED);
> >   	} else if (internal_lpm->rcu_mode == RTE_LPM_QSBR_MODE_DQ) {
> >   		/* Push into QSBR defer queue. */
> > -		rte_rcu_qsbr_dq_enqueue(internal_lpm->dq,
> > +		rc = rte_rcu_qsbr_dq_enqueue(internal_lpm->dq,
> >   				(void *)&tbl8_group_start);
> 
> On failure rte_rcu_qsbr_dq_enqueue() returns 1 and sets rte_errno.
> Consequently, rc value is propagated to delete_depth_big() ->
> rte_lpm_delete(), and on failure the latter returns "1" value, which conflicts
> with the LPM API:
> "0 on success, negative value otherwise"
> I would suggest here to return -rte_errno if rc is equal to 1.
> 
Yes, the return value is a little different from LPM APIs.
Will change it in next version to keep consistent with other LPM APIs.

Thanks.
/Ruifeng
> 
> > +		if (rc != 0)
> > +			RTE_LOG(ERR, LPM, "Failed to push QSBR FIFO\n");
> >   	}
> > +
> > +	return rc;
> >   }
> >
> >   static __rte_noinline int32_t
> > @@ -1041,6 +1046,7 @@ delete_depth_big(struct rte_lpm *lpm, uint32_t
> ip_masked,
> >   	uint32_t tbl24_index, tbl8_group_index, tbl8_group_start,
> tbl8_index,
> >   			tbl8_range, i;
> >   	int32_t tbl8_recycle_index;
> > +	int rc = 0;
> >
> >   	/*
> >   	 * Calculate the index into tbl24 and range. Note: All depths
> > larger @@ -1097,7 +1103,7 @@ delete_depth_big(struct rte_lpm *lpm,
> uint32_t ip_masked,
> >   		 */
> >   		lpm->tbl24[tbl24_index].valid = 0;
> >   		__atomic_thread_fence(__ATOMIC_RELEASE);
> > -		tbl8_free(lpm, tbl8_group_start);
> > +		rc = tbl8_free(lpm, tbl8_group_start);
> >   	} else if (tbl8_recycle_index > -1) {
> >   		/* Update tbl24 entry. */
> >   		struct rte_lpm_tbl_entry new_tbl24_entry = { @@ -1113,10
> +1119,10
> > @@ delete_depth_big(struct rte_lpm *lpm, uint32_t ip_masked,
> >   		__atomic_store(&lpm->tbl24[tbl24_index],
> &new_tbl24_entry,
> >   				__ATOMIC_RELAXED);
> >   		__atomic_thread_fence(__ATOMIC_RELEASE);
> > -		tbl8_free(lpm, tbl8_group_start);
> > +		rc = tbl8_free(lpm, tbl8_group_start);
> >   	}
> >   #undef group_idx
> > -	return 0;
> > +	return (int32_t)rc;
> >   }
> >
> >   /*
> >
> 
> --
> Regards,
> Vladimir