On 07/14/2011 11:14 AM, Lynch, Jonathan wrote:
> Hi Alex,
>
> Thanks, I was only looking at the 82599 datasheet. I'll also 
> crossreference with 82598 datasheet too.
>
> As regards QBTC and QBRC, in ixgbe_update_stats(), the reads for these 
> registers should be moved into seperate blocks depening on the mac 
> type. At the moment its done the same for 82598 or 82599. Also as 
> there are 16 of these registers and not 8, maybe this is in the wrong 
> loop. (I know these stats arent outputted to ethtool statistics, but 
> the wrong values will be dumped by ixgbe_get_regs() ).
>
> Thanks
> Jonathan
>

The loop is partially wrong.  I believe the original reason for 
capturing the per ring byte counts was because originally DCB would use 
only 8 queues so capturing the stats per queue on 82598 would allow you 
to break things down to a per traffic class basis.  On 82599 all of the 
statistics past the first set should be 0 since we never initialized the 
mapping.

In order to guarantee these are correct I will move the reading of QBTC, 
QBRC, QPTC, and QPRC into a separate loop of 16.  That way the 
statistics for those registers will be correct in the future.

Thanks,

Alex

------------------------------------------------------------------------------
AppSumo Presents a FREE Video for the SourceForge Community by Eric 
Ries, the creator of the Lean Startup Methodology on "Lean Startup 
Secrets Revealed." This video shows you how to validate your ideas, 
optimize your ideas and identify your business strategy.
http://p.sf.net/sfu/appsumosfdev2dev
_______________________________________________
E1000-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel® Ethernet, visit 
http://communities.intel.com/community/wired

Reply via email to