[dpdk-dev] [PATCH] net/i40e: exclude internal packet's byte count

[Zhang, Qi Z]

> -----Original Message-----
> From: Xing, Beilei
> Sent: Thursday, June 1, 2017 5:13 PM
> To: Zhang, Qi Z <qi.z.zhang at intel.com>; Wu, Jingjing
> <jingjing.wu at intel.com>; Zhang, Helin <helin.zhang at intel.com>
> Cc: dev at dpdk.org; Zhang, Qi Z <qi.z.zhang at intel.com>; stable at dpdk.org
> Subject: RE: [dpdk-dev] [PATCH] net/i40e: exclude internal packet's byte
> count
> 
> Hi Qi,
> 
> > -----Original Message-----
> > From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Qi Zhang
> > Sent: Thursday, June 1, 2017 6:53 AM
> > To: Wu, Jingjing <jingjing.wu at intel.com>; Zhang, Helin
> > <helin.zhang at intel.com>
> > Cc: dev at dpdk.org; Zhang, Qi Z <qi.z.zhang at intel.com>; stable at dpdk.org
> > Subject: [dpdk-dev] [PATCH] net/i40e: exclude internal packet's byte
> > count
> >
> > Tx/Rx byte counts of internal managed packet should be exluded from
> > the total rx/tx bytes.
> >
> > Fixes: 9aace75fc82e ("i40e: fix statistics")
> > Cc: stable at dpdk.org
> >
> > Signed-off-by: Qi Zhang <qi.z.zhang at intel.com>
> > ---
> >  drivers/net/i40e/i40e_ethdev.c | 13 +++++++++++--
> > drivers/net/i40e/i40e_ethdev.h |  5 +++++
> >  2 files changed, 16 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/net/i40e/i40e_ethdev.c
> > b/drivers/net/i40e/i40e_ethdev.c index 4c49673..c8802f2 100644
> > --- a/drivers/net/i40e/i40e_ethdev.c
> > +++ b/drivers/net/i40e/i40e_ethdev.c
> > @@ -2379,6 +2379,15 @@ i40e_read_stats_registers(struct i40e_pf *pf,
> > struct i40e_hw *hw)
> >  	struct i40e_hw_port_stats *ns = &pf->stats; /* new stats */
> >  	struct i40e_hw_port_stats *os = &pf->stats_offset; /* old stats */
> >
> > +	/* Get rx/tx bytes of internal transfer packets */
> > +	i40e_stat_update_48(hw, I40E_GLV_GORCH(hw->port),
> > I40E_GLV_GORCL(hw->port),
> > +			pf->offset_loaded, &pf->internal_rx_bytes_offset,
> > +			&pf->internal_rx_bytes);
> > +
> > +	i40e_stat_update_48(hw, I40E_GLV_GOTCH(hw->port),
> > I40E_GLV_GOTCL(hw->port),
> > +			pf->offset_loaded, &pf->internal_tx_bytes_offset,
> > +			&pf->internal_tx_bytes);
> > +
> >  	/* Get statistics of struct i40e_eth_stats */
> >  	i40e_stat_update_48(hw, I40E_GLPRT_GORCH(hw->port),
> >  			    I40E_GLPRT_GORCL(hw->port),
> > @@ -2400,7 +2409,7 @@ i40e_read_stats_registers(struct i40e_pf *pf,
> > struct i40e_hw *hw)
> >  	 * so subtract ETHER_CRC_LEN from the byte counter for each rx
> > packet.
> >  	 */
> >  	ns->eth.rx_bytes -= (ns->eth.rx_unicast + ns->eth.rx_multicast +
> > -		ns->eth.rx_broadcast) * ETHER_CRC_LEN;
> > +		ns->eth.rx_broadcast) * ETHER_CRC_LEN + pf-
> > >internal_rx_bytes;
> >
> >  	i40e_stat_update_32(hw, I40E_GLPRT_RDPC(hw->port),
> >  			    pf->offset_loaded, &os->eth.rx_discards, @@ -
> > 2428,7 +2437,7 @@ i40e_read_stats_registers(struct i40e_pf *pf, struct
> > i40e_hw *hw)
> >  			    pf->offset_loaded, &os->eth.tx_broadcast,
> >  			    &ns->eth.tx_broadcast);
> >  	ns->eth.tx_bytes -= (ns->eth.tx_unicast + ns->eth.tx_multicast +
> > -		ns->eth.tx_broadcast) * ETHER_CRC_LEN;
> > +		ns->eth.tx_broadcast) * ETHER_CRC_LEN + pf-
> > >internal_tx_bytes;
> >  	/* GLPRT_TEPC not supported */
> >
> >  	/* additional port specific stats */ diff --git
> > a/drivers/net/i40e/i40e_ethdev.h b/drivers/net/i40e/i40e_ethdev.h
> > index 2ff8282..b0d963c 100644
> > --- a/drivers/net/i40e/i40e_ethdev.h
> > +++ b/drivers/net/i40e/i40e_ethdev.h
> > @@ -639,6 +639,11 @@ struct i40e_pf {
> >
> >  	struct i40e_hw_port_stats stats_offset;
> >  	struct i40e_hw_port_stats stats;
> > +	/* internal packet byte count, it should be excluded from the total */
> > +	uint64_t internal_rx_bytes;
> > +	uint64_t internal_tx_bytes;
> > +	uint64_t internal_rx_bytes_offset;
> > +	uint64_t internal_tx_bytes_offset;
> >  	bool offset_loaded;
> >
> >  	struct rte_eth_dev_data *dev_data; /* Pointer to the device data */
> > --
> > 2.7.4
> 
> Do we need to clear the two new stats counters in i40e_pf_setup function?
> Just like clear pf->stats and pf->stats_offset.

Yes, it should be reset as other stats
Thanks for capture this.
Qi