[dpdk-dev] rte_ether: Driver-specific stats getting overwritten

[dpdk-dev] rte_ether: Driver-specific stats getting overwritten

[Remy Horton]

'noon,

In rte_eth_stats_get() after doing the driver callout to populate struct 
rte_eth_stats, the rx_nombuf member is overwritten with 
dev->data->rx_mbuf_alloc_failed even though some drivers will have 
filled rx_nombuf with a value from elsewhere. This makes assignment of 
rx_nombuf from within the driver callout redundant. Is this intentional?

..Remy

Re: [dpdk-dev] rte_ether: Driver-specific stats getting overwritten

[Thomas Monjalon]

2016-07-14 14:29, Remy Horton:
> 'noon,
> 
> In rte_eth_stats_get() after doing the driver callout to populate struct 
> rte_eth_stats, the rx_nombuf member is overwritten with 
> dev->data->rx_mbuf_alloc_failed even though some drivers will have 
> filled rx_nombuf with a value from elsewhere. This makes assignment of 
> rx_nombuf from within the driver callout redundant. Is this intentional?

Yes it is strange and has always been like that.
Why not moving the assignment before calling the driver callback?

Re: [dpdk-dev] rte_ether: Driver-specific stats getting overwritten

[Igor Ryzhov]

Hello.

How about deleting rx_nombuf from rte_eth_stats?
Do you think this counter is necessary? It just shows enormous numbers in case of a lack of processing speed.
But we already have imissed counter which shows real number of packets, dropped for the same reason.

> 14 июля 2016 г., в 16:37, Thomas Monjalon <thomas.monjalon@6wind.com> написал(а):
> 
> 2016-07-14 14:29, Remy Horton:
>> 'noon,
>> 
>> In rte_eth_stats_get() after doing the driver callout to populate struct 
>> rte_eth_stats, the rx_nombuf member is overwritten with 
>> dev->data->rx_mbuf_alloc_failed even though some drivers will have 
>> filled rx_nombuf with a value from elsewhere. This makes assignment of 
>> rx_nombuf from within the driver callout redundant. Is this intentional?
> 
> Yes it is strange and has always been like that.
> Why not moving the assignment before calling the driver callback?

Re: [dpdk-dev] rte_ether: Driver-specific stats getting overwritten

[Remy Horton]

On 14/07/2016 14:51, Igor Ryzhov wrote:
[..]
> How about deleting rx_nombuf from rte_eth_stats? Do you think this
> counter is necessary? It just shows enormous numbers in case of a
> lack of processing speed. But we already have imissed counter which
> shows real number of packets, dropped for the same reason.

Deleting it has API/ABI breakage issues. There is also lack of 
consistency between drivers as to what imissed includes, as some don't 
implement it at all whereas others include filtered packets as well.


>> 14 июля 2016 г., в 16:37, Thomas Monjalon
>> <thomas.monjalon@6wind.com> написал(а):
>>
[..]
>> Yes it is strange and has always been like that. Why not moving the
>> assignment before calling the driver callback?

Think I'll do that. Easier than updating all the drivers that don't fill 
it in..