On 2/25/19 11:25 AM, Sven Eckelmann wrote:
On Monday, 25 February 2019 15:08:15 CET Jeff Kletsky wrote:
Mesh is broken using ath10k-ct?
https://bugs.openwrt.org/index.php?do=details&task_id=2123
[...]
* The "classic" drivers/firmware fail on or after the indicated commit

      CONFIG_PACKAGE_ath10k-firmware-qca9887=y
      CONFIG_PACKAGE_kmod-ath10k=y
      # CONFIG_PACKAGE_kmod-ath10k-ct is not set


That the CT drivers/firmware don't support 802.11s mesh
isn't really an OpenWrt issue.

That there is a regression in overall system capability
that "kills" connectivity is a signficant concern.
* Can you confirm that it works with ath10k-ct but not with ath10k?
* And can you confirm that changing encryption to none has no effect on the
   problem.
* Can you confirm that reverting commit cd93b83ad927 ("ath10k: support for
   multicast rate control") [1] in ath10k fixes this problem?
* Can you confirm that manually setting mcast_rate to 18000  in the mesh wifi-
   iface fixes the problem?

     config wifi-iface 'mesh0'
         option device 'radio0'
         option ifname 'mesh0'
         option network 'nwi_mesh0'
         option mode 'mesh'
         option mesh_id 'TestMesh'
         option mesh_fwding '1'
         option mcast_rate '18000'
         option encryption 'none'

* can you confirm that it lo

Testing performed off `master` [1] on a mesh of Archer C7v2 units and one "test" unit.
* Setting `option mcast_rate '18000'` allows the mesh to connect and 
ping for ath10k "classic"
* https://github.com/openwrt/openwrt/pull/1861 allows the mesh to 
connect and ping for ath10k "classic"
* ath10k-ct drivers/firmware continue to be unable to support 802.11s on 
5 GHz on the Archer C7v2
    * The output of `iw phy` is ambiguous on mesh support for the CT 
drivers/firmware for 5 GHz phy
        * `Supported interface modes:` includes `* mesh point`
        * `valid interface combinations:` seems to contradict
            * `* #{ managed, P2P-client } <= 16, #{ P2P-GO } <= 3, #{ AP } <= 7, #{ IBSS } <= 1,`

Additional information at

https://github.com/openwrt/openwrt/pull/1861 -- shows significant benefit "classic"
https://github.com/openwrt/openwrt/pull/1862 -- "CT" drivers/firmware


Additional testing available on (reasonable) request


Jeff Kletsky


[1] "Current" builds based off

commit fe90e48c39c46b3c253b65f38e392c43f6abe2f0 (openwrt/master, openwrt/HEAD, master)
Author: Piotr Dymacz <pep...@gmail.com>
Date:   Tue Feb 26 13:04:54 2019 +0100


Previous reference points and other units are off master, at or prior to

commit b14289354adb1f07dc9904bfdbc43e6b0bf17a92 (tag: 80211s-good)
Author: John Crispin <j...@phrozen.org>
Date:   Wed Sep 5 14:51:44 2018 +0200





_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel

Reply via email to