On 12/6/21 11:25 PM, meichia chiu wrote:
On Mon, 2021-12-06 at 08:43 -0800, Ben Greear wrote:
On 12/6/21 6:10 AM, MeiChia Chiu wrote:
From: MeiChia Chiu <meichia.chiu@xxxxxxxxxxxx>
Add mu-mimo and ofdma packet counters statistics.
Hello,
Are these stats clear-on-read in the firmware? If so, then we need
to accumulate
them in the driver.
Is there any significant downside to enabling the stats? For
instance, a performance
degradation in the firmware/hardware?
Thanks,
Ben
Hi Ben,
Yes, the current design is clear-on-read and its sole purpose is for
debugging only. If you need statistics, this may have to be done in
another way.
So we could poll these stats in the other logic that polls firmware stats and accumulate
them in the driver and display with debugfs and/or ethtool. If you don't want
to do the accumulation, and lease add a comment to your patch about the
values being clear-on-read and someone else can add the accumulation logic
later.
Yes, a performance degradation is expected since enabling this debug
function would burden MCU with more calculation.
Have you done any testing to verify this? If it is just bumping a few
counters, it may not matter. But I don't know how the firmware is actually
implementing this, so maybe it is a lot more expensive than expected?
Thanks,
Ben
Thanks,
Meichia
--
Ben Greear <greearb@xxxxxxxxxxxxxxx>
Candela Technologies Inc http://www.candelatech.com
