[dpdk-dev,v2] net/i40e: exclude LLDP packet count
Checks
Commit Message
When use port stats register to calculate the packet count, LLDP packets
are counted in statistics which is not expected, the patch exclude this
number from total number.
Fixes: 763de290cbd1 ("net/i40e: fix packet count for PF")
Cc stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
---
v2:
- fix some typos
drivers/net/i40e/i40e_ethdev.c | 60 +++++++++++++++++++++++++++++++++++++++---
1 file changed, 56 insertions(+), 4 deletions(-)
Comments
> -----Original Message-----
> From: Zhang, Qi Z
> Sent: Tuesday, December 12, 2017 2:10 AM
> To: Xing, Beilei <beilei.xing@intel.com>
> Cc: dev@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>
> Subject: [PATCH v2] net/i40e: exclude LLDP packet count
>
> When use port stats register to calculate the packet count, LLDP packets are
> counted in statistics which is not expected, the patch exclude this number
> from total number.
>
> Fixes: 763de290cbd1 ("net/i40e: fix packet count for PF") Cc stable@dpdk.org
>
> Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>
> -----Original Message-----
> From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Xing, Beilei
> Sent: Monday, January 8, 2018 9:51 AM
> To: Zhang, Qi Z
> Cc: dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH v2] net/i40e: exclude LLDP packet count
>
>
>
> > -----Original Message-----
> > From: Zhang, Qi Z
> > Sent: Tuesday, December 12, 2017 2:10 AM
> > To: Xing, Beilei <beilei.xing@intel.com>
> > Cc: dev@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>
> > Subject: [PATCH v2] net/i40e: exclude LLDP packet count
> >
> > When use port stats register to calculate the packet count, LLDP
> > packets are counted in statistics which is not expected, the patch
> > exclude this number from total number.
> >
> > Fixes: 763de290cbd1 ("net/i40e: fix packet count for PF") Cc
> > stable@dpdk.org
> >
> > Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
>
> Acked-by: Beilei Xing <beilei.xing@intel.com>
Applied to dpdk-next-net-intel, thanks!
/Helin
@@ -2531,6 +2531,22 @@ i40e_read_stats_registers(struct i40e_pf *pf, struct i40e_hw *hw)
pf->offset_loaded,
&pf->internal_stats_offset.rx_broadcast,
&pf->internal_stats.rx_broadcast);
+ /* Get total internal tx packet count */
+ i40e_stat_update_48(hw, I40E_GLV_UPTCH(hw->port),
+ I40E_GLV_UPTCL(hw->port),
+ pf->offset_loaded,
+ &pf->internal_stats_offset.tx_unicast,
+ &pf->internal_stats.tx_unicast);
+ i40e_stat_update_48(hw, I40E_GLV_MPTCH(hw->port),
+ I40E_GLV_MPTCL(hw->port),
+ pf->offset_loaded,
+ &pf->internal_stats_offset.tx_multicast,
+ &pf->internal_stats.tx_multicast);
+ i40e_stat_update_48(hw, I40E_GLV_BPTCH(hw->port),
+ I40E_GLV_BPTCL(hw->port),
+ pf->offset_loaded,
+ &pf->internal_stats_offset.tx_broadcast,
+ &pf->internal_stats.tx_broadcast);
/* exclude CRC size */
pf->internal_stats.rx_bytes -= (pf->internal_stats.rx_unicast +
@@ -2560,16 +2576,32 @@ i40e_read_stats_registers(struct i40e_pf *pf, struct i40e_hw *hw)
ns->eth.rx_bytes -= (ns->eth.rx_unicast + ns->eth.rx_multicast +
ns->eth.rx_broadcast) * ETHER_CRC_LEN;
- /* Workaround: it is possible I40E_GLV_GORCH[H/L] is updated before
- * I40E_GLPRT_GORCH[H/L], so there is a small window that cause negtive
+ /* exclude internal rx bytes
+ * Workaround: it is possible I40E_GLV_GORCH[H/L] is updated before
+ * I40E_GLPRT_GORCH[H/L], so there is a small window that cause negative
* value.
+ * same to I40E_GLV_UPRC[H/L], I40E_GLV_MPRC[H/L], I40E_GLV_BPRC[H/L].
*/
if (ns->eth.rx_bytes < pf->internal_stats.rx_bytes)
ns->eth.rx_bytes = 0;
- /* exlude internal rx bytes */
else
ns->eth.rx_bytes -= pf->internal_stats.rx_bytes;
+ if (ns->eth.rx_unicast < pf->internal_stats.rx_unicast)
+ ns->eth.rx_unicast = 0;
+ else
+ ns->eth.rx_unicast -= pf->internal_stats.rx_unicast;
+
+ if (ns->eth.rx_multicast < pf->internal_stats.rx_multicast)
+ ns->eth.rx_multicast = 0;
+ else
+ ns->eth.rx_multicast -= pf->internal_stats.rx_multicast;
+
+ if (ns->eth.rx_broadcast < pf->internal_stats.rx_broadcast)
+ ns->eth.rx_broadcast = 0;
+ else
+ ns->eth.rx_broadcast -= pf->internal_stats.rx_broadcast;
+
i40e_stat_update_32(hw, I40E_GLPRT_RDPC(hw->port),
pf->offset_loaded, &os->eth.rx_discards,
&ns->eth.rx_discards);
@@ -2598,12 +2630,32 @@ i40e_read_stats_registers(struct i40e_pf *pf, struct i40e_hw *hw)
ns->eth.tx_bytes -= (ns->eth.tx_unicast + ns->eth.tx_multicast +
ns->eth.tx_broadcast) * ETHER_CRC_LEN;
- /* exclude internal tx bytes */
+ /* exclude internal tx bytes
+ * Workaround: it is possible I40E_GLV_GOTCH[H/L] is updated before
+ * I40E_GLPRT_GOTCH[H/L], so there is a small window that cause negative
+ * value.
+ * same to I40E_GLV_UPTC[H/L], I40E_GLV_MPTC[H/L], I40E_GLV_BPTC[H/L].
+ */
if (ns->eth.tx_bytes < pf->internal_stats.tx_bytes)
ns->eth.tx_bytes = 0;
else
ns->eth.tx_bytes -= pf->internal_stats.tx_bytes;
+ if (ns->eth.tx_unicast < pf->internal_stats.tx_unicast)
+ ns->eth.tx_unicast = 0;
+ else
+ ns->eth.tx_unicast -= pf->internal_stats.tx_unicast;
+
+ if (ns->eth.tx_multicast < pf->internal_stats.tx_multicast)
+ ns->eth.tx_multicast = 0;
+ else
+ ns->eth.tx_multicast -= pf->internal_stats.tx_multicast;
+
+ if (ns->eth.tx_broadcast < pf->internal_stats.tx_broadcast)
+ ns->eth.tx_broadcast = 0;
+ else
+ ns->eth.tx_broadcast -= pf->internal_stats.tx_broadcast;
+
/* GLPRT_TEPC not supported */
/* additional port specific stats */