the best ath10k driver

Michal Kazior michal.kazior at tieto.com
Wed Aug 12 00:13:10 PDT 2015


On 12 August 2015 at 08:11, Joe Qiao <qlnaaa at gmail.com> wrote:
> Dear experts,
>
> I’m testing the ath10k driver in AP mode now. Previously, I got many
> warning log with backports 4.1.1-1 + firmware 10.2.4.48, such as
>
> [30317.981129] ath10k_pci 0001:03:00.0: failed to transmit management
> frame via WMI: -11
> [31833.297127] ath10k_pci 0001:03:00.0: failed to delete peer
> xx:xx:xx:xx:xx:xx for vdev 0: -11
> [31836.297111] ath10k_pci 0001:03:00.0: failed to transmit management
> frame via WMI: -11
> [31836.913640] ath10k_pci 0001:03:00.0: reached WMI management
> transmit queue limit
> [31836.913662] ath10k_pci 0001:03:00.0: failed to transmit packet, dropping: -16
> [31836.923146] ath10k_pci 0001:03:00.0: reached WMI management
> transmit queue limit
> [31836.923164] ath10k_pci 0001:03:00.0: failed to transmit packet, dropping: -16
>
> But when I tried backports 4.2-rc1-1 + firmware 10.2.4.70.2 (I
> understand 10.2.4.70.2 still is under testing), it looks much better,
> there is no any warning log on my unit. That’s the big surprise to me!
>
> Would you please help shortly elaborate what difference between them?
> It looks the communication between target and host is improved?

10.2.48 uses mgmt tx over wmi. There's a problem how firmware handles
that wrt to client powersave states that can lead to long stalls and
trigger timeouts in ath10k.

The 10.2.4.70.2 (and most of the unstable/ images to my knowledge)
have this fixed and use htt service instead of wmi service to submit
mgmt frames to the device. This prevents the starvation/stall problem.

If you look at the archive of this mailing list you'll find some more
information on `mgmt tx starvation`.


Michal



More information about the ath10k mailing list